feat: Docling with new standard pipeline with threading (#428)

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

docling_serve/__main__.py

@@ -385,6 +385,11 @@ def rq_worker() -> Any:
         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(

docling_serve/orchestrator_factory.py

@@ -288,6 +288,11 @@ def get_async_orchestrator() -> BaseOrchestrator:
             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)
 

docling_serve/settings.py

@@ -57,6 +57,13 @@ class DoclingServeSettings(BaseSettings):
     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
 

docs/configuration.md

@@ -44,13 +44,17 @@ THe following table describes the options to configure the Docling Serve app.
 |  | `DOCLING_SERVE_SINGLE_USE_RESULTS` | `true` | If true, results can be accessed only once. If false, the results accumulate in the scratch directory. |
 |  | `DOCLING_SERVE_RESULT_REMOVAL_DELAY` | `300` | When `DOCLING_SERVE_SINGLE_USE_RESULTS` is active, this is the delay before results are removed from the task registry. |
 |  | `DOCLING_SERVE_MAX_DOCUMENT_TIMEOUT` | `604800` (7 days) | The maximum time for processing a document. |
-|  | `DOCLING_NUM_THREADS` | `4` | Number of concurrent threads 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. |

pyproject.toml

@@ -35,7 +35,7 @@ requires-python = ">=3.10"
 dependencies = [
     "docling~=2.38",
     "docling-core>=2.45.0",
-    "docling-jobkit[kfp,rq,vlm]>=1.6.0,<2.0.0",
+    "docling-jobkit[kfp,rq,vlm]>=1.8.0,<2.0.0",
     "fastapi[standard]<0.119.0",  # ~=0.115
     "httpx~=0.28",
     "pydantic~=2.10",