chore: bump version to 1.8.0 [skip ci]

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

.github/scripts/release.sh

@@ -3,32 +3,68 @@
 set -e  # trigger failure on error - do not remove!
 set -x  # display command on output
 
+## debug
+# TARGET_VERSION="1.2.x"
+
 if [ -z "${TARGET_VERSION}" ]; then
     >&2 echo "No TARGET_VERSION specified"
     exit 1
 fi
 CHGLOG_FILE="${CHGLOG_FILE:-CHANGELOG.md}"
 
-# update package version
+# Update package version
 uvx --from=toml-cli toml set --toml-path=pyproject.toml project.version "${TARGET_VERSION}"
 uv lock --upgrade-package docling-serve
 
-# collect release notes
+# Extract all docling packages and versions from uv.lock
+DOCVERSIONS=$(uvx --with toml python3 - <<'PY'
+import toml
+data = toml.load("uv.lock")
+for pkg in data.get("package", []):
+    if pkg["name"].startswith("docling"):
+        print(f"{pkg['name']} {pkg['version']}")
+PY
+)
+
+# Format docling versions list without trailing newline
+DOCLING_VERSIONS="### Docling libraries included in this release:"
+while IFS= read -r line; do
+  DOCLING_VERSIONS+="
+- $line"
+done <<< "$DOCVERSIONS"
+
+# Collect release notes
 REL_NOTES=$(mktemp)
 uv run --no-sync semantic-release changelog --unreleased >> "${REL_NOTES}"
 
-# update changelog
+# Strip trailing blank lines from release notes and append docling versions
+{
+  sed -e :a -e '/^\n*$/{$d;N;};/\n$/ba' "${REL_NOTES}"
+  printf "\n"
+  printf "%s" "${DOCLING_VERSIONS}"
+  printf "\n"
+} > "${REL_NOTES}.tmp" && mv "${REL_NOTES}.tmp" "${REL_NOTES}"
+
+# Update changelog
 TMP_CHGLOG=$(mktemp)
 TARGET_TAG_NAME="v${TARGET_VERSION}"
 RELEASE_URL="$(gh repo view --json url -q ".url")/releases/tag/${TARGET_TAG_NAME}"
-printf "## [${TARGET_TAG_NAME}](${RELEASE_URL}) - $(date -Idate)\n\n" >> "${TMP_CHGLOG}"
-cat "${REL_NOTES}" >> "${TMP_CHGLOG}"
-if [ -f "${CHGLOG_FILE}" ]; then
-    printf "\n" | cat - "${CHGLOG_FILE}" >> "${TMP_CHGLOG}"
-fi
+## debug
+#RELEASE_URL="myrepo/releases/tag/${TARGET_TAG_NAME}"
+
+# Strip leading blank lines from existing changelog to avoid multiple blank lines when appending
+EXISTING_CL=$(sed -e :a -e '/^\n*$/{$d;N;};/\n$/ba' "${CHGLOG_FILE}")
+
+{
+  printf "## [${TARGET_TAG_NAME}](${RELEASE_URL}) - $(date -Idate)\n\n"
+  cat "${REL_NOTES}"
+  printf "\n"
+  printf "%s\n" "${EXISTING_CL}"
+} >> "${TMP_CHGLOG}"
+
 mv "${TMP_CHGLOG}" "${CHGLOG_FILE}"
 
-# push changes
+# Push changes
 git config --global user.name 'github-actions[bot]'
 git config --global user.email 'github-actions[bot]@users.noreply.github.com'
 git add pyproject.toml uv.lock "${CHGLOG_FILE}"
@@ -36,5 +72,5 @@ COMMIT_MSG="chore: bump version to ${TARGET_VERSION} [skip ci]"
 git commit -m "${COMMIT_MSG}"
 git push origin main
 
-# create GitHub release (incl. Git tag)
+# Create GitHub release (incl. Git tag)
 gh release create "${TARGET_TAG_NAME}" -F "${REL_NOTES}"

.github/styles/config/vocabularies/Docling/accept.txt

@@ -0,0 +1,40 @@
+[Dd]ocling
+precommit
+asgi
+async
+(?i)urls
+uvicorn
+Config
+[Ww]ebserver
+RQ
+(?i)url
+keyfile
+[Ww]ebsocket(s?)
+[Kk]ubernetes
+UI
+(?i)vllm
+APIs
+[Ss]ubprocesses
+(?i)api
+Kubeflow
+(?i)Jobkit
+(?i)cpu
+(?i)PyTorch
+(?i)CUDA
+(?i)NVIDIA
+(?i)ROCm
+(?i)env
+Gradio
+Podman
+bool
+Ollama
+inbody
+LGTMs
+Dolfi
+Lysak
+Nikos
+Nassar
+Panos
+Vagenas
+Staar
+Livathinos

.github/vale.ini

@@ -0,0 +1,11 @@
+StylesPath = styles
+MinAlertLevel = suggestion
+; Packages = write-good, proselint
+
+Vocab = Docling
+
+[*.md]
+BasedOnStyles = Vale
+
+[CHANGELOG.md]
+BasedOnStyles = 

.github/workflows/actionlint.yml

@@ -13,7 +13,7 @@ jobs:
   actionlint:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
       - name: Download actionlint
         id: get_actionlint
         run: bash <(curl https://raw.githubusercontent.com/rhysd/actionlint/main/scripts/download-actionlint.bash)

.github/workflows/cd.yml

@@ -11,11 +11,11 @@ jobs:
     outputs:
       TARGET_TAG_V: ${{ steps.version_check.outputs.TRGT_VERSION }}
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
         with:
           fetch-depth: 0  # for fetching tags, required for semantic-release
       - name: Install uv and set the python version
-        uses: astral-sh/setup-uv@v5
+        uses: astral-sh/setup-uv@v6
         with:
           enable-cache: true
       - name: Install dependencies
@@ -40,12 +40,12 @@ jobs:
         with:
           app-id: ${{ vars.CI_APP_ID }}
           private-key: ${{ secrets.CI_PRIVATE_KEY }}
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
         with:
           token: ${{ steps.app-token.outputs.token }}
           fetch-depth: 0  # for fetching tags, required for semantic-release
       - name: Install uv and set the python version
-        uses: astral-sh/setup-uv@v5
+        uses: astral-sh/setup-uv@v6
         with:
           enable-cache: true
       - name: Install dependencies

.github/workflows/ci-images-dryrun.yml

@@ -21,10 +21,10 @@ jobs:
             build_args: |
               UV_SYNC_EXTRA_ARGS=--no-group pypi --group cpu --no-extra flash-attn
             platforms: linux/amd64, linux/arm64
-          - name: docling-project/docling-serve-cu124
-            build_args: |
-              UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu124
-            platforms: linux/amd64
+          # - name: docling-project/docling-serve-cu124
+          #   build_args: |
+          #     UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu124
+          #   platforms: linux/amd64
           - name: docling-project/docling-serve-cu126
             build_args: |
               UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu126
@@ -33,6 +33,10 @@ jobs:
             build_args: |
               UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu128
             platforms: linux/amd64
+          # - name: docling-project/docling-serve-rocm
+          #   build_args: |
+          #     UV_SYNC_EXTRA_ARGS=--no-group pypi --group rocm --no-extra flash-attn
+          #   platforms: linux/amd64
 
     permissions:
       packages: write

.github/workflows/discord-release.yml

@@ -0,0 +1,42 @@
+# .github/workflows/discord-release.yml
+name: Notify Discord on Release
+
+on:
+  release:
+    types: [published]
+
+jobs:
+  discord:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Send release info to Discord
+        env:
+          DISCORD_WEBHOOK: ${{ secrets.RELEASES_DISCORD_WEBHOOK }}
+        run: |
+          REPO_NAME=${{ github.repository }}
+          RELEASE_TAG=${{ github.event.release.tag_name }}
+          RELEASE_NAME="${{ github.event.release.name }}"
+          RELEASE_URL=${{ github.event.release.html_url }}
+
+          # Capture the body safely (handles backticks, $, ", etc.)
+          RELEASE_BODY=$(cat <<'EOF'
+            ${{ github.event.release.body }}
+          EOF
+          )
+
+          # Fallback if release name is empty
+          if [ -z "$RELEASE_NAME" ]; then
+            RELEASE_NAME=$RELEASE_TAG
+          fi
+
+          PAYLOAD=$(jq -n \
+          --arg title "🚀 New Release: $RELEASE_NAME" \
+          --arg url "$RELEASE_URL" \
+          --arg desc "$RELEASE_BODY" \
+          --arg author_name "$REPO_NAME" \
+          --arg author_icon "https://github.com/docling-project.png" \
+          '{embeds: [{title: $title, url: $url, description: $desc, color: 5814783, author: {name: $author_name, icon_url: $author_icon}}]}')
+
+          curl -H "Content-Type: application/json" \
+               -d "$PAYLOAD" \
+               "$DISCORD_WEBHOOK"

.github/workflows/images.yml

@@ -25,10 +25,10 @@ jobs:
             build_args: |
               UV_SYNC_EXTRA_ARGS=--no-group pypi --group cpu --no-extra flash-attn
             platforms: linux/amd64, linux/arm64
-          - name: docling-project/docling-serve-cu124
-            build_args: |
-              UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu124
-            platforms: linux/amd64
+          # - name: docling-project/docling-serve-cu124
+          #   build_args: |
+          #     UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu124
+          #   platforms: linux/amd64
           - name: docling-project/docling-serve-cu126
             build_args: |
               UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu126
@@ -37,7 +37,10 @@ jobs:
             build_args: |
               UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu128
             platforms: linux/amd64
-
+          # - name: docling-project/docling-serve-rocm
+          #   build_args: |
+          #     UV_SYNC_EXTRA_ARGS=--no-group pypi --group rocm --no-extra flash-attn
+          #   platforms: linux/amd64
     permissions:
       packages: write
       contents: read

.github/workflows/job-build.yml

@@ -10,9 +10,9 @@ jobs:
       matrix:
         python-version: ['3.12']
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
       - name: Install uv and set the python version
-        uses: astral-sh/setup-uv@v5
+        uses: astral-sh/setup-uv@v6
         with:
           python-version: ${{ matrix.python-version }}
           enable-cache: true

.github/workflows/job-checks.yml

@@ -10,9 +10,9 @@ jobs:
       matrix:
         python-version: ['3.12']
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v5
       - name: Install uv and set the python version
-        uses: astral-sh/setup-uv@v5
+        uses: astral-sh/setup-uv@v6
         with:
           python-version: ${{ matrix.python-version }}
           enable-cache: true
@@ -28,7 +28,7 @@ jobs:
         run: uv sync --frozen --all-extras --no-extra flash-attn
 
       - name: Run styling check
-        run: pre-commit run --all-files
+        run: uv run pre-commit run --all-files
 
   build-package:
     uses: ./.github/workflows/job-build.yml
@@ -47,21 +47,22 @@ jobs:
           name: python-package-distributions
           path: dist/
       - name: Install uv and set the python version
-        uses: astral-sh/setup-uv@v5
+        uses: astral-sh/setup-uv@v6
         with:
           python-version: ${{ matrix.python-version }}
           enable-cache: true
+      - name: Create virtual environment
+        run: uv venv
       - name: Install package
         run: uv pip install dist/*.whl
       - name: Create the server
-        run: python -c 'from docling_serve.app import create_app; create_app()'
-
-  markdown-lint:
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v4
-      - name: markdownlint-cli2-action
-        uses: DavidAnson/markdownlint-cli2-action@v16
-        with:
-          globs: "**/*.md"
+        run: .venv/bin/python -c 'from docling_serve.app import create_app; create_app()'
 
+  # markdown-lint:
+  #   runs-on: ubuntu-latest
+  #   steps:
+  #     - uses: actions/checkout@v5
+  #     - name: markdownlint-cli2-action
+  #       uses: DavidAnson/markdownlint-cli2-action@v16
+  #       with:
+  #         globs: "**/*.md"

.github/workflows/job-image.yml

@@ -53,7 +53,7 @@ jobs:
             df -h
 
       - name: Check out the repo
-        uses: actions/checkout@v4
+        uses: actions/checkout@v5
 
       - name: Log in to the GHCR container image registry
         if: ${{ inputs.publish }}
@@ -88,19 +88,115 @@ jobs:
         with:
           images: ${{ env.GHCR_REGISTRY }}/${{ inputs.ghcr_image_name }}
 
+      # # Local test
+      # - name: Set metadata outputs for local testing ## comment out Free up space, Log in to cr, Cache Docker, Extract metadata, and quay blocks and run act
+      #   id: ghcr_meta
+      #   run: |
+      #     echo "tags=ghcr.io/docling-project/docling-serve:pr-123" >> $GITHUB_OUTPUT
+      #     echo "labels=org.opencontainers.image.source=https://github.com/docling-project/docling-serve" >> $GITHUB_OUTPUT
+
       - name: Build and push image to ghcr.io
         id: ghcr_push
-        uses: docker/build-push-action@v5
+        uses: docker/build-push-action@v6
         with:
           context: .
-          push: ${{ inputs.publish }}
+          push: ${{ inputs.publish }} # set 'false' for local test
           tags: ${{ steps.ghcr_meta.outputs.tags }}
           labels: ${{ steps.ghcr_meta.outputs.labels }}
-          platforms: ${{ inputs.platforms}}
+          platforms: ${{ inputs.platforms }}
           cache-from: type=gha
           cache-to: type=gha,mode=max
           file: Containerfile
           build-args: ${{ inputs.build_args }}
+          pull: true
+      ##
+      ## This stage runs after the build, so it leverages all build cache
+      ## 
+      - name: Export built image for testing
+        id: ghcr_export_built_image
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          push: false
+          load: true
+          tags: ${{ env.GHCR_REGISTRY }}/${{ inputs.ghcr_image_name }}:${{ github.sha }}-test
+          labels: |
+            org.opencontainers.image.title=docling-serve
+            org.opencontainers.image.test=true
+          platforms: linux/amd64 # when 'load' is true, we can't use a list ${{ inputs.platforms }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
+          file: Containerfile
+          build-args: ${{ inputs.build_args }}
+
+      - name: Test image
+        if: steps.ghcr_export_built_image.outcome == 'success'
+        run: |
+          set -e
+
+          IMAGE_TAG="${{ env.GHCR_REGISTRY }}/${{ inputs.ghcr_image_name }}:${{ github.sha }}-test"
+          echo "Testing local image: $IMAGE_TAG"
+
+          # Remove existing container if any
+          docker rm -f docling-serve-test-container 2>/dev/null || true
+
+          echo "Starting container..."
+          docker run -d -p 5001:5001 --name docling-serve-test-container "$IMAGE_TAG"
+
+          echo "Waiting 15s for container to boot..."
+          sleep 15
+
+          # Health check
+          echo "Checking service health..."
+          for i in {1..20}; do
+            HEALTH_RESPONSE=$(curl -s http://localhost:5001/health || true)
+            echo "Health check response [$i]: $HEALTH_RESPONSE"
+
+            if echo "$HEALTH_RESPONSE" | grep -q '"status":"ok"'; then
+              echo "Service is healthy!"
+
+              # Install pytest and dependencies
+              echo "Installing pytest and dependencies..."
+              pip install uv
+              uv venv --allow-existing
+              source .venv/bin/activate
+              uv sync --all-extras --no-extra flash-attn
+
+              # Run pytest tests
+              echo "Running tests..."
+              # Test import
+              python -c 'from docling_serve.app import create_app; create_app()'
+
+              # Run pytest and check result directly
+              if ! pytest -sv -k "test_convert_url" tests/test_1-url-async.py \
+                --disable-warnings; then
+                echo "Tests failed!"
+                docker logs docling-serve-test-container
+                docker rm -f docling-serve-test-container
+                exit 1
+              fi
+
+              echo "Tests passed successfully!"
+              break
+            else
+              echo "Waiting for service... [$i/20]"
+              sleep 3
+            fi
+          done
+
+          # Final health check if service didn't pass earlier
+          if ! echo "$HEALTH_RESPONSE" | grep -q '"status":"ok"'; then
+            echo "Service did not become healthy in time."
+            docker logs docling-serve-test-container
+            docker rm -f docling-serve-test-container
+            exit 1
+          fi
+
+          # Cleanup
+          echo "Cleaning up test container..."
+          docker rm -f docling-serve-test-container
+          echo "Cleaning up test image..."
+          docker rmi "$IMAGE_TAG"
 
       - name: Generate artifact attestation
         if: ${{ inputs.publish }}
@@ -120,7 +216,7 @@ jobs:
       - name: Build and push image to quay.io
         if: ${{ inputs.publish }}
         # id: push-serve-cpu-quay
-        uses: docker/build-push-action@v5
+        uses: docker/build-push-action@v6
         with:
           context: .
           push: ${{ inputs.publish }}
@@ -131,11 +227,8 @@ jobs:
           cache-to: type=gha,mode=max
           file: Containerfile
           build-args: ${{ inputs.build_args }}
-      
-      # - name: Inspect the image details
-      #   run: |
-      #     echo "${{ steps.ghcr_push.outputs.metadata }}"
+          pull: true
 
-      - name: Remove Local Docker Images
+      - name: Remove local Docker images
         run: |
           docker image prune -af

.gitignore

@@ -445,4 +445,7 @@ pip-selfcheck.json
 .action-lint
 .markdown-lint
 
-cookies.txt
+cookies.txt
+
+# Examples
+/examples/splitted_pdf/*

.pre-commit-config.yaml

@@ -7,12 +7,12 @@ repos:
       - id: ruff-format
         name: "Ruff formatter"
         args: [--config=pyproject.toml]
-        files: '^(docling_serve|tests).*\.(py|ipynb)$'
+        files: '^(docling_serve|tests|examples|scripts).*\.(py|ipynb)$'
       # Run the Ruff linter.
       - id: ruff
         name: "Ruff linter"
         args: [--exit-non-zero-on-fix, --fix, --config=pyproject.toml]
-        files: '^(docling_serve|tests).*\.(py|ipynb)$'
+        files: '^(docling_serve|tests|examples|scripts).*\.(py|ipynb)$'
   - repo: local
     hooks:
       - id: system
@@ -21,8 +21,28 @@ repos:
         pass_filenames: false
         language: system
         files: '\.py$'
+  - repo: local
+    hooks:
+      - id: update-docs-common-parameters
+        name: Update Documentation File
+        entry: uv run scripts/update_doc_usage.py
+        language: python
+        pass_filenames: false
+        # Fail the commit if documentation generation fails
+        require_serial: true
+  - repo: https://github.com/errata-ai/vale
+    rev: v3.12.0  # Use latest stable version
+    hooks:
+      - id: vale
+        name: vale sync
+        pass_filenames: false
+        args: [sync, "--config=.github/vale.ini"]
+      - id: vale
+        name: Spell and Style Check with Vale
+        args: ["--config=.github/vale.ini"]
+        files: \.md$
   - repo: https://github.com/astral-sh/uv-pre-commit
-    # uv version.
-    rev: 0.7.13
+    # uv version, https://github.com/astral-sh/uv-pre-commit/releases
+    rev: 0.8.19
     hooks:
       - id: uv-lock

CHANGELOG.md

@@ -1,3 +1,264 @@
+## [v1.8.0](https://github.com/docling-project/docling-serve/releases/tag/v1.8.0) - 2025-10-31
+
+### Feature
+
+* Docling with new standard pipeline with threading ([#428](https://github.com/docling-project/docling-serve/issues/428)) ([`bf132a3`](https://github.com/docling-project/docling-serve/commit/bf132a3c3e615ddbe624841ea5b3a98593c00654))
+
+### Documentation
+
+* Expand automatic docs to nested objects. More complete usage docs. ([#426](https://github.com/docling-project/docling-serve/issues/426)) ([`35319b0`](https://github.com/docling-project/docling-serve/commit/35319b0da793a2a1a434fd2b60b7632e10ecced3))
+* Add docs for docling parameters like performance and debug ([#424](https://github.com/docling-project/docling-serve/issues/424)) ([`f3957ae`](https://github.com/docling-project/docling-serve/commit/f3957aeb577097121fe9d0d21f75a50643f03369))
+
+### Docling libraries included in this release:
+- docling 2.60.0
+- docling-core 2.50.0
+- docling-ibm-models 3.10.2
+- docling-jobkit 1.8.0
+- docling-mcp 1.3.2
+- docling-parse 4.7.0
+- docling-serve 1.8.0
+
+## [v1.7.2](https://github.com/docling-project/docling-serve/releases/tag/v1.7.2) - 2025-10-30
+
+### Fix
+
+* Update locked dependencies. Docling fixes, Expose temperature parameter for vlm models ([#423](https://github.com/docling-project/docling-serve/issues/423)) ([`e9b4140`](https://github.com/docling-project/docling-serve/commit/e9b41406c4116ff79a212877ff6484a1151e144d))
+* Temporary constrain fastapi version ([#418](https://github.com/docling-project/docling-serve/issues/418)) ([`7bf2e7b`](https://github.com/docling-project/docling-serve/commit/7bf2e7b366470e0cf1c4900df7c84becd6a96991))
+
+### Docling libraries included in this release:
+- docling 2.59.0
+- docling-core 2.50.0
+- docling-ibm-models 3.10.2
+- docling-jobkit 1.7.1
+- docling-mcp 1.3.2
+- docling-parse 4.7.0
+- docling-serve 1.7.2
+
+## [v1.7.1](https://github.com/docling-project/docling-serve/releases/tag/v1.7.1) - 2025-10-22
+
+### Fix
+
+* Upgrade dependencies ([#417](https://github.com/docling-project/docling-serve/issues/417)) ([`97613a1`](https://github.com/docling-project/docling-serve/commit/97613a19748e8c152db4a0f62b5a57fca807a33a))
+* Makes task status shared across multiple instances in RQ mode, resolves #378 ([#415](https://github.com/docling-project/docling-serve/issues/415)) ([`0961f2c`](https://github.com/docling-project/docling-serve/commit/0961f2c57425859c76130da3ea8a871d65df4b26))
+* `DOCLING_SERVE_SYNC_POLL_INTERVAL` controls the synchronous polling time ([#413](https://github.com/docling-project/docling-serve/issues/413)) ([`0f274ab`](https://github.com/docling-project/docling-serve/commit/0f274ab135a9bb41accd05db3c12a9dcce220ad9))
+
+### Documentation
+
+* Generate usage.md automatically ([#340](https://github.com/docling-project/docling-serve/issues/340)) ([`9672f31`](https://github.com/docling-project/docling-serve/commit/9672f310b1bb7030af8a276f14691e46f7da0e9e))
+
+### Docling libraries included in this release:
+- docling 2.58.0
+- docling-core 2.49.0
+- docling-ibm-models 3.10.1
+- docling-jobkit 1.7.0
+- docling-mcp 1.3.2
+- docling-parse 4.7.0
+- docling-serve 1.7.1
+
+## [v1.7.0](https://github.com/docling-project/docling-serve/releases/tag/v1.7.0) - 2025-10-17
+
+### Feature
+
+* **UI:** Add auto and orcmac options in demo UI ([#408](https://github.com/docling-project/docling-serve/issues/408)) ([`f5af71e`](https://github.com/docling-project/docling-serve/commit/f5af71e8f6de00d7dd702471a3eea2e94d882410))
+* Docling with auto-ocr ([#403](https://github.com/docling-project/docling-serve/issues/403)) ([`d95ea94`](https://github.com/docling-project/docling-serve/commit/d95ea940870af0d8df689061baa50f6026efce28))
+
+### Fix
+
+* Run docling ui behind a reverse proxy using a context path ([#396](https://github.com/docling-project/docling-serve/issues/396)) ([`5344505`](https://github.com/docling-project/docling-serve/commit/53445057184aa731ee7456b33b70bc0ecf82f2a6))
+
+### Docling libraries included in this release:
+- docling 2.57.0
+- docling-core 2.48.4
+- docling-ibm-models 3.9.1
+- docling-jobkit 1.6.0
+- docling-mcp 1.3.2
+- docling-parse 4.5.0
+- docling-serve 1.7.0
+
+## [v1.6.0](https://github.com/docling-project/docling-serve/releases/tag/v1.6.0) - 2025-10-03
+
+### Feature
+
+* Pin new version of jobkit with granite-docling and connectors ([#391](https://github.com/docling-project/docling-serve/issues/391)) ([`0595d31`](https://github.com/docling-project/docling-serve/commit/0595d31d5b357553426215ca6771796a47e41324))
+
+### Fix
+
+* Update locked dependencies ([#392](https://github.com/docling-project/docling-serve/issues/392)) ([`45f0f3c`](https://github.com/docling-project/docling-serve/commit/45f0f3c8f95d418ac30e3744d27d02a63f9e4490))
+* **UI:** Allow both lowercase and uppercase extensions ([#386](https://github.com/docling-project/docling-serve/issues/386)) ([`8b22a39`](https://github.com/docling-project/docling-serve/commit/8b22a391418d22c1a4d706f880341f28702057b5))
+* Correctly raise HTTPException for Gateway Timeout ([#382](https://github.com/docling-project/docling-serve/issues/382)) ([`d4eac05`](https://github.com/docling-project/docling-serve/commit/d4eac053f9ce0a60f9070127335bdd56e193d7fa))
+* Pinning of higher version of dependencies to fix potential security issues ([#363](https://github.com/docling-project/docling-serve/issues/363)) ([`ba61af2`](https://github.com/docling-project/docling-serve/commit/ba61af23591eff200481aa2e532cf7d0701f0ea4))
+
+### Documentation
+
+* Fix docs for websocket breaking condition ([#390](https://github.com/docling-project/docling-serve/issues/390)) ([`f6b5f0e`](https://github.com/docling-project/docling-serve/commit/f6b5f0e06354d2db7d03d274b114499e3407dccf))
+
+### Docling libraries included in this release:
+- docling 2.55.1
+- docling-core 2.48.4
+- docling-ibm-models 3.9.1
+- docling-jobkit 1.6.0
+- docling-mcp 1.3.2
+- docling-parse 4.5.0
+- docling-serve 1.6.0
+
+## [v1.5.1](https://github.com/docling-project/docling-serve/releases/tag/v1.5.1) - 2025-09-17
+
+### Fix
+
+* Remove old dependencies, fixes in docling-parse and more minor dependencies upgrade ([#362](https://github.com/docling-project/docling-serve/issues/362)) ([`513ae0c`](https://github.com/docling-project/docling-serve/commit/513ae0c119b66d3b17cf9a5d371a0f7971f43be7))
+* Updates rapidocr deps ([#361](https://github.com/docling-project/docling-serve/issues/361)) ([`bde0406`](https://github.com/docling-project/docling-serve/commit/bde040661fb65c67699326cd6281c0e6232e26f2))
+
+### Docling libraries included in this release:
+- docling 2.52.0
+- docling-core 2.48.1
+- docling-ibm-models 3.9.1
+- docling-jobkit 1.5.0
+- docling-mcp 1.2.0
+- docling-parse 4.5.0
+- docling-serve 1.5.1
+
+## [v1.5.0](https://github.com/docling-project/docling-serve/releases/tag/v1.5.0) - 2025-09-09
+
+### Feature
+
+* Add chunking endpoints ([#353](https://github.com/docling-project/docling-serve/issues/353)) ([`9d6def0`](https://github.com/docling-project/docling-serve/commit/9d6def0ec8b1804ad31aa71defa17658d73d29a1))
+
+### Docling libraries included in this release:
+- docling 2.46.0
+- docling 2.51.0
+- docling-core 2.47.0
+- docling-ibm-models 3.9.1
+- docling-jobkit 1.5.0
+- docling-mcp 1.2.0
+- docling-parse 4.4.0
+- docling-serve 1.5.0
+
+## [v1.4.1](https://github.com/docling-project/docling-serve/releases/tag/v1.4.1) - 2025-09-08
+
+### Fix
+
+* Trigger fix after ci fixes ([#355](https://github.com/docling-project/docling-serve/issues/355)) ([`b0360d7`](https://github.com/docling-project/docling-serve/commit/b0360d723bff202dcf44a25a3173ec1995945fc2))
+
+### Docling libraries included in this release:
+- docling 2.46.0
+- docling 2.51.0
+- docling-core 2.47.0
+- docling-ibm-models 3.9.1
+- docling-jobkit 1.4.1
+- docling-mcp 1.2.0
+- docling-parse 4.4.0
+- docling-serve 1.4.1
+
+## [v1.4.0](https://github.com/docling-project/docling-serve/releases/tag/v1.4.0) - 2025-09-05
+
+### Feature
+
+* **docling:** Perfomance improvements in parsing, new layout model, fixes in html processing ([#352](https://github.com/docling-project/docling-serve/issues/352)) ([`d64a2a9`](https://github.com/docling-project/docling-serve/commit/d64a2a974a276c7ae3b105c448fd79f77a653d20))
+
+### Fix
+
+* Upgrade to latest docling version with fixes ([#335](https://github.com/docling-project/docling-serve/issues/335)) ([`e544947`](https://github.com/docling-project/docling-serve/commit/e5449472b2a3e71796f41c8a58c251d8229305c1))
+
+### Documentation
+
+* Add split processing example ([#303](https://github.com/docling-project/docling-serve/issues/303)) ([`0d4545a`](https://github.com/docling-project/docling-serve/commit/0d4545a65a5a941fc1fdefda57e39cfb1ea106ab))
+* Document DOCLING_NUM_THREADS environment variable ([#341](https://github.com/docling-project/docling-serve/issues/341)) ([`27fdd7b`](https://github.com/docling-project/docling-serve/commit/27fdd7b85ab18b3eece428366f46dc5cf0995e38))
+* Fix parameters typo ([#333](https://github.com/docling-project/docling-serve/issues/333)) ([`81f0a8d`](https://github.com/docling-project/docling-serve/commit/81f0a8ddf80a532042d550ae4568f891458b45e7))
+* Describe how to use Docling MCP ([#332](https://github.com/docling-project/docling-serve/issues/332)) ([`a69cc86`](https://github.com/docling-project/docling-serve/commit/a69cc867f5a3fb76648803ca866d65cc3a75c6b8))
+
+### Docling libraries included in this release:
+- docling 2.46.0
+- docling 2.51.0
+- docling-core 2.47.0
+- docling-ibm-models 3.9.1
+- docling-jobkit 1.4.1
+- docling-mcp 1.2.0
+- docling-parse 4.4.0
+- docling-serve 1.4.0
+
+## [v1.3.1](https://github.com/docling-project/docling-serve/releases/tag/v1.3.1) - 2025-08-21
+
+### Fix
+
+* Configuration and performance fixes via upgrade of packages ([#328](https://github.com/docling-project/docling-serve/issues/328)) ([`f02dbc0`](https://github.com/docling-project/docling-serve/commit/f02dbc01449fe1caf3fb4a73c0a5f4adf8265faf))
+
+### Documentation
+
+* Fix parameter in api key docs ([#323](https://github.com/docling-project/docling-serve/issues/323)) ([`37fe022`](https://github.com/docling-project/docling-serve/commit/37fe02277b3e2358eced28e15b4360e7c82d3b43))
+
+## [v1.3.0](https://github.com/docling-project/docling-serve/releases/tag/v1.3.0) - 2025-08-14
+
+### Feature
+
+* Add configuration option for apikey security ([#322](https://github.com/docling-project/docling-serve/issues/322)) ([`9a64410`](https://github.com/docling-project/docling-serve/commit/9a644105523d312431993ded8dd88e064550a5db))
+* Add RQ engine ([#315](https://github.com/docling-project/docling-serve/issues/315)) ([`885f319`](https://github.com/docling-project/docling-serve/commit/885f319d3a3488a4090869560447437a4104f14e))
+
+### Documentation
+
+* Example of docling-serve deployment in the RQ engine mode ([#321](https://github.com/docling-project/docling-serve/issues/321)) ([`71edf41`](https://github.com/docling-project/docling-serve/commit/71edf4184960d8664ef9da20617e2d0f91793d36))
+* Handling models in docling-serve ([#319](https://github.com/docling-project/docling-serve/issues/319)) ([`6e9aa8c`](https://github.com/docling-project/docling-serve/commit/6e9aa8c759220458281c7fe4c87443ac41023eee))
+* Add Gradio cache usage ([#312](https://github.com/docling-project/docling-serve/issues/312)) ([`d584895`](https://github.com/docling-project/docling-serve/commit/d584895e1108d71a0f45deadcd3c669eb0a58133))
+
+## [v1.2.2](https://github.com/docling-project/docling-serve/releases/tag/v1.2.2) - 2025-08-13
+
+### Fix
+
+* Update of transformers module to 4.55.1 ([#316](https://github.com/docling-project/docling-serve/issues/316)) ([`7692eb2`](https://github.com/docling-project/docling-serve/commit/7692eb26006fd4deaa021180c99e23a1b65de506))
+
+## [v1.2.1](https://github.com/docling-project/docling-serve/releases/tag/v1.2.1) - 2025-08-13
+
+### Fix
+
+* Handling of vlm model options and update deps ([#314](https://github.com/docling-project/docling-serve/issues/314)) ([`8b470cb`](https://github.com/docling-project/docling-serve/commit/8b470cba8ef500c271eb84c8368c8a1a1a5a6d6a))
+* Add missing response type in sync endpoints ([#309](https://github.com/docling-project/docling-serve/issues/309)) ([`8048f45`](https://github.com/docling-project/docling-serve/commit/8048f4589a91de2b2b391ab33a326efd1b29f25b))
+
+### Documentation
+
+* Update readme to use v1 ([#306](https://github.com/docling-project/docling-serve/issues/306)) ([`b3058e9`](https://github.com/docling-project/docling-serve/commit/b3058e91e0c56e27110eb50f22cbdd89640bf398))
+* Update deployment examples to use v1 API ([#308](https://github.com/docling-project/docling-serve/issues/308)) ([`63da9ee`](https://github.com/docling-project/docling-serve/commit/63da9eedebae3ad31d04e65635e573194e413793))
+* Fix typo in v1 migration instructions ([#307](https://github.com/docling-project/docling-serve/issues/307)) ([`b15dc25`](https://github.com/docling-project/docling-serve/commit/b15dc2529f78d68a475e5221c37408c3f77d8588))
+
+## [v1.2.0](https://github.com/docling-project/docling-serve/releases/tag/v1.2.0) - 2025-08-07
+
+### Feature
+
+* Workers without shared models and convert params ([#304](https://github.com/docling-project/docling-serve/issues/304)) ([`db3fdb5`](https://github.com/docling-project/docling-serve/commit/db3fdb5bc1a0ae250afd420d737abc4071a7546c))
+* Add rocm image build support and fix cuda ([#292](https://github.com/docling-project/docling-serve/issues/292)) ([`fd1b987`](https://github.com/docling-project/docling-serve/commit/fd1b987e8dc174f1a6013c003dde33e9acbae39a))
+
+## [v1.1.0](https://github.com/docling-project/docling-serve/releases/tag/v1.1.0) - 2025-07-30
+
+### Feature
+
+* Add docling-mcp in the distribution ([#290](https://github.com/docling-project/docling-serve/issues/290)) ([`ecb1874`](https://github.com/docling-project/docling-serve/commit/ecb1874a507bef83d102e0e031e49fed34298637))
+* Add 3.0 openapi endpoint ([#287](https://github.com/docling-project/docling-serve/issues/287)) ([`ec594d8`](https://github.com/docling-project/docling-serve/commit/ec594d84fe36df23e7d010a2fcf769856c43600b))
+* Add new source and target ([#270](https://github.com/docling-project/docling-serve/issues/270)) ([`3771c1b`](https://github.com/docling-project/docling-serve/commit/3771c1b55403bd51966d07d8f760d5c4fbcc1760))
+
+### Fix
+
+* Referenced paths relative to zip root ([#289](https://github.com/docling-project/docling-serve/issues/289)) ([`1333f71`](https://github.com/docling-project/docling-serve/commit/1333f71c9c6495342b2169d574e921f828446f15))
+
+## [v1.0.1](https://github.com/docling-project/docling-serve/releases/tag/v1.0.1) - 2025-07-21
+
+### Fix
+
+* Docling update v2.42.0 ([#277](https://github.com/docling-project/docling-serve/issues/277)) ([`8706706`](https://github.com/docling-project/docling-serve/commit/8706706e8797b0a06ec4baa7cf87988311be68b6))
+
+### Documentation
+
+* Typo in README ([#276](https://github.com/docling-project/docling-serve/issues/276)) ([`766adb2`](https://github.com/docling-project/docling-serve/commit/766adb248113c7bd5144d14b3c82929a2ad29f8e))
+
+## [v1.0.0](https://github.com/docling-project/docling-serve/releases/tag/v1.0.0) - 2025-07-14
+
+### Feature
+
+* V1 api with list of sources and target ([#249](https://github.com/docling-project/docling-serve/issues/249)) ([`56e328b`](https://github.com/docling-project/docling-serve/commit/56e328baf76b4bb0476fc6ca820b52034e4f97bf))
+* Use orchestrators from jobkit ([#248](https://github.com/docling-project/docling-serve/issues/248)) ([`daa924a`](https://github.com/docling-project/docling-serve/commit/daa924a77e56d063ef17347dfd8a838872a70529))
+
+### Breaking
+
+* v1 api with list of sources and target ([#249](https://github.com/docling-project/docling-serve/issues/249)) ([`56e328b`](https://github.com/docling-project/docling-serve/commit/56e328baf76b4bb0476fc6ca820b52034e4f97bf))
+* use orchestrators from jobkit ([#248](https://github.com/docling-project/docling-serve/issues/248)) ([`daa924a`](https://github.com/docling-project/docling-serve/commit/daa924a77e56d063ef17347dfd8a838872a70529))
+
 ## [v0.16.1](https://github.com/docling-project/docling-serve/releases/tag/v0.16.1) - 2025-07-07
 
 ### Fix

Containerfile

@@ -1,13 +1,17 @@
 ARG BASE_IMAGE=quay.io/sclorg/python-312-c9s:c9s
 
-FROM ${BASE_IMAGE}
+ARG UV_IMAGE=ghcr.io/astral-sh/uv:0.8.19
 
-USER 0
+ARG UV_SYNC_EXTRA_ARGS=""
+
+FROM ${BASE_IMAGE} AS docling-base
 
 ###################################################################################################
 # OS Layer                                                                                        #
 ###################################################################################################
 
+USER 0
+
 RUN --mount=type=bind,source=os-packages.txt,target=/tmp/os-packages.txt \
     dnf -y install --best --nodocs --setopt=install_weak_deps=False dnf-plugins-core && \
     dnf config-manager --best --nodocs --setopt=install_weak_deps=False --save && \
@@ -21,16 +25,19 @@ RUN /usr/bin/fix-permissions /opt/app-root/src/.cache
 
 ENV TESSDATA_PREFIX=/usr/share/tesseract/tessdata/
 
+FROM ${UV_IMAGE} AS uv_stage
+
 ###################################################################################################
 # Docling layer                                                                                   #
 ###################################################################################################
 
+FROM docling-base
+
 USER 1001
 
 WORKDIR /opt/app-root/src
 
 ENV \
-    # On container environments, always set a thread budget to avoid undesired thread congestion.
     OMP_NUM_THREADS=4 \
     LANG=en_US.UTF-8 \
     LC_ALL=en_US.UTF-8 \
@@ -40,9 +47,9 @@ ENV \
     UV_PROJECT_ENVIRONMENT=/opt/app-root \
     DOCLING_SERVE_ARTIFACTS_PATH=/opt/app-root/src/.cache/docling/models
 
-ARG UV_SYNC_EXTRA_ARGS=""
+ARG UV_SYNC_EXTRA_ARGS
 
-RUN --mount=from=ghcr.io/astral-sh/uv:0.7.13,source=/uv,target=/bin/uv \
+RUN --mount=from=uv_stage,source=/uv,target=/bin/uv \
     --mount=type=cache,target=/opt/app-root/src/.cache/uv,uid=1001 \
     --mount=type=bind,source=uv.lock,target=uv.lock \
     --mount=type=bind,source=pyproject.toml,target=pyproject.toml \
@@ -51,7 +58,7 @@ RUN --mount=from=ghcr.io/astral-sh/uv:0.7.13,source=/uv,target=/bin/uv \
     uv sync ${UV_SYNC_ARGS} ${UV_SYNC_EXTRA_ARGS} --no-extra flash-attn && \
     FLASH_ATTENTION_SKIP_CUDA_BUILD=TRUE uv sync ${UV_SYNC_ARGS} ${UV_SYNC_EXTRA_ARGS} --no-build-isolation-package=flash-attn
 
-ARG MODELS_LIST="layout tableformer picture_classifier easyocr"
+ARG MODELS_LIST="layout tableformer picture_classifier rapidocr easyocr"
 
 RUN echo "Downloading models..." && \
     HF_HUB_DOWNLOAD_TIMEOUT="90" \
@@ -61,7 +68,8 @@ RUN echo "Downloading models..." && \
     chmod -R g=u ${DOCLING_SERVE_ARTIFACTS_PATH}
 
 COPY --chown=1001:0 ./docling_serve ./docling_serve
-RUN --mount=from=ghcr.io/astral-sh/uv:0.7.13,source=/uv,target=/bin/uv \
+
+RUN --mount=from=uv_stage,source=/uv,target=/bin/uv \
     --mount=type=cache,target=/opt/app-root/src/.cache/uv,uid=1001 \
     --mount=type=bind,source=uv.lock,target=uv.lock \
     --mount=type=bind,source=pyproject.toml,target=pyproject.toml \

MAINTAINERS.md

@@ -1,11 +1,11 @@
 # MAINTAINERS
 
-- Christoph Auer - [@cau-git](https://github.com/cau-git)
-- Michele Dolfi - [@dolfim-ibm](https://github.com/dolfim-ibm)
-- Maxim Lysak - [@maxmnemonic](https://github.com/maxmnemonic)
-- Nikos Livathinos - [@nikos-livathinos](https://github.com/nikos-livathinos)
-- Ahmed Nassar - [@nassarofficial](https://github.com/nassarofficial)
-- Panos Vagenas - [@vagenas](https://github.com/vagenas)
-- Peter Staar - [@PeterStaar-IBM](https://github.com/PeterStaar-IBM)
+- Christoph Auer - [`@cau-git`](https://github.com/cau-git)
+- Michele Dolfi - [`@dolfim-ibm`](https://github.com/dolfim-ibm)
+- Maxim Lysak - [`@maxmnemonic`](https://github.com/maxmnemonic)
+- Nikos Livathinos - [`@nikos-livathinos`](https://github.com/nikos-livathinos)
+- Ahmed Nassar - [`@nassarofficial`](https://github.com/nassarofficial)
+- Panos Vagenas - [`@vagenas`](https://github.com/vagenas)
+- Peter Staar - [`@PeterStaar-IBM`](https://github.com/PeterStaar-IBM)
 
 Maintainers can be contacted at [deepsearch-core@zurich.ibm.com](mailto:deepsearch-core@zurich.ibm.com).

Makefile

@@ -16,6 +16,9 @@ else
     PIPE_DEV_NULL=
 endif
 
+# Container runtime - can be overridden: make CONTAINER_RUNTIME=podman cmd
+CONTAINER_RUNTIME ?= docker
+
 TAG=$(shell git rev-parse HEAD)
 BRANCH_TAG=$(shell git rev-parse --abbrev-ref HEAD)
 
@@ -28,37 +31,44 @@ md-lint-file:
 .PHONY: docling-serve-image
 docling-serve-image: Containerfile ## Build docling-serve container image
 	$(ECHO_PREFIX) printf "  %-12s Containerfile\n" "[docling-serve]"
-	$(CMD_PREFIX) docker build --load -f Containerfile -t ghcr.io/docling-project/docling-serve:$(TAG) .
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve:$(TAG) ghcr.io/docling-project/docling-serve:$(BRANCH_TAG)
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve:$(TAG) quay.io/docling-project/docling-serve:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) build --load -f Containerfile -t ghcr.io/docling-project/docling-serve:$(TAG) .
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve:$(TAG) ghcr.io/docling-project/docling-serve:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve:$(TAG) quay.io/docling-project/docling-serve:$(BRANCH_TAG)
 
 .PHONY: docling-serve-cpu-image
 docling-serve-cpu-image: Containerfile ## Build docling-serve "cpu only" container image
 	$(ECHO_PREFIX) printf "  %-12s Containerfile\n" "[docling-serve CPU]"
-	$(CMD_PREFIX) docker build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cpu --no-extra flash-attn" -f Containerfile -t ghcr.io/docling-project/docling-serve-cpu:$(TAG) .
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cpu:$(TAG) ghcr.io/docling-project/docling-serve-cpu:$(BRANCH_TAG)
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cpu:$(TAG) quay.io/docling-project/docling-serve-cpu:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cpu --no-extra flash-attn" -f Containerfile -t ghcr.io/docling-project/docling-serve-cpu:$(TAG) .
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cpu:$(TAG) ghcr.io/docling-project/docling-serve-cpu:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cpu:$(TAG) quay.io/docling-project/docling-serve-cpu:$(BRANCH_TAG)
 
 .PHONY: docling-serve-cu124-image
 docling-serve-cu124-image: Containerfile ## Build docling-serve container image with CUDA 12.4 support
 	$(ECHO_PREFIX) printf "  %-12s Containerfile\n" "[docling-serve with Cuda 12.4]"
-	$(CMD_PREFIX) docker build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu124" -f Containerfile --platform linux/amd64 -t ghcr.io/docling-project/docling-serve-cu124:$(TAG) .
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cu124:$(TAG) ghcr.io/docling-project/docling-serve-cu124:$(BRANCH_TAG)
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cu124:$(TAG) quay.io/docling-project/docling-serve-cu124:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu124" -f Containerfile --platform linux/amd64 -t ghcr.io/docling-project/docling-serve-cu124:$(TAG) .
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cu124:$(TAG) ghcr.io/docling-project/docling-serve-cu124:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cu124:$(TAG) quay.io/docling-project/docling-serve-cu124:$(BRANCH_TAG)
 
 .PHONY: docling-serve-cu126-image
 docling-serve-cu126-image: Containerfile ## Build docling-serve container image with CUDA 12.6 support
 	$(ECHO_PREFIX) printf "  %-12s Containerfile\n" "[docling-serve with Cuda 12.6]"
-	$(CMD_PREFIX) docker build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu126" -f Containerfile --platform linux/amd64 -t ghcr.io/docling-project/docling-serve-cu126:$(TAG) .
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cu126:$(TAG) ghcr.io/docling-project/docling-serve-cu126:$(BRANCH_TAG)
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cu126:$(TAG) quay.io/docling-project/docling-serve-cu126:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu126" -f Containerfile --platform linux/amd64 -t ghcr.io/docling-project/docling-serve-cu126:$(TAG) .
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cu126:$(TAG) ghcr.io/docling-project/docling-serve-cu126:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cu126:$(TAG) quay.io/docling-project/docling-serve-cu126:$(BRANCH_TAG)
 
 .PHONY: docling-serve-cu128-image
 docling-serve-cu128-image: Containerfile ## Build docling-serve container image with CUDA 12.8 support
 	$(ECHO_PREFIX) printf "  %-12s Containerfile\n" "[docling-serve with Cuda 12.8]"
-	$(CMD_PREFIX) docker build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu128" -f Containerfile --platform linux/amd64 -t ghcr.io/docling-project/docling-serve-cu128:$(TAG) .
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cu128:$(TAG) ghcr.io/docling-project/docling-serve-cu128:$(BRANCH_TAG)
-	$(CMD_PREFIX) docker tag ghcr.io/docling-project/docling-serve-cu128:$(TAG) quay.io/docling-project/docling-serve-cu128:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group cu128" -f Containerfile --platform linux/amd64 -t ghcr.io/docling-project/docling-serve-cu128:$(TAG) .
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cu128:$(TAG) ghcr.io/docling-project/docling-serve-cu128:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-cu128:$(TAG) quay.io/docling-project/docling-serve-cu128:$(BRANCH_TAG)
+
+.PHONY: docling-serve-rocm-image
+docling-serve-rocm-image: Containerfile ## Build docling-serve container image with ROCm support
+	$(ECHO_PREFIX) printf "  %-12s Containerfile\n" "[docling-serve with ROCm 6.3]"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) build --load --build-arg "UV_SYNC_EXTRA_ARGS=--no-group pypi --group rocm --no-extra flash-attn" -f Containerfile --platform linux/amd64 -t ghcr.io/docling-project/docling-serve-rocm:$(TAG) .
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-rocm:$(TAG) ghcr.io/docling-project/docling-serve-rocm:$(BRANCH_TAG)
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) tag ghcr.io/docling-project/docling-serve-rocm:$(TAG) quay.io/docling-project/docling-serve-rocm:$(BRANCH_TAG)
 
 .PHONY: action-lint
 action-lint: .action-lint ##      Lint GitHub Action workflows
@@ -81,7 +91,7 @@ action-lint: .action-lint ##      Lint GitHub Action workflows
 md-lint: .md-lint ##      Lint markdown files
 .md-lint: $(wildcard */**/*.md) | md-lint-file
 	$(ECHO_PREFIX) printf "  %-12s ./...\n" "[MD LINT]"
-	$(CMD_PREFIX) docker run --rm -v $$(pwd):/workdir davidanson/markdownlint-cli2:v0.16.0 "**/*.md" "#.venv"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) run --rm -v $$(pwd):/workdir davidanson/markdownlint-cli2:v0.16.0 "**/*.md" "#.venv"
 	$(CMD_PREFIX) touch $@
 
 .PHONY: py-Lint
@@ -97,13 +107,34 @@ py-lint: ##      Lint Python files
 .PHONY: run-docling-cpu
 run-docling-cpu: ## Run the docling-serve container with CPU support and assign a container name
 	$(ECHO_PREFIX) printf "  %-12s Removing existing container if it exists...\n" "[CLEANUP]"
-	$(CMD_PREFIX) docker rm -f docling-serve-cpu 2>/dev/null || true
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) rm -f docling-serve-cpu 2>/dev/null || true
 	$(ECHO_PREFIX) printf "  %-12s Running docling-serve container with CPU support on port 5001...\n" "[RUN CPU]"
-	$(CMD_PREFIX) docker run -it --name docling-serve-cpu -p 5001:5001 ghcr.io/docling-project/docling-serve-cpu:main
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) run -it --name docling-serve-cpu -p 5001:5001 ghcr.io/docling-project/docling-serve-cpu:main
 
 .PHONY: run-docling-cu124
 run-docling-cu124: ## Run the docling-serve container with GPU support and assign a container name
 	$(ECHO_PREFIX) printf "  %-12s Removing existing container if it exists...\n" "[CLEANUP]"
-	$(CMD_PREFIX) docker rm -f docling-serve-cu124 2>/dev/null || true
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) rm -f docling-serve-cu124 2>/dev/null || true
 	$(ECHO_PREFIX) printf "  %-12s Running docling-serve container with GPU support on port 5001...\n" "[RUN CUDA 12.4]"
-	$(CMD_PREFIX) docker run -it --name docling-serve-cu124 -p 5001:5001 ghcr.io/docling-project/docling-serve-cu124:main
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) run -it --name docling-serve-cu124 -p 5001:5001 ghcr.io/docling-project/docling-serve-cu124:main
+
+.PHONY: run-docling-cu126
+run-docling-cu126: ## Run the docling-serve container with GPU support and assign a container name
+	$(ECHO_PREFIX) printf "  %-12s Removing existing container if it exists...\n" "[CLEANUP]"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) rm -f docling-serve-cu126 2>/dev/null || true
+	$(ECHO_PREFIX) printf "  %-12s Running docling-serve container with GPU support on port 5001...\n" "[RUN CUDA 12.6]"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) run -it --name docling-serve-cu126 -p 5001:5001 ghcr.io/docling-project/docling-serve-cu126:main
+
+.PHONY: run-docling-cu128
+run-docling-cu128: ## Run the docling-serve container with GPU support and assign a container name
+	$(ECHO_PREFIX) printf "  %-12s Removing existing container if it exists...\n" "[CLEANUP]"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) rm -f docling-serve-cu128 2>/dev/null || true
+	$(ECHO_PREFIX) printf "  %-12s Running docling-serve container with GPU support on port 5001...\n" "[RUN CUDA 12.8]"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) run -it --name docling-serve-cu128 -p 5001:5001 ghcr.io/docling-project/docling-serve-cu128:main
+
+.PHONY: run-docling-rocm
+run-docling-rocm: ## Run the docling-serve container with GPU support and assign a container name
+	$(ECHO_PREFIX) printf "  %-12s Removing existing container if it exists...\n" "[CLEANUP]"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) rm -f docling-serve-rocm 2>/dev/null || true
+	$(ECHO_PREFIX) printf "  %-12s Running docling-serve container with GPU support on port 5001...\n" "[RUN ROCm 6.3]"
+	$(CMD_PREFIX) $(CONTAINER_RUNTIME) run -it --name docling-serve-rocm -p 5001:5001 ghcr.io/docling-project/docling-serve-rocm:main

README.md

@@ -12,9 +12,12 @@ Running [Docling](https://github.com/docling-project/docling) as an API service.
 
 - Learning how to [configure the webserver](./docs/configuration.md)
 - Get to know all [runtime options](./docs/usage.md) of the API
-- Explore usefule [deployment examples](./docs/deployment.md)
+- Explore useful [deployment examples](./docs/deployment.md)
 - And more
 
+> [!NOTE]
+> **Migration to the `v1` API.** Docling Serve now has a stable v1 API. Read more on the [migration to v1](./docs/v1_migration.md).
+
 ## Getting started
 
 Install the `docling-serve` package and run the server.
@@ -33,31 +36,47 @@ The server is available at
 - API <http://127.0.0.1:5001>
 - API documentation <http://127.0.0.1:5001/docs>
 - UI playground <http://127.0.0.1:5001/ui>
-  ![swagger.png](img/swagger.png)
+
+![API documentation](img/fastapi-ui.png)
 
 Try it out with a simple conversion:
 
 ```bash
 curl -X 'POST' \
-  'http://localhost:5001/v1alpha/convert/source' \
+  'http://localhost:5001/v1/convert/source' \
   -H 'accept: application/json' \
   -H 'Content-Type: application/json' \
   -d '{
-    "http_sources": [{"url": "https://arxiv.org/pdf/2501.17887"}]
+    "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}]
   }'
 ```
 
-### Container images
+### Container Images
 
-Available container images:
+The following container images are available for running **Docling Serve** with different hardware and PyTorch configurations:
 
-| Name | Description | Arch | Size |
-| -----|-------------|------|------|
-| [`ghcr.io/docling-project/docling-serve`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve) <br /> [`quay.io/docling-project/docling-serve`](https://quay.io/repository/docling-project/docling-serve) | Simple image for Docling Serve, installing all packages from the official pypi.org index. | `linux/amd64`, `linux/arm64` | 3.6 GB (arm64) <br /> 8.7 GB (amd64) |
-| [`ghcr.io/docling-project/docling-serve-cpu`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve-cpu) <br /> [`quay.io/docling-project/docling-serve-cpu`](https://quay.io/repository/docling-project/docling-serve-cpu) | Cpu-only image which installs `torch` from the pytorch cpu index. | `linux/amd64`, `linux/arm64` | 3.6 GB |
-| [`ghcr.io/docling-project/docling-serve-cu124`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve-cu124) <br /> [`quay.io/docling-project/docling-serve-cu124`](https://quay.io/repository/docling-project/docling-serve-cu124) | Cuda 12.4 image which installs `torch` from the pytorch cu124 index. | `linux/amd64` | 8.7 GB |
-| [`ghcr.io/docling-project/docling-serve-cu126`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve-cu126) <br /> [`quay.io/docling-project/docling-serve-cu126`](https://quay.io/repository/docling-project/docling-serve-cu126) | Cuda 12.6 image which installs `torch` from the pytorch cu126 index. | `linux/amd64` | 8.7 GB |
-| [`ghcr.io/docling-project/docling-serve-cu128`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve-cu128) <br /> [`quay.io/docling-project/docling-serve-cu128`](https://quay.io/repository/docling-project/docling-serve-cu128) | Cuda 12.8 image which installs `torch` from the pytorch cu128 index. | `linux/amd64` | 8.7 GB |
+#### 📦 Distributed Images
+
+| Image | Description | Architectures | Size |
+|-------|-------------|----------------|------|
+| [`ghcr.io/docling-project/docling-serve`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve) <br> [`quay.io/docling-project/docling-serve`](https://quay.io/repository/docling-project/docling-serve) | Base image with all packages installed from the official PyPI index. | `linux/amd64`, `linux/arm64` | 4.4 GB (arm64) <br> 8.7 GB (amd64) |
+| [`ghcr.io/docling-project/docling-serve-cpu`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve-cpu) <br> [`quay.io/docling-project/docling-serve-cpu`](https://quay.io/repository/docling-project/docling-serve-cpu) | CPU-only variant, using `torch` from the PyTorch CPU index. | `linux/amd64`, `linux/arm64` | 4.4 GB |
+| [`ghcr.io/docling-project/docling-serve-cu126`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve-cu126) <br> [`quay.io/docling-project/docling-serve-cu126`](https://quay.io/repository/docling-project/docling-serve-cu126) | CUDA 12.6 build with `torch` from the cu126 index. | `linux/amd64` | 10.0 GB |
+| [`ghcr.io/docling-project/docling-serve-cu128`](https://github.com/docling-project/docling-serve/pkgs/container/docling-serve-cu128) <br> [`quay.io/docling-project/docling-serve-cu128`](https://quay.io/repository/docling-project/docling-serve-cu128) | CUDA 12.8 build with `torch` from the cu128 index. | `linux/amd64` | 11.4 GB |
+
+#### 🚫 Not Distributed
+
+An image for AMD ROCm 6.3 (`docling-serve-rocm`) is supported but **not published** due to its large size.
+
+To build it locally:
+
+```bash
+git clone --branch main git@github.com:docling-project/docling-serve.git
+cd docling-serve/
+make docling-serve-rocm-image
+```
+
+For deployment using Docker Compose, see [docs/deployment.md](docs/deployment.md).
 
 Coming soon: `docling-serve-slim` images will reduce the size by skipping the model weights download.
 
@@ -65,9 +84,9 @@ Coming soon: `docling-serve-slim` images will reduce the size by skipping the mo
 
 An easy to use UI is available at the `/ui` endpoint.
 
-![ui-input.png](img/ui-input.png)
+![Input controllers in the UI](img/ui-input.png)
 
-![ui-output.png](img/ui-output.png)
+![Output visualization in the UI](img/ui-output.png)
 
 ## Get help and support
 

docling_serve/__main__.py

@@ -11,6 +11,7 @@ import uvicorn
 from rich.console import Console
 
 from docling_serve.settings import docling_serve_settings, uvicorn_settings
+from docling_serve.storage import get_scratch
 
 warnings.filterwarnings(action="ignore", category=UserWarning, module="pydantic|torch")
 warnings.filterwarnings(action="ignore", category=FutureWarning, module="easyocr")
@@ -30,6 +31,7 @@ logger = logging.getLogger(__name__)
 def version_callback(value: bool) -> None:
     if value:
         docling_serve_version = importlib.metadata.version("docling_serve")
+        docling_jobkit_version = importlib.metadata.version("docling-jobkit")
         docling_version = importlib.metadata.version("docling")
         docling_core_version = importlib.metadata.version("docling-core")
         docling_ibm_models_version = importlib.metadata.version("docling-ibm-models")
@@ -38,6 +40,7 @@ def version_callback(value: bool) -> None:
         py_impl_version = sys.implementation.cache_tag
         py_lang_version = platform.python_version()
         console.print(f"Docling Serve version: {docling_serve_version}")
+        console.print(f"Docling Jobkit version: {docling_jobkit_version}")
         console.print(f"Docling version: {docling_version}")
         console.print(f"Docling Core version: {docling_core_version}")
         console.print(f"Docling IBM Models version: {docling_ibm_models_version}")
@@ -359,6 +362,42 @@ def run(
     )
 
 
+@app.command()
+def rq_worker() -> Any:
+    """
+    Run the [bold]Docling JobKit[/bold] RQ worker.
+    """
+    from docling_jobkit.convert.manager import DoclingConverterManagerConfig
+    from docling_jobkit.orchestrators.rq.orchestrator import RQOrchestratorConfig
+    from docling_jobkit.orchestrators.rq.worker import run_worker
+
+    rq_config = RQOrchestratorConfig(
+        redis_url=docling_serve_settings.eng_rq_redis_url,
+        results_prefix=docling_serve_settings.eng_rq_results_prefix,
+        sub_channel=docling_serve_settings.eng_rq_sub_channel,
+        scratch_dir=get_scratch(),
+    )
+
+    cm_config = DoclingConverterManagerConfig(
+        artifacts_path=docling_serve_settings.artifacts_path,
+        options_cache_size=docling_serve_settings.options_cache_size,
+        enable_remote_services=docling_serve_settings.enable_remote_services,
+        allow_external_plugins=docling_serve_settings.allow_external_plugins,
+        max_num_pages=docling_serve_settings.max_num_pages,
+        max_file_size=docling_serve_settings.max_file_size,
+        queue_max_size=docling_serve_settings.queue_max_size,
+        ocr_batch_size=docling_serve_settings.ocr_batch_size,
+        layout_batch_size=docling_serve_settings.layout_batch_size,
+        table_batch_size=docling_serve_settings.table_batch_size,
+        batch_polling_interval_seconds=docling_serve_settings.batch_polling_interval_seconds,
+    )
+
+    run_worker(
+        rq_config=rq_config,
+        cm_config=cm_config,
+    )
+
+
 def main() -> None:
     app()
 

docling_serve/app.py

@@ -1,4 +1,5 @@
 import asyncio
+import copy
 import importlib.metadata
 import logging
 import shutil
@@ -11,11 +12,13 @@ from fastapi import (
     BackgroundTasks,
     Depends,
     FastAPI,
+    Form,
     HTTPException,
     Query,
     UploadFile,
     WebSocket,
     WebSocketDisconnect,
+    status,
 )
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.openapi.docs import (
@@ -23,41 +26,62 @@ from fastapi.openapi.docs import (
     get_swagger_ui_html,
     get_swagger_ui_oauth2_redirect_html,
 )
-from fastapi.responses import RedirectResponse
+from fastapi.responses import JSONResponse, RedirectResponse
 from fastapi.staticfiles import StaticFiles
 from scalar_fastapi import get_scalar_api_reference
 
 from docling.datamodel.base_models import DocumentStream
-
-from docling_serve.datamodel.callback import (
+from docling_jobkit.datamodel.callback import (
     ProgressCallbackRequest,
     ProgressCallbackResponse,
 )
-from docling_serve.datamodel.convert import ConvertDocumentsOptions
+from docling_jobkit.datamodel.chunking import (
+    BaseChunkerOptions,
+    ChunkingExportOptions,
+    HierarchicalChunkerOptions,
+    HybridChunkerOptions,
+)
+from docling_jobkit.datamodel.http_inputs import FileSource, HttpSource
+from docling_jobkit.datamodel.s3_coords import S3Coordinates
+from docling_jobkit.datamodel.task import Task, TaskSource, TaskType
+from docling_jobkit.datamodel.task_targets import (
+    InBodyTarget,
+    ZipTarget,
+)
+from docling_jobkit.orchestrators.base_orchestrator import (
+    BaseOrchestrator,
+    ProgressInvalid,
+    TaskNotFoundError,
+)
+
+from docling_serve.auth import APIKeyAuth, AuthenticationResult
+from docling_serve.datamodel.convert import ConvertDocumentsRequestOptions
 from docling_serve.datamodel.requests import (
-    ConvertDocumentFileSourcesRequest,
-    ConvertDocumentHttpSourcesRequest,
     ConvertDocumentsRequest,
+    FileSourceRequest,
+    GenericChunkDocumentsRequest,
+    HttpSourceRequest,
+    S3SourceRequest,
+    TargetName,
+    TargetRequest,
+    make_request_model,
 )
 from docling_serve.datamodel.responses import (
+    ChunkDocumentResponse,
     ClearResponse,
     ConvertDocumentResponse,
     HealthCheckResponse,
     MessageKind,
+    PresignedUrlConvertDocumentResponse,
     TaskStatusResponse,
     WebsocketMessage,
 )
-from docling_serve.datamodel.task import Task, TaskSource
-from docling_serve.docling_conversion import _get_converter_from_hash
-from docling_serve.engines.async_orchestrator import (
-    BaseAsyncOrchestrator,
-    ProgressInvalid,
-)
-from docling_serve.engines.async_orchestrator_factory import get_async_orchestrator
-from docling_serve.engines.base_orchestrator import TaskNotFoundError
 from docling_serve.helper_functions import FormDepends
+from docling_serve.orchestrator_factory import get_async_orchestrator
+from docling_serve.response_preparation import prepare_response
 from docling_serve.settings import docling_serve_settings
 from docling_serve.storage import get_scratch
+from docling_serve.websocket_notifier import WebsocketNotifier
 
 
 # Set up custom logging as we'll be intermixes with FastAPI/Uvicorn's logging
@@ -95,9 +119,12 @@ _log = logging.getLogger(__name__)
 # Context manager to initialize and clean up the lifespan of the FastAPI app
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    orchestrator = get_async_orchestrator()
     scratch_dir = get_scratch()
 
+    orchestrator = get_async_orchestrator()
+    notifier = WebsocketNotifier(orchestrator)
+    orchestrator.bind_notifier(notifier)
+
     # Warm up processing cache
     if docling_serve_settings.load_models_at_boot:
         await orchestrator.warm_up_caches()
@@ -140,6 +167,7 @@ def create_app():  # noqa: C901
         offline_docs_assets = True
         _log.info("Found static assets.")
 
+    require_auth = APIKeyAuth(docling_serve_settings.api_key)
     app = FastAPI(
         title="Docling Serve",
         docs_url=None if offline_docs_assets else "/swagger",
@@ -166,16 +194,25 @@ def create_app():  # noqa: C901
             import gradio as gr
 
             from docling_serve.gradio_ui import ui as gradio_ui
+            from docling_serve.settings import uvicorn_settings
 
             tmp_output_dir = get_scratch() / "gradio"
             tmp_output_dir.mkdir(exist_ok=True, parents=True)
             gradio_ui.gradio_output_dir = tmp_output_dir
+
+            # Build the root_path for Gradio, accounting for UVICORN_ROOT_PATH
+            gradio_root_path = (
+                f"{uvicorn_settings.root_path}/ui"
+                if uvicorn_settings.root_path
+                else "/ui"
+            )
+
             app = gr.mount_gradio_app(
                 app,
                 gradio_ui,
                 path="/ui",
                 allowed_paths=["./logo.png", tmp_output_dir],
-                root_path="/ui",
+                root_path=gradio_root_path,
             )
         except ImportError:
             _log.warning(
@@ -230,23 +267,53 @@ def create_app():  # noqa: C901
     ########################
 
     async def _enque_source(
-        orchestrator: BaseAsyncOrchestrator, conversion_request: ConvertDocumentsRequest
+        orchestrator: BaseOrchestrator,
+        request: ConvertDocumentsRequest | GenericChunkDocumentsRequest,
     ) -> Task:
         sources: list[TaskSource] = []
-        if isinstance(conversion_request, ConvertDocumentFileSourcesRequest):
-            sources.extend(conversion_request.file_sources)
-        if isinstance(conversion_request, ConvertDocumentHttpSourcesRequest):
-            sources.extend(conversion_request.http_sources)
+        for s in request.sources:
+            if isinstance(s, FileSourceRequest):
+                sources.append(FileSource.model_validate(s))
+            elif isinstance(s, HttpSourceRequest):
+                sources.append(HttpSource.model_validate(s))
+            elif isinstance(s, S3SourceRequest):
+                sources.append(S3Coordinates.model_validate(s))
+
+        convert_options: ConvertDocumentsRequestOptions
+        chunking_options: BaseChunkerOptions | None = None
+        chunking_export_options = ChunkingExportOptions()
+        task_type: TaskType
+        if isinstance(request, ConvertDocumentsRequest):
+            task_type = TaskType.CONVERT
+            convert_options = request.options
+        elif isinstance(request, GenericChunkDocumentsRequest):
+            task_type = TaskType.CHUNK
+            convert_options = request.convert_options
+            chunking_options = request.chunking_options
+            chunking_export_options.include_converted_doc = (
+                request.include_converted_doc
+            )
+        else:
+            raise RuntimeError("Uknown request type.")
 
         task = await orchestrator.enqueue(
-            sources=sources, options=conversion_request.options
+            task_type=task_type,
+            sources=sources,
+            convert_options=convert_options,
+            chunking_options=chunking_options,
+            chunking_export_options=chunking_export_options,
+            target=request.target,
         )
         return task
 
     async def _enque_file(
-        orchestrator: BaseAsyncOrchestrator,
+        orchestrator: BaseOrchestrator,
         files: list[UploadFile],
-        options: ConvertDocumentsOptions,
+        task_type: TaskType,
+        convert_options: ConvertDocumentsRequestOptions,
+        chunking_options: BaseChunkerOptions | None,
+        chunking_export_options: ChunkingExportOptions | None,
+        target: TargetRequest,
     ) -> Task:
         _log.info(f"Received {len(files)} files for processing.")
 
@@ -258,26 +325,100 @@ def create_app():  # noqa: C901
             name = file.filename if file.filename else f"file{suffix}.pdf"
             file_sources.append(DocumentStream(name=name, stream=buf))
 
-        task = await orchestrator.enqueue(sources=file_sources, options=options)
+        task = await orchestrator.enqueue(
+            task_type=task_type,
+            sources=file_sources,
+            convert_options=convert_options,
+            chunking_options=chunking_options,
+            chunking_export_options=chunking_export_options,
+            target=target,
+        )
         return task
 
-    async def _wait_task_complete(
-        orchestrator: BaseAsyncOrchestrator, task_id: str
-    ) -> bool:
+    async def _wait_task_complete(orchestrator: BaseOrchestrator, task_id: str) -> bool:
         start_time = time.monotonic()
         while True:
             task = await orchestrator.task_status(task_id=task_id)
             if task.is_completed():
                 return True
-            await asyncio.sleep(5)
+            await asyncio.sleep(docling_serve_settings.sync_poll_interval)
             elapsed_time = time.monotonic() - start_time
             if elapsed_time > docling_serve_settings.max_sync_wait:
                 return False
 
+    ##########################################
+    # Downgrade openapi 3.1 to 3.0.x helpers #
+    ##########################################
+
+    def ensure_array_items(schema):
+        """Ensure that array items are defined."""
+        if "type" in schema and schema["type"] == "array":
+            if "items" not in schema or schema["items"] is None:
+                schema["items"] = {"type": "string"}
+            elif isinstance(schema["items"], dict):
+                if "type" not in schema["items"]:
+                    schema["items"]["type"] = "string"
+
+    def handle_discriminators(schema):
+        """Ensure that discriminator properties are included in required."""
+        if "discriminator" in schema and "propertyName" in schema["discriminator"]:
+            prop = schema["discriminator"]["propertyName"]
+            if "properties" in schema and prop in schema["properties"]:
+                if "required" not in schema:
+                    schema["required"] = []
+                if prop not in schema["required"]:
+                    schema["required"].append(prop)
+
+    def handle_properties(schema):
+        """Ensure that property 'kind' is included in required."""
+        if "properties" in schema and "kind" in schema["properties"]:
+            if "required" not in schema:
+                schema["required"] = []
+            if "kind" not in schema["required"]:
+                schema["required"].append("kind")
+
+    # Downgrade openapi 3.1 to 3.0.x
+    def downgrade_openapi31_to_30(spec):
+        def strip_unsupported(obj):
+            if isinstance(obj, dict):
+                obj = {
+                    k: strip_unsupported(v)
+                    for k, v in obj.items()
+                    if k not in ("const", "examples", "prefixItems")
+                }
+
+                handle_discriminators(obj)
+                ensure_array_items(obj)
+
+                # Check for oneOf and anyOf to handle nested schemas
+                for key in ["oneOf", "anyOf"]:
+                    if key in obj:
+                        for sub in obj[key]:
+                            handle_discriminators(sub)
+                            ensure_array_items(sub)
+
+                return obj
+            elif isinstance(obj, list):
+                return [strip_unsupported(i) for i in obj]
+            return obj
+
+        if "components" in spec and "schemas" in spec["components"]:
+            for schema_name, schema in spec["components"]["schemas"].items():
+                handle_properties(schema)
+
+        return strip_unsupported(copy.deepcopy(spec))
+
     #############################
     # API Endpoints definitions #
     #############################
 
+    @app.get("/openapi-3.0.json")
+    def openapi_30():
+        spec = app.openapi()
+        downgraded = downgrade_openapi31_to_30(spec)
+        downgraded["openapi"] = "3.0.3"
+        return JSONResponse(downgraded)
+
     # Favicon
     @app.get("/favicon.ico", include_in_schema=False)
     async def favicon():
@@ -287,7 +428,7 @@ def create_app():  # noqa: C901
         response = RedirectResponse(url=logo_url)
         return response
 
-    @app.get("/health")
+    @app.get("/health", tags=["health"])
     def health() -> HealthCheckResponse:
         return HealthCheckResponse()
 
@@ -298,8 +439,9 @@ def create_app():  # noqa: C901
 
     # Convert a document from URL(s)
     @app.post(
-        "/v1alpha/convert/source",
-        response_model=ConvertDocumentResponse,
+        "/v1/convert/source",
+        tags=["convert"],
+        response_model=ConvertDocumentResponse | PresignedUrlConvertDocumentResponse,
         responses={
             200: {
                 "content": {"application/zip": {}},
@@ -309,37 +451,43 @@ def create_app():  # noqa: C901
     )
     async def process_url(
         background_tasks: BackgroundTasks,
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         conversion_request: ConvertDocumentsRequest,
     ):
         task = await _enque_source(
-            orchestrator=orchestrator, conversion_request=conversion_request
+            orchestrator=orchestrator, request=conversion_request
         )
-        success = await _wait_task_complete(
+        completed = await _wait_task_complete(
             orchestrator=orchestrator, task_id=task.task_id
         )
 
-        if not success:
+        if not completed:
             # TODO: abort task!
-            return HTTPException(
+            raise HTTPException(
                 status_code=504,
                 detail=f"Conversion is taking too long. The maximum wait time is configure as DOCLING_SERVE_MAX_SYNC_WAIT={docling_serve_settings.max_sync_wait}.",
             )
 
-        result = await orchestrator.task_result(
-            task_id=task.task_id, background_tasks=background_tasks
-        )
-        if result is None:
+        task_result = await orchestrator.task_result(task_id=task.task_id)
+        if task_result is None:
             raise HTTPException(
                 status_code=404,
                 detail="Task result not found. Please wait for a completion status.",
             )
-        return result
+        response = await prepare_response(
+            task_id=task.task_id,
+            task_result=task_result,
+            orchestrator=orchestrator,
+            background_tasks=background_tasks,
+        )
+        return response
 
     # Convert a document from file(s)
     @app.post(
-        "/v1alpha/convert/file",
-        response_model=ConvertDocumentResponse,
+        "/v1/convert/file",
+        tags=["convert"],
+        response_model=ConvertDocumentResponse | PresignedUrlConvertDocumentResponse,
         responses={
             200: {
                 "content": {"application/zip": {}},
@@ -348,53 +496,69 @@ def create_app():  # noqa: C901
     )
     async def process_file(
         background_tasks: BackgroundTasks,
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         files: list[UploadFile],
         options: Annotated[
-            ConvertDocumentsOptions, FormDepends(ConvertDocumentsOptions)
+            ConvertDocumentsRequestOptions, FormDepends(ConvertDocumentsRequestOptions)
         ],
+        target_type: Annotated[TargetName, Form()] = TargetName.INBODY,
     ):
+        target = InBodyTarget() if target_type == TargetName.INBODY else ZipTarget()
         task = await _enque_file(
-            orchestrator=orchestrator, files=files, options=options
+            task_type=TaskType.CONVERT,
+            orchestrator=orchestrator,
+            files=files,
+            convert_options=options,
+            chunking_options=None,
+            chunking_export_options=None,
+            target=target,
         )
-        success = await _wait_task_complete(
+        completed = await _wait_task_complete(
             orchestrator=orchestrator, task_id=task.task_id
         )
 
-        if not success:
+        if not completed:
             # TODO: abort task!
-            return HTTPException(
+            raise HTTPException(
                 status_code=504,
                 detail=f"Conversion is taking too long. The maximum wait time is configure as DOCLING_SERVE_MAX_SYNC_WAIT={docling_serve_settings.max_sync_wait}.",
             )
 
-        result = await orchestrator.task_result(
-            task_id=task.task_id, background_tasks=background_tasks
-        )
-        if result is None:
+        task_result = await orchestrator.task_result(task_id=task.task_id)
+        if task_result is None:
             raise HTTPException(
                 status_code=404,
                 detail="Task result not found. Please wait for a completion status.",
             )
-        return result
+        response = await prepare_response(
+            task_id=task.task_id,
+            task_result=task_result,
+            orchestrator=orchestrator,
+            background_tasks=background_tasks,
+        )
+        return response
 
     # Convert a document from URL(s) using the async api
     @app.post(
-        "/v1alpha/convert/source/async",
+        "/v1/convert/source/async",
+        tags=["convert"],
         response_model=TaskStatusResponse,
     )
     async def process_url_async(
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         conversion_request: ConvertDocumentsRequest,
     ):
         task = await _enque_source(
-            orchestrator=orchestrator, conversion_request=conversion_request
+            orchestrator=orchestrator, request=conversion_request
         )
         task_queue_position = await orchestrator.get_queue_position(
             task_id=task.task_id
         )
         return TaskStatusResponse(
             task_id=task.task_id,
+            task_type=task.task_type,
             task_status=task.task_status,
             task_position=task_queue_position,
             task_meta=task.processing_meta,
@@ -402,40 +566,274 @@ def create_app():  # noqa: C901
 
     # Convert a document from file(s) using the async api
     @app.post(
-        "/v1alpha/convert/file/async",
+        "/v1/convert/file/async",
+        tags=["convert"],
         response_model=TaskStatusResponse,
     )
     async def process_file_async(
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         background_tasks: BackgroundTasks,
         files: list[UploadFile],
         options: Annotated[
-            ConvertDocumentsOptions, FormDepends(ConvertDocumentsOptions)
+            ConvertDocumentsRequestOptions, FormDepends(ConvertDocumentsRequestOptions)
         ],
+        target_type: Annotated[TargetName, Form()] = TargetName.INBODY,
     ):
+        target = InBodyTarget() if target_type == TargetName.INBODY else ZipTarget()
         task = await _enque_file(
-            orchestrator=orchestrator, files=files, options=options
+            task_type=TaskType.CONVERT,
+            orchestrator=orchestrator,
+            files=files,
+            convert_options=options,
+            chunking_options=None,
+            chunking_export_options=None,
+            target=target,
         )
         task_queue_position = await orchestrator.get_queue_position(
             task_id=task.task_id
         )
         return TaskStatusResponse(
             task_id=task.task_id,
+            task_type=task.task_type,
             task_status=task.task_status,
             task_position=task_queue_position,
             task_meta=task.processing_meta,
         )
 
+    # Chunking endpoints
+    for display_name, path_name, opt_cls in (
+        ("HybridChunker", "hybrid", HybridChunkerOptions),
+        ("HierarchicalChunker", "hierarchical", HierarchicalChunkerOptions),
+    ):
+        req_cls = make_request_model(opt_cls)
+
+        @app.post(
+            f"/v1/chunk/{path_name}/source/async",
+            name=f"Chunk sources with {display_name} as async task",
+            tags=["chunk"],
+            response_model=TaskStatusResponse,
+        )
+        async def chunk_source_async(
+            background_tasks: BackgroundTasks,
+            auth: Annotated[AuthenticationResult, Depends(require_auth)],
+            orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
+            request: req_cls,
+        ):
+            task = await _enque_source(orchestrator=orchestrator, request=request)
+            task_queue_position = await orchestrator.get_queue_position(
+                task_id=task.task_id
+            )
+            return TaskStatusResponse(
+                task_id=task.task_id,
+                task_type=task.task_type,
+                task_status=task.task_status,
+                task_position=task_queue_position,
+                task_meta=task.processing_meta,
+            )
+
+        @app.post(
+            f"/v1/chunk/{path_name}/file/async",
+            name=f"Chunk files with {display_name} as async task",
+            tags=["chunk"],
+            response_model=TaskStatusResponse,
+        )
+        async def chunk_file_async(
+            background_tasks: BackgroundTasks,
+            auth: Annotated[AuthenticationResult, Depends(require_auth)],
+            orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
+            files: list[UploadFile],
+            convert_options: Annotated[
+                ConvertDocumentsRequestOptions,
+                FormDepends(
+                    ConvertDocumentsRequestOptions,
+                    prefix="convert_",
+                    excluded_fields=[
+                        "to_formats",
+                    ],
+                ),
+            ],
+            chunking_options: Annotated[
+                opt_cls,
+                FormDepends(
+                    HybridChunkerOptions,
+                    prefix="chunking_",
+                    excluded_fields=["chunker"],
+                ),
+            ],
+            include_converted_doc: Annotated[
+                bool,
+                Form(
+                    description="If true, the output will include both the chunks and the converted document."
+                ),
+            ] = False,
+            target_type: Annotated[
+                TargetName,
+                Form(description="Specification for the type of output target."),
+            ] = TargetName.INBODY,
+        ):
+            target = InBodyTarget() if target_type == TargetName.INBODY else ZipTarget()
+            task = await _enque_file(
+                task_type=TaskType.CHUNK,
+                orchestrator=orchestrator,
+                files=files,
+                convert_options=convert_options,
+                chunking_options=chunking_options,
+                chunking_export_options=ChunkingExportOptions(
+                    include_converted_doc=include_converted_doc
+                ),
+                target=target,
+            )
+            task_queue_position = await orchestrator.get_queue_position(
+                task_id=task.task_id
+            )
+            return TaskStatusResponse(
+                task_id=task.task_id,
+                task_type=task.task_type,
+                task_status=task.task_status,
+                task_position=task_queue_position,
+                task_meta=task.processing_meta,
+            )
+
+        @app.post(
+            f"/v1/chunk/{path_name}/source",
+            name=f"Chunk sources with {display_name}",
+            tags=["chunk"],
+            response_model=ChunkDocumentResponse,
+            responses={
+                200: {
+                    "content": {"application/zip": {}},
+                    # "description": "Return the JSON item or an image.",
+                }
+            },
+        )
+        async def chunk_source(
+            background_tasks: BackgroundTasks,
+            auth: Annotated[AuthenticationResult, Depends(require_auth)],
+            orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
+            request: req_cls,
+        ):
+            task = await _enque_source(orchestrator=orchestrator, request=request)
+            completed = await _wait_task_complete(
+                orchestrator=orchestrator, task_id=task.task_id
+            )
+
+            if not completed:
+                # TODO: abort task!
+                raise HTTPException(
+                    status_code=504,
+                    detail=f"Conversion is taking too long. The maximum wait time is configure as DOCLING_SERVE_MAX_SYNC_WAIT={docling_serve_settings.max_sync_wait}.",
+                )
+
+            task_result = await orchestrator.task_result(task_id=task.task_id)
+            if task_result is None:
+                raise HTTPException(
+                    status_code=404,
+                    detail="Task result not found. Please wait for a completion status.",
+                )
+            response = await prepare_response(
+                task_id=task.task_id,
+                task_result=task_result,
+                orchestrator=orchestrator,
+                background_tasks=background_tasks,
+            )
+            return response
+
+        @app.post(
+            f"/v1/chunk/{path_name}/file",
+            name=f"Chunk files with {display_name}",
+            tags=["chunk"],
+            response_model=ChunkDocumentResponse,
+            responses={
+                200: {
+                    "content": {"application/zip": {}},
+                }
+            },
+        )
+        async def chunk_file(
+            background_tasks: BackgroundTasks,
+            auth: Annotated[AuthenticationResult, Depends(require_auth)],
+            orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
+            files: list[UploadFile],
+            convert_options: Annotated[
+                ConvertDocumentsRequestOptions,
+                FormDepends(
+                    ConvertDocumentsRequestOptions,
+                    prefix="convert_",
+                    excluded_fields=[
+                        "to_formats",
+                    ],
+                ),
+            ],
+            chunking_options: Annotated[
+                opt_cls,
+                FormDepends(
+                    HybridChunkerOptions,
+                    prefix="chunking_",
+                    excluded_fields=["chunker"],
+                ),
+            ],
+            include_converted_doc: Annotated[
+                bool,
+                Form(
+                    description="If true, the output will include both the chunks and the converted document."
+                ),
+            ] = False,
+            target_type: Annotated[
+                TargetName,
+                Form(description="Specification for the type of output target."),
+            ] = TargetName.INBODY,
+        ):
+            target = InBodyTarget() if target_type == TargetName.INBODY else ZipTarget()
+            task = await _enque_file(
+                task_type=TaskType.CHUNK,
+                orchestrator=orchestrator,
+                files=files,
+                convert_options=convert_options,
+                chunking_options=chunking_options,
+                chunking_export_options=ChunkingExportOptions(
+                    include_converted_doc=include_converted_doc
+                ),
+                target=target,
+            )
+            completed = await _wait_task_complete(
+                orchestrator=orchestrator, task_id=task.task_id
+            )
+
+            if not completed:
+                # TODO: abort task!
+                raise HTTPException(
+                    status_code=504,
+                    detail=f"Conversion is taking too long. The maximum wait time is configure as DOCLING_SERVE_MAX_SYNC_WAIT={docling_serve_settings.max_sync_wait}.",
+                )
+
+            task_result = await orchestrator.task_result(task_id=task.task_id)
+            if task_result is None:
+                raise HTTPException(
+                    status_code=404,
+                    detail="Task result not found. Please wait for a completion status.",
+                )
+            response = await prepare_response(
+                task_id=task.task_id,
+                task_result=task_result,
+                orchestrator=orchestrator,
+                background_tasks=background_tasks,
+            )
+            return response
+
     # Task status poll
     @app.get(
-        "/v1alpha/status/poll/{task_id}",
+        "/v1/status/poll/{task_id}",
+        tags=["tasks"],
         response_model=TaskStatusResponse,
     )
     async def task_status_poll(
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         task_id: str,
         wait: Annotated[
-            float, Query(help="Number of seconds to wait for a completed status.")
+            float,
+            Query(description="Number of seconds to wait for a completed status."),
         ] = 0.0,
     ):
         try:
@@ -445,6 +843,7 @@ def create_app():  # noqa: C901
             raise HTTPException(status_code=404, detail="Task not found.")
         return TaskStatusResponse(
             task_id=task.task_id,
+            task_type=task.task_type,
             task_status=task.task_status,
             task_position=task_queue_position,
             task_meta=task.processing_meta,
@@ -452,16 +851,28 @@ def create_app():  # noqa: C901
 
     # Task status websocket
     @app.websocket(
-        "/v1alpha/status/ws/{task_id}",
+        "/v1/status/ws/{task_id}",
     )
     async def task_status_ws(
         websocket: WebSocket,
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         task_id: str,
+        api_key: Annotated[str, Query()] = "",
     ):
+        if docling_serve_settings.api_key:
+            if api_key != docling_serve_settings.api_key:
+                raise HTTPException(
+                    status_code=status.HTTP_401_UNAUTHORIZED,
+                    detail="Api key is required as ?api_key=SECRET.",
+                )
+
+        assert isinstance(orchestrator.notifier, WebsocketNotifier)
         await websocket.accept()
 
-        if task_id not in orchestrator.tasks:
+        try:
+            # Get task status from Redis or RQ directly instead of checking in-memory registry
+            task = await orchestrator.task_status(task_id=task_id)
+        except TaskNotFoundError:
             await websocket.send_text(
                 WebsocketMessage(
                     message=MessageKind.ERROR, error="Task not found."
@@ -470,15 +881,14 @@ def create_app():  # noqa: C901
             await websocket.close()
             return
 
-        task = orchestrator.tasks[task_id]
-
         # Track active WebSocket connections for this job
-        orchestrator.task_subscribers[task_id].add(websocket)
+        orchestrator.notifier.task_subscribers[task_id].add(websocket)
 
         try:
             task_queue_position = await orchestrator.get_queue_position(task_id=task_id)
             task_response = TaskStatusResponse(
                 task_id=task.task_id,
+                task_type=task.task_type,
                 task_status=task.task_status,
                 task_position=task_queue_position,
                 task_meta=task.processing_meta,
@@ -494,6 +904,7 @@ def create_app():  # noqa: C901
                 )
                 task_response = TaskStatusResponse(
                     task_id=task.task_id,
+                    task_type=task.task_type,
                     task_status=task.task_status,
                     task_position=task_queue_position,
                     task_meta=task.processing_meta,
@@ -511,12 +922,15 @@ def create_app():  # noqa: C901
             _log.info(f"WebSocket disconnected for job {task_id}")
 
         finally:
-            orchestrator.task_subscribers[task_id].remove(websocket)
+            orchestrator.notifier.task_subscribers[task_id].remove(websocket)
 
     # Task result
     @app.get(
-        "/v1alpha/result/{task_id}",
-        response_model=ConvertDocumentResponse,
+        "/v1/result/{task_id}",
+        tags=["tasks"],
+        response_model=ConvertDocumentResponse
+        | PresignedUrlConvertDocumentResponse
+        | ChunkDocumentResponse,
         responses={
             200: {
                 "content": {"application/zip": {}},
@@ -524,27 +938,38 @@ def create_app():  # noqa: C901
         },
     )
     async def task_result(
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         background_tasks: BackgroundTasks,
         task_id: str,
     ):
-        result = await orchestrator.task_result(
-            task_id=task_id, background_tasks=background_tasks
-        )
-        if result is None:
-            raise HTTPException(
-                status_code=404,
-                detail="Task result not found. Please wait for a completion status.",
+        try:
+            task_result = await orchestrator.task_result(task_id=task_id)
+            if task_result is None:
+                raise HTTPException(
+                    status_code=404,
+                    detail="Task result not found. Please wait for a completion status.",
+                )
+            response = await prepare_response(
+                task_id=task_id,
+                task_result=task_result,
+                orchestrator=orchestrator,
+                background_tasks=background_tasks,
             )
-        return result
+            return response
+        except TaskNotFoundError:
+            raise HTTPException(status_code=404, detail="Task not found.")
 
     # Update task progress
     @app.post(
-        "/v1alpha/callback/task/progress",
+        "/v1/callback/task/progress",
+        tags=["internal"],
+        include_in_schema=False,
         response_model=ProgressCallbackResponse,
     )
     async def callback_task_progress(
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         request: ProgressCallbackRequest,
     ):
         try:
@@ -561,20 +986,26 @@ def create_app():  # noqa: C901
 
     # Offload models
     @app.get(
-        "/v1alpha/clear/converters",
+        "/v1/clear/converters",
+        tags=["clear"],
         response_model=ClearResponse,
     )
-    async def clear_converters():
-        _get_converter_from_hash.cache_clear()
+    async def clear_converters(
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
+    ):
+        await orchestrator.clear_converters()
         return ClearResponse()
 
     # Clean results
     @app.get(
-        "/v1alpha/clear/results",
+        "/v1/clear/results",
+        tags=["clear"],
         response_model=ClearResponse,
     )
     async def clear_results(
-        orchestrator: Annotated[BaseAsyncOrchestrator, Depends(get_async_orchestrator)],
+        auth: Annotated[AuthenticationResult, Depends(require_auth)],
+        orchestrator: Annotated[BaseOrchestrator, Depends(get_async_orchestrator)],
         older_then: float = 3600,
     ):
         await orchestrator.clear_results(older_than=older_then)

docling_serve/auth.py

@@ -0,0 +1,56 @@
+from typing import Any
+
+from fastapi import HTTPException, Request, status
+from fastapi.security import APIKeyHeader
+from pydantic import BaseModel
+
+
+class AuthenticationResult(BaseModel):
+    valid: bool
+    errors: list[str] = []
+    detail: Any | None = None
+
+
+class APIKeyAuth(APIKeyHeader):
+    """
+    FastAPI dependency which evaluates a status API Key.
+    """
+
+    def __init__(
+        self,
+        api_key: str,
+        header_name: str = "X-Api-Key",
+        fail_on_unauthorized: bool = True,
+    ) -> None:
+        self.api_key = api_key
+        self.header_name = header_name
+        super().__init__(name=self.header_name, auto_error=False)
+
+    async def _validate_api_key(self, header_api_key: str | None):
+        if header_api_key is None:
+            return AuthenticationResult(
+                valid=False, errors=[f"Missing header {self.header_name}."]
+            )
+
+        header_api_key = header_api_key.strip()
+
+        # Otherwise check the apikey
+        if header_api_key == self.api_key or self.api_key == "":
+            return AuthenticationResult(
+                valid=True,
+                detail=header_api_key,
+            )
+        else:
+            return AuthenticationResult(
+                valid=False,
+                errors=["The provided API Key is invalid."],
+            )
+
+    async def __call__(self, request: Request) -> AuthenticationResult:  # type: ignore
+        header_api_key = await super().__call__(request=request)
+        result = await self._validate_api_key(header_api_key)
+        if self.api_key and not result.valid:
+            raise HTTPException(
+                status_code=status.HTTP_401_UNAUTHORIZED, detail=result.detail
+            )
+        return result

docling_serve/datamodel/callback.py

@@ -1,50 +0,0 @@
-import enum
-from typing import Annotated, Literal
-
-from pydantic import BaseModel, Field
-
-
-class ProgressKind(str, enum.Enum):
-    SET_NUM_DOCS = "set_num_docs"
-    UPDATE_PROCESSED = "update_processed"
-
-
-class BaseProgress(BaseModel):
-    kind: ProgressKind
-
-
-class ProgressSetNumDocs(BaseProgress):
-    kind: Literal[ProgressKind.SET_NUM_DOCS] = ProgressKind.SET_NUM_DOCS
-
-    num_docs: int
-
-
-class SucceededDocsItem(BaseModel):
-    source: str
-
-
-class FailedDocsItem(BaseModel):
-    source: str
-    error: str
-
-
-class ProgressUpdateProcessed(BaseProgress):
-    kind: Literal[ProgressKind.UPDATE_PROCESSED] = ProgressKind.UPDATE_PROCESSED
-
-    num_processed: int
-    num_succeeded: int
-    num_failed: int
-
-    docs_succeeded: list[SucceededDocsItem]
-    docs_failed: list[FailedDocsItem]
-
-
-class ProgressCallbackRequest(BaseModel):
-    task_id: str
-    progress: Annotated[
-        ProgressSetNumDocs | ProgressUpdateProcessed, Field(discriminator="kind")
-    ]
-
-
-class ProgressCallbackResponse(BaseModel):
-    status: Literal["ack"] = "ack"

docling_serve/datamodel/convert.py

@@ -1,24 +1,13 @@
 # Define the input options for the API
-from typing import Annotated, Any, Optional
+from typing import Annotated
 
-from pydantic import AnyUrl, BaseModel, Field, model_validator
-from typing_extensions import Self
+from pydantic import Field
 
-from docling.datamodel.base_models import InputFormat, OutputFormat
 from docling.datamodel.pipeline_options import (
     EasyOcrOptions,
-    PdfBackend,
-    PictureDescriptionBaseOptions,
-    ProcessingPipeline,
-    TableFormerMode,
-    TableStructureOptions,
-)
-from docling.datamodel.settings import (
-    DEFAULT_PAGE_RANGE,
-    PageRange,
 )
 from docling.models.factories import get_ocr_factory
-from docling_core.types.doc import ImageRefMode
+from docling_jobkit.datamodel.convert import ConvertDocumentsOptions
 
 from docling_serve.settings import docling_serve_settings
 
@@ -28,154 +17,7 @@ ocr_factory = get_ocr_factory(
 ocr_engines_enum = ocr_factory.get_enum()
 
 
-class PictureDescriptionLocal(BaseModel):
-    repo_id: Annotated[
-        str,
-        Field(
-            description="Repository id from the Hugging Face Hub.",
-            examples=[
-                "HuggingFaceTB/SmolVLM-256M-Instruct",
-                "ibm-granite/granite-vision-3.2-2b",
-            ],
-        ),
-    ]
-    prompt: Annotated[
-        str,
-        Field(
-            description="Prompt used when calling the vision-language model.",
-            examples=[
-                "Describe this image in a few sentences.",
-                "This is a figure from a document. Provide a detailed description of it.",
-            ],
-        ),
-    ] = "Describe this image in a few sentences."
-    generation_config: Annotated[
-        dict[str, Any],
-        Field(
-            description="Config from https://huggingface.co/docs/transformers/en/main_classes/text_generation#transformers.GenerationConfig",
-            examples=[{"max_new_tokens": 200, "do_sample": False}],
-        ),
-    ] = {"max_new_tokens": 200, "do_sample": False}
-
-
-class PictureDescriptionApi(BaseModel):
-    url: Annotated[
-        AnyUrl,
-        Field(
-            description="Endpoint which accepts openai-api compatible requests.",
-            examples=[
-                AnyUrl(
-                    "http://localhost:8000/v1/chat/completions"
-                ),  # example of a local vllm api
-                AnyUrl(
-                    "http://localhost:11434/v1/chat/completions"
-                ),  # example of ollama
-            ],
-        ),
-    ]
-    headers: Annotated[
-        dict[str, str],
-        Field(
-            description="Headers used for calling the API endpoint. For example, it could include authentication headers."
-        ),
-    ] = {}
-    params: Annotated[
-        dict[str, Any],
-        Field(
-            description="Model parameters.",
-            examples=[
-                {  # on vllm
-                    "model": "HuggingFaceTB/SmolVLM-256M-Instruct",
-                    "max_completion_tokens": 200,
-                },
-                {  # on vllm
-                    "model": "ibm-granite/granite-vision-3.2-2b",
-                    "max_completion_tokens": 200,
-                },
-                {  # on ollama
-                    "model": "granite3.2-vision:2b"
-                },
-            ],
-        ),
-    ] = {}
-    timeout: Annotated[float, Field(description="Timeout for the API request.")] = 20
-    prompt: Annotated[
-        str,
-        Field(
-            description="Prompt used when calling the vision-language model.",
-            examples=[
-                "Describe this image in a few sentences.",
-                "This is a figures from a document. Provide a detailed description of it.",
-            ],
-        ),
-    ] = "Describe this image in a few sentences."
-
-
-class ConvertDocumentsOptions(BaseModel):
-    from_formats: Annotated[
-        list[InputFormat],
-        Field(
-            description=(
-                "Input format(s) to convert from. String or list of strings. "
-                f"Allowed values: {', '.join([v.value for v in InputFormat])}. "
-                "Optional, defaults to all formats."
-            ),
-            examples=[[v.value for v in InputFormat]],
-        ),
-    ] = list(InputFormat)
-
-    to_formats: Annotated[
-        list[OutputFormat],
-        Field(
-            description=(
-                "Output format(s) to convert to. String or list of strings. "
-                f"Allowed values: {', '.join([v.value for v in OutputFormat])}. "
-                "Optional, defaults to Markdown."
-            ),
-            examples=[
-                [OutputFormat.MARKDOWN],
-                [OutputFormat.MARKDOWN, OutputFormat.JSON],
-                [v.value for v in OutputFormat],
-            ],
-        ),
-    ] = [OutputFormat.MARKDOWN]
-
-    image_export_mode: Annotated[
-        ImageRefMode,
-        Field(
-            description=(
-                "Image export mode for the document (in case of JSON,"
-                " Markdown or HTML). "
-                f"Allowed values: {', '.join([v.value for v in ImageRefMode])}. "
-                "Optional, defaults to Embedded."
-            ),
-            examples=[ImageRefMode.EMBEDDED.value],
-            # pattern="embedded|placeholder|referenced",
-        ),
-    ] = ImageRefMode.EMBEDDED
-
-    do_ocr: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, the bitmap content will be processed using OCR. "
-                "Boolean. Optional, defaults to true"
-            ),
-            # examples=[True],
-        ),
-    ] = True
-
-    force_ocr: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, replace existing text with OCR-generated "
-                "text over content. Boolean. Optional, defaults to false."
-            ),
-            # examples=[False],
-        ),
-    ] = False
-
+class ConvertDocumentsRequestOptions(ConvertDocumentsOptions):
     ocr_engine: Annotated[  # type: ignore
         ocr_engines_enum,
         Field(
@@ -188,57 +30,6 @@ class ConvertDocumentsOptions(BaseModel):
         ),
     ] = ocr_engines_enum(EasyOcrOptions.kind)  # type: ignore
 
-    ocr_lang: Annotated[
-        Optional[list[str]],
-        Field(
-            description=(
-                "List of languages used by the OCR engine. "
-                "Note that each OCR engine has "
-                "different values for the language names. String or list of strings. "
-                "Optional, defaults to empty."
-            ),
-            examples=[["fr", "de", "es", "en"]],
-        ),
-    ] = None
-
-    pdf_backend: Annotated[
-        PdfBackend,
-        Field(
-            description=(
-                "The PDF backend to use. String. "
-                f"Allowed values: {', '.join([v.value for v in PdfBackend])}. "
-                f"Optional, defaults to {PdfBackend.DLPARSE_V4.value}."
-            ),
-            examples=[PdfBackend.DLPARSE_V4],
-        ),
-    ] = PdfBackend.DLPARSE_V4
-
-    table_mode: Annotated[
-        TableFormerMode,
-        Field(
-            description=(
-                "Mode to use for table structure, String. "
-                f"Allowed values: {', '.join([v.value for v in TableFormerMode])}. "
-                "Optional, defaults to fast."
-            ),
-            examples=[TableStructureOptions().mode],
-            # pattern="fast|accurate",
-        ),
-    ] = TableStructureOptions().mode
-
-    pipeline: Annotated[
-        ProcessingPipeline,
-        Field(description="Choose the pipeline to process PDF or image files."),
-    ] = ProcessingPipeline.STANDARD
-
-    page_range: Annotated[
-        PageRange,
-        Field(
-            description="Only convert a range of pages. The page number starts at 1.",
-            examples=[DEFAULT_PAGE_RANGE, (1, 4)],
-        ),
-    ] = DEFAULT_PAGE_RANGE
-
     document_timeout: Annotated[
         float,
         Field(
@@ -247,152 +38,3 @@ class ConvertDocumentsOptions(BaseModel):
             le=docling_serve_settings.max_document_timeout,
         ),
     ] = docling_serve_settings.max_document_timeout
-
-    abort_on_error: Annotated[
-        bool,
-        Field(
-            description=(
-                "Abort on error if enabled. Boolean. Optional, defaults to false."
-            ),
-            # examples=[False],
-        ),
-    ] = False
-
-    return_as_file: Annotated[
-        bool,
-        Field(
-            description=(
-                "Return the output as a zip file "
-                "(will happen anyway if multiple files are generated). "
-                "Boolean. Optional, defaults to false."
-            ),
-            examples=[False],
-        ),
-    ] = False
-
-    do_table_structure: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, the table structure will be extracted. "
-                "Boolean. Optional, defaults to true."
-            ),
-            examples=[True],
-        ),
-    ] = True
-
-    include_images: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, images will be extracted from the document. "
-                "Boolean. Optional, defaults to true."
-            ),
-            examples=[True],
-        ),
-    ] = True
-
-    images_scale: Annotated[
-        float,
-        Field(
-            description="Scale factor for images. Float. Optional, defaults to 2.0.",
-            examples=[2.0],
-        ),
-    ] = 2.0
-
-    md_page_break_placeholder: Annotated[
-        str,
-        Field(
-            description="Add this placeholder betweek pages in the markdown output.",
-            examples=["<!-- page-break -->", ""],
-        ),
-    ] = ""
-
-    do_code_enrichment: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, perform OCR code enrichment. "
-                "Boolean. Optional, defaults to false."
-            ),
-            examples=[False],
-        ),
-    ] = False
-
-    do_formula_enrichment: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, perform formula OCR, return LaTeX code. "
-                "Boolean. Optional, defaults to false."
-            ),
-            examples=[False],
-        ),
-    ] = False
-
-    do_picture_classification: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, classify pictures in documents. "
-                "Boolean. Optional, defaults to false."
-            ),
-            examples=[False],
-        ),
-    ] = False
-
-    do_picture_description: Annotated[
-        bool,
-        Field(
-            description=(
-                "If enabled, describe pictures in documents. "
-                "Boolean. Optional, defaults to false."
-            ),
-            examples=[False],
-        ),
-    ] = False
-
-    picture_description_area_threshold: Annotated[
-        float,
-        Field(
-            description="Minimum percentage of the area for a picture to be processed with the models.",
-            examples=[PictureDescriptionBaseOptions().picture_area_threshold],
-        ),
-    ] = PictureDescriptionBaseOptions().picture_area_threshold
-
-    picture_description_local: Annotated[
-        Optional[PictureDescriptionLocal],
-        Field(
-            description="Options for running a local vision-language model in the picture description. The parameters refer to a model hosted on Hugging Face. This parameter is mutually exclusive with picture_description_api.",
-            examples=[
-                PictureDescriptionLocal(repo_id="ibm-granite/granite-vision-3.2-2b"),
-                PictureDescriptionLocal(repo_id="HuggingFaceTB/SmolVLM-256M-Instruct"),
-            ],
-        ),
-    ] = None
-
-    picture_description_api: Annotated[
-        Optional[PictureDescriptionApi],
-        Field(
-            description="API details for using a vision-language model in the picture description. This parameter is mutually exclusive with picture_description_local.",
-            examples=[
-                PictureDescriptionApi(
-                    url="http://localhost:11434/v1/chat/completions",
-                    params={"model": "granite3.2-vision:2b"},
-                )
-            ],
-        ),
-    ] = None
-
-    @model_validator(mode="after")
-    def picture_description_exclusivity(self) -> Self:
-        # Validate picture description options
-        if (
-            self.picture_description_local is not None
-            and self.picture_description_api is not None
-        ):
-            raise ValueError(
-                "The parameters picture_description_local and picture_description_api are mutually exclusive, only one of them can be set."
-            )
-
-        return self

docling_serve/datamodel/engines.py

@@ -1,13 +0,0 @@
-import enum
-
-
-class TaskStatus(str, enum.Enum):
-    SUCCESS = "success"
-    PENDING = "pending"
-    STARTED = "started"
-    FAILURE = "failure"
-
-
-class AsyncEngine(str, enum.Enum):
-    LOCAL = "local"
-    KFP = "kfp"

docling_serve/datamodel/kfp.py

@@ -1,7 +0,0 @@
-from pydantic import AnyUrl, BaseModel
-
-
-class CallbackSpec(BaseModel):
-    url: AnyUrl
-    headers: dict[str, str] = {}
-    ca_cert: str = ""

docling_serve/datamodel/requests.py

@@ -1,62 +1,130 @@
-import base64
-from io import BytesIO
-from typing import Annotated, Any, Union
+import enum
+from functools import cache
+from typing import Annotated, Generic, Literal
 
-from pydantic import AnyHttpUrl, BaseModel, Field
+from pydantic import BaseModel, Field, model_validator
+from pydantic_core import PydanticCustomError
+from typing_extensions import Self, TypeVar
 
-from docling.datamodel.base_models import DocumentStream
+from docling_jobkit.datamodel.chunking import (
+    BaseChunkerOptions,
+)
+from docling_jobkit.datamodel.http_inputs import FileSource, HttpSource
+from docling_jobkit.datamodel.s3_coords import S3Coordinates
+from docling_jobkit.datamodel.task_targets import (
+    InBodyTarget,
+    PutTarget,
+    S3Target,
+    ZipTarget,
+)
 
-from docling_serve.datamodel.convert import ConvertDocumentsOptions
+from docling_serve.datamodel.convert import ConvertDocumentsRequestOptions
+from docling_serve.settings import AsyncEngine, docling_serve_settings
+
+## Sources
 
 
-class DocumentsConvertBase(BaseModel):
-    options: ConvertDocumentsOptions = ConvertDocumentsOptions()
+class FileSourceRequest(FileSource):
+    kind: Literal["file"] = "file"
 
 
-class HttpSource(BaseModel):
-    url: Annotated[
-        AnyHttpUrl,
-        Field(
-            description="HTTP url to process",
-            examples=["https://arxiv.org/pdf/2206.01062"],
-        ),
-    ]
-    headers: Annotated[
-        dict[str, Any],
-        Field(
-            description="Additional headers used to fetch the urls, "
-            "e.g. authorization, agent, etc"
-        ),
-    ] = {}
+class HttpSourceRequest(HttpSource):
+    kind: Literal["http"] = "http"
 
 
-class FileSource(BaseModel):
-    base64_string: Annotated[
-        str,
-        Field(
-            description="Content of the file serialized in base64. "
-            "For example it can be obtained via "
-            "`base64 -w 0 /path/to/file/pdf-to-convert.pdf`."
-        ),
-    ]
-    filename: Annotated[
-        str,
-        Field(description="Filename of the uploaded document", examples=["file.pdf"]),
-    ]
-
-    def to_document_stream(self) -> DocumentStream:
-        buf = BytesIO(base64.b64decode(self.base64_string))
-        return DocumentStream(stream=buf, name=self.filename)
+class S3SourceRequest(S3Coordinates):
+    kind: Literal["s3"] = "s3"
 
 
-class ConvertDocumentHttpSourcesRequest(DocumentsConvertBase):
-    http_sources: list[HttpSource]
+## Multipart targets
+class TargetName(str, enum.Enum):
+    INBODY = InBodyTarget().kind
+    ZIP = ZipTarget().kind
 
 
-class ConvertDocumentFileSourcesRequest(DocumentsConvertBase):
-    file_sources: list[FileSource]
-
-
-ConvertDocumentsRequest = Union[
-    ConvertDocumentFileSourcesRequest, ConvertDocumentHttpSourcesRequest
+## Aliases
+SourceRequestItem = Annotated[
+    FileSourceRequest | HttpSourceRequest | S3SourceRequest, Field(discriminator="kind")
 ]
+
+TargetRequest = Annotated[
+    InBodyTarget | ZipTarget | S3Target | PutTarget,
+    Field(discriminator="kind"),
+]
+
+
+## Complete Source request
+class ConvertDocumentsRequest(BaseModel):
+    options: ConvertDocumentsRequestOptions = ConvertDocumentsRequestOptions()
+    sources: list[SourceRequestItem]
+    target: TargetRequest = InBodyTarget()
+
+    @model_validator(mode="after")
+    def validate_s3_source_and_target(self) -> Self:
+        for source in self.sources:
+            if isinstance(source, S3SourceRequest):
+                if docling_serve_settings.eng_kind != AsyncEngine.KFP:
+                    raise PydanticCustomError(
+                        "error source", 'source kind "s3" requires engine kind "KFP"'
+                    )
+                if self.target.kind != "s3":
+                    raise PydanticCustomError(
+                        "error source", 'source kind "s3" requires target kind "s3"'
+                    )
+        if isinstance(self.target, S3Target):
+            for source in self.sources:
+                if isinstance(source, S3SourceRequest):
+                    return self
+            raise PydanticCustomError(
+                "error target", 'target kind "s3" requires source kind "s3"'
+            )
+        return self
+
+
+## Source chunking requests
+
+
+class BaseChunkDocumentsRequest(BaseModel):
+    convert_options: Annotated[
+        ConvertDocumentsRequestOptions, Field(description="Conversion options.")
+    ] = ConvertDocumentsRequestOptions()
+    sources: Annotated[
+        list[SourceRequestItem],
+        Field(description="List of input document sources to process."),
+    ]
+    include_converted_doc: Annotated[
+        bool,
+        Field(
+            description="If true, the output will include both the chunks and the converted document."
+        ),
+    ] = False
+    target: Annotated[
+        TargetRequest, Field(description="Specification for the type of output target.")
+    ] = InBodyTarget()
+
+
+ChunkingOptT = TypeVar("ChunkingOptT", bound=BaseChunkerOptions)
+
+
+class GenericChunkDocumentsRequest(BaseChunkDocumentsRequest, Generic[ChunkingOptT]):
+    chunking_options: ChunkingOptT
+
+
+@cache
+def make_request_model(
+    opt_type: type[ChunkingOptT],
+) -> type[GenericChunkDocumentsRequest[ChunkingOptT]]:
+    """
+    Dynamically create (and cache) a subclass of GenericChunkDocumentsRequest[opt_type]
+    with chunking_options having a default factory.
+    """
+    return type(
+        f"{opt_type.__name__}DocumentsRequest",
+        (GenericChunkDocumentsRequest[opt_type],),  # type: ignore[valid-type]
+        {
+            "__annotations__": {"chunking_options": opt_type},
+            "chunking_options": Field(
+                default_factory=opt_type, description="Options specific to the chunker."
+            ),
+        },
+    )

docling_serve/datamodel/responses.py

@@ -5,9 +5,12 @@ from pydantic import BaseModel
 
 from docling.datamodel.document import ConversionStatus, ErrorItem
 from docling.utils.profiling import ProfilingItem
-from docling_core.types.doc import DoclingDocument
-
-from docling_serve.datamodel.task_meta import TaskProcessingMeta
+from docling_jobkit.datamodel.result import (
+    ChunkedDocumentResultItem,
+    ExportDocumentResponse,
+    ExportResult,
+)
+from docling_jobkit.datamodel.task_meta import TaskProcessingMeta, TaskType
 
 
 # Status
@@ -19,29 +22,34 @@ class ClearResponse(BaseModel):
     status: str = "ok"
 
 
-class DocumentResponse(BaseModel):
-    filename: str
-    md_content: Optional[str] = None
-    json_content: Optional[DoclingDocument] = None
-    html_content: Optional[str] = None
-    text_content: Optional[str] = None
-    doctags_content: Optional[str] = None
-
-
 class ConvertDocumentResponse(BaseModel):
-    document: DocumentResponse
+    document: ExportDocumentResponse
     status: ConversionStatus
     errors: list[ErrorItem] = []
     processing_time: float
     timings: dict[str, ProfilingItem] = {}
 
 
+class PresignedUrlConvertDocumentResponse(BaseModel):
+    processing_time: float
+    num_converted: int
+    num_succeeded: int
+    num_failed: int
+
+
 class ConvertDocumentErrorResponse(BaseModel):
     status: ConversionStatus
 
 
+class ChunkDocumentResponse(BaseModel):
+    chunks: list[ChunkedDocumentResultItem]
+    documents: list[ExportResult]
+    processing_time: float
+
+
 class TaskStatusResponse(BaseModel):
     task_id: str
+    task_type: TaskType
     task_status: str
     task_position: Optional[int] = None
     task_meta: Optional[TaskProcessingMeta] = None

docling_serve/datamodel/task.py

@@ -1,55 +0,0 @@
-import datetime
-from functools import partial
-from pathlib import Path
-from typing import Optional, Union
-
-from fastapi.responses import FileResponse
-from pydantic import BaseModel, ConfigDict, Field
-
-from docling.datamodel.base_models import DocumentStream
-
-from docling_serve.datamodel.convert import ConvertDocumentsOptions
-from docling_serve.datamodel.engines import TaskStatus
-from docling_serve.datamodel.requests import FileSource, HttpSource
-from docling_serve.datamodel.responses import ConvertDocumentResponse
-from docling_serve.datamodel.task_meta import TaskProcessingMeta
-
-TaskSource = Union[HttpSource, FileSource, DocumentStream]
-
-
-class Task(BaseModel):
-    model_config = ConfigDict(arbitrary_types_allowed=True)
-
-    task_id: str
-    task_status: TaskStatus = TaskStatus.PENDING
-    sources: list[TaskSource] = []
-    options: Optional[ConvertDocumentsOptions]
-    result: Optional[Union[ConvertDocumentResponse, FileResponse]] = None
-    scratch_dir: Optional[Path] = None
-    processing_meta: Optional[TaskProcessingMeta] = None
-    created_at: datetime.datetime = Field(
-        default_factory=partial(datetime.datetime.now, datetime.timezone.utc)
-    )
-    started_at: Optional[datetime.datetime] = None
-    finished_at: Optional[datetime.datetime] = None
-    last_update_at: datetime.datetime = Field(
-        default_factory=partial(datetime.datetime.now, datetime.timezone.utc)
-    )
-
-    def set_status(self, status: TaskStatus):
-        now = datetime.datetime.now(datetime.timezone.utc)
-        if status == TaskStatus.STARTED and self.started_at is None:
-            self.started_at = now
-        if (
-            status in [TaskStatus.SUCCESS, TaskStatus.FAILURE]
-            and self.finished_at is None
-        ):
-            self.finished_at = now
-
-        self.last_update_at = now
-        self.task_status = status
-
-    def is_completed(self) -> bool:
-        if self.task_status in [TaskStatus.SUCCESS, TaskStatus.FAILURE]:
-            return True
-        return False

docling_serve/datamodel/task_meta.py

@@ -1,8 +0,0 @@
-from pydantic import BaseModel
-
-
-class TaskProcessingMeta(BaseModel):
-    num_docs: int
-    num_processed: int = 0
-    num_succeeded: int = 0
-    num_failed: int = 0

docling_serve/docling_conversion.py

@@ -1,256 +0,0 @@
-import hashlib
-import json
-import logging
-import sys
-from collections.abc import Iterable, Iterator
-from functools import lru_cache
-from pathlib import Path
-from typing import Any, Optional, Union
-
-from fastapi import HTTPException
-
-from docling.backend.docling_parse_backend import DoclingParseDocumentBackend
-from docling.backend.docling_parse_v2_backend import DoclingParseV2DocumentBackend
-from docling.backend.docling_parse_v4_backend import DoclingParseV4DocumentBackend
-from docling.backend.pdf_backend import PdfDocumentBackend
-from docling.backend.pypdfium2_backend import PyPdfiumDocumentBackend
-from docling.datamodel.base_models import DocumentStream, InputFormat
-from docling.datamodel.document import ConversionResult
-from docling.datamodel.pipeline_options import (
-    OcrOptions,
-    PdfBackend,
-    PdfPipelineOptions,
-    PictureDescriptionApiOptions,
-    PictureDescriptionVlmOptions,
-    ProcessingPipeline,
-    TableFormerMode,
-    VlmPipelineOptions,
-    smoldocling_vlm_conversion_options,
-    smoldocling_vlm_mlx_conversion_options,
-)
-from docling.document_converter import DocumentConverter, FormatOption, PdfFormatOption
-from docling.pipeline.vlm_pipeline import VlmPipeline
-from docling_core.types.doc import ImageRefMode
-
-from docling_serve.datamodel.convert import ConvertDocumentsOptions, ocr_factory
-from docling_serve.helper_functions import _to_list_of_strings
-from docling_serve.settings import docling_serve_settings
-
-_log = logging.getLogger(__name__)
-
-
-# Custom serializer for PdfFormatOption
-# (model_dump_json does not work with some classes)
-def _hash_pdf_format_option(pdf_format_option: PdfFormatOption) -> bytes:
-    data = pdf_format_option.model_dump(serialize_as_any=True)
-
-    # pipeline_options are not fully serialized by model_dump, dedicated pass
-    if pdf_format_option.pipeline_options:
-        data["pipeline_options"] = pdf_format_option.pipeline_options.model_dump(
-            serialize_as_any=True, mode="json"
-        )
-
-    # Replace `pipeline_cls` with a string representation
-    data["pipeline_cls"] = repr(data["pipeline_cls"])
-
-    # Replace `backend` with a string representation
-    data["backend"] = repr(data["backend"])
-
-    # Serialize the dictionary to JSON with sorted keys to have consistent hashes
-    serialized_data = json.dumps(data, sort_keys=True)
-    options_hash = hashlib.sha1(
-        serialized_data.encode(), usedforsecurity=False
-    ).digest()
-    return options_hash
-
-
-# Cache of DocumentConverter objects
-_options_map: dict[bytes, PdfFormatOption] = {}
-
-
-@lru_cache(maxsize=docling_serve_settings.options_cache_size)
-def _get_converter_from_hash(options_hash: bytes) -> DocumentConverter:
-    pdf_format_option = _options_map[options_hash]
-    format_options: dict[InputFormat, FormatOption] = {
-        InputFormat.PDF: pdf_format_option,
-        InputFormat.IMAGE: pdf_format_option,
-    }
-
-    return DocumentConverter(format_options=format_options)
-
-
-def get_converter(pdf_format_option: PdfFormatOption) -> DocumentConverter:
-    options_hash = _hash_pdf_format_option(pdf_format_option)
-    _options_map[options_hash] = pdf_format_option
-    return _get_converter_from_hash(options_hash)
-
-
-def _parse_standard_pdf_opts(
-    request: ConvertDocumentsOptions, artifacts_path: Optional[Path]
-) -> PdfPipelineOptions:
-    try:
-        ocr_options: OcrOptions = ocr_factory.create_options(
-            kind=request.ocr_engine.value,  # type: ignore
-            force_full_page_ocr=request.force_ocr,
-        )
-    except ImportError as err:
-        raise HTTPException(
-            status_code=400,
-            detail="The requested OCR engine"
-            f" (ocr_engine={request.ocr_engine.value})"  # type: ignore
-            " is not available on this system. Please choose another OCR engine "
-            "or contact your system administrator.\n"
-            f"{err}",
-        )
-
-    if request.ocr_lang is not None:
-        if isinstance(request.ocr_lang, str):
-            ocr_options.lang = _to_list_of_strings(request.ocr_lang)
-        else:
-            ocr_options.lang = request.ocr_lang
-
-    pipeline_options = PdfPipelineOptions(
-        artifacts_path=artifacts_path,
-        enable_remote_services=docling_serve_settings.enable_remote_services,
-        document_timeout=request.document_timeout,
-        do_ocr=request.do_ocr,
-        ocr_options=ocr_options,
-        do_table_structure=request.do_table_structure,
-        do_code_enrichment=request.do_code_enrichment,
-        do_formula_enrichment=request.do_formula_enrichment,
-        do_picture_classification=request.do_picture_classification,
-        do_picture_description=request.do_picture_description,
-    )
-    pipeline_options.table_structure_options.mode = TableFormerMode(request.table_mode)
-
-    if request.image_export_mode != ImageRefMode.PLACEHOLDER:
-        pipeline_options.generate_page_images = True
-        if request.image_export_mode == ImageRefMode.REFERENCED:
-            pipeline_options.generate_picture_images = True
-        if request.images_scale:
-            pipeline_options.images_scale = request.images_scale
-
-    if request.picture_description_local is not None:
-        pipeline_options.picture_description_options = (
-            PictureDescriptionVlmOptions.model_validate(
-                request.picture_description_local.model_dump()
-            )
-        )
-
-    if request.picture_description_api is not None:
-        pipeline_options.picture_description_options = (
-            PictureDescriptionApiOptions.model_validate(
-                request.picture_description_api.model_dump()
-            )
-        )
-    pipeline_options.picture_description_options.picture_area_threshold = (
-        request.picture_description_area_threshold
-    )
-
-    return pipeline_options
-
-
-def _parse_backend(request: ConvertDocumentsOptions) -> type[PdfDocumentBackend]:
-    if request.pdf_backend == PdfBackend.DLPARSE_V1:
-        backend: type[PdfDocumentBackend] = DoclingParseDocumentBackend
-    elif request.pdf_backend == PdfBackend.DLPARSE_V2:
-        backend = DoclingParseV2DocumentBackend
-    elif request.pdf_backend == PdfBackend.DLPARSE_V4:
-        backend = DoclingParseV4DocumentBackend
-    elif request.pdf_backend == PdfBackend.PYPDFIUM2:
-        backend = PyPdfiumDocumentBackend
-    else:
-        raise RuntimeError(f"Unexpected PDF backend type {request.pdf_backend}")
-
-    return backend
-
-
-def _parse_vlm_pdf_opts(
-    request: ConvertDocumentsOptions, artifacts_path: Optional[Path]
-) -> VlmPipelineOptions:
-    pipeline_options = VlmPipelineOptions(
-        artifacts_path=artifacts_path,
-        document_timeout=request.document_timeout,
-    )
-    pipeline_options.vlm_options = smoldocling_vlm_conversion_options
-    if sys.platform == "darwin":
-        try:
-            import mlx_vlm  # noqa: F401
-
-            pipeline_options.vlm_options = smoldocling_vlm_mlx_conversion_options
-        except ImportError:
-            _log.warning(
-                "To run SmolDocling faster, please install mlx-vlm:\n"
-                "pip install mlx-vlm"
-            )
-    return pipeline_options
-
-
-# Computes the PDF pipeline options and returns the PdfFormatOption and its hash
-def get_pdf_pipeline_opts(
-    request: ConvertDocumentsOptions,
-) -> PdfFormatOption:
-    artifacts_path: Optional[Path] = None
-    if docling_serve_settings.artifacts_path is not None:
-        if str(docling_serve_settings.artifacts_path.absolute()) == "":
-            _log.info(
-                "artifacts_path is an empty path, model weights will be downloaded "
-                "at runtime."
-            )
-            artifacts_path = None
-        elif docling_serve_settings.artifacts_path.is_dir():
-            _log.info(
-                "artifacts_path is set to a valid directory. "
-                "No model weights will be downloaded at runtime."
-            )
-            artifacts_path = docling_serve_settings.artifacts_path
-        else:
-            _log.warning(
-                "artifacts_path is set to an invalid directory. "
-                "The system will download the model weights at runtime."
-            )
-            artifacts_path = None
-    else:
-        _log.info(
-            "artifacts_path is unset. "
-            "The system will download the model weights at runtime."
-        )
-
-    pipeline_options: Union[PdfPipelineOptions, VlmPipelineOptions]
-    if request.pipeline == ProcessingPipeline.STANDARD:
-        pipeline_options = _parse_standard_pdf_opts(request, artifacts_path)
-        backend = _parse_backend(request)
-        pdf_format_option = PdfFormatOption(
-            pipeline_options=pipeline_options,
-            backend=backend,
-        )
-
-    elif request.pipeline == ProcessingPipeline.VLM:
-        pipeline_options = _parse_vlm_pdf_opts(request, artifacts_path)
-        pdf_format_option = PdfFormatOption(
-            pipeline_cls=VlmPipeline, pipeline_options=pipeline_options
-        )
-    else:
-        raise NotImplementedError(
-            f"The pipeline {request.pipeline} is not implemented."
-        )
-
-    return pdf_format_option
-
-
-def convert_documents(
-    sources: Iterable[Union[Path, str, DocumentStream]],
-    options: ConvertDocumentsOptions,
-    headers: Optional[dict[str, Any]] = None,
-):
-    pdf_format_option = get_pdf_pipeline_opts(options)
-    converter = get_converter(pdf_format_option)
-    results: Iterator[ConversionResult] = converter.convert_all(
-        sources,
-        headers=headers,
-        page_range=options.page_range,
-        max_file_size=docling_serve_settings.max_file_size,
-        max_num_pages=docling_serve_settings.max_num_pages,
-    )
-
-    return results

docling_serve/engines/async_kfp/kfp_pipeline.py

@@ -1,137 +0,0 @@
-# ruff: noqa: E402, UP006, UP035
-
-from typing import Any, Dict, List
-
-from kfp import dsl
-
-PYTHON_BASE_IMAGE = "python:3.12"
-
-
-@dsl.component(
-    base_image=PYTHON_BASE_IMAGE,
-    packages_to_install=[
-        "pydantic",
-        "docling-serve @ git+https://github.com/docling-project/docling-serve@feat-kfp-engine",
-    ],
-    pip_index_urls=["https://download.pytorch.org/whl/cpu", "https://pypi.org/simple"],
-)
-def generate_chunks(
-    run_name: str,
-    request: Dict[str, Any],
-    batch_size: int,
-    callbacks: List[Dict[str, Any]],
-) -> List[List[Dict[str, Any]]]:
-    from pydantic import TypeAdapter
-
-    from docling_serve.datamodel.callback import (
-        ProgressCallbackRequest,
-        ProgressSetNumDocs,
-    )
-    from docling_serve.datamodel.kfp import CallbackSpec
-    from docling_serve.engines.async_kfp.notify import notify_callbacks
-
-    CallbacksListType = TypeAdapter(list[CallbackSpec])
-
-    sources = request["http_sources"]
-    splits = [sources[i : i + batch_size] for i in range(0, len(sources), batch_size)]
-
-    total = sum(len(chunk) for chunk in splits)
-    payload = ProgressCallbackRequest(
-        task_id=run_name, progress=ProgressSetNumDocs(num_docs=total)
-    )
-    notify_callbacks(
-        payload=payload,
-        callbacks=CallbacksListType.validate_python(callbacks),
-    )
-
-    return splits
-
-
-@dsl.component(
-    base_image=PYTHON_BASE_IMAGE,
-    packages_to_install=[
-        "pydantic",
-        "docling-serve @ git+https://github.com/docling-project/docling-serve@feat-kfp-engine",
-    ],
-    pip_index_urls=["https://download.pytorch.org/whl/cpu", "https://pypi.org/simple"],
-)
-def convert_batch(
-    run_name: str,
-    data_splits: List[Dict[str, Any]],
-    request: Dict[str, Any],
-    callbacks: List[Dict[str, Any]],
-    output_path: dsl.OutputPath("Directory"),  # type: ignore
-):
-    from pathlib import Path
-
-    from pydantic import AnyUrl, TypeAdapter
-
-    from docling_serve.datamodel.callback import (
-        FailedDocsItem,
-        ProgressCallbackRequest,
-        ProgressUpdateProcessed,
-        SucceededDocsItem,
-    )
-    from docling_serve.datamodel.convert import ConvertDocumentsOptions
-    from docling_serve.datamodel.kfp import CallbackSpec
-    from docling_serve.datamodel.requests import HttpSource
-    from docling_serve.engines.async_kfp.notify import notify_callbacks
-
-    CallbacksListType = TypeAdapter(list[CallbackSpec])
-
-    convert_options = ConvertDocumentsOptions.model_validate(request["options"])
-    print(convert_options)
-
-    output_dir = Path(output_path)
-    output_dir.mkdir(exist_ok=True, parents=True)
-    docs_succeeded: list[SucceededDocsItem] = []
-    docs_failed: list[FailedDocsItem] = []
-    for source_dict in data_splits:
-        source = HttpSource.model_validate(source_dict)
-        filename = Path(str(AnyUrl(source.url).path)).name
-        output_filename = output_dir / filename
-        print(f"Writing {output_filename}")
-        with output_filename.open("w") as f:
-            f.write(source.model_dump_json())
-        docs_succeeded.append(SucceededDocsItem(source=source.url))
-
-    payload = ProgressCallbackRequest(
-        task_id=run_name,
-        progress=ProgressUpdateProcessed(
-            num_failed=len(docs_failed),
-            num_processed=len(docs_succeeded) + len(docs_failed),
-            num_succeeded=len(docs_succeeded),
-            docs_succeeded=docs_succeeded,
-            docs_failed=docs_failed,
-        ),
-    )
-
-    print(payload)
-    notify_callbacks(
-        payload=payload,
-        callbacks=CallbacksListType.validate_python(callbacks),
-    )
-
-
-@dsl.pipeline()
-def process(
-    batch_size: int,
-    request: Dict[str, Any],
-    callbacks: List[Dict[str, Any]] = [],
-    run_name: str = "",
-):
-    chunks_task = generate_chunks(
-        run_name=run_name,
-        request=request,
-        batch_size=batch_size,
-        callbacks=callbacks,
-    )
-    chunks_task.set_caching_options(False)
-
-    with dsl.ParallelFor(chunks_task.output, parallelism=4) as data_splits:
-        convert_batch(
-            run_name=run_name,
-            data_splits=data_splits,
-            request=request,
-            callbacks=callbacks,
-        )

docling_serve/engines/async_kfp/notify.py

@@ -1,32 +0,0 @@
-import ssl
-
-import certifi
-import httpx
-
-from docling_serve.datamodel.callback import ProgressCallbackRequest
-from docling_serve.datamodel.kfp import CallbackSpec
-
-
-def notify_callbacks(
-    payload: ProgressCallbackRequest,
-    callbacks: list[CallbackSpec],
-):
-    if len(callbacks) == 0:
-        return
-
-    for callback in callbacks:
-        # https://www.python-httpx.org/advanced/ssl/#configuring-client-instances
-        if callback.ca_cert:
-            ctx = ssl.create_default_context(cadata=callback.ca_cert)
-        else:
-            ctx = ssl.create_default_context(cafile=certifi.where())
-
-        try:
-            httpx.post(
-                str(callback.url),
-                headers=callback.headers,
-                json=payload.model_dump(mode="json"),
-                verify=ctx,
-            )
-        except httpx.HTTPError as err:
-            print(f"Error notifying callback {callback.url}: {err}")

docling_serve/engines/async_kfp/orchestrator.py

@@ -1,235 +0,0 @@
-import datetime
-import json
-import logging
-import uuid
-from pathlib import Path
-from typing import Optional
-
-from kfp_server_api.models import V2beta1RuntimeState
-from pydantic import BaseModel, TypeAdapter
-from pydantic_settings import SettingsConfigDict
-
-from docling_serve.datamodel.callback import (
-    ProgressCallbackRequest,
-    ProgressSetNumDocs,
-    ProgressUpdateProcessed,
-)
-from docling_serve.datamodel.convert import ConvertDocumentsOptions
-from docling_serve.datamodel.engines import TaskStatus
-from docling_serve.datamodel.kfp import CallbackSpec
-from docling_serve.datamodel.requests import HttpSource
-from docling_serve.datamodel.task import Task, TaskSource
-from docling_serve.datamodel.task_meta import TaskProcessingMeta
-from docling_serve.engines.async_kfp.kfp_pipeline import process
-from docling_serve.engines.async_orchestrator import (
-    BaseAsyncOrchestrator,
-    ProgressInvalid,
-)
-from docling_serve.settings import docling_serve_settings
-
-_log = logging.getLogger(__name__)
-
-
-class _RunItem(BaseModel):
-    model_config = SettingsConfigDict(arbitrary_types_allowed=True)
-
-    run_id: str
-    state: str
-    created_at: datetime.datetime
-    scheduled_at: datetime.datetime
-    finished_at: datetime.datetime
-
-
-class AsyncKfpOrchestrator(BaseAsyncOrchestrator):
-    def __init__(self):
-        super().__init__()
-        import kfp
-
-        kfp_endpoint = docling_serve_settings.eng_kfp_endpoint
-        if kfp_endpoint is None:
-            raise ValueError("KFP endpoint is required when using the KFP engine.")
-
-        kube_sa_token_path = Path("/run/secrets/kubernetes.io/serviceaccount/token")
-        kube_sa_ca_cert_path = Path(
-            "/run/secrets/kubernetes.io/serviceaccount/service-ca.crt"
-        )
-
-        ssl_ca_cert = docling_serve_settings.eng_kfp_ca_cert_path
-        token = docling_serve_settings.eng_kfp_token
-        if (
-            ssl_ca_cert is None
-            and ".svc" in kfp_endpoint.host
-            and kube_sa_ca_cert_path.exists()
-        ):
-            ssl_ca_cert = str(kube_sa_ca_cert_path)
-        if token is None and kube_sa_token_path.exists():
-            token = kube_sa_token_path.read_text()
-
-        self._client = kfp.Client(
-            host=str(kfp_endpoint),
-            existing_token=token,
-            ssl_ca_cert=ssl_ca_cert,
-            # verify_ssl=False,
-        )
-
-    async def enqueue(
-        self, sources: list[TaskSource], options: ConvertDocumentsOptions
-    ) -> Task:
-        callbacks = []
-        if docling_serve_settings.eng_kfp_self_callback_endpoint is not None:
-            headers = {}
-            if docling_serve_settings.eng_kfp_self_callback_token_path is not None:
-                token = (
-                    docling_serve_settings.eng_kfp_self_callback_token_path.read_text()
-                )
-                headers["Authorization"] = f"Bearer {token}"
-            ca_cert = ""
-            if docling_serve_settings.eng_kfp_self_callback_ca_cert_path is not None:
-                ca_cert = docling_serve_settings.eng_kfp_self_callback_ca_cert_path.read_text()
-            callbacks.append(
-                CallbackSpec(
-                    url=docling_serve_settings.eng_kfp_self_callback_endpoint,
-                    headers=headers,
-                    ca_cert=ca_cert,
-                )
-            )
-
-        CallbacksType = TypeAdapter(list[CallbackSpec])
-        SourcesListType = TypeAdapter(list[HttpSource])
-        http_sources = [s for s in sources if isinstance(s, HttpSource)]
-        # hack: since the current kfp backend is not resolving the job_id placeholder,
-        # we set the run_name and pass it as argument to the job itself.
-        run_name = f"docling-job-{uuid.uuid4()}"
-        kfp_run = self._client.create_run_from_pipeline_func(
-            process,
-            arguments={
-                "batch_size": 10,
-                "sources": SourcesListType.dump_python(http_sources, mode="json"),
-                "options": options.model_dump(mode="json"),
-                "callbacks": CallbacksType.dump_python(callbacks, mode="json"),
-                "run_name": run_name,
-            },
-            run_name=run_name,
-        )
-        task_id = kfp_run.run_id
-
-        task = Task(task_id=task_id, sources=sources, options=options)
-        await self.init_task_tracking(task)
-        return task
-
-    async def _update_task_from_run(self, task_id: str, wait: float = 0.0):
-        run_info = self._client.get_run(run_id=task_id)
-        task = await self.get_raw_task(task_id=task_id)
-        # RUNTIME_STATE_UNSPECIFIED = "RUNTIME_STATE_UNSPECIFIED"
-        # PENDING = "PENDING"
-        # RUNNING = "RUNNING"
-        # SUCCEEDED = "SUCCEEDED"
-        # SKIPPED = "SKIPPED"
-        # FAILED = "FAILED"
-        # CANCELING = "CANCELING"
-        # CANCELED = "CANCELED"
-        # PAUSED = "PAUSED"
-        if run_info.state == V2beta1RuntimeState.SUCCEEDED:
-            task.set_status(TaskStatus.SUCCESS)
-        elif run_info.state == V2beta1RuntimeState.PENDING:
-            task.set_status(TaskStatus.PENDING)
-        elif run_info.state == V2beta1RuntimeState.RUNNING:
-            task.set_status(TaskStatus.STARTED)
-        else:
-            task.set_status(TaskStatus.FAILURE)
-
-    async def task_status(self, task_id: str, wait: float = 0.0) -> Task:
-        await self._update_task_from_run(task_id=task_id, wait=wait)
-        return await self.get_raw_task(task_id=task_id)
-
-    async def _get_pending(self) -> list[_RunItem]:
-        runs: list[_RunItem] = []
-        next_page: Optional[str] = None
-        while True:
-            res = self._client.list_runs(
-                page_token=next_page,
-                page_size=20,
-                filter=json.dumps(
-                    {
-                        "predicates": [
-                            {
-                                "operation": "EQUALS",
-                                "key": "state",
-                                "stringValue": "PENDING",
-                            }
-                        ]
-                    }
-                ),
-            )
-            if res.runs is not None:
-                for run in res.runs:
-                    runs.append(
-                        _RunItem(
-                            run_id=run.run_id,
-                            state=run.state,
-                            created_at=run.created_at,
-                            scheduled_at=run.scheduled_at,
-                            finished_at=run.finished_at,
-                        )
-                    )
-            if res.next_page_token is None:
-                break
-            next_page = res.next_page_token
-        return runs
-
-    async def queue_size(self) -> int:
-        runs = await self._get_pending()
-        return len(runs)
-
-    async def get_queue_position(self, task_id: str) -> Optional[int]:
-        runs = await self._get_pending()
-        for pos, run in enumerate(runs, start=1):
-            if run.run_id == task_id:
-                return pos
-        return None
-
-    async def process_queue(self):
-        return
-
-    async def warm_up_caches(self):
-        return
-
-    async def _get_run_id(self, run_name: str) -> str:
-        res = self._client.list_runs(
-            filter=json.dumps(
-                {
-                    "predicates": [
-                        {
-                            "operation": "EQUALS",
-                            "key": "name",
-                            "stringValue": run_name,
-                        }
-                    ]
-                }
-            ),
-        )
-        if res.runs is not None and len(res.runs) > 0:
-            return res.runs[0].run_id
-        raise RuntimeError(f"Run with {run_name=} not found.")
-
-    async def receive_task_progress(self, request: ProgressCallbackRequest):
-        task_id = await self._get_run_id(run_name=request.task_id)
-        progress = request.progress
-        task = await self.get_raw_task(task_id=task_id)
-
-        if isinstance(progress, ProgressSetNumDocs):
-            task.processing_meta = TaskProcessingMeta(num_docs=progress.num_docs)
-            task.task_status = TaskStatus.STARTED
-
-        elif isinstance(progress, ProgressUpdateProcessed):
-            if task.processing_meta is None:
-                raise ProgressInvalid(
-                    "UpdateProcessed was called before setting the expected number of documents."
-                )
-            task.processing_meta.num_processed += progress.num_processed
-            task.processing_meta.num_succeeded += progress.num_succeeded
-            task.processing_meta.num_failed += progress.num_failed
-            task.task_status = TaskStatus.STARTED
-
-        # TODO: could be moved to BackgroundTask
-        await self.notify_task_subscribers(task_id=task_id)

docling_serve/engines/async_local/orchestrator.py

@@ -1,60 +0,0 @@
-import asyncio
-import logging
-import uuid
-from typing import Optional
-
-from docling.datamodel.base_models import InputFormat
-
-from docling_serve.datamodel.convert import ConvertDocumentsOptions
-from docling_serve.datamodel.task import Task, TaskSource
-from docling_serve.docling_conversion import get_converter, get_pdf_pipeline_opts
-from docling_serve.engines.async_local.worker import AsyncLocalWorker
-from docling_serve.engines.async_orchestrator import BaseAsyncOrchestrator
-from docling_serve.settings import docling_serve_settings
-
-_log = logging.getLogger(__name__)
-
-
-class AsyncLocalOrchestrator(BaseAsyncOrchestrator):
-    def __init__(self):
-        super().__init__()
-        self.task_queue = asyncio.Queue()
-        self.queue_list: list[str] = []
-
-    async def enqueue(
-        self, sources: list[TaskSource], options: ConvertDocumentsOptions
-    ) -> Task:
-        task_id = str(uuid.uuid4())
-        task = Task(task_id=task_id, sources=sources, options=options)
-        await self.init_task_tracking(task)
-
-        self.queue_list.append(task_id)
-        await self.task_queue.put(task_id)
-        return task
-
-    async def queue_size(self) -> int:
-        return self.task_queue.qsize()
-
-    async def get_queue_position(self, task_id: str) -> Optional[int]:
-        return (
-            self.queue_list.index(task_id) + 1 if task_id in self.queue_list else None
-        )
-
-    async def process_queue(self):
-        # Create a pool of workers
-        workers = []
-        for i in range(docling_serve_settings.eng_loc_num_workers):
-            _log.debug(f"Starting worker {i}")
-            w = AsyncLocalWorker(i, self)
-            worker_task = asyncio.create_task(w.loop())
-            workers.append(worker_task)
-
-        # Wait for all workers to complete (they won't, as they run indefinitely)
-        await asyncio.gather(*workers)
-        _log.debug("All workers completed.")
-
-    async def warm_up_caches(self):
-        # Converter with default options
-        pdf_format_option = get_pdf_pipeline_opts(ConvertDocumentsOptions())
-        converter = get_converter(pdf_format_option)
-        converter.initialize_pipeline(InputFormat.PDF)

docling_serve/engines/async_local/worker.py

@@ -1,124 +0,0 @@
-import asyncio
-import logging
-import shutil
-import time
-from typing import TYPE_CHECKING, Any, Optional, Union
-
-from fastapi.responses import FileResponse
-
-from docling.datamodel.base_models import DocumentStream
-
-from docling_serve.datamodel.engines import TaskStatus
-from docling_serve.datamodel.requests import FileSource, HttpSource
-from docling_serve.docling_conversion import convert_documents
-from docling_serve.response_preparation import process_results
-from docling_serve.storage import get_scratch
-
-if TYPE_CHECKING:
-    from docling_serve.engines.async_local.orchestrator import AsyncLocalOrchestrator
-
-_log = logging.getLogger(__name__)
-
-
-class AsyncLocalWorker:
-    def __init__(self, worker_id: int, orchestrator: "AsyncLocalOrchestrator"):
-        self.worker_id = worker_id
-        self.orchestrator = orchestrator
-
-    async def loop(self):
-        _log.debug(f"Starting loop for worker {self.worker_id}")
-        while True:
-            task_id: str = await self.orchestrator.task_queue.get()
-            self.orchestrator.queue_list.remove(task_id)
-
-            if task_id not in self.orchestrator.tasks:
-                raise RuntimeError(f"Task {task_id} not found.")
-            task = self.orchestrator.tasks[task_id]
-
-            try:
-                task.set_status(TaskStatus.STARTED)
-                _log.info(f"Worker {self.worker_id} processing task {task_id}")
-
-                # Notify clients about task updates
-                await self.orchestrator.notify_task_subscribers(task_id)
-
-                # Notify clients about queue updates
-                await self.orchestrator.notify_queue_positions()
-
-                # Define a callback function to send progress updates to the client.
-                # TODO: send partial updates, e.g. when a document in the batch is done
-                def run_conversion():
-                    convert_sources: list[Union[str, DocumentStream]] = []
-                    headers: Optional[dict[str, Any]] = None
-                    for source in task.sources:
-                        if isinstance(source, DocumentStream):
-                            convert_sources.append(source)
-                        elif isinstance(source, FileSource):
-                            convert_sources.append(source.to_document_stream())
-                        elif isinstance(source, HttpSource):
-                            convert_sources.append(str(source.url))
-                            if headers is None and source.headers:
-                                headers = source.headers
-
-                    # Note: results are only an iterator->lazy evaluation
-                    results = convert_documents(
-                        sources=convert_sources,
-                        options=task.options,
-                        headers=headers,
-                    )
-
-                    # The real processing will happen here
-                    work_dir = get_scratch() / task_id
-                    response = process_results(
-                        conversion_options=task.options,
-                        conv_results=results,
-                        work_dir=work_dir,
-                    )
-
-                    if work_dir.exists():
-                        task.scratch_dir = work_dir
-                        if not isinstance(response, FileResponse):
-                            _log.warning(
-                                f"Task {task_id=} produced content in {work_dir=} but the response is not a file."
-                            )
-                            shutil.rmtree(work_dir, ignore_errors=True)
-
-                    return response
-
-                start_time = time.monotonic()
-
-                # Run the prediction in a thread to avoid blocking the event loop.
-                # Get the current event loop
-                # loop = asyncio.get_event_loop()
-                # future = asyncio.run_coroutine_threadsafe(
-                #     run_conversion(),
-                #     loop=loop
-                # )
-                # response = future.result()
-
-                # Run in a thread
-                response = await asyncio.to_thread(
-                    run_conversion,
-                )
-                processing_time = time.monotonic() - start_time
-
-                task.result = response
-                task.sources = []
-                task.options = None
-
-                task.set_status(TaskStatus.SUCCESS)
-                _log.info(
-                    f"Worker {self.worker_id} completed job {task_id} "
-                    f"in {processing_time:.2f} seconds"
-                )
-
-            except Exception as e:
-                _log.error(
-                    f"Worker {self.worker_id} failed to process job {task_id}: {e}"
-                )
-                task.set_status(TaskStatus.FAILURE)
-
-            finally:
-                await self.orchestrator.notify_task_subscribers(task_id)
-                self.orchestrator.task_queue.task_done()
-                _log.debug(f"Worker {self.worker_id} completely done with {task_id}")

docling_serve/engines/async_orchestrator.py

@@ -1,127 +0,0 @@
-import asyncio
-import datetime
-import logging
-import shutil
-from typing import Union
-
-from fastapi import BackgroundTasks, WebSocket
-from fastapi.responses import FileResponse
-
-from docling_serve.datamodel.callback import ProgressCallbackRequest
-from docling_serve.datamodel.engines import TaskStatus
-from docling_serve.datamodel.responses import (
-    ConvertDocumentResponse,
-    MessageKind,
-    TaskStatusResponse,
-    WebsocketMessage,
-)
-from docling_serve.datamodel.task import Task
-from docling_serve.engines.base_orchestrator import (
-    BaseOrchestrator,
-    OrchestratorError,
-    TaskNotFoundError,
-)
-from docling_serve.settings import docling_serve_settings
-
-_log = logging.getLogger(__name__)
-
-
-class ProgressInvalid(OrchestratorError):
-    pass
-
-
-class BaseAsyncOrchestrator(BaseOrchestrator):
-    def __init__(self):
-        self.tasks: dict[str, Task] = {}
-        self.task_subscribers: dict[str, set[WebSocket]] = {}
-
-    async def init_task_tracking(self, task: Task):
-        task_id = task.task_id
-        self.tasks[task.task_id] = task
-        self.task_subscribers[task_id] = set()
-
-    async def get_raw_task(self, task_id: str) -> Task:
-        if task_id not in self.tasks:
-            raise TaskNotFoundError()
-        return self.tasks[task_id]
-
-    async def task_status(self, task_id: str, wait: float = 0.0) -> Task:
-        return await self.get_raw_task(task_id=task_id)
-
-    async def task_result(
-        self, task_id: str, background_tasks: BackgroundTasks
-    ) -> Union[ConvertDocumentResponse, FileResponse, None]:
-        try:
-            task = await self.get_raw_task(task_id=task_id)
-            if task.is_completed() and docling_serve_settings.single_use_results:
-                if task.scratch_dir is not None:
-                    background_tasks.add_task(
-                        shutil.rmtree, task.scratch_dir, ignore_errors=True
-                    )
-
-                async def _remove_task_impl():
-                    await asyncio.sleep(docling_serve_settings.result_removal_delay)
-                    await self.delete_task(task_id=task.task_id)
-
-                async def _remove_task():
-                    asyncio.create_task(_remove_task_impl())  # noqa: RUF006
-
-                background_tasks.add_task(_remove_task)
-
-            return task.result
-        except TaskNotFoundError:
-            return None
-
-    async def delete_task(self, task_id: str):
-        _log.info(f"Deleting {task_id=}")
-        if task_id in self.task_subscribers:
-            for websocket in self.task_subscribers[task_id]:
-                await websocket.close()
-
-            del self.task_subscribers[task_id]
-
-        if task_id in self.tasks:
-            del self.tasks[task_id]
-
-    async def clear_results(self, older_than: float = 0.0):
-        cutoff_time = datetime.datetime.now(datetime.timezone.utc) - datetime.timedelta(
-            seconds=older_than
-        )
-
-        tasks_to_delete = [
-            task_id
-            for task_id, task in self.tasks.items()
-            if task.finished_at is not None and task.finished_at < cutoff_time
-        ]
-        for task_id in tasks_to_delete:
-            await self.delete_task(task_id=task_id)
-
-    async def notify_task_subscribers(self, task_id: str):
-        if task_id not in self.task_subscribers:
-            raise RuntimeError(f"Task {task_id} does not have a subscribers list.")
-
-        task = await self.get_raw_task(task_id=task_id)
-        task_queue_position = await self.get_queue_position(task_id)
-        msg = TaskStatusResponse(
-            task_id=task.task_id,
-            task_status=task.task_status,
-            task_position=task_queue_position,
-            task_meta=task.processing_meta,
-        )
-        for websocket in self.task_subscribers[task_id]:
-            await websocket.send_text(
-                WebsocketMessage(message=MessageKind.UPDATE, task=msg).model_dump_json()
-            )
-            if task.is_completed():
-                await websocket.close()
-
-    async def notify_queue_positions(self):
-        for task_id in self.task_subscribers.keys():
-            # notify only pending tasks
-            if self.tasks[task_id].task_status != TaskStatus.PENDING:
-                continue
-
-            await self.notify_task_subscribers(task_id)
-
-    async def receive_task_progress(self, request: ProgressCallbackRequest):
-        raise NotImplementedError()

docling_serve/engines/async_orchestrator_factory.py

@@ -1,21 +0,0 @@
-from functools import lru_cache
-
-from docling_serve.datamodel.engines import AsyncEngine
-from docling_serve.engines.async_orchestrator import BaseAsyncOrchestrator
-from docling_serve.settings import docling_serve_settings
-
-
-@lru_cache
-def get_async_orchestrator() -> BaseAsyncOrchestrator:
-    if docling_serve_settings.eng_kind == AsyncEngine.LOCAL:
-        from docling_serve.engines.async_local.orchestrator import (
-            AsyncLocalOrchestrator,
-        )
-
-        return AsyncLocalOrchestrator()
-    elif docling_serve_settings.eng_kind == AsyncEngine.KFP:
-        from docling_serve.engines.async_kfp.orchestrator import AsyncKfpOrchestrator
-
-        return AsyncKfpOrchestrator()
-
-    raise RuntimeError(f"Engine {docling_serve_settings.eng_kind} not recognized.")

docling_serve/engines/base_orchestrator.py

@@ -1,55 +0,0 @@
-from abc import ABC, abstractmethod
-from typing import Optional, Union
-
-from fastapi import BackgroundTasks
-from fastapi.responses import FileResponse
-
-from docling_serve.datamodel.convert import ConvertDocumentsOptions
-from docling_serve.datamodel.responses import ConvertDocumentResponse
-from docling_serve.datamodel.task import Task, TaskSource
-
-
-class OrchestratorError(Exception):
-    pass
-
-
-class TaskNotFoundError(OrchestratorError):
-    pass
-
-
-class BaseOrchestrator(ABC):
-    @abstractmethod
-    async def enqueue(
-        self, sources: list[TaskSource], options: ConvertDocumentsOptions
-    ) -> Task:
-        pass
-
-    @abstractmethod
-    async def queue_size(self) -> int:
-        pass
-
-    @abstractmethod
-    async def get_queue_position(self, task_id: str) -> Optional[int]:
-        pass
-
-    @abstractmethod
-    async def task_status(self, task_id: str, wait: float = 0.0) -> Task:
-        pass
-
-    @abstractmethod
-    async def task_result(
-        self, task_id: str, background_tasks: BackgroundTasks
-    ) -> Union[ConvertDocumentResponse, FileResponse, None]:
-        pass
-
-    @abstractmethod
-    async def clear_results(self, older_than: float = 0.0):
-        pass
-
-    @abstractmethod
-    async def process_queue(self):
-        pass
-
-    @abstractmethod
-    async def warm_up_caches(self):
-        pass

docling_serve/gradio_ui.py

@@ -4,6 +4,7 @@ import itertools
 import json
 import logging
 import ssl
+import sys
 import tempfile
 import time
 from pathlib import Path
@@ -224,24 +225,34 @@ def auto_set_return_as_file(
 
 def change_ocr_lang(ocr_engine):
     if ocr_engine == "easyocr":
-        return "en,fr,de,es"
+        return gr.update(visible=True, value="en,fr,de,es")
     elif ocr_engine == "tesseract_cli":
-        return "eng,fra,deu,spa"
+        return gr.update(visible=True, value="eng,fra,deu,spa")
     elif ocr_engine == "tesseract":
-        return "eng,fra,deu,spa"
+        return gr.update(visible=True, value="eng,fra,deu,spa")
     elif ocr_engine == "rapidocr":
-        return "english,chinese"
+        return gr.update(visible=True, value="english,chinese")
+    elif ocr_engine == "ocrmac":
+        return gr.update(visible=True, value="fr-FR,de-DE,es-ES,en-US")
+
+    return gr.update(visible=False, value="")
 
 
-def wait_task_finish(task_id: str, return_as_file: bool):
+def wait_task_finish(auth: str, task_id: str, return_as_file: bool):
     conversion_sucess = False
     task_finished = False
     task_status = ""
+
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = str(auth)
+
     ssl_ctx = get_ssl_context()
     while not task_finished:
         try:
             response = httpx.get(
-                f"{get_api_endpoint()}/v1alpha/status/poll/{task_id}?wait=5",
+                f"{get_api_endpoint()}/v1/status/poll/{task_id}?wait=5",
+                headers=headers,
                 verify=ssl_ctx,
                 timeout=15,
             )
@@ -264,7 +275,8 @@ def wait_task_finish(task_id: str, return_as_file: bool):
     if conversion_sucess:
         try:
             response = httpx.get(
-                f"{get_api_endpoint()}/v1alpha/result/{task_id}",
+                f"{get_api_endpoint()}/v1/result/{task_id}",
+                headers=headers,
                 timeout=15,
                 verify=ssl_ctx,
             )
@@ -279,6 +291,7 @@ def wait_task_finish(task_id: str, return_as_file: bool):
 
 
 def process_url(
+    auth,
     input_sources,
     to_formats,
     image_export_mode,
@@ -296,8 +309,11 @@ def process_url(
     do_picture_classification,
     do_picture_description,
 ):
+    target = {"kind": "zip" if return_as_file else "inbody"}
     parameters = {
-        "http_sources": [{"url": source} for source in input_sources.split(",")],
+        "sources": [
+            {"kind": "http", "url": source} for source in input_sources.split(",")
+        ],
         "options": {
             "to_formats": to_formats,
             "image_export_mode": image_export_mode,
@@ -309,25 +325,32 @@ def process_url(
             "pdf_backend": pdf_backend,
             "table_mode": table_mode,
             "abort_on_error": abort_on_error,
-            "return_as_file": return_as_file,
             "do_code_enrichment": do_code_enrichment,
             "do_formula_enrichment": do_formula_enrichment,
             "do_picture_classification": do_picture_classification,
             "do_picture_description": do_picture_description,
         },
+        "target": target,
     }
     if (
-        not parameters["http_sources"]
-        or len(parameters["http_sources"]) == 0
-        or parameters["http_sources"][0]["url"] == ""
+        not parameters["sources"]
+        or len(parameters["sources"]) == 0
+        or parameters["sources"][0]["url"] == ""
     ):
         logger.error("No input sources provided.")
         raise gr.Error("No input sources provided.", print_exception=False)
+
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = str(auth)
+
+    print(f"{headers=}")
     try:
         ssl_ctx = get_ssl_context()
         response = httpx.post(
-            f"{get_api_endpoint()}/v1alpha/convert/source/async",
+            f"{get_api_endpoint()}/v1/convert/source/async",
             json=parameters,
+            headers=headers,
             verify=ssl_ctx,
             timeout=60,
         )
@@ -351,6 +374,7 @@ def file_to_base64(file):
 
 
 def process_file(
+    auth,
     files,
     to_formats,
     image_export_mode,
@@ -372,11 +396,13 @@ def process_file(
         logger.error("No files provided.")
         raise gr.Error("No files provided.", print_exception=False)
     files_data = [
-        {"base64_string": file_to_base64(file), "filename": file.name} for file in files
+        {"kind": "file", "base64_string": file_to_base64(file), "filename": file.name}
+        for file in files
     ]
+    target = {"kind": "zip" if return_as_file else "inbody"}
 
     parameters = {
-        "file_sources": files_data,
+        "sources": files_data,
         "options": {
             "to_formats": to_formats,
             "image_export_mode": image_export_mode,
@@ -394,13 +420,19 @@ def process_file(
             "do_picture_classification": do_picture_classification,
             "do_picture_description": do_picture_description,
         },
+        "target": target,
     }
 
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = str(auth)
+
     try:
         ssl_ctx = get_ssl_context()
         response = httpx.post(
-            f"{get_api_endpoint()}/v1alpha/convert/source/async",
+            f"{get_api_endpoint()}/v1/convert/source/async",
             json=parameters,
+            headers=headers,
             verify=ssl_ctx,
             timeout=60,
         )
@@ -474,7 +506,7 @@ with gr.Blocks(
     css=css,
     theme=theme,
     title="Docling Serve",
-    delete_cache=(3600, 3600),  # Delete all files older than 1 hour every hour
+    delete_cache=(3600, 36000),  # Delete all files older than 10 hour every hour
 ) as ui:
     # Constants stored in states to be able to pass them as inputs to functions
     processing_text = gr.State("Processing your document(s), please wait...")
@@ -543,14 +575,17 @@ with gr.Blocks(
     with gr.Tab("Convert File"):
         with gr.Row():
             with gr.Column(scale=4):
+                raw_exts = itertools.chain.from_iterable(FormatToExtensions.values())
                 file_input = gr.File(
                     elem_id="file_input_zone",
                     label="Upload File",
                     file_types=[
-                        f".{v}"
-                        for v in itertools.chain.from_iterable(
-                            FormatToExtensions.values()
-                        )
+                        f".{v.lower()}"
+                        for v in raw_exts  # lowercase
+                    ]
+                    + [
+                        f".{v.upper()}"
+                        for v in raw_exts  # uppercase
                     ],
                     file_count="multiple",
                     scale=4,
@@ -559,6 +594,15 @@ with gr.Blocks(
                 file_process_btn = gr.Button("Process File", scale=1)
                 file_reset_btn = gr.Button("Reset", scale=1)
 
+    # Auth
+    with gr.Row(visible=bool(docling_serve_settings.api_key)):
+        with gr.Column():
+            auth = gr.Textbox(
+                label="Authentication",
+                placeholder="API Key",
+                type="password",
+            )
+
     # Options
     with gr.Accordion("Options") as options:
         with gr.Row():
@@ -584,6 +628,7 @@ with gr.Blocks(
                     label="Image Export Mode",
                     value="embedded",
                 )
+
         with gr.Row():
             with gr.Column(scale=1, min_width=200):
                 pipeline = gr.Radio(
@@ -596,18 +641,25 @@ with gr.Blocks(
                 ocr = gr.Checkbox(label="Enable OCR", value=True)
                 force_ocr = gr.Checkbox(label="Force OCR", value=False)
             with gr.Column(scale=1):
+                engines_list = [
+                    ("Auto", "auto"),
+                    ("EasyOCR", "easyocr"),
+                    ("Tesseract", "tesseract"),
+                    ("RapidOCR", "rapidocr"),
+                ]
+                if sys.platform == "darwin":
+                    engines_list.append(("OCRMac", "ocrmac"))
+
                 ocr_engine = gr.Radio(
-                    [
-                        ("EasyOCR", "easyocr"),
-                        ("Tesseract", "tesseract"),
-                        ("RapidOCR", "rapidocr"),
-                    ],
+                    engines_list,
                     label="OCR Engine",
-                    value="easyocr",
+                    value="auto",
                 )
             with gr.Column(scale=1, min_width=200):
                 ocr_lang = gr.Textbox(
-                    label="OCR Language (beware of the format)", value="en,fr,de,es"
+                    label="OCR Language (beware of the format)",
+                    value="en,fr,de,es",
+                    visible=False,
                 )
             ocr_engine.change(change_ocr_lang, inputs=[ocr_engine], outputs=[ocr_lang])
         with gr.Row():
@@ -718,6 +770,7 @@ with gr.Blocks(
     ).then(
         process_url,
         inputs=[
+            auth,
             url_input,
             to_formats,
             image_export_mode,
@@ -744,7 +797,7 @@ with gr.Blocks(
         outputs=[content_output, file_output],
     ).then(
         wait_task_finish,
-        inputs=[task_id_rendered, return_as_file],
+        inputs=[auth, task_id_rendered, return_as_file],
         outputs=[
             output_markdown,
             output_markdown_rendered,
@@ -805,6 +858,7 @@ with gr.Blocks(
     ).then(
         process_file,
         inputs=[
+            auth,
             file_input,
             to_formats,
             image_export_mode,
@@ -831,7 +885,7 @@ with gr.Blocks(
         outputs=[content_output, file_output],
     ).then(
         wait_task_finish,
-        inputs=[task_id_rendered, return_as_file],
+        inputs=[auth, task_id_rendered, return_as_file],
         outputs=[
             output_markdown,
             output_markdown_rendered,

docling_serve/helper_functions.py

@@ -29,10 +29,15 @@ def is_pydantic_model(type_):
 
 # Adapted from
 # https://github.com/fastapi/fastapi/discussions/8971#discussioncomment-7892972
-def FormDepends(cls: type[BaseModel]):
+def FormDepends(
+    cls: type[BaseModel], prefix: str = "", excluded_fields: list[str] = []
+):
     new_parameters = []
 
     for field_name, model_field in cls.model_fields.items():
+        if field_name in excluded_fields:
+            continue
+
         annotation = model_field.annotation
         description = model_field.description
         default = (
@@ -63,7 +68,7 @@ def FormDepends(cls: type[BaseModel]):
 
         new_parameters.append(
             inspect.Parameter(
-                name=field_name,
+                name=f"{prefix}{field_name}",
                 kind=inspect.Parameter.POSITIONAL_ONLY,
                 default=default,
                 annotation=annotation,
@@ -71,19 +76,23 @@ def FormDepends(cls: type[BaseModel]):
         )
 
     async def as_form_func(**data):
+        newdata = {}
         for field_name, model_field in cls.model_fields.items():
-            value = data.get(field_name)
+            if field_name in excluded_fields:
+                continue
+            value = data.get(f"{prefix}{field_name}")
+            newdata[field_name] = value
             annotation = model_field.annotation
 
             # Parse nested models from JSON string
             if value is not None and is_pydantic_model(annotation):
                 try:
                     validator = TypeAdapter(annotation)
-                    data[field_name] = validator.validate_json(value)
+                    newdata[field_name] = validator.validate_json(value)
                 except Exception as e:
                     raise ValueError(f"Invalid JSON for field '{field_name}': {e}")
 
-        return cls(**data)
+        return cls(**newdata)
 
     sig = inspect.signature(as_form_func)
     sig = sig.replace(parameters=new_parameters)

docling_serve/orchestrator_factory.py

@@ -0,0 +1,336 @@
+import json
+import logging
+from functools import lru_cache
+from typing import Any, Optional
+
+import redis.asyncio as redis
+
+from docling_jobkit.datamodel.task import Task
+from docling_jobkit.datamodel.task_meta import TaskStatus
+from docling_jobkit.orchestrators.base_orchestrator import (
+    BaseOrchestrator,
+    TaskNotFoundError,
+)
+
+from docling_serve.settings import AsyncEngine, docling_serve_settings
+from docling_serve.storage import get_scratch
+
+_log = logging.getLogger(__name__)
+
+
+class RedisTaskStatusMixin:
+    tasks: dict[str, Task]
+    _task_result_keys: dict[str, str]
+    config: Any
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.redis_prefix = "docling:tasks:"
+        self._redis_pool = redis.ConnectionPool.from_url(
+            self.config.redis_url,
+            max_connections=10,
+            socket_timeout=2.0,
+        )
+
+    async def task_status(self, task_id: str, wait: float = 0.0) -> Task:
+        """
+        Get task status by checking Redis first, then falling back to RQ verification.
+
+        When Redis shows 'pending' but RQ shows 'success', we update Redis
+        and return the RQ status for cross-instance consistency.
+        """
+        _log.info(f"Task {task_id} status check")
+
+        # Always check RQ directly first - this is the most reliable source
+        rq_task = await self._get_task_from_rq_direct(task_id)
+        if rq_task:
+            _log.info(f"Task {task_id} in RQ: {rq_task.task_status}")
+
+            # Update memory registry
+            self.tasks[task_id] = rq_task
+
+            # Store/update in Redis for other instances
+            await self._store_task_in_redis(rq_task)
+            return rq_task
+
+        # If not in RQ, check Redis (maybe it's cached from another instance)
+        task = await self._get_task_from_redis(task_id)
+        if task:
+            _log.info(f"Task {task_id} in Redis: {task.task_status}")
+
+            # CRITICAL FIX: Check if Redis status might be stale
+            # STARTED tasks might have completed since they were cached
+            if task.task_status in [TaskStatus.PENDING, TaskStatus.STARTED]:
+                _log.debug(f"Task {task_id} verifying stale status")
+
+                # Try to get fresh status from RQ
+                fresh_rq_task = await self._get_task_from_rq_direct(task_id)
+                if fresh_rq_task and fresh_rq_task.task_status != task.task_status:
+                    _log.info(
+                        f"Task {task_id} status updated: {fresh_rq_task.task_status}"
+                    )
+
+                    # Update memory and Redis with fresh status
+                    self.tasks[task_id] = fresh_rq_task
+                    await self._store_task_in_redis(fresh_rq_task)
+                    return fresh_rq_task
+                else:
+                    _log.debug(f"Task {task_id} status consistent")
+
+            return task
+
+        # Fall back to parent implementation
+        try:
+            parent_task = await super().task_status(task_id, wait)  # type: ignore[misc]
+            _log.debug(f"Task {task_id} from parent: {parent_task.task_status}")
+
+            # Store in Redis for other instances to find
+            await self._store_task_in_redis(parent_task)
+            return parent_task
+        except TaskNotFoundError:
+            _log.warning(f"Task {task_id} not found")
+            raise
+
+    async def _get_task_from_redis(self, task_id: str) -> Optional[Task]:
+        try:
+            async with redis.Redis(connection_pool=self._redis_pool) as r:
+                task_data = await r.get(f"{self.redis_prefix}{task_id}:metadata")
+                if not task_data:
+                    return None
+
+                data: dict[str, Any] = json.loads(task_data)
+                meta = data.get("processing_meta") or {}
+                meta.setdefault("num_docs", 0)
+                meta.setdefault("num_processed", 0)
+                meta.setdefault("num_succeeded", 0)
+                meta.setdefault("num_failed", 0)
+
+                return Task(
+                    task_id=data["task_id"],
+                    task_type=data["task_type"],
+                    task_status=TaskStatus(data["task_status"]),
+                    processing_meta=meta,
+                )
+        except Exception as e:
+            _log.error(f"Redis get task {task_id}: {e}")
+            return None
+
+    async def _get_task_from_rq_direct(self, task_id: str) -> Optional[Task]:
+        try:
+            _log.debug(f"Checking RQ for task {task_id}")
+
+            temp_task = Task(
+                task_id=task_id,
+                task_type="convert",
+                task_status=TaskStatus.PENDING,
+                processing_meta={
+                    "num_docs": 0,
+                    "num_processed": 0,
+                    "num_succeeded": 0,
+                    "num_failed": 0,
+                },
+            )
+
+            original_task = self.tasks.get(task_id)
+            self.tasks[task_id] = temp_task
+
+            try:
+                await super()._update_task_from_rq(task_id)  # type: ignore[misc]
+
+                updated_task = self.tasks.get(task_id)
+                if updated_task and updated_task.task_status != TaskStatus.PENDING:
+                    _log.debug(f"RQ task {task_id}: {updated_task.task_status}")
+
+                    # Store result key if available
+                    if task_id in self._task_result_keys:
+                        try:
+                            async with redis.Redis(
+                                connection_pool=self._redis_pool
+                            ) as r:
+                                await r.set(
+                                    f"{self.redis_prefix}{task_id}:result_key",
+                                    self._task_result_keys[task_id],
+                                    ex=86400,
+                                )
+                                _log.debug(f"Stored result key for {task_id}")
+                        except Exception as e:
+                            _log.error(f"Store result key {task_id}: {e}")
+
+                    return updated_task
+                return None
+
+            finally:
+                # Restore original task state
+                if original_task:
+                    self.tasks[task_id] = original_task
+                elif task_id in self.tasks and self.tasks[task_id] == temp_task:
+                    # Only remove if it's still our temp task
+                    del self.tasks[task_id]
+
+        except Exception as e:
+            _log.error(f"RQ check {task_id}: {e}")
+            return None
+
+    async def get_raw_task(self, task_id: str) -> Task:
+        if task_id in self.tasks:
+            return self.tasks[task_id]
+
+        task = await self._get_task_from_redis(task_id)
+        if task:
+            self.tasks[task_id] = task
+            return task
+
+        try:
+            parent_task = await super().get_raw_task(task_id)  # type: ignore[misc]
+            await self._store_task_in_redis(parent_task)
+            return parent_task
+        except TaskNotFoundError:
+            raise
+
+    async def _store_task_in_redis(self, task: Task) -> None:
+        try:
+            meta: Any = task.processing_meta
+            if hasattr(meta, "model_dump"):
+                meta = meta.model_dump()
+            elif not isinstance(meta, dict):
+                meta = {
+                    "num_docs": 0,
+                    "num_processed": 0,
+                    "num_succeeded": 0,
+                    "num_failed": 0,
+                }
+
+            data: dict[str, Any] = {
+                "task_id": task.task_id,
+                "task_type": task.task_type.value
+                if hasattr(task.task_type, "value")
+                else str(task.task_type),
+                "task_status": task.task_status.value,
+                "processing_meta": meta,
+            }
+            async with redis.Redis(connection_pool=self._redis_pool) as r:
+                await r.set(
+                    f"{self.redis_prefix}{task.task_id}:metadata",
+                    json.dumps(data),
+                    ex=86400,
+                )
+        except Exception as e:
+            _log.error(f"Store task {task.task_id}: {e}")
+
+    async def enqueue(self, **kwargs):  # type: ignore[override]
+        task = await super().enqueue(**kwargs)  # type: ignore[misc]
+        await self._store_task_in_redis(task)
+        return task
+
+    async def task_result(self, task_id: str):  # type: ignore[override]
+        result = await super().task_result(task_id)  # type: ignore[misc]
+        if result is not None:
+            return result
+
+        try:
+            async with redis.Redis(connection_pool=self._redis_pool) as r:
+                result_key = await r.get(f"{self.redis_prefix}{task_id}:result_key")
+                if result_key:
+                    self._task_result_keys[task_id] = result_key.decode("utf-8")
+                    return await super().task_result(task_id)  # type: ignore[misc]
+        except Exception as e:
+            _log.error(f"Redis result key {task_id}: {e}")
+
+        return None
+
+    async def _update_task_from_rq(self, task_id: str) -> None:
+        original_status = (
+            self.tasks[task_id].task_status if task_id in self.tasks else None
+        )
+
+        await super()._update_task_from_rq(task_id)  # type: ignore[misc]
+
+        if task_id in self.tasks:
+            new_status = self.tasks[task_id].task_status
+            if original_status != new_status:
+                _log.debug(f"Task {task_id} status: {original_status} -> {new_status}")
+                await self._store_task_in_redis(self.tasks[task_id])
+
+        if task_id in self._task_result_keys:
+            try:
+                async with redis.Redis(connection_pool=self._redis_pool) as r:
+                    await r.set(
+                        f"{self.redis_prefix}{task_id}:result_key",
+                        self._task_result_keys[task_id],
+                        ex=86400,
+                    )
+            except Exception as e:
+                _log.error(f"Store result key {task_id}: {e}")
+
+
+@lru_cache
+def get_async_orchestrator() -> BaseOrchestrator:
+    if docling_serve_settings.eng_kind == AsyncEngine.LOCAL:
+        from docling_jobkit.convert.manager import (
+            DoclingConverterManager,
+            DoclingConverterManagerConfig,
+        )
+        from docling_jobkit.orchestrators.local.orchestrator import (
+            LocalOrchestrator,
+            LocalOrchestratorConfig,
+        )
+
+        local_config = LocalOrchestratorConfig(
+            num_workers=docling_serve_settings.eng_loc_num_workers,
+            shared_models=docling_serve_settings.eng_loc_share_models,
+            scratch_dir=get_scratch(),
+        )
+
+        cm_config = DoclingConverterManagerConfig(
+            artifacts_path=docling_serve_settings.artifacts_path,
+            options_cache_size=docling_serve_settings.options_cache_size,
+            enable_remote_services=docling_serve_settings.enable_remote_services,
+            allow_external_plugins=docling_serve_settings.allow_external_plugins,
+            max_num_pages=docling_serve_settings.max_num_pages,
+            max_file_size=docling_serve_settings.max_file_size,
+            queue_max_size=docling_serve_settings.queue_max_size,
+            ocr_batch_size=docling_serve_settings.ocr_batch_size,
+            layout_batch_size=docling_serve_settings.layout_batch_size,
+            table_batch_size=docling_serve_settings.table_batch_size,
+            batch_polling_interval_seconds=docling_serve_settings.batch_polling_interval_seconds,
+        )
+        cm = DoclingConverterManager(config=cm_config)
+
+        return LocalOrchestrator(config=local_config, converter_manager=cm)
+
+    elif docling_serve_settings.eng_kind == AsyncEngine.RQ:
+        from docling_jobkit.orchestrators.rq.orchestrator import (
+            RQOrchestrator,
+            RQOrchestratorConfig,
+        )
+
+        class RedisAwareRQOrchestrator(RedisTaskStatusMixin, RQOrchestrator):  # type: ignore[misc]
+            pass
+
+        rq_config = RQOrchestratorConfig(
+            redis_url=docling_serve_settings.eng_rq_redis_url,
+            results_prefix=docling_serve_settings.eng_rq_results_prefix,
+            sub_channel=docling_serve_settings.eng_rq_sub_channel,
+            scratch_dir=get_scratch(),
+        )
+
+        return RedisAwareRQOrchestrator(config=rq_config)
+
+    elif docling_serve_settings.eng_kind == AsyncEngine.KFP:
+        from docling_jobkit.orchestrators.kfp.orchestrator import (
+            KfpOrchestrator,
+            KfpOrchestratorConfig,
+        )
+
+        kfp_config = KfpOrchestratorConfig(
+            endpoint=docling_serve_settings.eng_kfp_endpoint,
+            token=docling_serve_settings.eng_kfp_token,
+            ca_cert_path=docling_serve_settings.eng_kfp_ca_cert_path,
+            self_callback_endpoint=docling_serve_settings.eng_kfp_self_callback_endpoint,
+            self_callback_token_path=docling_serve_settings.eng_kfp_self_callback_token_path,
+            self_callback_ca_cert_path=docling_serve_settings.eng_kfp_self_callback_ca_cert_path,
+        )
+
+        return KfpOrchestrator(config=kfp_config)
+
+    raise RuntimeError(f"Engine {docling_serve_settings.eng_kind} not recognized.")

docling_serve/response_preparation.py

@@ -1,232 +1,82 @@
+import asyncio
 import logging
-import os
-import shutil
-import time
-from collections.abc import Iterable
-from pathlib import Path
-from typing import Union
 
-from fastapi import HTTPException
-from fastapi.responses import FileResponse
+from fastapi import BackgroundTasks, Response
 
-from docling.datamodel.base_models import OutputFormat
-from docling.datamodel.document import ConversionResult, ConversionStatus
-from docling_core.types.doc import ImageRefMode
+from docling_jobkit.datamodel.result import (
+    ChunkedDocumentResult,
+    DoclingTaskResult,
+    ExportResult,
+    RemoteTargetResult,
+    ZipArchiveResult,
+)
+from docling_jobkit.orchestrators.base_orchestrator import (
+    BaseOrchestrator,
+)
 
-from docling_serve.datamodel.convert import ConvertDocumentsOptions
-from docling_serve.datamodel.responses import ConvertDocumentResponse, DocumentResponse
+from docling_serve.datamodel.responses import (
+    ChunkDocumentResponse,
+    ConvertDocumentResponse,
+    PresignedUrlConvertDocumentResponse,
+)
+from docling_serve.settings import docling_serve_settings
 
 _log = logging.getLogger(__name__)
 
 
-def _export_document_as_content(
-    conv_res: ConversionResult,
-    export_json: bool,
-    export_html: bool,
-    export_md: bool,
-    export_txt: bool,
-    export_doctags: bool,
-    image_mode: ImageRefMode,
-    md_page_break_placeholder: str,
+async def prepare_response(
+    task_id: str,
+    task_result: DoclingTaskResult,
+    orchestrator: BaseOrchestrator,
+    background_tasks: BackgroundTasks,
 ):
-    document = DocumentResponse(filename=conv_res.input.file.name)
-
-    if conv_res.status == ConversionStatus.SUCCESS:
-        new_doc = conv_res.document._make_copy_with_refmode(Path(), image_mode)
-
-        # Create the different formats
-        if export_json:
-            document.json_content = new_doc
-        if export_html:
-            document.html_content = new_doc.export_to_html(image_mode=image_mode)
-        if export_txt:
-            document.text_content = new_doc.export_to_markdown(
-                strict_text=True,
-                image_mode=image_mode,
-            )
-        if export_md:
-            document.md_content = new_doc.export_to_markdown(
-                image_mode=image_mode,
-                page_break_placeholder=md_page_break_placeholder or None,
-            )
-        if export_doctags:
-            document.doctags_content = new_doc.export_to_doctags()
-    elif conv_res.status == ConversionStatus.SKIPPED:
-        raise HTTPException(status_code=400, detail=conv_res.errors)
-    else:
-        raise HTTPException(status_code=500, detail=conv_res.errors)
-
-    return document
-
-
-def _export_documents_as_files(
-    conv_results: Iterable[ConversionResult],
-    output_dir: Path,
-    export_json: bool,
-    export_html: bool,
-    export_md: bool,
-    export_txt: bool,
-    export_doctags: bool,
-    image_export_mode: ImageRefMode,
-    md_page_break_placeholder: str,
-):
-    success_count = 0
-    failure_count = 0
-
-    for conv_res in conv_results:
-        if conv_res.status == ConversionStatus.SUCCESS:
-            success_count += 1
-            doc_filename = conv_res.input.file.stem
-
-            # Export JSON format:
-            if export_json:
-                fname = output_dir / f"{doc_filename}.json"
-                _log.info(f"writing JSON output to {fname}")
-                conv_res.document.save_as_json(
-                    filename=fname, image_mode=image_export_mode
-                )
-
-            # Export HTML format:
-            if export_html:
-                fname = output_dir / f"{doc_filename}.html"
-                _log.info(f"writing HTML output to {fname}")
-                conv_res.document.save_as_html(
-                    filename=fname, image_mode=image_export_mode
-                )
-
-            # Export Text format:
-            if export_txt:
-                fname = output_dir / f"{doc_filename}.txt"
-                _log.info(f"writing TXT output to {fname}")
-                conv_res.document.save_as_markdown(
-                    filename=fname,
-                    strict_text=True,
-                    image_mode=ImageRefMode.PLACEHOLDER,
-                )
-
-            # Export Markdown format:
-            if export_md:
-                fname = output_dir / f"{doc_filename}.md"
-                _log.info(f"writing Markdown output to {fname}")
-                conv_res.document.save_as_markdown(
-                    filename=fname,
-                    image_mode=image_export_mode,
-                    page_break_placeholder=md_page_break_placeholder or None,
-                )
-
-            # Export Document Tags format:
-            if export_doctags:
-                fname = output_dir / f"{doc_filename}.doctags"
-                _log.info(f"writing Doc Tags output to {fname}")
-                conv_res.document.save_as_document_tokens(filename=fname)
-
-        else:
-            _log.warning(f"Document {conv_res.input.file} failed to convert.")
-            failure_count += 1
-
-    _log.info(
-        f"Processed {success_count + failure_count} docs, "
-        f"of which {failure_count} failed"
+    response: (
+        Response
+        | ConvertDocumentResponse
+        | PresignedUrlConvertDocumentResponse
+        | ChunkDocumentResponse
     )
-
-
-def process_results(
-    conversion_options: ConvertDocumentsOptions,
-    conv_results: Iterable[ConversionResult],
-    work_dir: Path,
-) -> Union[ConvertDocumentResponse, FileResponse]:
-    # Let's start by processing the documents
-    try:
-        start_time = time.monotonic()
-
-        # Convert the iterator to a list to count the number of results and get timings
-        # As it's an iterator (lazy evaluation), it will also start the conversion
-        conv_results = list(conv_results)
-
-        processing_time = time.monotonic() - start_time
-
-        _log.info(
-            f"Processed {len(conv_results)} docs in {processing_time:.2f} seconds."
-        )
-
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e))
-
-    if len(conv_results) == 0:
-        raise HTTPException(
-            status_code=500, detail="No documents were generated by Docling."
-        )
-
-    # We have some results, let's prepare the response
-    response: Union[FileResponse, ConvertDocumentResponse]
-
-    # Booleans to know what to export
-    export_json = OutputFormat.JSON in conversion_options.to_formats
-    export_html = OutputFormat.HTML in conversion_options.to_formats
-    export_md = OutputFormat.MARKDOWN in conversion_options.to_formats
-    export_txt = OutputFormat.TEXT in conversion_options.to_formats
-    export_doctags = OutputFormat.DOCTAGS in conversion_options.to_formats
-
-    # Only 1 document was processed, and we are not returning it as a file
-    if len(conv_results) == 1 and not conversion_options.return_as_file:
-        conv_res = conv_results[0]
-        document = _export_document_as_content(
-            conv_res,
-            export_json=export_json,
-            export_html=export_html,
-            export_md=export_md,
-            export_txt=export_txt,
-            export_doctags=export_doctags,
-            image_mode=conversion_options.image_export_mode,
-            md_page_break_placeholder=conversion_options.md_page_break_placeholder,
-        )
-
+    if isinstance(task_result.result, ExportResult):
         response = ConvertDocumentResponse(
-            document=document,
-            status=conv_res.status,
-            processing_time=processing_time,
-            timings=conv_res.timings,
+            document=task_result.result.content,
+            status=task_result.result.status,
+            processing_time=task_result.processing_time,
+            timings=task_result.result.timings,
+            errors=task_result.result.errors,
+        )
+    elif isinstance(task_result.result, ZipArchiveResult):
+        response = Response(
+            content=task_result.result.content,
+            media_type="application/zip",
+            headers={
+                "Content-Disposition": 'attachment; filename="converted_docs.zip"'
+            },
+        )
+    elif isinstance(task_result.result, RemoteTargetResult):
+        response = PresignedUrlConvertDocumentResponse(
+            processing_time=task_result.processing_time,
+            num_converted=task_result.num_converted,
+            num_succeeded=task_result.num_succeeded,
+            num_failed=task_result.num_failed,
+        )
+    elif isinstance(task_result.result, ChunkedDocumentResult):
+        response = ChunkDocumentResponse(
+            chunks=task_result.result.chunks,
+            documents=task_result.result.documents,
+            processing_time=task_result.processing_time,
         )
-
-    # Multiple documents were processed, or we are forced returning as a file
     else:
-        # Temporary directory to store the outputs
-        output_dir = work_dir / "output"
-        output_dir.mkdir(parents=True, exist_ok=True)
+        raise ValueError("Unknown result type")
 
-        # Worker pid to use in archive identification as we may have multiple workers
-        os.getpid()
+    if docling_serve_settings.single_use_results:
 
-        # Export the documents
-        _export_documents_as_files(
-            conv_results=conv_results,
-            output_dir=output_dir,
-            export_json=export_json,
-            export_html=export_html,
-            export_md=export_md,
-            export_txt=export_txt,
-            export_doctags=export_doctags,
-            image_export_mode=conversion_options.image_export_mode,
-            md_page_break_placeholder=conversion_options.md_page_break_placeholder,
-        )
+        async def _remove_task_impl():
+            await asyncio.sleep(docling_serve_settings.result_removal_delay)
+            await orchestrator.delete_task(task_id=task_id)
 
-        files = os.listdir(output_dir)
-        if len(files) == 0:
-            raise HTTPException(status_code=500, detail="No documents were exported.")
+        async def _remove_task():
+            asyncio.create_task(_remove_task_impl())  # noqa: RUF006
 
-        file_path = work_dir / "converted_docs.zip"
-        shutil.make_archive(
-            base_name=str(file_path.with_suffix("")),
-            format="zip",
-            root_dir=output_dir,
-        )
-
-        # Other cleanups after the response is sent
-        # Output directory
-        # background_tasks.add_task(shutil.rmtree, work_dir, ignore_errors=True)
-
-        response = FileResponse(
-            file_path, filename=file_path.name, media_type="application/zip"
-        )
+        background_tasks.add_task(_remove_task)
 
     return response

docling_serve/settings.py

@@ -1,3 +1,4 @@
+import enum
 import sys
 from pathlib import Path
 from typing import Optional, Union
@@ -6,8 +7,6 @@ from pydantic import AnyUrl, model_validator
 from pydantic_settings import BaseSettings, SettingsConfigDict
 from typing_extensions import Self
 
-from docling_serve.datamodel.engines import AsyncEngine
-
 
 class UvicornSettings(BaseSettings):
     model_config = SettingsConfigDict(
@@ -26,6 +25,12 @@ class UvicornSettings(BaseSettings):
     workers: Union[int, None] = None
 
 
+class AsyncEngine(str, enum.Enum):
+    LOCAL = "local"
+    KFP = "kfp"
+    RQ = "rq"
+
+
 class DoclingServeSettings(BaseSettings):
     model_config = SettingsConfigDict(
         env_prefix="DOCLING_SERVE_",
@@ -46,10 +51,20 @@ class DoclingServeSettings(BaseSettings):
     enable_remote_services: bool = False
     allow_external_plugins: bool = False
 
+    api_key: str = ""
+
     max_document_timeout: float = 3_600 * 24 * 7  # 7 days
     max_num_pages: int = sys.maxsize
     max_file_size: int = sys.maxsize
 
+    # Threading pipeline
+    queue_max_size: Optional[int] = None
+    ocr_batch_size: Optional[int] = None
+    layout_batch_size: Optional[int] = None
+    table_batch_size: Optional[int] = None
+    batch_polling_interval_seconds: Optional[float] = None
+
+    sync_poll_interval: int = 2  # seconds
     max_sync_wait: int = 120  # 2 minutes
 
     cors_origins: list[str] = ["*"]
@@ -59,6 +74,11 @@ class DoclingServeSettings(BaseSettings):
     eng_kind: AsyncEngine = AsyncEngine.LOCAL
     # Local engine
     eng_loc_num_workers: int = 2
+    eng_loc_share_models: bool = False
+    # RQ engine
+    eng_rq_redis_url: str = ""
+    eng_rq_results_prefix: str = "docling:results"
+    eng_rq_sub_channel: str = "docling:updates"
     # KFP engine
     eng_kfp_endpoint: Optional[AnyUrl] = None
     eng_kfp_token: Optional[str] = None
@@ -82,6 +102,10 @@ class DoclingServeSettings(BaseSettings):
                     "KFP is not yet working. To enable the development version, you must set DOCLING_SERVE_ENG_KFP_EXPERIMENTAL=true."
                 )
 
+        if self.eng_kind == AsyncEngine.RQ:
+            if not self.eng_rq_redis_url:
+                raise ValueError("RQ Redis url is required when using the RQ engine.")
+
         return self
 
 

docling_serve/websocket_notifier.py

@@ -0,0 +1,76 @@
+from fastapi import WebSocket
+
+from docling_jobkit.datamodel.task_meta import TaskStatus
+from docling_jobkit.orchestrators.base_notifier import BaseNotifier
+from docling_jobkit.orchestrators.base_orchestrator import BaseOrchestrator
+
+from docling_serve.datamodel.responses import (
+    MessageKind,
+    TaskStatusResponse,
+    WebsocketMessage,
+)
+
+
+class WebsocketNotifier(BaseNotifier):
+    def __init__(self, orchestrator: BaseOrchestrator):
+        super().__init__(orchestrator)
+        self.task_subscribers: dict[str, set[WebSocket]] = {}
+
+    async def add_task(self, task_id: str):
+        self.task_subscribers[task_id] = set()
+
+    async def remove_task(self, task_id: str):
+        if task_id in self.task_subscribers:
+            for websocket in self.task_subscribers[task_id]:
+                await websocket.close()
+
+            del self.task_subscribers[task_id]
+
+    async def notify_task_subscribers(self, task_id: str):
+        if task_id not in self.task_subscribers:
+            raise RuntimeError(f"Task {task_id} does not have a subscribers list.")
+
+        try:
+            # Get task status from Redis or RQ directly instead of in-memory registry
+            task = await self.orchestrator.task_status(task_id=task_id)
+            task_queue_position = await self.orchestrator.get_queue_position(task_id)
+            msg = TaskStatusResponse(
+                task_id=task.task_id,
+                task_type=task.task_type,
+                task_status=task.task_status,
+                task_position=task_queue_position,
+                task_meta=task.processing_meta,
+            )
+            for websocket in self.task_subscribers[task_id]:
+                await websocket.send_text(
+                    WebsocketMessage(
+                        message=MessageKind.UPDATE, task=msg
+                    ).model_dump_json()
+                )
+                if task.is_completed():
+                    await websocket.close()
+        except Exception as e:
+            # Log the error but don't crash the notifier
+            import logging
+
+            _log = logging.getLogger(__name__)
+            _log.error(f"Error notifying subscribers for task {task_id}: {e}")
+
+    async def notify_queue_positions(self):
+        """Notify all subscribers of pending tasks about queue position updates."""
+        for task_id in self.task_subscribers.keys():
+            try:
+                # Check task status directly from Redis or RQ
+                task = await self.orchestrator.task_status(task_id)
+
+                # Notify only pending tasks
+                if task.task_status == TaskStatus.PENDING:
+                    await self.notify_task_subscribers(task_id)
+            except Exception as e:
+                # Log the error but don't crash the notifier
+                import logging
+
+                _log = logging.getLogger(__name__)
+                _log.error(
+                    f"Error checking task {task_id} status for queue position notification: {e}"
+                )

docs/README.md

@@ -3,6 +3,9 @@
 This documentation pages explore the webserver configurations, runtime options, deployment examples as well as development best practices.
 
 - [Configuration](./configuration.md)
-- [Advance usage](./usage.md)
+- [Handling models](./models.md)
+- [Usage](./usage.md)
 - [Deployment](./deployment.md)
+- [MCP](./mcp.md)
 - [Development](./development.md)
+- [`v1` migration](./v1_migration.md)

docs/configuration.md

@@ -7,7 +7,7 @@ server and the actual app-specific configurations.
 
  > [!WARNING]
 > When the server is running with `reload` or with multiple `workers`, uvicorn
-> will spawn multiple subprocessed. This invalidates all the values configured
+> will spawn multiple subprocesses. This invalidates all the values configured
 > via the CLI command line options. Please use environment variables in this
 > type of deployments.
 
@@ -36,7 +36,7 @@ THe following table describes the options to configure the Docling Serve app.
 | CLI option | ENV | Default | Description |
 | -----------|-----|---------|-------------|
 | `--artifacts-path` | `DOCLING_SERVE_ARTIFACTS_PATH` | unset | If set to a valid directory, the model weights will be loaded from this path |
-|  | `DOCLING_SERVE_STATIC_PATH` | unset | If set to a valid directory, the static assets for the docs and ui will be loaded from this path |
+|  | `DOCLING_SERVE_STATIC_PATH` | unset | If set to a valid directory, the static assets for the docs and UI will be loaded from this path |
 |  | `DOCLING_SERVE_SCRATCH_PATH` |  | If set, this directory will be used as scratch workspace, e.g. storing the results before they get requested. If unset, a temporary created is created for this purpose. |
 | `--enable-ui` | `DOCLING_SERVE_ENABLE_UI` | `false` | Enable the demonstrator UI. |
 |  | `DOCLING_SERVE_ENABLE_REMOTE_SERVICES` | `false` | Allow pipeline components making remote connections. For example, this is needed when using a vision-language model via APIs. |
@@ -46,13 +46,33 @@ THe following table describes the options to configure the Docling Serve app.
 |  | `DOCLING_SERVE_MAX_DOCUMENT_TIMEOUT` | `604800` (7 days) | The maximum time for processing a document. |
 |  | `DOCLING_SERVE_MAX_NUM_PAGES` |  | The maximum number of pages for a document to be processed. |
 |  | `DOCLING_SERVE_MAX_FILE_SIZE` |  | The maximum file size for a document to be processed. |
+|  | `DOCLING_SERVE_SYNC_POLL_INTERVAL` | `2` | Number of seconds to sleep between polling the task status in the sync endpoints. |
 |  | `DOCLING_SERVE_MAX_SYNC_WAIT` | `120` | Max number of seconds a synchronous endpoint is waiting for the task completion. |
 |  | `DOCLING_SERVE_LOAD_MODELS_AT_BOOT` | `True` | If enabled, the models for the default options will be loaded at boot. |
 |  | `DOCLING_SERVE_OPTIONS_CACHE_SIZE` | `2` | How many DocumentConveter objects (including their loaded models) to keep in the cache. |
+|  | `DOCLING_SERVE_QUEUE_MAX_SIZE` | | Size of the pages queue. Potentially so many pages opened at the same time. |
+|  | `DOCLING_SERVE_OCR_BATCH_SIZE` | | Batch size for the OCR stage. |
+|  | `DOCLING_SERVE_LAYOUT_BATCH_SIZE` | | Batch size for the layout detection stage. |
+|  | `DOCLING_SERVE_TABLE_BATCH_SIZE` | | Batch size for the table structure stage. |
+|  | `DOCLING_SERVE_BATCH_POLLING_INTERVAL_SECONDS` | | Wait time for gathering pages before starting a stage processing. |
 |  | `DOCLING_SERVE_CORS_ORIGINS` | `["*"]` | A list of origins that should be permitted to make cross-origin requests. |
 |  | `DOCLING_SERVE_CORS_METHODS` | `["*"]` | A list of HTTP methods that should be allowed for cross-origin requests. |
 |  | `DOCLING_SERVE_CORS_HEADERS` | `["*"]` | A list of HTTP request headers that should be supported for cross-origin requests. |
-|  | `DOCLING_SERVE_ENG_KIND` | `local` | The compute engine to use for the async tasks. Possible values are `local` and `kfp`. See below for more configurations of the engines. |
+|  | `DOCLING_SERVE_API_KEY` | | If specified, all the API requests must contain the header `X-Api-Key` with this value. |
+|  | `DOCLING_SERVE_ENG_KIND` | `local` | The compute engine to use for the async tasks. Possible values are `local`, `rq` and `kfp`. See below for more configurations of the engines. |
+
+### Docling configuration
+
+Some Docling settings, mostly about performance, are exposed as environment variable which can be used also when running Docling Serve.
+
+| ENV | Default | Description |
+| ----|---------|-------------|
+| `DOCLING_NUM_THREADS` | `4` | Number of concurrent threads used for the `torch` CPU execution. |
+| `DOCLING_DEVICE` | | Device used for the model execution. Valid values are `cpu`, `cude`, `mps`. When unset, the best device is chosen. For CUDA-enabled environments, you can choose which GPU using the syntax `cuda:0`, `cuda:1`, ... |
+| `DOCLING_PERF_PAGE_BATCH_SIZE` | `4` | Number of pages processed in the same batch. |
+| `DOCLING_PERF_ELEMENTS_BATCH_SIZE` | `8` | Number of document items/elements processed in the same batch during enrichment. |
+| `DOCLING_DEBUG_PROFILE_PIPELINE_TIMINGS` | `false` | When enabled, Docling will provide detailed timings information. |
+
 
 ### Compute engine
 
@@ -66,6 +86,17 @@ The following table describes the options to configure the Docling Serve local e
 | ENV | Default | Description |
 |-----|---------|-------------|
 | `DOCLING_SERVE_ENG_LOC_NUM_WORKERS` | 2 | Number of workers/threads processing the incoming tasks. |
+| `DOCLING_SERVE_ENG_LOC_SHARE_MODELS` | False | If true, each process will share the same models among all thread workers. Otherwise, one instance of the models is allocated for each worker thread. |
+
+#### RQ engine
+
+The following table describes the options to configure the Docling Serve RQ engine.
+
+| ENV | Default | Description |
+|-----|---------|-------------|
+| `DOCLING_SERVE_ENG_RQ_REDIS_URL` | (required) | The connection Redis url, e.g. `redis://localhost:6373/` |
+| `DOCLING_SERVE_ENG_RQ_RESULTS_PREFIX` | `docling:results` | The prefix used for storing the results in Redis. |
+| `DOCLING_SERVE_ENG_RQ_SUB_CHANNEL` | `docling:updates` | The channel key name used for storing communicating updates between the workers and the orchestrator. |
 
 #### KFP engine
 
@@ -76,6 +107,13 @@ The following table describes the options to configure the Docling Serve KFP eng
 | `DOCLING_SERVE_ENG_KFP_ENDPOINT` |  | Must be set to the Kubeflow Pipeline endpoint. When using the in-cluster deployment, make sure to use the cluster endpoint, e.g. `https://NAME.NAMESPACE.svc.cluster.local:8888`  |
 | `DOCLING_SERVE_ENG_KFP_TOKEN` |  | The authentication token for KFP. For in-cluster deployment, the app will load automatically the token of the ServiceAccount. |
 | `DOCLING_SERVE_ENG_KFP_CA_CERT_PATH` |  | Path to the CA certificates for the KFP endpoint. For in-cluster deployment, the app will load automatically the internal CA. |
-| `DOCLING_SERVE_ENG_KFP_SELF_CALLBACK_ENDPOINT` |  | If set, it enables internal callbacks providing status update of the KFP job. Usually something like `https://NAME.NAMESPACE.svc.cluster.local:5001/v1alpha/callback/task/progress`. |
+| `DOCLING_SERVE_ENG_KFP_SELF_CALLBACK_ENDPOINT` |  | If set, it enables internal callbacks providing status update of the KFP job. Usually something like `https://NAME.NAMESPACE.svc.cluster.local:5001/v1/callback/task/progress`. |
 | `DOCLING_SERVE_ENG_KFP_SELF_CALLBACK_TOKEN_PATH` |  | The token used for authenticating the progress callback. For cluster-internal workloads, use `/run/secrets/kubernetes.io/serviceaccount/token`. |
 | `DOCLING_SERVE_ENG_KFP_SELF_CALLBACK_CA_CERT_PATH` |  | The CA certificate for the progress callback. For cluster-inetrnal workloads, use `/var/run/secrets/kubernetes.io/serviceaccount/service-ca.crt`. |
+
+#### Gradio UI
+
+When using Gradio UI and using the option to output conversion as file, Gradio uses cache to prevent files to be overwritten ([more info here](https://www.gradio.app/guides/file-access#the-gradio-cache)), and we defined the cache clean frequency of one hour to clean files older than 10hours. For situations that files need to be available to download from UI older than 10 hours, there is two options:
+
+- Increase the older age of files to clean [here](https://github.com/docling-project/docling-serve/blob/main/docling_serve/gradio_ui.py#L483) to suffice the age desired;
+- Or set the clean up manually by defining the temporary dir of Gradio to use the same as `DOCLING_SERVE_SCRATCH_PATH` absolute path. This can be achieved by setting the environment variable `GRADIO_TEMP_DIR`, that can be done via command line `export GRADIO_TEMP_DIR="<same_path_as_scratch>"` or in `Dockerfile` using `ENV GRADIO_TEMP_DIR="<same_path_as_scratch>"`. After this, set the clean of cache to `None` [here](https://github.com/docling-project/docling-serve/blob/main/docling_serve/gradio_ui.py#L483). Now, the clean up of `DOCLING_SERVE_SCRATCH_PATH` will also clean the Gradio temporary dir. (If you use this option, please remember when reversing changes to remove the environment variable `GRADIO_TEMP_DIR`, otherwise may lead to files not be available to download).

docs/deploy-examples/compose-amd.yaml

@@ -0,0 +1,21 @@
+# AMD ROCm deployment
+
+services:
+  docling-serve:
+    image: ghcr.io/docling-project/docling-serve-rocm:main
+    container_name: docling-serve
+    ports:
+      - "5001:5001"
+    environment:
+      DOCLING_SERVE_ENABLE_UI: "true"
+      ROCR_VISIBLE_DEVICES: "0" # https://rocm.docs.amd.com/en/latest/conceptual/gpu-isolation.html#rocr-visible-devices
+      ## This section is for compatibility with older cards
+      # HSA_OVERRIDE_GFX_VERSION: "11.0.0"
+      # HSA_ENABLE_SDMA: "0"
+    devices:
+      - /dev/kfd:/dev/kfd
+      - /dev/dri:/dev/dri
+    group_add:
+      - 44    # video group GID from host
+      - 992   # render group GID from host
+    restart: always

docs/deploy-examples/compose-gpu.yaml

@@ -1,15 +0,0 @@
-services:
-  docling:
-    image: ghcr.io/docling-project/docling-serve-cu124
-    container_name: docling-serve
-    ports:
-      - 5001:5001
-    environment:
-      - DOCLING_SERVE_ENABLE_UI=true
-    deploy:
-      resources:
-        reservations:
-          devices:
-          - driver: nvidia
-            count: all # nvidia-smi 
-            capabilities: [gpu]

docs/deploy-examples/compose-nvidia.yaml

@@ -0,0 +1,20 @@
+# NVIDIA CUDA deployment
+
+services:
+  docling-serve:
+    image: ghcr.io/docling-project/docling-serve-cu126:main
+    container_name: docling-serve
+    ports:
+      - "5001:5001"
+    environment:
+      DOCLING_SERVE_ENABLE_UI: "true"
+      NVIDIA_VISIBLE_DEVICES: "all" # https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/docker-specialized.html
+    # deploy:  # This section is for compatibility with Swarm
+    #   resources:
+    #     reservations:
+    #       devices:
+    #         - driver: nvidia
+    #           count: all
+    #           capabilities: [gpu]
+    runtime: nvidia
+    restart: always

docs/deploy-examples/docling-serve-rq-workers.yaml

@@ -0,0 +1,192 @@
+# This example deployment configures Docling Serve with a Service and RQ workers
+
+# Create following secret
+# kubectl create secret generic docling-serve-rq-secrets --from-literal=REDIS_PASSWORD=myredispassword --from-literal=RQ_REDIS_URL=redis://:myredispassword@docling-serve-redis-service:6373/
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: docling-serve
+  labels:
+    app: docling-serve
+    component: docling-serve-api
+spec:
+  ports:
+  - name: http
+    port: 5001
+    targetPort: http
+  selector:
+    app: docling-serve
+    component: docling-serve-api
+---
+kind: Deployment
+apiVersion: apps/v1
+metadata:
+  name: docling-serve
+  labels:
+    app: docling-serve
+    component: docling-serve-api
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: docling-serve
+      component: docling-serve-api
+  template:
+    metadata:
+      labels:
+        app: docling-serve
+        component: docling-serve-api
+    spec:
+      restartPolicy: Always
+      containers:
+        - name: api
+          resources:
+            limits:
+              cpu: 1
+              memory: 8Gi
+            requests:
+              cpu: 250m
+              memory: 1Gi
+          env:
+            - name: DOCLING_SERVE_ENABLE_UI
+              value: 'true'
+            - name: DOCLING_SERVE_ENG_KIND
+              value: 'rq'
+            - name: DOCLING_SERVE_ENG_RQ_REDIS_URL
+              valueFrom:
+                secretKeyRef:
+                  name: docling-serve-rq-secrets
+                  key: RQ_REDIS_URL
+          ports:
+            - name: http
+              containerPort: 5001
+              protocol: TCP
+          imagePullPolicy: Always
+          image: 'ghcr.io/docling-project/docling-serve-cpu'
+---
+kind: Deployment
+apiVersion: apps/v1
+metadata:
+  name: docling-serve-rq-workers
+  labels:
+    app: docling-serve-rq-workers
+    component: docling-serve-rq-worker
+spec:
+  replicas: 2
+  selector:
+    matchLabels:
+      app: docling-serve-rq-workers
+      component: docling-serve-rq-worker
+  template:
+    metadata:
+      labels:
+        app: docling-serve-rq-workers
+        component: docling-serve-rq-worker
+    spec:
+      restartPolicy: Always
+      containers:
+        - name: worker
+          resources:
+            limits:
+              cpu: 1
+              memory: 4Gi
+            requests:
+              cpu: 250m
+              memory: 1Gi
+          env:
+            - name: DOCLING_SERVE_ENG_KIND
+              value: 'rq'
+            - name: DOCLING_SERVE_ENG_RQ_REDIS_URL
+              valueFrom:
+                secretKeyRef:
+                  name: docling-serve-rq-secrets
+                  key: RQ_REDIS_URL
+          ports:
+            - name: http
+              containerPort: 5001
+              protocol: TCP
+          imagePullPolicy: Always
+          image: 'ghcr.io/docling-project/docling-serve-cpu'
+          command: ["docling-serve"]
+          args: ["rq-worker"]
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: docling-serve-redis
+  labels:
+    app: docling-serve-redis
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: docling-serve-redis
+  template:
+    metadata:
+      labels:
+        app: docling-serve-redis
+    spec:
+      restartPolicy: Always
+      terminationGracePeriodSeconds: 30
+      containers:
+        - name: redis
+          resources:
+            limits:
+              cpu: 1
+              memory: 1Gi
+            requests:
+              cpu: 250m
+              memory: 100Mi
+          image: redis:latest
+          command: ["redis-server"]
+          args:
+            - "--port"
+            - "6373"
+            - "--dir"
+            - "/mnt/redis/data"
+            - "--appendonly"
+            - "yes"
+            - "--requirepass"
+            - "$(REDIS_PASSWORD)"
+          ports:
+            - containerPort: 6373
+          env:
+            - name: REDIS_PASSWORD
+              valueFrom:
+                secretKeyRef:
+                  name: docling-serve-rq-secrets
+                  key: REDIS_PASSWORD
+          volumeMounts:
+            - name: redis-data
+              mountPath: /mnt/redis/data
+          securityContext:
+            fsGroup: 1004
+            runAsNonRoot: true
+            allowPrivilegeEscalation: false
+            capabilities:
+              drop:
+                - ALL
+            seccompProfile:
+              type: RuntimeDefault
+      volumes:
+        - name: redis-data
+          emptyDir:
+            medium: Memory
+            sizeLimit: 2Gi
+---
+apiVersion: v1
+kind: Service
+metadata:
+  name: docling-serve-redis-service
+  labels:
+      app: docling-serve-redis
+spec:
+  type: NodePort
+  ports:
+    - name: redis-service
+      protocol: TCP
+      port: 6373
+      targetPort: 6373
+  selector:
+    app: docling-serve-redis

docs/deployment.md

@@ -4,16 +4,17 @@ This document provides deployment examples for running the application in differ
 
 Choose the deployment option that best fits your setup.
 
-- **[Local GPU](#local-gpu)**: For deploying the application locally on a machine with a NVIDIA GPU (using Docker Compose).
+- **[Local GPU NVIDIA](#local-gpu-nvidia)**: For deploying the application locally on a machine with a supported NVIDIA GPU (using Docker Compose).
+- **[Local GPU AMD](#local-gpu-amd)**: For deploying the application locally on a machine with a supported AMD GPU (using Docker Compose).
 - **[OpenShift](#openshift)**: For deploying the application on an OpenShift cluster, designed for cloud-native environments.
 
 ---
 
-## Local GPU
+## Local GPU NVIDIA
 
 ### Docker compose
 
-Manifest example: [compose-gpu.yaml](./deploy-examples/compose-gpu.yaml)
+Manifest example: [compose-nvidia.yaml](./deploy-examples/compose-nvidia.yaml)
 
 This deployment has the following features:
 
@@ -22,7 +23,7 @@ This deployment has the following features:
 Install the app with:
 
 ```sh
-docker compose -f docs/deploy-examples/compose-gpu.yaml up -d
+docker compose -f docs/deploy-examples/compose-nvidia.yaml up -d
 ```
 
 For using the API:
@@ -30,11 +31,11 @@ For using the API:
 ```sh
 # Make a test query
 curl -X 'POST' \
-  "localhost:5001/v1alpha/convert/source/async" \
+  "localhost:5001/v1/convert/source/async" \
   -H "accept: application/json" \
   -H "Content-Type: application/json" \
   -d '{
-    "http_sources": [{"url": "https://arxiv.org/pdf/2501.17887"}]
+    "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}]
   }'
 ```
 
@@ -56,7 +57,7 @@ Docs:
 <details>
 <summary><b>Steps</b></summary>
 
-1. Check driver version and which GPU you want to use (0/1/2/3.. and update [compose-gpu.yaml](./deploy-examples/compose-gpu.yaml) file or use `count: all`)
+1. Check driver version and which GPU you want to use 0/1/2/n (and update [compose-nvidia.yaml](./deploy-examples/compose-nvidia.yaml) file or use `count: all`)
 
     ```sh
     nvidia-smi
@@ -117,7 +118,75 @@ Docs:
 5. Run the container:
 
     ```sh
-    docker compose -f docs/deploy-examples/compose-gpu.yaml up -d
+    docker compose -f docs/deploy-examples/compose-nvidia.yaml up -d
+    ```
+
+</details>
+
+## Local GPU AMD
+
+### Docker compose
+
+Manifest example: [compose-amd.yaml](./deploy-examples/compose-amd.yaml)
+
+This deployment has the following features:
+
+- AMD rocm enabled
+
+Install the app with:
+
+```sh
+docker compose -f docs/deploy-examples/compose-amd.yaml up -d
+```
+
+For using the API:
+
+```sh
+# Make a test query
+curl -X 'POST' \
+  "localhost:5001/v1/convert/source/async" \
+  -H "accept: application/json" \
+  -H "Content-Type: application/json" \
+  -d '{
+    "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}]
+  }'
+```
+
+<details>
+<summary><b>Requirements</b></summary>
+
+- debian/ubuntu/rhel/fedora/opensuse
+- docker
+- AMDGPU driver >=6.3
+- AMD ROCm >=6.3
+
+Docs:
+
+- [AMD ROCm installation](https://rocm.docs.amd.com/projects/install-on-linux/en/latest/install/quick-start.html)
+
+</details>
+
+<details>
+<summary><b>Steps</b></summary>
+
+1. Check driver version and which GPU you want to use 0/1/2/n (and update [compose-amd.yaml](./deploy-examples/compose-amd.yaml) file)
+
+    ```sh
+    rocm-smi --showdriverversion
+    rocminfo | grep -i "ROCm version"
+    ```
+
+2. Find both video group GID and render group GID from host (and update [compose-amd.yaml](./deploy-examples/compose-amd.yaml) file)
+
+    ```sh
+    getent group video
+    getent group render
+    ```
+
+3. Build the image locally (and update [compose-amd.yaml](./deploy-examples/compose-amd.yaml) file)
+
+    ```sh
+    make docling-serve-rocm-image
     ```
 
 </details>
@@ -148,14 +217,39 @@ oc port-forward svc/docling-serve 5001:5001
 
 # Make a test query
 curl -X 'POST' \
-  "localhost:5001/v1alpha/convert/source/async" \
+  "localhost:5001/v1/convert/source/async" \
   -H "accept: application/json" \
   -H "Content-Type: application/json" \
   -d '{
-    "http_sources": [{"url": "https://arxiv.org/pdf/2501.17887"}]
+    "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}]
   }'
 ```
 
+### Multiple workers with RQ
+
+Manifest example: [`docling-serve-rq-workers.yaml`](./deploy-examples/docling-serve-rq-workers.yaml)
+
+This deployment example has the following features:
+
+- Deployment configuration
+- Service configuration
+- Redis deployment
+- Multiple (2 by default) worker Pods
+
+Install the app with:
+
+- create k8s secret:
+
+```sh
+kubectl create secret generic docling-serve-rq-secrets --from-literal=REDIS_PASSWORD=myredispassword --from-literal=RQ_REDIS_URL=redis://:myredispassword@docling-serve-redis-service:6373/
+```
+
+- apply deployment manifest:
+
+```sh
+oc apply -f docs/deploy-examples/docling-serve-rq-workers.yaml
+```
+
 ### Secure deployment with `oauth-proxy`
 
 Manifest example: [docling-serve-oauth.yaml](./deploy-examples/docling-serve-oauth.yaml)
@@ -184,12 +278,12 @@ OCP_AUTH_TOKEN=$(oc whoami --show-token)
 
 # Make a test query
 curl -X 'POST' \
-  "${DOCLING_ROUTE}/v1alpha/convert/source/async" \
+  "${DOCLING_ROUTE}/v1/convert/source/async" \
   -H "Authorization: Bearer ${OCP_AUTH_TOKEN}" \
   -H "accept: application/json" \
   -H "Content-Type: application/json" \
   -d '{
-    "http_sources": [{"url": "https://arxiv.org/pdf/2501.17887"}]
+    "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}]
   }'
 ```
 
@@ -218,11 +312,11 @@ DOCLING_ROUTE="https://$(oc get routes $DOCLING_NAME --template={{.spec.host}})"
 
 # Make a test query, store the cookie and taskid
 task_id=$(curl -s -X 'POST' \
-    "${DOCLING_ROUTE}/v1alpha/convert/source/async" \
+    "${DOCLING_ROUTE}/v1/convert/source/async" \
     -H "accept: application/json" \
     -H "Content-Type: application/json" \
     -d '{
-    "http_sources": [{"url": "https://arxiv.org/pdf/2501.17887"}]
+      "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}]
     }' \
     -c cookies.txt | grep -oP '"task_id":"\K[^"]+')
 ```
@@ -230,7 +324,7 @@ task_id=$(curl -s -X 'POST' \
 ```sh
 # Grab the taskid and cookie to check the task status
 curl -v -X 'GET' \
-  "${DOCLING_ROUTE}/v1alpha/status/poll/$task_id?wait=0" \
+  "${DOCLING_ROUTE}/v1/status/poll/$task_id?wait=0" \
   -H "accept: application/json" \
   -b "cookies.txt"
 ```

docs/examples.md

@@ -0,0 +1,22 @@
+# Examples
+
+## Split processing
+
+The example of provided of split processing demonstrates how to split a PDF into chunks of pages and send them for conversion. At the end, it concatenates all split pages into a single conversion `JSON`.
+
+At beginning of file there's variables to be used (and modified) such as:
+| Variable | Description |
+| ---------|-------------|
+| `path_to_pdf`| Path to PDF file to be split |
+| `pages_per_file`| The number of pages per chunk to split PDF |
+| `base_url`| Base url of the `docling-serve` host |
+| `out_dir`| The output folder of each conversion `JSON` of split PDF and the final concatenated `JSON` |
+
+The example follows the following logic:
+- Get the number of pages of the `PDF`
+- Based on the number of chunks of pages, send each chunk to conversion using `page_range` parameter
+- Wait all conversions to finish
+- Get all conversion results
+- Save each conversion `JSON` result into a `JSON` file
+- Concatenate all `JSONs` into a single `JSON` using `docling` concatenate method
+- Save concatenated `JSON` into a `JSON` file

docs/mcp.md

@@ -0,0 +1,39 @@
+# Docling MCP in Docling Serve
+
+The `docling-serve` container image includes all MCP (Model Communication Protocol) features starting from version v1.1.0. To leverage these features, you simply need to use a different entrypoint—no custom image builds or additional installations are required. The image provides the `docling-mcp-server` executable, which enables MCP functionality out of the box as of version v1.1.0 ([changelog](https://github.com/docling-project/docling-serve/blob/624f65d41b734e8b39ff267bc8bf6e766c376d6d/CHANGELOG.md)).
+
+Read more on [Docling MCP](https://github.com/docling-project/docling-mcp) in its dedicated repository.
+
+## Launching the MCP Service
+
+By default, the container runs `docling-serve run` and exposes port 5001. To start the MCP service, override the entrypoint and specify your desired port mapping. For example:
+
+```sh
+podman run -p 8000:8000 quay.io/docling-project/docling-serve -- docling-mcp-server --transport streamable-http --port 8000 --host 0.0.0.0
+```
+
+This command starts the MCP server on port 8000, accessible at `http://localhost:8000/mcp`. Adjust the port and host as needed. Key arguments for `docling-mcp-server` include `--transport streamable-http` (HTTP transport for client connections), `--port <PORT>`, and `--host <HOST>` (use `0.0.0.0` to accept connections from any interface).
+
+## Configuring MCP Clients
+
+Most MCP-compatible clients, such as LM Studio and Claude Desktop, allow you to specify custom MCP server endpoints. The standard configuration uses a JSON block to define available MCP servers. For example, to connect to the Docling MCP server running on port 8000:
+
+```json
+{
+  "mcpServers": {
+    "docling": {
+      "url": "http://localhost:8000/mcp"
+    }
+  }
+}
+```
+
+Insert this configuration in your client's settings where MCP servers are defined. Update the URL if you use a different port.
+
+### LM Studio and Claude Desktop
+
+Both LM Studio and Claude Desktop support MCP endpoints via configuration files or UI settings. Paste the above JSON block into the appropriate configuration section. For Claude Desktop, add the MCP server in the "Custom Model" or "MCP Server" section. For LM Studio, refer to its documentation for the location of the MCP server configuration.
+
+### Other MCP Clients
+
+Other clients, such as Continue Coding Assistant, also support custom MCP endpoints. Use the same configuration pattern: provide the MCP server URL ending with `/mcp` and ensure the port matches your container setup. See the [Docling MCP docs](https://github.com/docling-project/docling-mcp/tree/main/docs/integrations) for more details.

docs/models.md

@@ -0,0 +1,175 @@
+# Handling Models in Docling Serve
+
+When enabling steps in Docling Serve that require extra models (such as picture classification, picture description, table detection, code recognition, formula extraction, or vision-language modules), you must ensure those models are available in the runtime environment. The standard container image includes only the default models. Any additional models must be downloaded and made available before use. If required models are missing, Docling Serve will raise runtime errors rather than downloading them automatically. This default choice wants to guarantee the system is not calling external services.
+
+## Model Storage Location
+
+Docling Serve loads models from the directory specified by the `DOCLING_SERVE_ARTIFACTS_PATH` environment variable. This path must be consistent across model download and runtime. When running with multiple workers or reload enabled, you must use the environment variable rather than the CLI argument for configuration [[source]](./configuration.md).
+
+## Approaches for Making Extra Models Available
+
+There are several ways to ensure required models are present:
+
+### 1. Disable Local Models (Trigger Auto-Download)
+
+You can configure the container to download all models at startup by clearing the artifacts path:
+
+```sh
+podman run -d -p 5001:5001 --name docling-serve \
+  -e DOCLING_SERVE_ARTIFACTS_PATH="" \
+  -e DOCLING_SERVE_ENABLE_UI=true \
+  quay.io/docling-project/docling-serve
+```
+
+This approach is simple for local development but not recommended for production, as it increases startup time and depends on network availability.
+
+### 2. Build a Custom Image with Pre-Downloaded Models
+
+You can create a new image that includes the required models:
+
+```Dockerfile
+FROM quay.io/docling-project/docling-serve
+RUN docling-tools models download smolvlm
+```
+
+This method is suitable for production, as it ensures all models are present in the image and avoids runtime downloads.
+
+### 3. Update the Entrypoint to Download Models Before Startup
+
+You can override the entrypoint to download models before starting the service:
+
+```sh
+podman run -p 5001:5001 -e DOCLING_SERVE_ENABLE_UI=true \
+  quay.io/docling-project/docling-serve \
+  -- sh -c 'exec docling-tools models download smolvlm && exec docling-serve run'
+```
+
+This is useful for environments where you want to keep the base image unchanged but still automate model preparation.
+
+### 4. Mount a Volume with Pre-Downloaded Models
+
+Download models locally and mount them into the container:
+
+```sh
+# Download the models locally
+docling-tools models download --all -o models
+
+# Start the container with the local models folder
+podman run -p 5001:5001 \
+  -v $(pwd)/models:/opt/app-root/src/models \
+  -e DOCLING_SERVE_ARTIFACTS_PATH="/opt/app-root/src/models" \
+  -e DOCLING_SERVE_ENABLE_UI=true \
+  quay.io/docling-project/docling-serve
+```
+
+This approach is robust for both local and production deployments, especially when using persistent storage.
+
+## Kubernetes/Cluster Deployments
+
+For Kubernetes or OpenShift clusters, the recommended approach is to use a PersistentVolumeClaim (PVC) for model storage, a Kubernetes Job to download models, and mount the volume into the deployment. This ensures models persist across pod restarts and scale-out scenarios.
+
+### Example: PersistentVolumeClaim
+
+```yaml
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: docling-model-cache-pvc
+spec:
+  accessModes:
+    - ReadWriteOnce
+  volumeMode: Filesystem
+  resources:
+    requests:
+      storage: 10Gi
+```
+
+If you don't want to use default storage class, set your custom storage class with following:
+
+```yaml
+spec:
+    ...
+    storageClassName: <Storage Class Name>
+```
+
+Manifest example: [docling-model-cache-pvc.yaml](./deploy-examples/docling-model-cache-pvc.yaml)
+
+### Example: Model Download Job
+
+```yaml
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: docling-model-cache-load
+spec:
+  template:
+    spec:
+      containers:
+        - name: loader
+          image: ghcr.io/docling-project/docling-serve-cpu:main
+          command:
+            - docling-tools
+            - models
+            - download
+            - '--output-dir=/modelcache'
+            - 'layout'
+            - 'tableformer'
+            - 'code_formula'
+            - 'picture_classifier'
+            - 'smolvlm'
+            - 'granite_vision'
+            - 'easyocr'
+          volumeMounts:
+            - name: docling-model-cache
+              mountPath: /modelcache
+      volumes:
+        - name: docling-model-cache
+          persistentVolumeClaim:
+            claimName: docling-model-cache-pvc
+      restartPolicy: Never
+```
+
+The job will mount the previously created persistent volume and execute command similar to how we would load models locally:
+`docling-tools models download --output-dir <MOUNT-PATH> [LIST_OF_MODELS]`
+
+In manifest, we specify desired models individually, or we can use `--all` parameter to download all models.
+
+Manifest example: [docling-model-cache-job.yaml](./deploy-examples/docling-model-cache-job.yaml)
+
+### Example: Deployment with Mounted Volume
+
+```yaml
+spec:
+  template:
+    spec:
+      containers:
+        - name: api
+          env:
+            - name: DOCLING_SERVE_ARTIFACTS_PATH
+              value: '/modelcache'
+          volumeMounts:
+            - name: docling-model-cache
+              mountPath: /modelcache
+      volumes:
+        - name: docling-model-cache
+          persistentVolumeClaim:
+            claimName: docling-model-cache-pvc
+```
+
+The value of `DOCLING_SERVE_ARTIFACTS_PATH` must match the mount path where models are stored.
+
+Now, when docling-serve is executing tasks, the underlying docling installation will load model weights from mounted volume.
+
+Manifest example: [docling-model-cache-deployment.yaml](./deploy-examples/docling-model-cache-deployment.yaml)
+
+## Local Docker Execution
+
+For local Docker or Podman execution, you can use any of the approaches above. Mounting a local directory with pre-downloaded models is the most reliable for repeated runs and avoids network dependencies.
+
+## Troubleshooting and Best Practices
+
+- If a required model is missing from the artifacts path, Docling Serve will raise a runtime error.
+- Always ensure the value of `DOCLING_SERVE_ARTIFACTS_PATH` matches the directory where models are stored and mounted.
+- For production and cluster environments, prefer persistent storage and pre-loading models via a dedicated job.
+
+For more details and YAML manifest examples, see the [deployment documentation](./deployment.md).

docs/pre-loading-models.md

@@ -1,103 +0,0 @@
-# Pre-loading models for docling
-
-This document provides examples for pre-loading docling models to a persistent volume and re-using it for docling-serve deployments.
-
-1. We need to create a persistent volume that will store models weights:
-
-    ```yaml
-    apiVersion: v1
-    kind: PersistentVolumeClaim
-    metadata:
-      name: docling-model-cache-pvc
-    spec:
-      accessModes:
-        - ReadWriteOnce
-      volumeMode: Filesystem
-      resources:
-        requests:
-          storage: 10Gi
-    ```
-
-    If you don't want to use default storage class, set your custom storage class with following:
-
-    ```yaml
-    spec:
-      ...
-      storageClassName: <Storage Class Name>
-    ```
-
-    Manifest example: [docling-model-cache-pvc.yaml](./deploy-examples/docling-model-cache-pvc.yaml)
-
-2. In order to load model weights, we can use docling-toolkit to download them, as this is a one time operation we can use kubernetes job for this:
-
-    ```yaml
-    apiVersion: batch/v1
-    kind: Job
-    metadata:
-      name: docling-model-cache-load
-    spec:
-      selector: {}
-      template:
-        metadata:
-          name: docling-model-load
-        spec:
-          containers:
-            - name: loader
-              image: ghcr.io/docling-project/docling-serve-cpu:main
-              command:
-                - docling-tools
-                - models
-                - download
-                - '--output-dir=/modelcache'
-                - 'layout'
-                - 'tableformer'
-                - 'code_formula'
-                - 'picture_classifier'
-                - 'smolvlm'
-                - 'granite_vision'
-                - 'easyocr'
-              volumeMounts:
-                - name: docling-model-cache
-                  mountPath: /modelcache
-          volumes:
-            - name: docling-model-cache
-              persistentVolumeClaim:
-                claimName: docling-model-cache-pvc
-          restartPolicy: Never
-    ```
-
-    The job will mount previously created persistent volume and execute command similar to how we would load models locally:
-    `docling-tools models download --output-dir <MOUNT-PATH> [LIST_OF_MODELS]`
-
-    In manifest, we specify desired models individually, or we can use `--all` parameter to download all models.
-
-    Manifest example: [docling-model-cache-job.yaml](./deploy-examples/docling-model-cache-job.yaml)
-
-3. Now we can mount volume in the docling-serve deployment and set env `DOCLING_SERVE_ARTIFACTS_PATH` to point to it.
-    Following additions to deploymeny should be made:
-
-    ```yaml
-    spec:
-      template:
-        spec:
-          containers:
-            - name: api
-              env:
-              ...
-                - name: DOCLING_SERVE_ARTIFACTS_PATH
-                  value: '/modelcache'
-              volumeMounts:
-                - name: docling-model-cache
-                  mountPath: /modelcache
-          ...
-          volumes:
-            - name: docling-model-cache
-              persistentVolumeClaim:
-                claimName: docling-model-cache-pvc
-    ```
-
-    Make sure that value of `DOCLING_SERVE_ARTIFACTS_PATH` is the same as where models were downloaded and where volume is mounted.
-
-    Now when docling-serve is executing tasks, the underlying docling installation will load model weights from mouted volume.
-
-    Manifest example: [docling-model-cache-deployment.yaml](./deploy-examples/docling-model-cache-deployment.yaml)

docs/usage.md

@@ -4,38 +4,99 @@ The API provides two endpoints: one for urls, one for files. This is necessary t
 
 ## Common parameters
 
-On top of the source of file (see below), both endpoints support the same parameters, which are almost the same as the Docling CLI.
+On top of the source of file (see below), both endpoints support the same parameters.
 
-- `from_formats` (List[str]): Input format(s) to convert from. Allowed values: `docx`, `pptx`, `html`, `image`, `pdf`, `asciidoc`, `md`. Defaults to all formats.
-- `to_formats` (List[str]): Output format(s) to convert to. Allowed values: `md`, `json`, `html`, `text`, `doctags`. Defaults to `md`.
-- `pipeline` (str). The choice of which pipeline to use. Allowed values are `standard` and `vlm`. Defaults to `standard`.
-- `page_range` (tuple). If speficied, only convert a range of pages. The page number starts at 1.
-- `do_ocr` (bool): If enabled, the bitmap content will be processed using OCR. Defaults to `True`.
-- `image_export_mode`: Image export mode for the document (only in case of JSON, Markdown or HTML). Allowed values: embedded, placeholder, referenced. Optional, defaults to `embedded`.
-- `force_ocr` (bool): If enabled, replace any existing text with OCR-generated text over the full content. Defaults to `False`.
-- `ocr_engine` (str): OCR engine to use. Allowed values: `easyocr`, `tesserocr`, `tesseract`, `rapidocr`, `ocrmac`. Defaults to `easyocr`. To use the `tesserocr` engine, `tesserocr` must be installed where docling-serve is running: `pip install tesserocr`
-- `ocr_lang` (List[str]): List of languages used by the OCR engine. Note that each OCR engine has different values for the language names. Defaults to empty.
-- `pdf_backend` (str): PDF backend to use. Allowed values: `pypdfium2`, `dlparse_v1`, `dlparse_v2`, `dlparse_v4`. Defaults to `dlparse_v4`.
-- `table_mode` (str): Table mode to use. Allowed values: `fast`, `accurate`. Defaults to `fast`.
-- `abort_on_error` (bool): If enabled, abort on error. Defaults to false.
-- `return_as_file` (boo): If enabled, return the output as a file. Defaults to false.
-- `md_page_break_placeholder` (str): Add this placeholder between pages in the markdown output.
-- `do_table_structure` (bool): If enabled, the table structure will be extracted. Defaults to true.
-- `do_code_enrichment` (bool): If enabled, perform OCR code enrichment. Defaults to false.
-- `do_formula_enrichment` (bool): If enabled, perform formula OCR, return LaTeX code. Defaults to false.
-- `do_picture_classification` (bool): If enabled, classify pictures in documents. Defaults to false.
-- `do_picture_description` (bool): If enabled, describe pictures in documents. Defaults to false.
-- `picture_description_area_threshold` (float): Minimum percentage of the area for a picture to be processed with the models. Defaults to 0.05.
-- `picture_description_local` (dict): Options for running a local vision-language model in the picture description. The parameters refer to a model hosted on Hugging Face. This parameter is mutually exclusive with picture_description_api.
-- `picture_description_api` (dict): API details for using a vision-language model in the picture description. This parameter is mutually exclusive with picture_description_local.
-- `include_images` (bool): If enabled, images will be extracted from the document. Defaults to false.
-- `images_scale` (float): Scale factor for images. Defaults to 2.0.
+<!-- begin: parameters-docs -->
+<h4>ConvertDocumentsRequestOptions</h4>
+
+| Field Name | Type | Description |
+|------------|------|-------------|
+| `from_formats` | List[InputFormat] | Input format(s) to convert from. String or list of strings. Allowed values: `docx`, `pptx`, `html`, `image`, `pdf`, `asciidoc`, `md`, `csv`, `xlsx`, `xml_uspto`, `xml_jats`, `mets_gbs`, `json_docling`, `audio`, `vtt`. Optional, defaults to all formats. |
+| `to_formats` | List[OutputFormat] | Output format(s) to convert to. String or list of strings. Allowed values: `md`, `json`, `html`, `html_split_page`, `text`, `doctags`. Optional, defaults to Markdown. |
+| `image_export_mode` | ImageRefMode | Image export mode for the document (in case of JSON, Markdown or HTML). Allowed values: `placeholder`, `embedded`, `referenced`. Optional, defaults to Embedded. |
+| `do_ocr` | bool | If enabled, the bitmap content will be processed using OCR. Boolean. Optional, defaults to true |
+| `force_ocr` | bool | If enabled, replace existing text with OCR-generated text over content. Boolean. Optional, defaults to false. |
+| `ocr_engine` | `ocr_engines_enum` | The OCR engine to use. String. Allowed values: `auto`, `easyocr`, `ocrmac`, `rapidocr`, `tesserocr`, `tesseract`. Optional, defaults to `easyocr`. |
+| `ocr_lang` | List[str] or NoneType | List of languages used by the OCR engine. Note that each OCR engine has different values for the language names. String or list of strings. Optional, defaults to empty. |
+| `pdf_backend` | PdfBackend | The PDF backend to use. String. Allowed values: `pypdfium2`, `dlparse_v1`, `dlparse_v2`, `dlparse_v4`. Optional, defaults to `dlparse_v4`. |
+| `table_mode` | TableFormerMode | Mode to use for table structure, String. Allowed values: `fast`, `accurate`. Optional, defaults to accurate. |
+| `table_cell_matching` | bool | If true, matches table cells predictions back to PDF cells. Can break table output if PDF cells are merged across table columns. If false, let table structure model define the text cells, ignore PDF cells. |
+| `pipeline` | ProcessingPipeline | Choose the pipeline to process PDF or image files. |
+| `page_range` | Tuple | Only convert a range of pages. The page number starts at 1. |
+| `document_timeout` | float | The timeout for processing each document, in seconds. |
+| `abort_on_error` | bool | Abort on error if enabled. Boolean. Optional, defaults to false. |
+| `do_table_structure` | bool | If enabled, the table structure will be extracted. Boolean. Optional, defaults to true. |
+| `include_images` | bool | If enabled, images will be extracted from the document. Boolean. Optional, defaults to true. |
+| `images_scale` | float | Scale factor for images. Float. Optional, defaults to 2.0. |
+| `md_page_break_placeholder` | str | Add this placeholder between pages in the markdown output. |
+| `do_code_enrichment` | bool | If enabled, perform OCR code enrichment. Boolean. Optional, defaults to false. |
+| `do_formula_enrichment` | bool | If enabled, perform formula OCR, return LaTeX code. Boolean. Optional, defaults to false. |
+| `do_picture_classification` | bool | If enabled, classify pictures in documents. Boolean. Optional, defaults to false. |
+| `do_picture_description` | bool | If enabled, describe pictures in documents. Boolean. Optional, defaults to false. |
+| `picture_description_area_threshold` | float | Minimum percentage of the area for a picture to be processed with the models. |
+| `picture_description_local` | PictureDescriptionLocal or NoneType | Options for running a local vision-language model in the picture description. The parameters refer to a model hosted on Hugging Face. This parameter is mutually exclusive with `picture_description_api`. |
+| `picture_description_api` | PictureDescriptionApi or NoneType | API details for using a vision-language model in the picture description. This parameter is mutually exclusive with `picture_description_local`. |
+| `vlm_pipeline_model` | VlmModelType or NoneType | Preset of local and API models for the `vlm` pipeline. This parameter is mutually exclusive with `vlm_pipeline_model_local` and `vlm_pipeline_model_api`. Use the other options for more parameters. |
+| `vlm_pipeline_model_local` | VlmModelLocal or NoneType | Options for running a local vision-language model for the `vlm` pipeline. The parameters refer to a model hosted on Hugging Face. This parameter is mutually exclusive with `vlm_pipeline_model_api` and `vlm_pipeline_model`. |
+| `vlm_pipeline_model_api` | VlmModelApi or NoneType | API details for using a vision-language model for the `vlm` pipeline. This parameter is mutually exclusive with `vlm_pipeline_model_local` and `vlm_pipeline_model`. |
+
+<h4>VlmModelApi</h4>
+
+| Field Name | Type | Description |
+|------------|------|-------------|
+| `url` | AnyUrl | Endpoint which accepts openai-api compatible requests. |
+| `headers` | Dict[str, str] | Headers used for calling the API endpoint. For example, it could include authentication headers. |
+| `params` | Dict[str, Any] | Model parameters. |
+| `timeout` | float | Timeout for the API request. |
+| `concurrency` | int | Maximum number of concurrent requests to the API. |
+| `prompt` | str | Prompt used when calling the vision-language model. |
+| `scale` | float | Scale factor of the images used. |
+| `response_format` | ResponseFormat | Type of response generated by the model. |
+| `temperature` | float | Temperature parameter controlling the reproducibility of the result. |
+
+<h4>VlmModelLocal</h4>
+
+| Field Name | Type | Description |
+|------------|------|-------------|
+| `repo_id` | str | Repository id from the Hugging Face Hub. |
+| `prompt` | str | Prompt used when calling the vision-language model. |
+| `scale` | float | Scale factor of the images used. |
+| `response_format` | ResponseFormat | Type of response generated by the model. |
+| `inference_framework` | InferenceFramework | Inference framework to use. |
+| `transformers_model_type` | TransformersModelType | Type of transformers auto-model to use. |
+| `extra_generation_config` | Dict[str, Any] | Config from https://huggingface.co/docs/transformers/en/main_classes/text_generation#transformers.GenerationConfig |
+| `temperature` | float | Temperature parameter controlling the reproducibility of the result. |
+
+<h4>PictureDescriptionApi</h4>
+
+| Field Name | Type | Description |
+|------------|------|-------------|
+| `url` | AnyUrl | Endpoint which accepts openai-api compatible requests. |
+| `headers` | Dict[str, str] | Headers used for calling the API endpoint. For example, it could include authentication headers. |
+| `params` | Dict[str, Any] | Model parameters. |
+| `timeout` | float | Timeout for the API request. |
+| `concurrency` | int | Maximum number of concurrent requests to the API. |
+| `prompt` | str | Prompt used when calling the vision-language model. |
+
+<h4>PictureDescriptionLocal</h4>
+
+| Field Name | Type | Description |
+|------------|------|-------------|
+| `repo_id` | str | Repository id from the Hugging Face Hub. |
+| `prompt` | str | Prompt used when calling the vision-language model. |
+| `generation_config` | Dict[str, Any] | Config from https://huggingface.co/docs/transformers/en/main_classes/text_generation#transformers.GenerationConfig |
+
+<!-- end: parameters-docs -->
+
+### Authentication
+
+When authentication is activated (see the parameter `DOCLING_SERVE_API_KEY` in [configuration.md](./configuration.md)), all the API requests **must** provide the header `X-Api-Key` with the correct secret key.
 
 ## Convert endpoints
 
 ### Source endpoint
 
-The endpoint is `/v1alpha/convert/source`, listening for POST requests of JSON payloads.
+The endpoint is `/v1/convert/source`, listening for POST requests of JSON payloads.
 
 On top of the above parameters, you must send the URL(s) of the document you want process with either the `http_sources` or `file_sources` fields.
 The first is fetching URL(s) (optionally using with extra headers), the second allows to provide documents as base64-encoded strings.
@@ -66,7 +127,6 @@ Simple payload example:
     "pdf_backend": "dlparse_v2",
     "table_mode": "fast",
     "abort_on_error": false,
-    "return_as_file": false,
   },
   "http_sources": [{"url": "https://arxiv.org/pdf/2206.01062"}]
 }
@@ -80,7 +140,7 @@ Simple payload example:
 
 ```sh
 curl -X 'POST' \
-  'http://localhost:5001/v1alpha/convert/source' \
+  'http://localhost:5001/v1/convert/source' \
   -H 'accept: application/json' \
   -H 'Content-Type: application/json' \
   -d '{
@@ -109,7 +169,6 @@ curl -X 'POST' \
     "pdf_backend": "dlparse_v2",
     "table_mode": "fast",
     "abort_on_error": false,
-    "return_as_file": false,
     "do_table_structure": true,
     "include_images": true,
     "images_scale": 2
@@ -127,7 +186,7 @@ curl -X 'POST' \
 import httpx
 
 async_client = httpx.AsyncClient(timeout=60.0)
-url = "http://localhost:5001/v1alpha/convert/source"
+url = "http://localhost:5001/v1/convert/source"
 payload = {
   "options": {
     "from_formats": ["docx", "pptx", "html", "image", "pdf", "asciidoc", "md", "xlsx"],
@@ -140,7 +199,6 @@ payload = {
     "pdf_backend": "dlparse_v2",
     "table_mode": "fast",
     "abort_on_error": False,
-    "return_as_file": False,
   },
   "http_sources": [{"url": "https://arxiv.org/pdf/2206.01062"}]
 }
@@ -179,7 +237,7 @@ cat <<EOF > /tmp/request_body.json
 EOF
 
 # 3. POST the request to the docling service
-curl -X POST "localhost:5001/v1alpha/convert/source" \
+curl -X POST "localhost:5001/v1/convert/source" \
      -H "Content-Type: application/json" \
      -d @/tmp/request_body.json
 ```
@@ -188,14 +246,14 @@ curl -X POST "localhost:5001/v1alpha/convert/source" \
 
 ### File endpoint
 
-The endpoint is: `/v1alpha/convert/file`, listening for POST requests of Form payloads (necessary as the files are sent as multipart/form data). You can send one or multiple files.
+The endpoint is: `/v1/convert/file`, listening for POST requests of Form payloads (necessary as the files are sent as multipart/form data). You can send one or multiple files.
 
 <details>
 <summary>CURL example:</summary>
 
 ```sh
 curl -X 'POST' \
-  'http://127.0.0.1:5001/v1alpha/convert/file' \
+  'http://127.0.0.1:5001/v1/convert/file' \
   -H 'accept: application/json' \
   -H 'Content-Type: multipart/form-data' \
   -F 'ocr_engine=easyocr' \
@@ -211,7 +269,6 @@ curl -X 'POST' \
   -F 'abort_on_error=false' \
   -F 'to_formats=md' \
   -F 'to_formats=text' \
-  -F 'return_as_file=false' \
   -F 'do_ocr=true'
 ```
 
@@ -224,7 +281,7 @@ curl -X 'POST' \
 import httpx
 
 async_client = httpx.AsyncClient(timeout=60.0)
-url = "http://localhost:5001/v1alpha/convert/file"
+url = "http://localhost:5001/v1/convert/file"
 parameters = {
 "from_formats": ["docx", "pptx", "html", "image", "pdf", "asciidoc", "md", "xlsx"],
 "to_formats": ["md", "json", "html", "text", "doctags"],
@@ -236,7 +293,6 @@ parameters = {
 "pdf_backend": "dlparse_v2",
 "table_mode": "fast",
 "abort_on_error": False,
-"return_as_file": False
 }
 
 current_dir = os.path.dirname(__file__)
@@ -313,7 +369,7 @@ Example URLs are:
     }
     ```
 
-- `http://localhost:11434/v1/chat/completions` for the local ollama api, with example `picture_description_api`:
+- `http://localhost:11434/v1/chat/completions` for the local Ollama api, with example `picture_description_api`:
   - the `granite3.2-vision:2b` model
 
     ```json
@@ -354,19 +410,19 @@ The response can be a JSON Document or a File.
   `processing_time` is the Docling processing time in seconds, and `timings` (when enabled in the backend) provides the detailed
   timing of all the internal Docling components.
 
-- If you set the parameter `return_as_file` to True, the response will be a zip file.
-- If multiple files are generated (multiple inputs, or one input but multiple outputs with `return_as_file` True), the response will be a zip file.
+- If you set the parameter `target` to the zip mode, the response will be a zip file.
+- If multiple files are generated (multiple inputs, or one input but multiple outputs with the zip target mode), the response will be a zip file.
 
 ## Asynchronous API
 
-Both `/v1alpha/convert/source` and `/v1alpha/convert/file` endpoints are available as asynchronous variants.
+Both `/v1/convert/source` and `/v1/convert/file` endpoints are available as asynchronous variants.
 The advantage of the asynchronous endpoints is the possible to interrupt the connection, check for the progress update and fetch the result.
-This approach is more resilient against network stabilities and allows the client application logic to easily interleave conversion with other tasks.
+This approach is more resilient against network instabilities and allows the client application logic to easily interleave conversion with other tasks.
 
 Launch an asynchronous conversion with:
 
-- `POST /v1alpha/convert/source/async` when providing the input as sources.
-- `POST /v1alpha/convert/file/async` when providing the input as multipart-form files.
+- `POST /v1/convert/source/async` when providing the input as sources.
+- `POST /v1/convert/file/async` when providing the input as multipart-form files.
 
 The response format is a task detail:
 
@@ -383,7 +439,7 @@ The response format is a task detail:
 
 For checking the progress of the conversion task and wait for its completion, use the endpoint:
 
-- `GET /v1alpha/status/poll/{task_id}`
+- `GET /v1/status/poll/{task_id}`
 
 <details>
 <summary>Example waiting loop:</summary>
@@ -408,9 +464,9 @@ while task["task_status"] not in ("success", "failure"):
 ### Subscribe with websockets
 
 Using websocket you can get the client application being notified about updates of the conversion task.
-To start the websocker connection, use the endpoint:
+To start the websocket connection, use the endpoint:
 
-- `/v1alpha/status/ws/{task_id}`
+- `/v1/status/ws/{task_id}`
 
 Websocket messages are JSON object with the following structure:
 
@@ -423,19 +479,19 @@ Websocket messages are JSON object with the following structure:
 ```
 
 <details>
-<summary>Example websocker usage:</summary>
+<summary>Example websocket usage:</summary>
 
 ```python
 from websockets.sync.client import connect
 
-uri = f"ws://{base_url}/v1alpha/status/ws/{task['task_id']}"
+uri = f"ws://{base_url}/v1/status/ws/{task['task_id']}"
 with connect(uri) as websocket:
     for message in websocket:
         try:
             payload = json.loads(message)
             if payload["message"] == "error":
                 break
-            if payload["message"] == "error" and payload["task"]["task_status"] in ("success", "failure"):
+            if payload["message"] == "update" and payload["task"]["task_status"] in ("success", "failure"):
                 break
         except:
           break
@@ -447,4 +503,4 @@ with connect(uri) as websocket:
 
 When the task is completed, the result can be fetched with the endpoint:
 
-- `GET /v1alpha/result/{task_id}`
+- `GET /v1/result/{task_id}`

docs/v1_migration.md

@@ -0,0 +1,80 @@
+# Migration to the `v1` API
+
+Docling Serve from the initial prototype `v1alpha` API to the stable `v1` API.
+This page provides simple instructions to upgrade your application to the new API.
+
+## API changes
+
+The breaking changes introduced in the `v1` release of Docling Serve are designed to provide a stable schema which
+allows the project to provide new capabilities as new type of input sources, targets and also the definition of callback for event-driven applications.
+
+### Endpoint names
+
+All endpoints are renamed from `/v1alpha/` to `/v1/`.
+
+### Sources
+
+When using the `/v1/convert/source` endpoint, input documents have to be specified with the `sources: []` argument, which is replacing the usage of `file_sources` and `http_sources`.
+
+Old version:
+
+```jsonc
+{
+    "options": {},  // conversion options
+    "file_sources": [  // input documents provided as base64-encoded strings
+        {"base64_string": "abc123...", "filename": "file.pdf"}
+    ],
+    "http_sources": [  // input documents provided as http urls
+        {"url": "https://..."}
+    ]
+}
+```
+
+New version:
+
+```jsonc
+{
+    "options": {},  // conversion options
+    "sources": [
+        // input document provided as base64-encoded string
+        {"kind": "file", "base64_string": "abc123...", "filename": "file.pdf"},
+        // input document provided as http urls
+        {"kind": "http", "url": "https://..."},
+    ]
+}
+```
+
+### Targets
+
+Switching between output formats, i.e. from the JSON inbody response to the zip archive response, users have to specify the `target` argument, which is replacing the usage of `options.return_as_file`.
+
+Old version:
+
+```jsonc
+{
+    "options": {
+        "return_as_file": true  // <-- to be removed
+    },
+    // ...
+}
+```
+
+New version:
+
+```jsonc
+{
+    "options": {},
+    "target": {"kind": "zip"},  // <-- add this
+    // ...
+}
+```
+
+## Continue with the old API
+
+If you are not able to apply the changes above to your application, please consider pinning of the previous `v0.x` container images, e.g.
+
+```sh
+podman run -p 5001:5001 -e DOCLING_SERVE_ENABLE_UI=1 quay.io/docling-project/docling-serve:v0.16.1
+```
+
+_Note that the old prototype API will not be supported in new `v1.x` versions._

examples/split_processing.py

@@ -0,0 +1,124 @@
+import json
+import time
+from pathlib import Path
+
+import httpx
+from pydantic import BaseModel
+from pypdf import PdfReader
+
+from docling_core.types.doc.document import DoclingDocument
+
+# Variables to use
+path_to_pdf = Path("./tests/2206.01062v1.pdf")
+pages_per_file = 4
+base_url = "http://localhost:5001/v1"
+out_dir = Path("examples/splitted_pdf/")
+
+
+class ConvertedSplittedPdf(BaseModel):
+    task_id: str
+    conversion_finished: bool = False
+    result: dict | None = None
+
+
+def get_task_result(task_id: str):
+    response = httpx.get(
+        f"{base_url}/result/{task_id}",
+        timeout=15,
+    )
+    return response.json()
+
+
+def check_task_status(task_id: str):
+    response = httpx.get(f"{base_url}/status/poll/{task_id}", timeout=15)
+    task = response.json()
+    task_status = task["task_status"]
+
+    task_finished = False
+    if task_status == "success":
+        task_finished = True
+
+    if task_status in ("failure", "revoked"):
+        raise RuntimeError("A conversion failed")
+
+    time.sleep(5)
+
+    return task_finished
+
+
+def post_file(file_path: Path, start_page: int, end_page: int):
+    payload = {
+        "to_formats": ["json"],
+        "image_export_mode": "placeholder",
+        "ocr": False,
+        "abort_on_error": False,
+        "page_range": [start_page, end_page],
+    }
+
+    files = {
+        "files": (file_path.name, file_path.open("rb"), "application/pdf"),
+    }
+    response = httpx.post(
+        f"{base_url}/convert/file/async",
+        files=files,
+        data=payload,
+        timeout=15,
+    )
+
+    task = response.json()
+
+    return task["task_id"]
+
+
+def main():
+    filename = path_to_pdf
+
+    splitted_pdfs: list[ConvertedSplittedPdf] = []
+
+    with open(filename, "rb") as input_pdf_file:
+        pdf_reader = PdfReader(input_pdf_file)
+        total_pages = len(pdf_reader.pages)
+
+        for start_page in range(0, total_pages, pages_per_file):
+            task_id = post_file(
+                filename, start_page + 1, min(start_page + pages_per_file, total_pages)
+            )
+            splitted_pdfs.append(ConvertedSplittedPdf(task_id=task_id))
+
+    all_files_converted = False
+    while not all_files_converted:
+        found_conversion_running = False
+        for splitted_pdf in splitted_pdfs:
+            if not splitted_pdf.conversion_finished:
+                found_conversion_running = True
+                print("checking conversion status...")
+                splitted_pdf.conversion_finished = check_task_status(
+                    splitted_pdf.task_id
+                )
+        if not found_conversion_running:
+            all_files_converted = True
+
+    for splitted_pdf in splitted_pdfs:
+        splitted_pdf.result = get_task_result(splitted_pdf.task_id)
+
+    files = []
+    for i, splitted_pdf in enumerate(splitted_pdfs):
+        json_content = json.dumps(
+            splitted_pdf.result.get("document").get("json_content"), indent=2
+        )
+        doc = DoclingDocument.model_validate_json(json_content)
+        filename = f"{out_dir}/splited_json_{i}.json"
+        doc.save_as_json(filename=filename)
+        files.append(filename)
+
+    docs = [DoclingDocument.load_from_json(filename=f) for f in files]
+    concate_doc = DoclingDocument.concatenate(docs=docs)
+
+    exp_json_file = Path(f"{out_dir}/concatenated.json")
+    concate_doc.save_as_json(exp_json_file)
+
+    print("Finished")
+
+
+if __name__ == "__main__":
+    main()

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "docling-serve"
-version = "0.16.1"  # DO NOT EDIT, updated automatically
+version = "1.8.0"  # DO NOT EDIT, updated automatically
 description = "Running Docling as a service"
 license = {text = "MIT"}
 authors = [
@@ -8,7 +8,6 @@ authors = [
     {name="Guillaume Moutier", email="gmoutier@redhat.com"},
     {name="Anil Vishnoi", email="avishnoi@redhat.com"},
     {name="Panos Vagenas", email="pva@zurich.ibm.com"},
-    {name="Panos Vagenas", email="pva@zurich.ibm.com"},
     {name="Christoph Auer", email="cau@zurich.ibm.com"},
     {name="Peter Staar", email="taa@zurich.ibm.com"},
 ]
@@ -23,7 +22,7 @@ readme = "README.md"
 classifiers = [
     "License :: OSI Approved :: MIT License",
     "Operating System :: OS Independent",
-    # "Development Status :: 5 - Production/Stable",
+    "Development Status :: 5 - Production/Stable",
     "Intended Audience :: Developers",
     "Typing :: Typed",
     "Programming Language :: Python :: 3",
@@ -34,12 +33,11 @@ classifiers = [
 ]
 requires-python = ">=3.10"
 dependencies = [
-    "docling[vlm]~=2.38",
-    "docling-core>=2.32.0",
-    "mlx-vlm~=0.1.12; sys_platform == 'darwin' and platform_machine == 'arm64'",
-    "fastapi[standard]~=0.115",
+    "docling~=2.38",
+    "docling-core>=2.45.0",
+    "docling-jobkit[kfp,rq,vlm]>=1.8.0,<2.0.0",
+    "fastapi[standard]<0.119.0",  # ~=0.115
     "httpx~=0.28",
-    "kfp[kubernetes]>=2.10.0",
     "pydantic~=2.10",
     "pydantic-settings~=2.4",
     "python-multipart>=0.0.14,<0.1.0",
@@ -47,22 +45,25 @@ dependencies = [
     "uvicorn[standard]>=0.29.0,<1.0.0",
     "websockets~=14.0",
     "scalar-fastapi>=1.0.3",
+    "docling-mcp>=1.0.0",
 ]
 
 [project.optional-dependencies]
 ui = [
-    "gradio~=5.9",
-    "pydantic<2.11.0",  # fix compatibility between gradio and new pydantic 2.11
+    "gradio>=5.23.2,<6.0.0",
 ]
 tesserocr = [
     "tesserocr~=2.7"
 ]
+easyocr = [
+    "easyocr>=1.7",
+]
 rapidocr = [
-    "rapidocr-onnxruntime~=1.4; python_version<'3.13'",
-    "onnxruntime~=1.7",
+    "rapidocr (>=3.3,<4.0.0) ; python_version < '3.14'",
+    "onnxruntime (>=1.7.0,<2.0.0)",
 ]
 flash-attn = [
-  "flash-attn~=2.7.0; sys_platform == 'linux' and platform_machine == 'x86_64'"
+  "flash-attn~=2.8.2; sys_platform == 'linux' and platform_machine == 'x86_64'"
 ]
 
 [dependency-groups]
@@ -70,31 +71,43 @@ dev = [
     "asgi-lifespan~=2.0",
     "mypy~=1.11",
     "pre-commit-uv~=4.1",
+    "pypdf>=6.0.0",
     "pytest~=8.3",
     "pytest-asyncio~=0.24",
     "pytest-check~=2.4",
     "python-semantic-release~=7.32",
     "ruff>=0.9.6",
 ]
+
 pypi = [
-  "torch>=2.6.0",
-  "torchvision>=0.21.0",
+  "torch>=2.7.1",
+  "torchvision>=0.22.1",
 ]
+
 cpu = [
-  "torch>=2.6.0",
-  "torchvision>=0.21.0",
-]
-cu124 = [
-  "torch>=2.6.0",
-  "torchvision>=0.21.0",
+  "torch>=2.7.1",
+  "torchvision>=0.22.1",
 ]
+
+# cu124 = [
+#   "torch>=2.6.0",
+#   "torchvision>=0.21.0",
+# ]
+
 cu126 = [
-  "torch>=2.6.0",
-  "torchvision>=0.21.0",
+  "torch>=2.7.1",
+  "torchvision>=0.22.1",
 ]
+
 cu128 = [
-  "torch>=2.7.0",
-  "torchvision>=0.22.0",
+  "torch>=2.7.1",
+  "torchvision>=0.22.1",
+]
+
+rocm = [
+  "torch>=2.7.1",
+  "torchvision>=0.22.1",
+  "pytorch-triton-rocm>=3.3.1 ; sys_platform == 'linux' and platform_machine == 'x86_64'",
 ]
 
 [tool.uv]
@@ -104,32 +117,44 @@ conflicts = [
   [
     { group = "pypi" },
     { group = "cpu" },
-    { group = "cu124" },
+    # { group = "cu124" },
     { group = "cu126" },
     { group = "cu128" },
+    { group = "rocm" },
   ],
 ]
 environments = ["sys_platform != 'darwin' or platform_machine != 'x86_64'"]
 override-dependencies = [
-  "urllib3~=2.0"
+  "urllib3~=2.0",
+  "xgrammar>=0.1.24"
 ]
 
 [tool.uv.sources]
 torch = [
   { index = "pytorch-pypi", group = "pypi" },
   { index = "pytorch-cpu", group = "cpu" },
-  { index = "pytorch-cu124", group = "cu124" },
-  { index = "pytorch-cu126", group = "cu126" },
-  { index = "pytorch-cu128", group = "cu128" },
+  # { index = "pytorch-cu124", group = "cu124", marker = "sys_platform == 'linux'" },
+  { index = "pytorch-cu126", group = "cu126", marker = "sys_platform == 'linux'" },
+  { index = "pytorch-cu128", group = "cu128", marker = "sys_platform == 'linux'" },
+  { index = "pytorch-rocm", group = "rocm", marker = "sys_platform == 'linux'" },
 ]
+
 torchvision = [
   { index = "pytorch-pypi", group = "pypi" },
   { index = "pytorch-cpu", group = "cpu" },
-  { index = "pytorch-cu124", group = "cu124" },
-  { index = "pytorch-cu126", group = "cu126" },
-  { index = "pytorch-cu128", group = "cu128" },
+  # { index = "pytorch-cu124", group = "cu124", marker = "sys_platform == 'linux'" },
+  { index = "pytorch-cu126", group = "cu126", marker = "sys_platform == 'linux'" },
+  { index = "pytorch-cu128", group = "cu128", marker = "sys_platform == 'linux'" },
+  { index = "pytorch-rocm", group = "rocm", marker = "sys_platform == 'linux'" },
 ]
 
+pytorch-triton-rocm = [
+  { index = "pytorch-rocm", marker = "sys_platform == 'linux'" },
+]
+
+# docling-jobkit = { git = "https://github.com/docling-project/docling-jobkit/", rev = "main" }
+# docling-jobkit = { path = "../docling-jobkit", editable = true }
+
 [[tool.uv.index]]
 name = "pytorch-pypi"
 url = "https://pypi.org/simple"
@@ -140,10 +165,10 @@ name = "pytorch-cpu"
 url = "https://download.pytorch.org/whl/cpu"
 explicit = true
 
-[[tool.uv.index]]
-name = "pytorch-cu124"
-url = "https://download.pytorch.org/whl/cu124"
-explicit = true
+# [[tool.uv.index]]
+# name = "pytorch-cu124"
+# url = "https://download.pytorch.org/whl/cu124"
+# explicit = true
 
 [[tool.uv.index]]
 name = "pytorch-cu126"
@@ -155,6 +180,11 @@ name = "pytorch-cu128"
 url = "https://download.pytorch.org/whl/cu128"
 explicit = true
 
+[[tool.uv.index]]
+name = "pytorch-rocm"
+url = "https://download.pytorch.org/whl/rocm6.3"
+explicit = true
+
 [tool.setuptools.packages.find]
 include = ["docling_serve*"]
 namespaces = true
@@ -223,7 +253,7 @@ ignore = [
 max-complexity = 15
 
 [tool.ruff.lint.isort.sections]
-"docling" = ["docling", "docling_core"]
+"docling" = ["docling", "docling_core", "docling_jobkit"]
 
 [tool.ruff.lint.isort]
 combine-as-imports = true
@@ -252,6 +282,7 @@ module = [
     "kfp.*",
     "kfp_server_api.*",
     "mlx_vlm.*",
+    "mlx.*",
     "scalar_fastapi.*",
 ]
 ignore_missing_imports = true

scripts/update_doc_usage.py

@@ -0,0 +1,199 @@
+import re
+from typing import Annotated, Any, Union, get_args, get_origin
+
+from pydantic import BaseModel
+
+from docling_serve.datamodel.convert import ConvertDocumentsRequestOptions
+
+DOCS_FILE = "docs/usage.md"
+
+VARIABLE_WORDS: list[str] = [
+    "picture_description_local",
+    "vlm_pipeline_model",
+    "vlm",
+    "vlm_pipeline_model_api",
+    "ocr_engines_enum",
+    "easyocr",
+    "dlparse_v4",
+    "fast",
+    "picture_description_api",
+    "vlm_pipeline_model_local",
+]
+
+
+def format_variable_names(text: str) -> str:
+    """Format specific words in description to be code-formatted."""
+    sorted_words = sorted(VARIABLE_WORDS, key=len, reverse=True)
+
+    escaped_words = [re.escape(word) for word in sorted_words]
+
+    for word in escaped_words:
+        pattern = rf"(?<!`)\b{word}\b(?!`)"
+        text = re.sub(pattern, f"`{word}`", text)
+
+    return text
+
+
+def format_allowed_values_description(description: str) -> str:
+    """Format description to code-format allowed values."""
+    # Regex pattern to find text after "Allowed values:"
+    match = re.search(r"Allowed values:(.+?)(?:\.|$)", description, re.DOTALL)
+
+    if match:
+        # Extract the allowed values
+        values_str = match.group(1).strip()
+
+        # Split values, handling both comma and 'and' separators
+        values = re.split(r"\s*(?:,\s*|\s+and\s+)", values_str)
+
+        # Remove any remaining punctuation and whitespace
+        values = [value.strip("., ") for value in values]
+
+        # Create code-formatted values
+        formatted_values = ", ".join(f"`{value}`" for value in values)
+
+        # Replace the original allowed values with formatted version
+        formatted_description = re.sub(
+            r"(Allowed values:)(.+?)(?:\.|$)",
+            f"\\1 {formatted_values}.",
+            description,
+            flags=re.DOTALL,
+        )
+
+        return formatted_description
+
+    return description
+
+
+def _format_type(type_hint: Any) -> str:
+    """Format type ccrrectly, like Annotation or Union."""
+    if get_origin(type_hint) is Annotated:
+        base_type = get_args(type_hint)[0]
+        return _format_type(base_type)
+
+    if hasattr(type_hint, "__origin__"):
+        origin = type_hint.__origin__
+        args = get_args(type_hint)
+
+        if origin is list:
+            return f"List[{_format_type(args[0])}]"
+        elif origin is dict:
+            return f"Dict[{_format_type(args[0])}, {_format_type(args[1])}]"
+        elif str(origin).__contains__("Union") or str(origin).__contains__("Optional"):
+            return " or ".join(_format_type(arg) for arg in args)
+        elif origin is None:
+            return "null"
+
+    if hasattr(type_hint, "__name__"):
+        return type_hint.__name__
+
+    return str(type_hint)
+
+
+def _unroll_types(tp) -> list[type]:
+    """
+    Unrolls typing.Union and typing.Optional types into a flat list of types.
+    """
+    origin = get_origin(tp)
+    if origin is Union:
+        # Recursively unroll each type inside the Union
+        types = []
+        for arg in get_args(tp):
+            types.extend(_unroll_types(arg))
+        # Remove duplicates while preserving order
+        return list(dict.fromkeys(types))
+    else:
+        # If it's not a Union, just return it as a single-element list
+        return [tp]
+
+
+def generate_model_doc(model: type[BaseModel]) -> str:
+    """Generate documentation for a Pydantic model."""
+
+    models_stack = [model]
+
+    doc = ""
+    while models_stack:
+        current_model = models_stack.pop()
+
+        doc += f"<h4>{current_model.__name__}</h4>\n"
+
+        doc += "\n| Field Name | Type | Description |\n"
+        doc += "|------------|------|-------------|\n"
+
+        base_models = []
+        if hasattr(current_model, "__mro__"):
+            base_models = current_model.__mro__
+        else:
+            base_models = [current_model]
+
+        for base_model in base_models:
+            # Check if this is a Pydantic model
+            if hasattr(base_model, "model_fields"):
+                # Iterate through fields of this model
+                for field_name, field in base_model.model_fields.items():
+                    # Extract description from Annotated field if possible
+                    description = field.description or "No description provided."
+                    description = format_allowed_values_description(description)
+                    description = format_variable_names(description)
+
+                    # Handle Annotated types
+                    original_type = field.annotation
+                    if get_origin(original_type) is Annotated:
+                        # Extract base type and additional metadata
+                        type_args = get_args(original_type)
+                        base_type = type_args[0]
+                    else:
+                        base_type = original_type
+
+                    field_type = _format_type(base_type)
+                    field_type = format_variable_names(field_type)
+
+                    doc += f"| `{field_name}` | {field_type} | {description} |\n"
+
+                    for field_type in _unroll_types(base_type):
+                        if issubclass(field_type, BaseModel):
+                            models_stack.append(field_type)
+
+                # stop iterating the base classes
+                break
+
+        doc += "\n"
+    return doc
+
+
+def update_documentation():
+    """Update the documentation file with model information."""
+    doc_request = generate_model_doc(ConvertDocumentsRequestOptions)
+
+    with open(DOCS_FILE) as f:
+        content = f.readlines()
+
+    # Prepare to update the content
+    new_content = []
+    in_cp_section = False
+
+    for line in content:
+        if line.startswith("<!-- begin: parameters-docs -->"):
+            in_cp_section = True
+            new_content.append(line)
+            new_content.append(doc_request)
+            continue
+
+        if in_cp_section and line.strip() == "<!-- end: parameters-docs -->":
+            in_cp_section = False
+
+        if not in_cp_section:
+            new_content.append(line)
+
+    # Only write to the file if new_content is different from content
+    if "".join(new_content) != "".join(content):
+        with open(DOCS_FILE, "w") as f:
+            f.writelines(new_content)
+        print(f"Documentation updated in {DOCS_FILE}")
+    else:
+        print("No changes detected. Documentation file remains unchanged.")
+
+
+if __name__ == "__main__":
+    update_documentation()

tests/test_1-file-all-outputs.py

@@ -6,17 +6,22 @@ import pytest
 import pytest_asyncio
 from pytest_check import check
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
 @pytest.mark.asyncio
 async def test_convert_file(async_client):
     """Test convert single file to all outputs"""
-    url = "http://localhost:5001/v1alpha/convert/file"
+    url = "http://localhost:5001/v1/convert/file"
     options = {
         "from_formats": [
             "docx",
@@ -37,7 +42,6 @@ async def test_convert_file(async_client):
         "pdf_backend": "dlparse_v2",
         "table_mode": "fast",
         "abort_on_error": False,
-        "return_as_file": False,
     }
 
     current_dir = os.path.dirname(__file__)

tests/test_1-file-async.py

@@ -6,10 +6,15 @@ import httpx
 import pytest
 import pytest_asyncio
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
@@ -17,13 +22,12 @@ async def async_client():
 async def test_convert_url(async_client):
     """Test convert URL to all outputs"""
 
-    base_url = "http://localhost:5001/v1alpha"
+    base_url = "http://localhost:5001/v1"
     payload = {
         "to_formats": ["md", "json", "html"],
         "image_export_mode": "placeholder",
         "ocr": False,
         "abort_on_error": False,
-        "return_as_file": False,
     }
 
     file_path = Path(__file__).parent / "2206.01062v1.pdf"

tests/test_1-url-all-outputs.py

@@ -5,17 +5,22 @@ import pytest
 import pytest_asyncio
 from pytest_check import check
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
 @pytest.mark.asyncio
 async def test_convert_url(async_client):
     """Test convert URL to all outputs"""
-    url = "http://localhost:5001/v1alpha/convert/source"
+    url = "http://localhost:5001/v1/convert/source"
     payload = {
         "options": {
             "from_formats": [
@@ -37,9 +42,8 @@ async def test_convert_url(async_client):
             "pdf_backend": "dlparse_v2",
             "table_mode": "fast",
             "abort_on_error": False,
-            "return_as_file": False,
         },
-        "http_sources": [{"url": "https://arxiv.org/pdf/2206.01062"}],
+        "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2206.01062"}],
     }
     print(json.dumps(payload, indent=2))
 

tests/test_1-url-async-ws.py

@@ -6,28 +6,35 @@ import pytest
 import pytest_asyncio
 from websockets.sync.client import connect
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
 @pytest.mark.asyncio
 async def test_convert_url(async_client: httpx.AsyncClient):
     """Test convert URL to all outputs"""
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
 
     doc_filename = Path("tests/2408.09869v5.pdf")
     encoded_doc = base64.b64encode(doc_filename.read_bytes()).decode()
 
-    base_url = "http://localhost:5001/v1alpha"
+    base_url = "http://localhost:5001/v1"
     payload = {
         "options": {
             "to_formats": ["md", "json"],
             "image_export_mode": "placeholder",
             "ocr": True,
             "abort_on_error": False,
-            "return_as_file": False,
             # "do_picture_description": True,
             # "picture_description_api": {
             #     "url": "http://localhost:11434/v1/chat/completions",
@@ -39,8 +46,14 @@ async def test_convert_url(async_client: httpx.AsyncClient):
             #     "repo_id": "HuggingFaceTB/SmolVLM-256M-Instruct",
             # },
         },
-        # "http_sources": [{"url": "https://arxiv.org/pdf/2501.17887"}],
-        "file_sources": [{"base64_string": encoded_doc, "filename": doc_filename.name}],
+        # "sources": [{"kind": "http", "url": "https://arxiv.org/pdf/2501.17887"}],
+        "sources": [
+            {
+                "kind": "file",
+                "base64_string": encoded_doc,
+                "filename": doc_filename.name,
+            }
+        ],
     }
     # print(json.dumps(payload, indent=2))
 
@@ -52,7 +65,13 @@ async def test_convert_url(async_client: httpx.AsyncClient):
 
     task = response.json()
 
-    uri = f"ws://localhost:5001/v1alpha/status/ws/{task['task_id']}"
+    uri = f"ws://localhost:5001/v1/status/ws/{task['task_id']}?api_key={docling_serve_settings.api_key}"
     with connect(uri) as websocket:
         for message in websocket:
             print(message)
+
+    result_resp = await async_client.get(f"{base_url}/result/{task['task_id']}")
+    assert result_resp.status_code == 200, "Response should be 200 OK"
+    result = result_resp.json()
+    print(f"{result['processing_time']=}")
+    assert result["processing_time"] > 1.0

tests/test_1-url-async.py

@@ -6,10 +6,15 @@ import httpx
 import pytest
 import pytest_asyncio
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
@@ -25,16 +30,15 @@ async def test_convert_url(async_client):
         "https://arxiv.org/pdf/2311.18481",
     ]
 
-    base_url = "http://localhost:5001/v1alpha"
+    base_url = "http://localhost:5001/v1"
     payload = {
         "options": {
             "to_formats": ["md", "json"],
             "image_export_mode": "placeholder",
             "ocr": True,
             "abort_on_error": False,
-            "return_as_file": False,
         },
-        "http_sources": [{"url": random.choice(example_docs)}],
+        "sources": [{"kind": "http", "url": random.choice(example_docs)}],
     }
     print(json.dumps(payload, indent=2))
 
@@ -58,3 +62,60 @@ async def test_convert_url(async_client):
         time.sleep(2)
 
     assert task["task_status"] == "success"
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("include_converted_doc", [False, True])
+async def test_chunk_url(async_client, include_converted_doc: bool):
+    """Test chunk URL"""
+
+    example_docs = [
+        "https://arxiv.org/pdf/2311.18481",
+    ]
+
+    base_url = "http://localhost:5001/v1"
+    payload = {
+        "sources": [{"kind": "http", "url": random.choice(example_docs)}],
+        "include_converted_doc": include_converted_doc,
+    }
+
+    response = await async_client.post(
+        f"{base_url}/chunk/hybrid/source/async", json=payload
+    )
+    assert response.status_code == 200, "Response should be 200 OK"
+
+    task = response.json()
+
+    print(json.dumps(task, indent=2))
+
+    while task["task_status"] not in ("success", "failure"):
+        response = await async_client.get(f"{base_url}/status/poll/{task['task_id']}")
+        assert response.status_code == 200, "Response should be 200 OK"
+        task = response.json()
+        print(f"{task['task_status']=}")
+        print(f"{task['task_position']=}")
+
+        time.sleep(2)
+
+    assert task["task_status"] == "success"
+
+    result_resp = await async_client.get(f"{base_url}/result/{task['task_id']}")
+    assert result_resp.status_code == 200, "Response should be 200 OK"
+    result = result_resp.json()
+    print("Got result.")
+
+    assert "chunks" in result
+    assert len(result["chunks"]) > 0
+
+    assert "documents" in result
+    assert len(result["documents"]) > 0
+    assert result["documents"][0]["status"] == "success"
+
+    if include_converted_doc:
+        assert result["documents"][0]["content"]["json_content"] is not None
+        assert (
+            result["documents"][0]["content"]["json_content"]["schema_name"]
+            == "DoclingDocument"
+        )
+    else:
+        assert result["documents"][0]["content"]["json_content"] is None

tests/test_2-files-all-outputs.py

@@ -5,17 +5,22 @@ import pytest
 import pytest_asyncio
 from pytest_check import check
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
 @pytest.mark.asyncio
 async def test_convert_file(async_client):
     """Test convert single file to all outputs"""
-    url = "http://localhost:5001/v1alpha/convert/file"
+    url = "http://localhost:5001/v1/convert/file"
     options = {
         "from_formats": [
             "docx",
@@ -36,7 +41,6 @@ async def test_convert_file(async_client):
         "pdf_backend": "dlparse_v2",
         "table_mode": "fast",
         "abort_on_error": False,
-        "return_as_file": False,
     }
 
     current_dir = os.path.dirname(__file__)

tests/test_2-urls-all-outputs.py

@@ -3,17 +3,22 @@ import pytest
 import pytest_asyncio
 from pytest_check import check
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
 @pytest.mark.asyncio
 async def test_convert_url(async_client):
     """Test convert URL to all outputs"""
-    url = "http://localhost:5001/v1alpha/convert/source"
+    url = "http://localhost:5001/v1/convert/source"
     payload = {
         "options": {
             "from_formats": [
@@ -35,12 +40,12 @@ async def test_convert_url(async_client):
             "pdf_backend": "dlparse_v2",
             "table_mode": "fast",
             "abort_on_error": False,
-            "return_as_file": False,
         },
-        "http_sources": [
-            {"url": "https://arxiv.org/pdf/2206.01062"},
-            {"url": "https://arxiv.org/pdf/2408.09869"},
+        "sources": [
+            {"kind": "http", "url": "https://arxiv.org/pdf/2206.01062"},
+            {"kind": "http", "url": "https://arxiv.org/pdf/2408.09869"},
         ],
+        "target": {"kind": "zip"},
     }
 
     response = await async_client.post(url, json=payload)

tests/test_2-urls-async-all-outputs.py

@@ -6,17 +6,22 @@ import pytest
 import pytest_asyncio
 from pytest_check import check
 
+from docling_serve.settings import docling_serve_settings
+
 
 @pytest_asyncio.fixture
 async def async_client():
-    async with httpx.AsyncClient(timeout=60.0) as client:
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    async with httpx.AsyncClient(timeout=60.0, headers=headers) as client:
         yield client
 
 
 @pytest.mark.asyncio
 async def test_convert_url(async_client):
     """Test convert URL to all outputs"""
-    base_url = "http://localhost:5001/v1alpha"
+    base_url = "http://localhost:5001/v1"
     payload = {
         "options": {
             "from_formats": [
@@ -38,12 +43,12 @@ async def test_convert_url(async_client):
             "pdf_backend": "dlparse_v2",
             "table_mode": "fast",
             "abort_on_error": False,
-            "return_as_file": False,
         },
-        "http_sources": [
-            {"url": "https://arxiv.org/pdf/2206.01062"},
-            {"url": "https://arxiv.org/pdf/2408.09869"},
+        "sources": [
+            {"kind": "http", "url": "https://arxiv.org/pdf/2206.01062"},
+            {"kind": "http", "url": "https://arxiv.org/pdf/2408.09869"},
         ],
+        "target": {"kind": "zip"},
     }
 
     response = await async_client.post(f"{base_url}/convert/source/async", json=payload)

tests/test_fastapi_endpoints.py

@@ -1,6 +1,8 @@
 import asyncio
+import io
 import json
 import os
+import zipfile
 
 import pytest
 import pytest_asyncio
@@ -8,7 +10,10 @@ from asgi_lifespan import LifespanManager
 from httpx import ASGITransport, AsyncClient
 from pytest_check import check
 
+from docling_core.types.doc import DoclingDocument, PictureItem
+
 from docling_serve.app import create_app
+from docling_serve.settings import docling_serve_settings
 
 
 @pytest.fixture(scope="session")
@@ -16,6 +21,14 @@ def event_loop():
     return asyncio.get_event_loop()
 
 
+@pytest.fixture(scope="session")
+def auth_headers():
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    return headers
+
+
 @pytest_asyncio.fixture(scope="session")
 async def app():
     app = create_app()
@@ -42,10 +55,18 @@ async def test_health(client: AsyncClient):
 
 
 @pytest.mark.asyncio
-async def test_convert_file(client: AsyncClient):
+async def test_openapijson(client: AsyncClient):
+    response = await client.get("/openapi.json")
+    assert response.status_code == 200
+    schema = response.json()
+    assert "openapi" in schema
+
+
+@pytest.mark.asyncio
+async def test_convert_file(client: AsyncClient, auth_headers: dict):
     """Test convert single file to all outputs"""
 
-    endpoint = "/v1alpha/convert/file"
+    endpoint = "/v1/convert/file"
     options = {
         "from_formats": [
             "docx",
@@ -66,7 +87,6 @@ async def test_convert_file(client: AsyncClient):
         "pdf_backend": "dlparse_v2",
         "table_mode": "fast",
         "abort_on_error": False,
-        "return_as_file": False,
     }
 
     current_dir = os.path.dirname(__file__)
@@ -76,7 +96,9 @@ async def test_convert_file(client: AsyncClient):
         "files": ("2206.01062v1.pdf", open(file_path, "rb"), "application/pdf"),
     }
 
-    response = await client.post(endpoint, files=files, data=options)
+    response = await client.post(
+        endpoint, files=files, data=options, headers=auth_headers
+    )
     assert response.status_code == 200, "Response should be 200 OK"
 
     data = response.json()
@@ -154,3 +176,39 @@ async def test_convert_file(client: AsyncClient):
             data["document"]["doctags_content"],
             msg=f"DocTags document should contain '<doctag><page_header>'. Received: {safe_slice(data['document']['doctags_content'])}",
         )
+
+
+@pytest.mark.asyncio
+async def test_referenced_artifacts(client: AsyncClient, auth_headers: dict):
+    """Test that paths in the zip file are relative to the zip file root."""
+
+    endpoint = "/v1/convert/file"
+    options = {
+        "to_formats": ["json"],
+        "image_export_mode": "referenced",
+        "target_type": "zip",
+        "ocr": False,
+    }
+
+    current_dir = os.path.dirname(__file__)
+    file_path = os.path.join(current_dir, "2206.01062v1.pdf")
+
+    files = {
+        "files": ("2206.01062v1.pdf", open(file_path, "rb"), "application/pdf"),
+    }
+
+    response = await client.post(
+        endpoint, files=files, data=options, headers=auth_headers
+    )
+    assert response.status_code == 200, "Response should be 200 OK"
+
+    with zipfile.ZipFile(io.BytesIO(response.content)) as zip_file:
+        namelist = zip_file.namelist()
+        for file in namelist:
+            if file.endswith(".json"):
+                doc = DoclingDocument.model_validate(json.loads(zip_file.read(file)))
+                for item, _level in doc.iterate_items():
+                    if isinstance(item, PictureItem):
+                        assert item.image is not None
+                        print(f"{item.image.uri}=")
+                        assert str(item.image.uri) in namelist

tests/test_file_opts.py

@@ -11,6 +11,7 @@ from docling_core.types import DoclingDocument
 from docling_core.types.doc.document import PictureDescriptionData
 
 from docling_serve.app import create_app
+from docling_serve.settings import docling_serve_settings
 
 
 @pytest.fixture(scope="session")
@@ -18,6 +19,14 @@ def event_loop():
     return asyncio.get_event_loop()
 
 
+@pytest.fixture(scope="session")
+def auth_headers():
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    return headers
+
+
 @pytest_asyncio.fixture(scope="session")
 async def app():
     app = create_app()
@@ -37,10 +46,10 @@ async def client(app):
 
 
 @pytest.mark.asyncio
-async def test_convert_file(client: AsyncClient):
+async def test_convert_file(client: AsyncClient, auth_headers: dict):
     """Test convert single file to all outputs"""
 
-    endpoint = "/v1alpha/convert/file"
+    endpoint = "/v1/convert/file"
     options = {
         "to_formats": ["md", "json"],
         "image_export_mode": "placeholder",
@@ -63,7 +72,9 @@ async def test_convert_file(client: AsyncClient):
         "files": ("2206.01062v1.pdf", open(file_path, "rb"), "application/pdf"),
     }
 
-    response = await client.post(endpoint, files=files, data=options)
+    response = await client.post(
+        endpoint, files=files, data=options, headers=auth_headers
+    )
     assert response.status_code == 200, "Response should be 200 OK"
 
     data = response.json()

tests/test_options_serialization.py

@@ -1,54 +0,0 @@
-from docling_serve.datamodel.convert import (
-    ConvertDocumentsOptions,
-    PictureDescriptionApi,
-)
-from docling_serve.docling_conversion import (
-    _hash_pdf_format_option,
-    get_pdf_pipeline_opts,
-)
-
-
-def test_options_cache_key():
-    hashes = set()
-
-    opts = ConvertDocumentsOptions()
-    pipeline_opts = get_pdf_pipeline_opts(opts)
-    hash = _hash_pdf_format_option(pipeline_opts)
-    assert hash not in hashes
-    hashes.add(hash)
-
-    opts.do_picture_description = True
-    pipeline_opts = get_pdf_pipeline_opts(opts)
-    hash = _hash_pdf_format_option(pipeline_opts)
-    # pprint(pipeline_opts.pipeline_options.model_dump(serialize_as_any=True))
-    assert hash not in hashes
-    hashes.add(hash)
-
-    opts.picture_description_api = PictureDescriptionApi(
-        url="http://localhost",
-        params={"model": "mymodel"},
-        prompt="Hello 1",
-    )
-    pipeline_opts = get_pdf_pipeline_opts(opts)
-    hash = _hash_pdf_format_option(pipeline_opts)
-    # pprint(pipeline_opts.pipeline_options.model_dump(serialize_as_any=True))
-    assert hash not in hashes
-    hashes.add(hash)
-
-    opts.picture_description_api = PictureDescriptionApi(
-        url="http://localhost",
-        params={"model": "your-model"},
-        prompt="Hello 1",
-    )
-    pipeline_opts = get_pdf_pipeline_opts(opts)
-    hash = _hash_pdf_format_option(pipeline_opts)
-    # pprint(pipeline_opts.pipeline_options.model_dump(serialize_as_any=True))
-    assert hash not in hashes
-    hashes.add(hash)
-
-    opts.picture_description_api.prompt = "World"
-    pipeline_opts = get_pdf_pipeline_opts(opts)
-    hash = _hash_pdf_format_option(pipeline_opts)
-    # pprint(pipeline_opts.pipeline_options.model_dump(serialize_as_any=True))
-    assert hash not in hashes
-    hashes.add(hash)

tests/test_results_clear.py

@@ -17,6 +17,14 @@ def event_loop():
     return asyncio.get_event_loop()
 
 
+@pytest.fixture(scope="session")
+def auth_headers():
+    headers = {}
+    if docling_serve_settings.api_key:
+        headers["X-Api-Key"] = docling_serve_settings.api_key
+    return headers
+
+
 @pytest_asyncio.fixture(scope="session")
 async def app():
     app = create_app()
@@ -35,7 +43,7 @@ async def client(app):
         yield client
 
 
-async def convert_file(client: AsyncClient):
+async def convert_file(client: AsyncClient, auth_headers: dict):
     doc_filename = Path("tests/2408.09869v5.pdf")
     encoded_doc = base64.b64encode(doc_filename.read_bytes()).decode()
 
@@ -43,10 +51,18 @@ async def convert_file(client: AsyncClient):
         "options": {
             "to_formats": ["json"],
         },
-        "file_sources": [{"base64_string": encoded_doc, "filename": doc_filename.name}],
+        "sources": [
+            {
+                "kind": "file",
+                "base64_string": encoded_doc,
+                "filename": doc_filename.name,
+            }
+        ],
     }
 
-    response = await client.post("/v1alpha/convert/source/async", json=payload)
+    response = await client.post(
+        "/v1/convert/source/async", json=payload, headers=auth_headers
+    )
     assert response.status_code == 200, "Response should be 200 OK"
 
     task = response.json()
@@ -54,7 +70,9 @@ async def convert_file(client: AsyncClient):
     print(json.dumps(task, indent=2))
 
     while task["task_status"] not in ("success", "failure"):
-        response = await client.get(f"/v1alpha/status/poll/{task['task_id']}")
+        response = await client.get(
+            f"/v1/status/poll/{task['task_id']}", headers=auth_headers
+        )
         assert response.status_code == 200, "Response should be 200 OK"
         task = response.json()
         print(f"{task['task_status']=}")
@@ -68,52 +86,62 @@ async def convert_file(client: AsyncClient):
 
 
 @pytest.mark.asyncio
-async def test_clear_results(client: AsyncClient):
+async def test_clear_results(client: AsyncClient, auth_headers: dict):
     """Test removal of task."""
 
     # Set long delay deletion
     docling_serve_settings.result_removal_delay = 100
 
     # Convert and wait for completion
-    task = await convert_file(client)
+    task = await convert_file(client, auth_headers=auth_headers)
 
     # Get result once
-    result_response = await client.get(f"/v1alpha/result/{task['task_id']}")
+    result_response = await client.get(
+        f"/v1/result/{task['task_id']}", headers=auth_headers
+    )
     assert result_response.status_code == 200, "Response should be 200 OK"
     print("Result 1 ok.")
     result = result_response.json()
     assert result["document"]["json_content"]["schema_name"] == "DoclingDocument"
 
     # Get result twice
-    result_response = await client.get(f"/v1alpha/result/{task['task_id']}")
+    result_response = await client.get(
+        f"/v1/result/{task['task_id']}", headers=auth_headers
+    )
     assert result_response.status_code == 200, "Response should be 200 OK"
     print("Result 2 ok.")
     result = result_response.json()
     assert result["document"]["json_content"]["schema_name"] == "DoclingDocument"
 
     # Clear
-    clear_response = await client.get("/v1alpha/clear/results?older_then=0")
+    clear_response = await client.get(
+        "/v1/clear/results?older_then=0", headers=auth_headers
+    )
     assert clear_response.status_code == 200, "Response should be 200 OK"
     print("Clear ok.")
 
     # Get deleted result
-    result_response = await client.get(f"/v1alpha/result/{task['task_id']}")
+    result_response = await client.get(
+        f"/v1/result/{task['task_id']}", headers=auth_headers
+    )
     assert result_response.status_code == 404, "Response should be removed"
     print("Result was no longer found.")
 
 
 @pytest.mark.asyncio
-async def test_delay_remove(client: AsyncClient):
+async def test_delay_remove(client: AsyncClient, auth_headers: dict):
     """Test automatic removal of task with delay."""
 
     # Set short delay deletion
     docling_serve_settings.result_removal_delay = 5
 
     # Convert and wait for completion
-    task = await convert_file(client)
+    task = await convert_file(client, auth_headers=auth_headers)
 
     # Get result once
-    result_response = await client.get(f"/v1alpha/result/{task['task_id']}")
+    result_response = await client.get(
+        f"/v1/result/{task['task_id']}", headers=auth_headers
+    )
     assert result_response.status_code == 200, "Response should be 200 OK"
     print("Result ok.")
     result = result_response.json()
@@ -123,5 +151,7 @@ async def test_delay_remove(client: AsyncClient):
     await asyncio.sleep(10)
 
     # Get deleted result
-    result_response = await client.get(f"/v1alpha/result/{task['task_id']}")
+    result_response = await client.get(
+        f"/v1/result/{task['task_id']}", headers=auth_headers
+    )
     assert result_response.status_code == 404, "Response should be removed"