Merge pull request #926 from siiddhantt/feature

Feature: Logging token usage info to MongoDB

SECURITY.md

@@ -0,0 +1,14 @@
+# Security Policy
+
+## Supported Versions
+
+Supported Versions:
+
+Currently, we support security patches by committing changes and bumping the version published on Github.
+
+## Reporting a Vulnerability
+
+Found a vulnerability? Please email us:
+
+security@arc53.com
+

application/api/answer/routes.py

@@ -8,17 +8,14 @@ import traceback
 
 from pymongo import MongoClient
 from bson.objectid import ObjectId
-from transformers import GPT2TokenizerFast
-
 
 
 from application.core.settings import settings
-from application.vectorstore.vector_creator import VectorCreator
 from application.llm.llm_creator import LLMCreator
+from application.retriever.retriever_creator import RetrieverCreator
 from application.error import bad_request
 
 
-
 logger = logging.getLogger(__name__)
 
 mongo = MongoClient(settings.MONGO_URI)
@@ -27,20 +24,22 @@ conversations_collection = db["conversations"]
 vectors_collection = db["vectors"]
 prompts_collection = db["prompts"]
 api_key_collection = db["api_keys"]
-answer = Blueprint('answer', __name__)
+answer = Blueprint("answer", __name__)
 
 gpt_model = ""
 # to have some kind of default behaviour
 if settings.LLM_NAME == "openai":
-    gpt_model = 'gpt-3.5-turbo'
+    gpt_model = "gpt-3.5-turbo"
 elif settings.LLM_NAME == "anthropic":
-    gpt_model = 'claude-2'
+    gpt_model = "claude-2"
 
 if settings.MODEL_NAME:  # in case there is particular model name configured
     gpt_model = settings.MODEL_NAME
 
 # load the prompts
-current_dir = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+current_dir = os.path.dirname(
+    os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+)
 with open(os.path.join(current_dir, "prompts", "chat_combine_default.txt"), "r") as f:
     chat_combine_template = f.read()
 
@@ -51,7 +50,7 @@ with open(os.path.join(current_dir, "prompts", "chat_combine_creative.txt"), "r"
     chat_combine_creative = f.read()
 
 with open(os.path.join(current_dir, "prompts", "chat_combine_strict.txt"), "r") as f:
-    chat_combine_strict = f.read()    
+    chat_combine_strict = f.read()
 
 api_key_set = settings.API_KEY is not None
 embeddings_key_set = settings.EMBEDDINGS_KEY is not None
@@ -62,11 +61,6 @@ async def async_generate(chain, question, chat_history):
     return result
 
 
-def count_tokens(string):
-    tokenizer = GPT2TokenizerFast.from_pretrained('gpt2')
-    return len(tokenizer(string)['input_ids'])
-
-
 def run_async_chain(chain, question, chat_history):
     loop = asyncio.new_event_loop()
     asyncio.set_event_loop(loop)
@@ -78,17 +72,18 @@ def run_async_chain(chain, question, chat_history):
     result["answer"] = answer
     return result
 
+
 def get_data_from_api_key(api_key):
     data = api_key_collection.find_one({"key": api_key})
     if data is None:
         return bad_request(401, "Invalid API key")
     return data
-    
+
 
 def get_vectorstore(data):
     if "active_docs" in data:
         if data["active_docs"].split("/")[0] == "default":
-                vectorstore = ""
+            vectorstore = ""
         elif data["active_docs"].split("/")[0] == "local":
             vectorstore = "indexes/" + data["active_docs"]
         else:
@@ -102,85 +97,99 @@ def get_vectorstore(data):
 
 
 def is_azure_configured():
-    return settings.OPENAI_API_BASE and settings.OPENAI_API_VERSION and settings.AZURE_DEPLOYMENT_NAME
+    return (
+        settings.OPENAI_API_BASE
+        and settings.OPENAI_API_VERSION
+        and settings.AZURE_DEPLOYMENT_NAME
+    )
 
 
-def complete_stream(question, docsearch, chat_history, prompt_id, conversation_id, chunks=2):
-    llm = LLMCreator.create_llm(settings.LLM_NAME, api_key=settings.API_KEY)
-    if prompt_id == 'default':
-        prompt = chat_combine_template
-    elif prompt_id == 'creative':
-        prompt = chat_combine_creative
-    elif prompt_id == 'strict':
-        prompt = chat_combine_strict
-    else:
-        prompt = prompts_collection.find_one({"_id": ObjectId(prompt_id)})["content"]
-        
-    if chunks == 0:
-        docs = []
-    else:
-        docs = docsearch.search(question, k=chunks)
-    if settings.LLM_NAME == "llama.cpp":
-        docs = [docs[0]]
-    # join all page_content together with a newline
-    docs_together = "\n".join([doc.page_content for doc in docs])
-    p_chat_combine = prompt.replace("{summaries}", docs_together)
-    messages_combine = [{"role": "system", "content": p_chat_combine}]
-    source_log_docs = []
-    for doc in docs:
-        if doc.metadata:
-            source_log_docs.append({"title": doc.metadata['title'].split('/')[-1], "text": doc.page_content})
-        else:
-            source_log_docs.append({"title": doc.page_content, "text": doc.page_content})
-
-    if len(chat_history) > 1:
-        tokens_current_history = 0
-        # count tokens in history
-        chat_history.reverse()
-        for i in chat_history:
-            if "prompt" in i and "response" in i:
-                tokens_batch = count_tokens(i["prompt"]) + count_tokens(i["response"])
-                if tokens_current_history + tokens_batch < settings.TOKENS_MAX_HISTORY:
-                    tokens_current_history += tokens_batch
-                    messages_combine.append({"role": "user", "content": i["prompt"]})
-                    messages_combine.append({"role": "system", "content": i["response"]})
-    messages_combine.append({"role": "user", "content": question})
-
-    response_full = ""
-    completion = llm.gen_stream(model=gpt_model, engine=settings.AZURE_DEPLOYMENT_NAME,
-                                messages=messages_combine)
-    for line in completion:
-        data = json.dumps({"answer": str(line)})
-        response_full += str(line)
-        yield f"data: {data}\n\n"
-
-    # save conversation to database
-    if conversation_id is not None:
+def save_conversation(conversation_id, question, response, source_log_docs, llm):
+    if conversation_id is not None and conversation_id != "None":
         conversations_collection.update_one(
             {"_id": ObjectId(conversation_id)},
-            {"$push": {"queries": {"prompt": question, "response": response_full, "sources": source_log_docs}}},
+            {
+                "$push": {
+                    "queries": {
+                        "prompt": question,
+                        "response": response,
+                        "sources": source_log_docs,
+                    }
+                }
+            },
         )
 
     else:
         # create new conversation
         # generate summary
-        messages_summary = [{"role": "assistant", "content": "Summarise following conversation in no more than 3 "
-                                                             "words, respond ONLY with the summary, use the same "
-                                                             "language as the system \n\nUser: " + question + "\n\n" +
-                                                             "AI: " +
-                                                             response_full},
-                            {"role": "user", "content": "Summarise following conversation in no more than 3 words, "
-                                                        "respond ONLY with the summary, use the same language as the "
-                                                        "system"}]
+        messages_summary = [
+            {
+                "role": "assistant",
+                "content": "Summarise following conversation in no more than 3 "
+                "words, respond ONLY with the summary, use the same "
+                "language as the system \n\nUser: "
+                + question
+                + "\n\n"
+                + "AI: "
+                + response,
+            },
+            {
+                "role": "user",
+                "content": "Summarise following conversation in no more than 3 words, "
+                "respond ONLY with the summary, use the same language as the "
+                "system",
+            },
+        ]
 
-        completion = llm.gen(model=gpt_model, engine=settings.AZURE_DEPLOYMENT_NAME,
-                             messages=messages_summary, max_tokens=30)
+        completion = llm.gen(model=gpt_model, messages=messages_summary, max_tokens=30)
         conversation_id = conversations_collection.insert_one(
-            {"user": "local",
-             "date": datetime.datetime.utcnow(),
-             "name": completion,
-             "queries": [{"prompt": question, "response": response_full, "sources": source_log_docs}]}
+            {
+                "user": "local",
+                "date": datetime.datetime.utcnow(),
+                "name": completion,
+                "queries": [
+                    {
+                        "prompt": question,
+                        "response": response,
+                        "sources": source_log_docs,
+                    }
+                ],
+            }
         ).inserted_id
+    return conversation_id
+
+
+def get_prompt(prompt_id):
+    if prompt_id == "default":
+        prompt = chat_combine_template
+    elif prompt_id == "creative":
+        prompt = chat_combine_creative
+    elif prompt_id == "strict":
+        prompt = chat_combine_strict
+    else:
+        prompt = prompts_collection.find_one({"_id": ObjectId(prompt_id)})["content"]
+    return prompt
+
+
+def complete_stream(question, retriever, conversation_id, user_api_key):
+
+    response_full = ""
+    source_log_docs = []
+    answer = retriever.gen()
+    for line in answer:
+        if "answer" in line:
+            response_full += str(line["answer"])
+            data = json.dumps(line)
+            yield f"data: {data}\n\n"
+        elif "source" in line:
+            source_log_docs.append(line["source"])
+
+    llm = LLMCreator.create_llm(
+        settings.LLM_NAME, api_key=settings.API_KEY, user_api_key=user_api_key
+    )
+    conversation_id = save_conversation(
+        conversation_id, question, response_full, source_log_docs, llm
+    )
 
     # send data.type = "end" to indicate that the stream has ended as json
     data = json.dumps({"type": "id", "id": str(conversation_id)})
@@ -203,34 +212,59 @@ def stream():
         conversation_id = None
     else:
         conversation_id = data["conversation_id"]
-    if 'prompt_id' in data:
+    if "prompt_id" in data:
         prompt_id = data["prompt_id"]
     else:
-        prompt_id = 'default'
-    if 'selectedDocs' in data and data['selectedDocs'] is None:
+        prompt_id = "default"
+    if "selectedDocs" in data and data["selectedDocs"] is None:
         chunks = 0
-    elif 'chunks' in data:
+    elif "chunks" in data:
         chunks = int(data["chunks"])
     else:
         chunks = 2
 
+    prompt = get_prompt(prompt_id)
+
     # check if active_docs is set
 
     if "api_key" in data:
         data_key = get_data_from_api_key(data["api_key"])
-        vectorstore = get_vectorstore({"active_docs": data_key["source"]})
+        source = {"active_docs": data_key["source"]}
+        user_api_key = data["api_key"]
     elif "active_docs" in data:
-        vectorstore = get_vectorstore({"active_docs": data["active_docs"]})
+        source = {"active_docs": data["active_docs"]}
+        user_api_key = None
     else:
-        vectorstore = ""
-    docsearch = VectorCreator.create_vectorstore(settings.VECTOR_STORE, vectorstore, settings.EMBEDDINGS_KEY)
+        source = {}
+        user_api_key = None
+
+    if (
+        source["active_docs"].split("/")[0] == "default"
+        or source["active_docs"].split("/")[0] == "local"
+    ):
+        retriever_name = "classic"
+    else:
+        retriever_name = source["active_docs"]
+
+    retriever = RetrieverCreator.create_retriever(
+        retriever_name,
+        question=question,
+        source=source,
+        chat_history=history,
+        prompt=prompt,
+        chunks=chunks,
+        gpt_model=gpt_model,
+        user_api_key=user_api_key,
+    )
 
     return Response(
-        complete_stream(question, docsearch,
-                        chat_history=history,
-                        prompt_id=prompt_id,
-                        conversation_id=conversation_id,
-                        chunks=chunks), mimetype="text/event-stream"
+        complete_stream(
+            question=question,
+            retriever=retriever,
+            conversation_id=conversation_id,
+            user_api_key=user_api_key,
+        ),
+        mimetype="text/event-stream",
     )
 
 
@@ -247,118 +281,63 @@ def api_answer():
     else:
         conversation_id = data["conversation_id"]
     print("-" * 5)
-    if 'prompt_id' in data:
+    if "prompt_id" in data:
         prompt_id = data["prompt_id"]
     else:
-        prompt_id = 'default'
-    if 'chunks' in data:
+        prompt_id = "default"
+    if "chunks" in data:
         chunks = int(data["chunks"])
     else:
         chunks = 2
 
-    if prompt_id == 'default':
-        prompt = chat_combine_template
-    elif prompt_id == 'creative':
-        prompt = chat_combine_creative
-    elif prompt_id == 'strict':
-        prompt = chat_combine_strict
-    else:
-        prompt = prompts_collection.find_one({"_id": ObjectId(prompt_id)})["content"]
+    prompt = get_prompt(prompt_id)
 
     # use try and except  to check for exception
     try:
         # check if the vectorstore is set
         if "api_key" in data:
             data_key = get_data_from_api_key(data["api_key"])
-            vectorstore = get_vectorstore({"active_docs": data_key["source"]})
+            source = {"active_docs": data_key["source"]}
+            user_api_key = data["api_key"]
         else:
-            vectorstore = get_vectorstore(data)
-        # loading the index and the store and the prompt template
-        # Note if you have used other embeddings than OpenAI, you need to change the embeddings
-        docsearch = VectorCreator.create_vectorstore(settings.VECTOR_STORE, vectorstore, settings.EMBEDDINGS_KEY)
+            source = {data}
+            user_api_key = None
 
-
-        llm = LLMCreator.create_llm(settings.LLM_NAME, api_key=settings.API_KEY)
-
-
-
-        if chunks == 0:
-            docs = []
+        if (
+            source["active_docs"].split("/")[0] == "default"
+            or source["active_docs"].split("/")[0] == "local"
+        ):
+            retriever_name = "classic"
         else:
-            docs = docsearch.search(question, k=chunks)
-        # join all page_content together with a newline
-        docs_together = "\n".join([doc.page_content for doc in docs])
-        p_chat_combine = prompt.replace("{summaries}", docs_together)
-        messages_combine = [{"role": "system", "content": p_chat_combine}]
+            retriever_name = source["active_docs"]
+
+        retriever = RetrieverCreator.create_retriever(
+            retriever_name,
+            question=question,
+            source=source,
+            chat_history=history,
+            prompt=prompt,
+            chunks=chunks,
+            gpt_model=gpt_model,
+            user_api_key=user_api_key,
+        )
         source_log_docs = []
-        for doc in docs:
-            if doc.metadata:
-                source_log_docs.append({"title": doc.metadata['title'].split('/')[-1], "text": doc.page_content})
-            else:
-                source_log_docs.append({"title": doc.page_content, "text": doc.page_content})
-        # join all page_content together with a newline
+        response_full = ""
+        for line in retriever.gen():
+            if "source" in line:
+                source_log_docs.append(line["source"])
+            elif "answer" in line:
+                response_full += line["answer"]
 
+        llm = LLMCreator.create_llm(
+            settings.LLM_NAME, api_key=settings.API_KEY, user_api_key=user_api_key
+        )
 
-        if len(history) > 1:
-            tokens_current_history = 0
-            # count tokens in history
-            history.reverse()
-            for i in history:
-                if "prompt" in i and "response" in i:
-                    tokens_batch = count_tokens(i["prompt"]) + count_tokens(i["response"])
-                    if tokens_current_history + tokens_batch < settings.TOKENS_MAX_HISTORY:
-                        tokens_current_history += tokens_batch
-                        messages_combine.append({"role": "user", "content": i["prompt"]})
-                        messages_combine.append({"role": "system", "content": i["response"]})
-        messages_combine.append({"role": "user", "content": question})
+        result = {"answer": response_full, "sources": source_log_docs}
+        result["conversation_id"] = save_conversation(
+            conversation_id, question, response_full, source_log_docs, llm
+        )
 
-
-        completion = llm.gen(model=gpt_model, engine=settings.AZURE_DEPLOYMENT_NAME,
-                                    messages=messages_combine)
-
-
-        result = {"answer": completion, "sources": source_log_docs}
-        logger.debug(result)
-
-        # generate conversationId
-        if conversation_id is not None:
-            conversations_collection.update_one(
-                {"_id": ObjectId(conversation_id)},
-                {"$push": {"queries": {"prompt": question,
-                                       "response": result["answer"], "sources": result['sources']}}},
-            )
-
-        else:
-            # create new conversation
-            # generate summary
-            messages_summary = [
-                {"role": "assistant", "content": "Summarise following conversation in no more than 3 words, "
-                    "respond ONLY with the summary, use the same language as the system \n\n"
-                    "User: " + question + "\n\n" + "AI: " + result["answer"]},
-                {"role": "user", "content": "Summarise following conversation in no more than 3 words, "
-                    "respond ONLY with the summary, use the same language as the system"}
-            ]
-
-            completion = llm.gen(
-                model=gpt_model,
-                engine=settings.AZURE_DEPLOYMENT_NAME,
-                messages=messages_summary,
-                max_tokens=30
-            )
-            conversation_id = conversations_collection.insert_one(
-                {"user": "local",
-                "date": datetime.datetime.utcnow(),
-                "name": completion,
-                "queries": [{"prompt": question, "response": result["answer"], "sources": source_log_docs}]}
-            ).inserted_id
-
-        result["conversation_id"] = str(conversation_id)
-
-        # mock result
-        # result = {
-        #     "answer": "The answer is 42",
-        #     "sources": ["https://en.wikipedia.org/wiki/42_(number)", "https://en.wikipedia.org/wiki/42_(number)"]
-        # }
         return result
     except Exception as e:
         # print whole traceback
@@ -375,27 +354,36 @@ def api_search():
 
     if "api_key" in data:
         data_key = get_data_from_api_key(data["api_key"])
-        vectorstore = data_key["source"]
+        source = {"active_docs": data_key["source"]}
+        user_api_key = data["api_key"]
     elif "active_docs" in data:
-        vectorstore = get_vectorstore({"active_docs": data["active_docs"]})
+        source = {"active_docs": data["active_docs"]}
+        user_api_key = None
     else:
-        vectorstore = ""
-    if 'chunks' in data:
+        source = {}
+        user_api_key = None
+    if "chunks" in data:
         chunks = int(data["chunks"])
     else:
         chunks = 2
-    docsearch = VectorCreator.create_vectorstore(settings.VECTOR_STORE, vectorstore, settings.EMBEDDINGS_KEY)
-    if chunks == 0:
-        docs = []
+
+    if (
+        source["active_docs"].split("/")[0] == "default"
+        or source["active_docs"].split("/")[0] == "local"
+    ):
+        retriever_name = "classic"
     else:
-        docs = docsearch.search(question, k=chunks)
-
-    source_log_docs = []
-    for doc in docs:
-        if doc.metadata:
-            source_log_docs.append({"title": doc.metadata['title'].split('/')[-1], "text": doc.page_content})
-        else:
-            source_log_docs.append({"title": doc.page_content, "text": doc.page_content})
-        #yield f"data:{data}\n\n"
-    return source_log_docs
+        retriever_name = source["active_docs"]
 
+    retriever = RetrieverCreator.create_retriever(
+        retriever_name,
+        question=question,
+        source=source,
+        chat_history=[],
+        prompt="default",
+        chunks=chunks,
+        gpt_model=gpt_model,
+        user_api_key=user_api_key,
+    )
+    docs = retriever.search()
+    return docs

application/api/user/routes.py

@@ -1,5 +1,6 @@
 import os
 import uuid
+import shutil
 from flask import Blueprint, request, jsonify
 from urllib.parse import urlparse
 import requests
@@ -136,30 +137,43 @@ def upload_file():
         return {"status": "no name"}
     job_name = secure_filename(request.form["name"])
     # check if the post request has the file part
-    if "file" not in request.files:
-        print("No file part")
-        return {"status": "no file"}
-    file = request.files["file"]
-    if file.filename == "":
+    files = request.files.getlist("file")
+        
+    if not files or all(file.filename == '' for file in files):
         return {"status": "no file name"}
 
-    if file:
-        filename = secure_filename(file.filename)
-        # save dir
-        save_dir = os.path.join(current_dir, settings.UPLOAD_FOLDER, user, job_name)
-        # create dir if not exists
-        if not os.path.exists(save_dir):
-            os.makedirs(save_dir)
-
-        file.save(os.path.join(save_dir, filename))
-        task = ingest.delay(settings.UPLOAD_FOLDER, [".rst", ".md", ".pdf", ".txt", ".docx", 
-        ".csv", ".epub", ".html", ".mdx"],
-         job_name, filename, user)
-        # task id
-        task_id = task.id
-        return {"status": "ok", "task_id": task_id}
+    # Directory where files will be saved
+    save_dir = os.path.join(current_dir, settings.UPLOAD_FOLDER, user, job_name)
+    os.makedirs(save_dir, exist_ok=True)
+    
+    if len(files) > 1:
+        # Multiple files; prepare them for zip
+        temp_dir = os.path.join(save_dir, "temp")
+        os.makedirs(temp_dir, exist_ok=True)
+        
+        for file in files:
+            filename = secure_filename(file.filename)
+            file.save(os.path.join(temp_dir, filename))
+        
+        # Use shutil.make_archive to zip the temp directory
+        zip_path = shutil.make_archive(base_name=os.path.join(save_dir, job_name), format='zip', root_dir=temp_dir)
+        final_filename = os.path.basename(zip_path)
+        
+        # Clean up the temporary directory after zipping
+        shutil.rmtree(temp_dir)
     else:
-        return {"status": "error"}
+        # Single file
+        file = files[0]
+        final_filename = secure_filename(file.filename)
+        file_path = os.path.join(save_dir, final_filename)
+        file.save(file_path)
+    
+    # Call ingest with the single file or zipped file
+    task = ingest.delay(settings.UPLOAD_FOLDER, [".rst", ".md", ".pdf", ".txt", ".docx", 
+    ".csv", ".epub", ".html", ".mdx"],
+    job_name, final_filename, user)
+    
+    return {"status": "ok", "task_id": task.id}
     
 @user.route("/api/remote", methods=["POST"])
 def upload_remote():
@@ -237,6 +251,34 @@ def combined_json():
         for index in data_remote:
             index["location"] = "remote"
             data.append(index)
+    if 'duckduck_search' in settings.RETRIEVERS_ENABLED:
+        data.append(
+            {
+                "name": "DuckDuckGo Search",
+                "language": "en",
+                "version": "",
+                "description": "duckduck_search",
+                "fullName": "DuckDuckGo Search",
+                "date": "duckduck_search",
+                "docLink": "duckduck_search",
+                "model": settings.EMBEDDINGS_NAME,
+                "location": "custom",
+            }
+        )
+    if 'brave_search' in settings.RETRIEVERS_ENABLED:
+        data.append(
+            {
+                "name": "Brave Search",
+                "language": "en",
+                "version": "",
+                "description": "brave_search",
+                "fullName": "Brave Search",
+                "date": "brave_search",
+                "docLink": "brave_search",
+                "model": settings.EMBEDDINGS_NAME,
+                "location": "custom",
+            }
+        )
 
     return jsonify(data)
 
@@ -255,10 +297,12 @@ def check_docs():
     else:
         file_url = urlparse(base_path + vectorstore + "index.faiss")
         
-        if file_url.scheme in ['https'] and file_url.netloc == 'raw.githubusercontent.com' and file_url.path.startswith('/arc53/DocsHUB/main/'):
-            
+        if (
+            file_url.scheme in ['https'] and 
+            file_url.netloc == 'raw.githubusercontent.com' and 
+            file_url.path.startswith('/arc53/DocsHUB/main/')
+        ):
             r = requests.get(file_url.geturl())
-
             if r.status_code != 200:
                 return {"status": "null"}
             else:
@@ -267,7 +311,6 @@ def check_docs():
                 with open(vectorstore + "index.faiss", "wb") as f:
                     f.write(r.content)
 
-                # download the store
                 r = requests.get(base_path + vectorstore + "index.pkl")
                 with open(vectorstore + "index.pkl", "wb") as f:
                     f.write(r.content)

application/core/settings.py

@@ -9,7 +9,7 @@ current_dir = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__
 
 class Settings(BaseSettings):
     LLM_NAME: str = "docsgpt"
-    MODEL_NAME: Optional[str] = None # when LLM_NAME is openai, MODEL_NAME can be e.g. gpt-4-turbo-preview or gpt-3.5-turbo
+    MODEL_NAME: Optional[str] = None # if LLM_NAME is openai, MODEL_NAME can be gpt-4 or gpt-3.5-turbo
     EMBEDDINGS_NAME: str = "huggingface_sentence-transformers/all-mpnet-base-v2"
     CELERY_BROKER_URL: str = "redis://localhost:6379/0"
     CELERY_RESULT_BACKEND: str = "redis://localhost:6379/1"
@@ -18,6 +18,7 @@ class Settings(BaseSettings):
     TOKENS_MAX_HISTORY: int = 150
     UPLOAD_FOLDER: str = "inputs"
     VECTOR_STORE: str = "faiss"  # "faiss" or "elasticsearch" or "qdrant"
+    RETRIEVERS_ENABLED: list = ["classic_rag", "duckduck_search"] # also brave_search
 
     API_URL: str = "http://localhost:7091"  # backend url for celery worker
 
@@ -59,6 +60,8 @@ class Settings(BaseSettings):
     QDRANT_PATH: Optional[str] = None
     QDRANT_DISTANCE_FUNC: str = "Cosine"
 
+    BRAVE_SEARCH_API_KEY: Optional[str] = None
+
     FLASK_DEBUG_MODE: bool = False
 
 

application/llm/anthropic.py

@@ -1,21 +1,29 @@
 from application.llm.base import BaseLLM
 from application.core.settings import settings
 
+
 class AnthropicLLM(BaseLLM):
 
-    def __init__(self, api_key=None):
+    def __init__(self, api_key=None, user_api_key=None, *args, **kwargs):
         from anthropic import Anthropic, HUMAN_PROMPT, AI_PROMPT
-        self.api_key = api_key or settings.ANTHROPIC_API_KEY  # If not provided, use a default from settings
+
+        super().__init__(*args, **kwargs)
+        self.api_key = (
+            api_key or settings.ANTHROPIC_API_KEY
+        )  # If not provided, use a default from settings
+        self.user_api_key = user_api_key
         self.anthropic = Anthropic(api_key=self.api_key)
         self.HUMAN_PROMPT = HUMAN_PROMPT
         self.AI_PROMPT = AI_PROMPT
 
-    def gen(self, model, messages, engine=None, max_tokens=300, stream=False, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+    def _raw_gen(
+        self, baseself, model, messages, stream=False, max_tokens=300, **kwargs
+    ):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Context \n {context} \n ### Question \n {user_question}"
         if stream:
-            return self.gen_stream(model, prompt, max_tokens, **kwargs)
+            return self.gen_stream(model, prompt, stream, max_tokens, **kwargs)
 
         completion = self.anthropic.completions.create(
             model=model,
@@ -25,9 +33,11 @@ class AnthropicLLM(BaseLLM):
         )
         return completion.completion
 
-    def gen_stream(self, model, messages, engine=None, max_tokens=300, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+    def _raw_gen_stream(
+        self, baseself, model, messages, stream=True, max_tokens=300, **kwargs
+    ):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Context \n {context} \n ### Question \n {user_question}"
         stream_response = self.anthropic.completions.create(
             model=model,
@@ -37,4 +47,4 @@ class AnthropicLLM(BaseLLM):
         )
 
         for completion in stream_response:
-            yield completion.completion
+            yield completion.completion

application/llm/base.py

@@ -1,14 +1,28 @@
 from abc import ABC, abstractmethod
+from application.usage import gen_token_usage, stream_token_usage
 
 
 class BaseLLM(ABC):
     def __init__(self):
-        pass
+        self.token_usage = {"prompt_tokens": 0, "generated_tokens": 0}
+
+    def _apply_decorator(self, method, decorator, *args, **kwargs):
+        return decorator(method, *args, **kwargs)
 
     @abstractmethod
-    def gen(self, *args, **kwargs):
+    def _raw_gen(self, model, messages, stream, *args, **kwargs):
         pass
 
+    def gen(self, model, messages, stream=False, *args, **kwargs):
+        return self._apply_decorator(self._raw_gen, gen_token_usage)(
+            self, model=model, messages=messages, stream=stream, *args, **kwargs
+        )
+
     @abstractmethod
-    def gen_stream(self, *args, **kwargs):
+    def _raw_gen_stream(self, model, messages, stream, *args, **kwargs):
         pass
+
+    def gen_stream(self, model, messages, stream=True, *args, **kwargs):
+        return self._apply_decorator(self._raw_gen_stream, stream_token_usage)(
+            self, model=model, messages=messages, stream=stream, *args, **kwargs
+        )

application/llm/docsgpt_provider.py

@@ -2,48 +2,43 @@ from application.llm.base import BaseLLM
 import json
 import requests
 
+
 class DocsGPTAPILLM(BaseLLM):
 
-    def __init__(self, *args, **kwargs):
-        self.endpoint =  "https://llm.docsgpt.co.uk"
+    def __init__(self, api_key=None, user_api_key=None, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.api_key = api_key
+        self.user_api_key = user_api_key
+        self.endpoint = "https://llm.docsgpt.co.uk"
 
-
-    def gen(self, model, engine, messages, stream=False, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+    def _raw_gen(self, baseself, model, messages, stream=False, *args, **kwargs):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Instruction \n {user_question} \n ### Context \n {context} \n ### Answer \n"
 
         response = requests.post(
-            f"{self.endpoint}/answer",
-            json={
-                "prompt": prompt,
-                "max_new_tokens": 30
-            }
+            f"{self.endpoint}/answer", json={"prompt": prompt, "max_new_tokens": 30}
         )
-        response_clean = response.json()['a'].replace("###", "")
+        response_clean = response.json()["a"].replace("###", "")
 
         return response_clean
 
-    def gen_stream(self, model, engine, messages, stream=True, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+    def _raw_gen_stream(self, baseself, model, messages, stream=True, *args, **kwargs):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Instruction \n {user_question} \n ### Context \n {context} \n ### Answer \n"
 
         # send prompt to endpoint /stream
         response = requests.post(
             f"{self.endpoint}/stream",
-            json={
-                "prompt": prompt,
-                "max_new_tokens": 256
-            },
-            stream=True
+            json={"prompt": prompt, "max_new_tokens": 256},
+            stream=True,
         )
-    
+
         for line in response.iter_lines():
             if line:
-                #data = json.loads(line)
-                data_str = line.decode('utf-8')
+                # data = json.loads(line)
+                data_str = line.decode("utf-8")
                 if data_str.startswith("data: "):
                     data = json.loads(data_str[6:])
-                    yield data['a']
-                    
+                    yield data["a"]

application/llm/huggingface.py

@@ -1,44 +1,68 @@
 from application.llm.base import BaseLLM
 
+
 class HuggingFaceLLM(BaseLLM):
 
-    def __init__(self, api_key, llm_name='Arc53/DocsGPT-7B',q=False):
+    def __init__(
+        self,
+        api_key=None,
+        user_api_key=None,
+        llm_name="Arc53/DocsGPT-7B",
+        q=False,
+        *args,
+        **kwargs,
+    ):
         global hf
-        
+
         from langchain.llms import HuggingFacePipeline
+
         if q:
             import torch
-            from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline, BitsAndBytesConfig
+            from transformers import (
+                AutoModelForCausalLM,
+                AutoTokenizer,
+                pipeline,
+                BitsAndBytesConfig,
+            )
+
             tokenizer = AutoTokenizer.from_pretrained(llm_name)
             bnb_config = BitsAndBytesConfig(
-                            load_in_4bit=True,
-                            bnb_4bit_use_double_quant=True,
-                            bnb_4bit_quant_type="nf4",
-                            bnb_4bit_compute_dtype=torch.bfloat16
-                        )
-            model = AutoModelForCausalLM.from_pretrained(llm_name,quantization_config=bnb_config)
+                load_in_4bit=True,
+                bnb_4bit_use_double_quant=True,
+                bnb_4bit_quant_type="nf4",
+                bnb_4bit_compute_dtype=torch.bfloat16,
+            )
+            model = AutoModelForCausalLM.from_pretrained(
+                llm_name, quantization_config=bnb_config
+            )
         else:
             from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
+
             tokenizer = AutoTokenizer.from_pretrained(llm_name)
             model = AutoModelForCausalLM.from_pretrained(llm_name)
-        
+
+        super().__init__(*args, **kwargs)
+        self.api_key = api_key
+        self.user_api_key = user_api_key
         pipe = pipeline(
-            "text-generation", model=model,
-            tokenizer=tokenizer, max_new_tokens=2000,
-            device_map="auto", eos_token_id=tokenizer.eos_token_id
+            "text-generation",
+            model=model,
+            tokenizer=tokenizer,
+            max_new_tokens=2000,
+            device_map="auto",
+            eos_token_id=tokenizer.eos_token_id,
         )
         hf = HuggingFacePipeline(pipeline=pipe)
 
-    def gen(self, model, engine, messages, stream=False, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+    def _raw_gen(self, baseself, model, messages, stream=False, **kwargs):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Instruction \n {user_question} \n ### Context \n {context} \n ### Answer \n"
 
         result = hf(prompt)
 
         return result.content
 
-    def gen_stream(self, model, engine, messages, stream=True, **kwargs):
+    def _raw_gen_stream(self, baseself, model, messages, stream=True, **kwargs):
 
         raise NotImplementedError("HuggingFaceLLM Streaming is not implemented yet.")
-

application/llm/llama_cpp.py

@@ -1,32 +1,45 @@
 from application.llm.base import BaseLLM
 from application.core.settings import settings
 
+
 class LlamaCpp(BaseLLM):
 
-    def __init__(self, api_key, llm_name=settings.MODEL_PATH, **kwargs):
+    def __init__(
+        self,
+        api_key=None,
+        user_api_key=None,
+        llm_name=settings.MODEL_PATH,
+        *args,
+        **kwargs,
+    ):
         global llama
         try:
             from llama_cpp import Llama
         except ImportError:
-            raise ImportError("Please install llama_cpp using pip install llama-cpp-python")
+            raise ImportError(
+                "Please install llama_cpp using pip install llama-cpp-python"
+            )
 
+        super().__init__(*args, **kwargs)
+        self.api_key = api_key
+        self.user_api_key = user_api_key
         llama = Llama(model_path=llm_name, n_ctx=2048)
 
-    def gen(self, model, engine, messages, stream=False, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+    def _raw_gen(self, baseself, model, messages, stream=False, **kwargs):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Instruction \n {user_question} \n ### Context \n {context} \n ### Answer \n"
 
         result = llama(prompt, max_tokens=150, echo=False)
 
         # import sys
         # print(result['choices'][0]['text'].split('### Answer \n')[-1], file=sys.stderr)
-        
-        return result['choices'][0]['text'].split('### Answer \n')[-1]
 
-    def gen_stream(self, model, engine, messages, stream=True, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+        return result["choices"][0]["text"].split("### Answer \n")[-1]
+
+    def _raw_gen_stream(self, baseself, model, messages, stream=True, **kwargs):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Instruction \n {user_question} \n ### Context \n {context} \n ### Answer \n"
 
         result = llama(prompt, max_tokens=150, echo=False, stream=stream)
@@ -35,5 +48,5 @@ class LlamaCpp(BaseLLM):
         # print(list(result), file=sys.stderr)
 
         for item in result:
-            for choice in item['choices']:
-                yield choice['text']
+            for choice in item["choices"]:
+                yield choice["text"]

application/llm/llm_creator.py

@@ -7,22 +7,21 @@ from application.llm.docsgpt_provider import DocsGPTAPILLM
 from application.llm.premai import PremAILLM
 
 
-
 class LLMCreator:
     llms = {
-        'openai': OpenAILLM,
-        'azure_openai': AzureOpenAILLM,
-        'sagemaker': SagemakerAPILLM,
-        'huggingface': HuggingFaceLLM,
-        'llama.cpp': LlamaCpp,
-        'anthropic': AnthropicLLM,
-        'docsgpt': DocsGPTAPILLM,
-        'premai': PremAILLM,
+        "openai": OpenAILLM,
+        "azure_openai": AzureOpenAILLM,
+        "sagemaker": SagemakerAPILLM,
+        "huggingface": HuggingFaceLLM,
+        "llama.cpp": LlamaCpp,
+        "anthropic": AnthropicLLM,
+        "docsgpt": DocsGPTAPILLM,
+        "premai": PremAILLM,
     }
 
     @classmethod
-    def create_llm(cls, type, *args, **kwargs):
+    def create_llm(cls, type, api_key, user_api_key, *args, **kwargs):
         llm_class = cls.llms.get(type.lower())
         if not llm_class:
             raise ValueError(f"No LLM class found for type {type}")
-        return llm_class(*args, **kwargs)
+        return llm_class(api_key, user_api_key, *args, **kwargs)

application/llm/openai.py

@@ -1,36 +1,53 @@
 from application.llm.base import BaseLLM
 from application.core.settings import settings
 
+
 class OpenAILLM(BaseLLM):
 
-    def __init__(self, api_key):
+    def __init__(self, api_key=None, user_api_key=None, *args, **kwargs):
         global openai
         from openai import OpenAI
-        
+
+        super().__init__(*args, **kwargs)
         self.client = OpenAI(
-                api_key=api_key, 
-            )
+            api_key=api_key,
+        )
         self.api_key = api_key
+        self.user_api_key = user_api_key
 
     def _get_openai(self):
         # Import openai when needed
         import openai
-        
+
         return openai
 
-    def gen(self, model, engine, messages, stream=False, **kwargs):
-        response = self.client.chat.completions.create(model=model,
-            messages=messages,
-            stream=stream,
-            **kwargs)
+    def _raw_gen(
+        self,
+        baseself,
+        model,
+        messages,
+        stream=False,
+        engine=settings.AZURE_DEPLOYMENT_NAME,
+        **kwargs
+    ):
+        response = self.client.chat.completions.create(
+            model=model, messages=messages, stream=stream, **kwargs
+        )
 
         return response.choices[0].message.content
 
-    def gen_stream(self, model, engine, messages, stream=True, **kwargs):
-        response = self.client.chat.completions.create(model=model,
-            messages=messages,
-            stream=stream,
-            **kwargs)
+    def _raw_gen_stream(
+        self,
+        baseself,
+        model,
+        messages,
+        stream=True,
+        engine=settings.AZURE_DEPLOYMENT_NAME,
+        **kwargs
+    ):
+        response = self.client.chat.completions.create(
+            model=model, messages=messages, stream=stream, **kwargs
+        )
 
         for line in response:
             # import sys
@@ -41,14 +58,17 @@ class OpenAILLM(BaseLLM):
 
 class AzureOpenAILLM(OpenAILLM):
 
-    def __init__(self, openai_api_key, openai_api_base, openai_api_version, deployment_name):
+    def __init__(
+        self, openai_api_key, openai_api_base, openai_api_version, deployment_name
+    ):
         super().__init__(openai_api_key)
-        self.api_base = settings.OPENAI_API_BASE,
-        self.api_version = settings.OPENAI_API_VERSION,
-        self.deployment_name = settings.AZURE_DEPLOYMENT_NAME,
+        self.api_base = (settings.OPENAI_API_BASE,)
+        self.api_version = (settings.OPENAI_API_VERSION,)
+        self.deployment_name = (settings.AZURE_DEPLOYMENT_NAME,)
         from openai import AzureOpenAI
+
         self.client = AzureOpenAI(
-            api_key=openai_api_key,  
+            api_key=openai_api_key,
             api_version=settings.OPENAI_API_VERSION,
             api_base=settings.OPENAI_API_BASE,
             deployment_name=settings.AZURE_DEPLOYMENT_NAME,

application/llm/premai.py

@@ -1,32 +1,37 @@
 from application.llm.base import BaseLLM
 from application.core.settings import settings
 
+
 class PremAILLM(BaseLLM):
 
-    def __init__(self, api_key):
+    def __init__(self, api_key=None, user_api_key=None, *args, **kwargs):
         from premai import Prem
-        
-        self.client = Prem(
-            api_key=api_key
-        )
+
+        super().__init__(*args, **kwargs)
+        self.client = Prem(api_key=api_key)
         self.api_key = api_key
+        self.user_api_key = user_api_key
         self.project_id = settings.PREMAI_PROJECT_ID
 
-    def gen(self, model, engine, messages, stream=False, **kwargs):
-        response = self.client.chat.completions.create(model=model,
+    def _raw_gen(self, baseself, model, messages, stream=False, **kwargs):
+        response = self.client.chat.completions.create(
+            model=model,
             project_id=self.project_id,
             messages=messages,
             stream=stream,
-            **kwargs)
+            **kwargs
+        )
 
         return response.choices[0].message["content"]
 
-    def gen_stream(self, model, engine, messages, stream=True, **kwargs):
-        response = self.client.chat.completions.create(model=model,
+    def _raw_gen_stream(self, baseself, model, messages, stream=True, **kwargs):
+        response = self.client.chat.completions.create(
+            model=model,
             project_id=self.project_id,
             messages=messages,
             stream=stream,
-            **kwargs)
+            **kwargs
+        )
 
         for line in response:
             if line.choices[0].delta["content"] is not None:

application/llm/sagemaker.py

@@ -4,11 +4,10 @@ import json
 import io
 
 
-
 class LineIterator:
     """
-    A helper class for parsing the byte stream input. 
-    
+    A helper class for parsing the byte stream input.
+
     The output of the model will be in the following format:
     ```
     b'{"outputs": [" a"]}\n'
@@ -16,21 +15,21 @@ class LineIterator:
     b'{"outputs": [" problem"]}\n'
     ...
     ```
-    
-    While usually each PayloadPart event from the event stream will contain a byte array 
+
+    While usually each PayloadPart event from the event stream will contain a byte array
     with a full json, this is not guaranteed and some of the json objects may be split across
     PayloadPart events. For example:
     ```
     {'PayloadPart': {'Bytes': b'{"outputs": '}}
     {'PayloadPart': {'Bytes': b'[" problem"]}\n'}}
     ```
-    
+
     This class accounts for this by concatenating bytes written via the 'write' function
     and then exposing a method which will return lines (ending with a '\n' character) within
-    the buffer via the 'scan_lines' function. It maintains the position of the last read 
-    position to ensure that previous bytes are not exposed again. 
+    the buffer via the 'scan_lines' function. It maintains the position of the last read
+    position to ensure that previous bytes are not exposed again.
     """
-    
+
     def __init__(self, stream):
         self.byte_iterator = iter(stream)
         self.buffer = io.BytesIO()
@@ -43,7 +42,7 @@ class LineIterator:
         while True:
             self.buffer.seek(self.read_pos)
             line = self.buffer.readline()
-            if line and line[-1] == ord('\n'):
+            if line and line[-1] == ord("\n"):
                 self.read_pos += len(line)
                 return line[:-1]
             try:
@@ -52,33 +51,35 @@ class LineIterator:
                 if self.read_pos < self.buffer.getbuffer().nbytes:
                     continue
                 raise
-            if 'PayloadPart' not in chunk:
-                print('Unknown event type:' + chunk)
+            if "PayloadPart" not in chunk:
+                print("Unknown event type:" + chunk)
                 continue
             self.buffer.seek(0, io.SEEK_END)
-            self.buffer.write(chunk['PayloadPart']['Bytes'])
+            self.buffer.write(chunk["PayloadPart"]["Bytes"])
+
 
 class SagemakerAPILLM(BaseLLM):
 
-    def __init__(self, *args, **kwargs):
+    def __init__(self, api_key=None, user_api_key=None, *args, **kwargs):
         import boto3
+
         runtime = boto3.client(
-            'runtime.sagemaker',
-            aws_access_key_id='xxx',
-            aws_secret_access_key='xxx',
-            region_name='us-west-2'
+            "runtime.sagemaker",
+            aws_access_key_id="xxx",
+            aws_secret_access_key="xxx",
+            region_name="us-west-2",
         )
 
-        
-        self.endpoint =  settings.SAGEMAKER_ENDPOINT
+        super().__init__(*args, **kwargs)
+        self.api_key = api_key
+        self.user_api_key = user_api_key
+        self.endpoint = settings.SAGEMAKER_ENDPOINT
         self.runtime = runtime
 
-
-    def gen(self, model, engine, messages, stream=False, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+    def _raw_gen(self, baseself, model, messages, stream=False, **kwargs):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Instruction \n {user_question} \n ### Context \n {context} \n ### Answer \n"
-    
 
         # Construct payload for endpoint
         payload = {
@@ -89,25 +90,25 @@ class SagemakerAPILLM(BaseLLM):
                 "temperature": 0.1,
                 "max_new_tokens": 30,
                 "repetition_penalty": 1.03,
-                "stop": ["</s>", "###"]
-            }
+                "stop": ["</s>", "###"],
+            },
         }
-        body_bytes = json.dumps(payload).encode('utf-8')
+        body_bytes = json.dumps(payload).encode("utf-8")
 
         # Invoke the endpoint
-        response = self.runtime.invoke_endpoint(EndpointName=self.endpoint,
-                                        ContentType='application/json',
-                                        Body=body_bytes)
-        result = json.loads(response['Body'].read().decode())
+        response = self.runtime.invoke_endpoint(
+            EndpointName=self.endpoint, ContentType="application/json", Body=body_bytes
+        )
+        result = json.loads(response["Body"].read().decode())
         import sys
-        print(result[0]['generated_text'], file=sys.stderr)
-        return result[0]['generated_text'][len(prompt):]
 
-    def gen_stream(self, model, engine, messages, stream=True, **kwargs):
-        context = messages[0]['content']
-        user_question = messages[-1]['content']
+        print(result[0]["generated_text"], file=sys.stderr)
+        return result[0]["generated_text"][len(prompt) :]
+
+    def _raw_gen_stream(self, baseself, model, messages, stream=True, **kwargs):
+        context = messages[0]["content"]
+        user_question = messages[-1]["content"]
         prompt = f"### Instruction \n {user_question} \n ### Context \n {context} \n ### Answer \n"
-    
 
         # Construct payload for endpoint
         payload = {
@@ -118,22 +119,22 @@ class SagemakerAPILLM(BaseLLM):
                 "temperature": 0.1,
                 "max_new_tokens": 512,
                 "repetition_penalty": 1.03,
-                "stop": ["</s>", "###"]
-            }
+                "stop": ["</s>", "###"],
+            },
         }
-        body_bytes = json.dumps(payload).encode('utf-8')
+        body_bytes = json.dumps(payload).encode("utf-8")
 
         # Invoke the endpoint
-        response = self.runtime.invoke_endpoint_with_response_stream(EndpointName=self.endpoint,
-                                        ContentType='application/json',
-                                        Body=body_bytes)
-        #result = json.loads(response['Body'].read().decode())
-        event_stream = response['Body']
-        start_json = b'{'
+        response = self.runtime.invoke_endpoint_with_response_stream(
+            EndpointName=self.endpoint, ContentType="application/json", Body=body_bytes
+        )
+        # result = json.loads(response['Body'].read().decode())
+        event_stream = response["Body"]
+        start_json = b"{"
         for line in LineIterator(event_stream):
-            if line != b'' and start_json in line:
-                #print(line)
-                data = json.loads(line[line.find(start_json):].decode('utf-8'))
-                if data['token']['text'] not in ["</s>", "###"]:
-                    print(data['token']['text'],end='')
-                    yield data['token']['text']
+            if line != b"" and start_json in line:
+                # print(line)
+                data = json.loads(line[line.find(start_json) :].decode("utf-8"))
+                if data["token"]["text"] not in ["</s>", "###"]:
+                    print(data["token"]["text"], end="")
+                    yield data["token"]["text"]

application/parser/token_func.py

@@ -22,7 +22,10 @@ def group_documents(documents: List[Document], min_tokens: int, max_tokens: int)
         doc_len = len(tiktoken.get_encoding("cl100k_base").encode(doc.text))
 
         # Check if current group is empty or if the document can be added based on token count and matching metadata
-        if current_group is None or (len(tiktoken.get_encoding("cl100k_base").encode(current_group.text)) + doc_len < max_tokens and doc_len < min_tokens and current_group.extra_info == doc.extra_info):
+        if (current_group is None or 
+            (len(tiktoken.get_encoding("cl100k_base").encode(current_group.text)) + doc_len < max_tokens and 
+             doc_len < min_tokens and 
+             current_group.extra_info == doc.extra_info)):
             if current_group is None:
                 current_group = doc  # Use the document directly to retain its metadata
             else:

application/requirements.txt

@@ -3,6 +3,7 @@ boto3==1.34.6
 celery==5.3.6
 dataclasses_json==0.6.3
 docx2txt==0.8
+duckduckgo-search==5.3.0
 EbookLib==0.18
 elasticsearch==8.12.0
 escodegen==1.0.11
@@ -18,10 +19,10 @@ nltk==3.8.1
 openapi3_parser==1.1.16
 pandas==2.2.0
 pydantic_settings==2.1.0
-pymongo==4.6.1
+pymongo==4.6.3
 PyPDF2==3.0.1
 python-dotenv==1.0.1
-qdrant-client==1.7.3
+qdrant-client==1.8.2
 redis==5.0.1
 Requests==2.31.0
 retry==0.9.2

application/retriever/base.py

@@ -0,0 +1,14 @@
+from abc import ABC, abstractmethod
+
+
+class BaseRetriever(ABC):
+    def __init__(self):
+        pass
+
+    @abstractmethod
+    def gen(self, *args, **kwargs):
+        pass
+
+    @abstractmethod
+    def search(self, *args, **kwargs):
+        pass

application/retriever/brave_search.py

@@ -0,0 +1,95 @@
+import json
+from application.retriever.base import BaseRetriever
+from application.core.settings import settings
+from application.llm.llm_creator import LLMCreator
+from application.utils import count_tokens
+from langchain_community.tools import BraveSearch
+
+
+class BraveRetSearch(BaseRetriever):
+
+    def __init__(
+        self,
+        question,
+        source,
+        chat_history,
+        prompt,
+        chunks=2,
+        gpt_model="docsgpt",
+        user_api_key=None,
+    ):
+        self.question = question
+        self.source = source
+        self.chat_history = chat_history
+        self.prompt = prompt
+        self.chunks = chunks
+        self.gpt_model = gpt_model
+        self.user_api_key = user_api_key
+
+    def _get_data(self):
+        if self.chunks == 0:
+            docs = []
+        else:
+            search = BraveSearch.from_api_key(
+                api_key=settings.BRAVE_SEARCH_API_KEY,
+                search_kwargs={"count": int(self.chunks)},
+            )
+            results = search.run(self.question)
+            results = json.loads(results)
+
+            docs = []
+            for i in results:
+                try:
+                    title = i["title"]
+                    link = i["link"]
+                    snippet = i["snippet"]
+                    docs.append({"text": snippet, "title": title, "link": link})
+                except IndexError:
+                    pass
+        if settings.LLM_NAME == "llama.cpp":
+            docs = [docs[0]]
+
+        return docs
+
+    def gen(self):
+        docs = self._get_data()
+
+        # join all page_content together with a newline
+        docs_together = "\n".join([doc["text"] for doc in docs])
+        p_chat_combine = self.prompt.replace("{summaries}", docs_together)
+        messages_combine = [{"role": "system", "content": p_chat_combine}]
+        for doc in docs:
+            yield {"source": doc}
+
+        if len(self.chat_history) > 1:
+            tokens_current_history = 0
+            # count tokens in history
+            self.chat_history.reverse()
+            for i in self.chat_history:
+                if "prompt" in i and "response" in i:
+                    tokens_batch = count_tokens(i["prompt"]) + count_tokens(
+                        i["response"]
+                    )
+                    if (
+                        tokens_current_history + tokens_batch
+                        < settings.TOKENS_MAX_HISTORY
+                    ):
+                        tokens_current_history += tokens_batch
+                        messages_combine.append(
+                            {"role": "user", "content": i["prompt"]}
+                        )
+                        messages_combine.append(
+                            {"role": "system", "content": i["response"]}
+                        )
+        messages_combine.append({"role": "user", "content": self.question})
+
+        llm = LLMCreator.create_llm(
+            settings.LLM_NAME, api_key=settings.API_KEY, user_api_key=self.user_api_key
+        )
+
+        completion = llm.gen_stream(model=self.gpt_model, messages=messages_combine)
+        for line in completion:
+            yield {"answer": str(line)}
+
+    def search(self):
+        return self._get_data()

application/retriever/classic_rag.py

@@ -0,0 +1,110 @@
+import os
+from application.retriever.base import BaseRetriever
+from application.core.settings import settings
+from application.vectorstore.vector_creator import VectorCreator
+from application.llm.llm_creator import LLMCreator
+
+from application.utils import count_tokens
+
+
+class ClassicRAG(BaseRetriever):
+
+    def __init__(
+        self,
+        question,
+        source,
+        chat_history,
+        prompt,
+        chunks=2,
+        gpt_model="docsgpt",
+        user_api_key=None,
+    ):
+        self.question = question
+        self.vectorstore = self._get_vectorstore(source=source)
+        self.chat_history = chat_history
+        self.prompt = prompt
+        self.chunks = chunks
+        self.gpt_model = gpt_model
+        self.user_api_key = user_api_key
+
+    def _get_vectorstore(self, source):
+        if "active_docs" in source:
+            if source["active_docs"].split("/")[0] == "default":
+                vectorstore = ""
+            elif source["active_docs"].split("/")[0] == "local":
+                vectorstore = "indexes/" + source["active_docs"]
+            else:
+                vectorstore = "vectors/" + source["active_docs"]
+            if source["active_docs"] == "default":
+                vectorstore = ""
+        else:
+            vectorstore = ""
+        vectorstore = os.path.join("application", vectorstore)
+        return vectorstore
+
+    def _get_data(self):
+        if self.chunks == 0:
+            docs = []
+        else:
+            docsearch = VectorCreator.create_vectorstore(
+                settings.VECTOR_STORE, self.vectorstore, settings.EMBEDDINGS_KEY
+            )
+            docs_temp = docsearch.search(self.question, k=self.chunks)
+            docs = [
+                {
+                    "title": (
+                        i.metadata["title"].split("/")[-1]
+                        if i.metadata
+                        else i.page_content
+                    ),
+                    "text": i.page_content,
+                }
+                for i in docs_temp
+            ]
+        if settings.LLM_NAME == "llama.cpp":
+            docs = [docs[0]]
+
+        return docs
+
+    def gen(self):
+        docs = self._get_data()
+
+        # join all page_content together with a newline
+        docs_together = "\n".join([doc["text"] for doc in docs])
+        p_chat_combine = self.prompt.replace("{summaries}", docs_together)
+        messages_combine = [{"role": "system", "content": p_chat_combine}]
+        for doc in docs:
+            yield {"source": doc}
+
+        if len(self.chat_history) > 1:
+            tokens_current_history = 0
+            # count tokens in history
+            self.chat_history.reverse()
+            for i in self.chat_history:
+                if "prompt" in i and "response" in i:
+                    tokens_batch = count_tokens(i["prompt"]) + count_tokens(
+                        i["response"]
+                    )
+                    if (
+                        tokens_current_history + tokens_batch
+                        < settings.TOKENS_MAX_HISTORY
+                    ):
+                        tokens_current_history += tokens_batch
+                        messages_combine.append(
+                            {"role": "user", "content": i["prompt"]}
+                        )
+                        messages_combine.append(
+                            {"role": "system", "content": i["response"]}
+                        )
+        messages_combine.append({"role": "user", "content": self.question})
+
+        llm = LLMCreator.create_llm(
+            settings.LLM_NAME, api_key=settings.API_KEY, user_api_key=self.user_api_key
+        )
+
+        completion = llm.gen_stream(model=self.gpt_model, messages=messages_combine)
+        for line in completion:
+            yield {"answer": str(line)}
+
+    def search(self):
+        return self._get_data()

application/retriever/duckduck_search.py

@@ -0,0 +1,112 @@
+from application.retriever.base import BaseRetriever
+from application.core.settings import settings
+from application.llm.llm_creator import LLMCreator
+from application.utils import count_tokens
+from langchain_community.tools import DuckDuckGoSearchResults
+from langchain_community.utilities import DuckDuckGoSearchAPIWrapper
+
+
+class DuckDuckSearch(BaseRetriever):
+
+    def __init__(
+        self,
+        question,
+        source,
+        chat_history,
+        prompt,
+        chunks=2,
+        gpt_model="docsgpt",
+        user_api_key=None,
+    ):
+        self.question = question
+        self.source = source
+        self.chat_history = chat_history
+        self.prompt = prompt
+        self.chunks = chunks
+        self.gpt_model = gpt_model
+        self.user_api_key = user_api_key
+
+    def _parse_lang_string(self, input_string):
+        result = []
+        current_item = ""
+        inside_brackets = False
+        for char in input_string:
+            if char == "[":
+                inside_brackets = True
+            elif char == "]":
+                inside_brackets = False
+                result.append(current_item)
+                current_item = ""
+            elif inside_brackets:
+                current_item += char
+
+        if inside_brackets:
+            result.append(current_item)
+
+        return result
+
+    def _get_data(self):
+        if self.chunks == 0:
+            docs = []
+        else:
+            wrapper = DuckDuckGoSearchAPIWrapper(max_results=self.chunks)
+            search = DuckDuckGoSearchResults(api_wrapper=wrapper)
+            results = search.run(self.question)
+            results = self._parse_lang_string(results)
+
+            docs = []
+            for i in results:
+                try:
+                    text = i.split("title:")[0]
+                    title = i.split("title:")[1].split("link:")[0]
+                    link = i.split("link:")[1]
+                    docs.append({"text": text, "title": title, "link": link})
+                except IndexError:
+                    pass
+        if settings.LLM_NAME == "llama.cpp":
+            docs = [docs[0]]
+
+        return docs
+
+    def gen(self):
+        docs = self._get_data()
+
+        # join all page_content together with a newline
+        docs_together = "\n".join([doc["text"] for doc in docs])
+        p_chat_combine = self.prompt.replace("{summaries}", docs_together)
+        messages_combine = [{"role": "system", "content": p_chat_combine}]
+        for doc in docs:
+            yield {"source": doc}
+
+        if len(self.chat_history) > 1:
+            tokens_current_history = 0
+            # count tokens in history
+            self.chat_history.reverse()
+            for i in self.chat_history:
+                if "prompt" in i and "response" in i:
+                    tokens_batch = count_tokens(i["prompt"]) + count_tokens(
+                        i["response"]
+                    )
+                    if (
+                        tokens_current_history + tokens_batch
+                        < settings.TOKENS_MAX_HISTORY
+                    ):
+                        tokens_current_history += tokens_batch
+                        messages_combine.append(
+                            {"role": "user", "content": i["prompt"]}
+                        )
+                        messages_combine.append(
+                            {"role": "system", "content": i["response"]}
+                        )
+        messages_combine.append({"role": "user", "content": self.question})
+
+        llm = LLMCreator.create_llm(
+            settings.LLM_NAME, api_key=settings.API_KEY, user_api_key=self.user_api_key
+        )
+
+        completion = llm.gen_stream(model=self.gpt_model, messages=messages_combine)
+        for line in completion:
+            yield {"answer": str(line)}
+
+    def search(self):
+        return self._get_data()

application/retriever/retriever_creator.py

@@ -0,0 +1,19 @@
+from application.retriever.classic_rag import ClassicRAG
+from application.retriever.duckduck_search import DuckDuckSearch
+from application.retriever.brave_search import BraveRetSearch
+
+
+
+class RetrieverCreator:
+    retievers = {
+        'classic': ClassicRAG,
+        'duckduck_search': DuckDuckSearch,
+        'brave_search': BraveRetSearch
+    }
+
+    @classmethod
+    def create_retriever(cls, type, *args, **kwargs):
+        retiever_class = cls.retievers.get(type.lower())
+        if not retiever_class:
+            raise ValueError(f"No retievers class found for type {type}")
+        return retiever_class(*args, **kwargs)

application/usage.py

@@ -0,0 +1,49 @@
+import sys
+from pymongo import MongoClient
+from datetime import datetime
+from application.core.settings import settings
+from application.utils import count_tokens
+
+mongo = MongoClient(settings.MONGO_URI)
+db = mongo["docsgpt"]
+usage_collection = db["token_usage"]
+
+
+def update_token_usage(user_api_key, token_usage):
+    if "pytest" in sys.modules:
+        return
+    usage_data = {
+        "api_key": user_api_key,
+        "prompt_tokens": token_usage["prompt_tokens"],
+        "generated_tokens": token_usage["generated_tokens"],
+        "timestamp": datetime.now(),
+    }
+    usage_collection.insert_one(usage_data)
+
+
+def gen_token_usage(func):
+    def wrapper(self, model, messages, stream, **kwargs):
+        for message in messages:
+            self.token_usage["prompt_tokens"] += count_tokens(message["content"])
+        result = func(self, model, messages, stream, **kwargs)
+        self.token_usage["generated_tokens"] += count_tokens(result)
+        update_token_usage(self.user_api_key, self.token_usage)
+        return result
+
+    return wrapper
+
+
+def stream_token_usage(func):
+    def wrapper(self, model, messages, stream, **kwargs):
+        for message in messages:
+            self.token_usage["prompt_tokens"] += count_tokens(message["content"])
+        batch = []
+        result = func(self, model, messages, stream, **kwargs)
+        for r in result:
+            batch.append(r)
+            yield r
+        for line in batch:
+            self.token_usage["generated_tokens"] += count_tokens(line)
+        update_token_usage(self.user_api_key, self.token_usage)
+
+    return wrapper

application/utils.py

@@ -0,0 +1,6 @@
+from transformers import GPT2TokenizerFast
+
+
+def count_tokens(string):
+    tokenizer = GPT2TokenizerFast.from_pretrained('gpt2')
+    return len(tokenizer(string)['input_ids'])

application/worker.py

@@ -36,6 +36,32 @@ current_dir = os.path.dirname(
     os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 )
 
+def extract_zip_recursive(zip_path, extract_to, current_depth=0, max_depth=5):
+    """
+    Recursively extract zip files with a limit on recursion depth.
+
+    Args:
+        zip_path (str): Path to the zip file to be extracted.
+        extract_to (str): Destination path for extracted files.
+        current_depth (int): Current depth of recursion.
+        max_depth (int): Maximum allowed depth of recursion to prevent infinite loops.
+    """
+    if current_depth > max_depth:
+        print(f"Reached maximum recursion depth of {max_depth}")
+        return
+
+    with zipfile.ZipFile(zip_path, 'r') as zip_ref:
+        zip_ref.extractall(extract_to)
+    os.remove(zip_path)  # Remove the zip file after extracting
+
+    # Check for nested zip files and extract them
+    for root, dirs, files in os.walk(extract_to):
+        for file in files:
+            if file.endswith(".zip"):
+                # If a nested zip file is found, extract it recursively
+                file_path = os.path.join(root, file)
+                extract_zip_recursive(file_path, root, current_depth + 1, max_depth)
+
 
 # Define the main function for ingesting and processing documents.
 def ingest_worker(self, directory, formats, name_job, filename, user):
@@ -66,9 +92,11 @@ def ingest_worker(self, directory, formats, name_job, filename, user):
     token_check = True
     min_tokens = 150
     max_tokens = 1250
-    full_path = directory + "/" + user + "/" + name_job
+    recursion_depth = 2
+    full_path = os.path.join(directory, user, name_job)
     import sys
 
+
     print(full_path, file=sys.stderr)
     # check if API_URL env variable is set
     file_data = {"name": name_job, "file": filename, "user": user}
@@ -81,14 +109,12 @@ def ingest_worker(self, directory, formats, name_job, filename, user):
 
     if not os.path.exists(full_path):
         os.makedirs(full_path)
-    with open(full_path + "/" + filename, "wb") as f:
+    with open(os.path.join(full_path, filename), "wb") as f:
         f.write(file)
 
     # check if file is .zip and extract it
     if filename.endswith(".zip"):
-        with zipfile.ZipFile(full_path + "/" + filename, "r") as zip_ref:
-            zip_ref.extractall(full_path)
-        os.remove(full_path + "/" + filename)
+        extract_zip_recursive(os.path.join(full_path, filename), full_path, 0, recursion_depth)
 
     self.update_state(state="PROGRESS", meta={"current": 1})
 

docs/package-lock.json

@@ -8143,9 +8143,9 @@
       "integrity": "sha512-gkXMxRzUH+PB0ax9dUN0yYF0S25BqeAYqhgMaLUFmpXLEk7Fcu8f4emJuOAY0V8kjDICxROIKsTAKsV/v355xw=="
     },
     "node_modules/npm": {
-      "version": "10.5.0",
-      "resolved": "https://registry.npmjs.org/npm/-/npm-10.5.0.tgz",
-      "integrity": "sha512-Ejxwvfh9YnWVU2yA5FzoYLTW52vxHCz+MHrOFg9Cc8IFgF/6f5AGPAvb5WTay5DIUP1NIfN3VBZ0cLlGO0Ys+A==",
+      "version": "10.5.1",
+      "resolved": "https://registry.npmjs.org/npm/-/npm-10.5.1.tgz",
+      "integrity": "sha512-RozZuGuWbbhDM2sRhOSLIRb3DLyof6TREi0TW5b3xUEBropDhDqEHv0iAjA1zsIwXKgfIkR8GvQMd4oeKKg9eQ==",
       "bundleDependencies": [
         "@isaacs/string-locale-compare",
         "@npmcli/arborist",
@@ -8154,6 +8154,7 @@
         "@npmcli/map-workspaces",
         "@npmcli/package-json",
         "@npmcli/promise-spawn",
+        "@npmcli/redact",
         "@npmcli/run-script",
         "@sigstore/tuf",
         "abbrev",
@@ -8226,23 +8227,24 @@
         "@npmcli/map-workspaces": "^3.0.4",
         "@npmcli/package-json": "^5.0.0",
         "@npmcli/promise-spawn": "^7.0.1",
+        "@npmcli/redact": "^1.1.0",
         "@npmcli/run-script": "^7.0.4",
-        "@sigstore/tuf": "^2.3.1",
+        "@sigstore/tuf": "^2.3.2",
         "abbrev": "^2.0.0",
         "archy": "~1.0.0",
         "cacache": "^18.0.2",
         "chalk": "^5.3.0",
         "ci-info": "^4.0.0",
         "cli-columns": "^4.0.0",
-        "cli-table3": "^0.6.3",
+        "cli-table3": "^0.6.4",
         "columnify": "^1.6.0",
         "fastest-levenshtein": "^1.0.16",
         "fs-minipass": "^3.0.3",
-        "glob": "^10.3.10",
+        "glob": "^10.3.12",
         "graceful-fs": "^4.2.11",
         "hosted-git-info": "^7.0.1",
-        "ini": "^4.1.1",
-        "init-package-json": "^6.0.0",
+        "ini": "^4.1.2",
+        "init-package-json": "^6.0.2",
         "is-cidr": "^5.0.3",
         "json-parse-even-better-errors": "^3.0.1",
         "libnpmaccess": "^8.0.1",
@@ -8257,11 +8259,11 @@
         "libnpmteam": "^6.0.0",
         "libnpmversion": "^5.0.1",
         "make-fetch-happen": "^13.0.0",
-        "minimatch": "^9.0.3",
+        "minimatch": "^9.0.4",
         "minipass": "^7.0.4",
         "minipass-pipeline": "^1.2.4",
         "ms": "^2.1.2",
-        "node-gyp": "^10.0.1",
+        "node-gyp": "^10.1.0",
         "nopt": "^7.2.0",
         "normalize-package-data": "^6.0.0",
         "npm-audit-report": "^5.0.0",
@@ -8269,7 +8271,7 @@
         "npm-package-arg": "^11.0.1",
         "npm-pick-manifest": "^9.0.0",
         "npm-profile": "^9.0.0",
-        "npm-registry-fetch": "^16.1.0",
+        "npm-registry-fetch": "^16.2.0",
         "npm-user-validate": "^2.0.0",
         "npmlog": "^7.0.1",
         "p-map": "^4.0.0",
@@ -8277,12 +8279,12 @@
         "parse-conflict-json": "^3.0.1",
         "proc-log": "^3.0.0",
         "qrcode-terminal": "^0.12.0",
-        "read": "^2.1.0",
+        "read": "^3.0.1",
         "semver": "^7.6.0",
         "spdx-expression-parse": "^3.0.1",
         "ssri": "^10.0.5",
         "supports-color": "^9.4.0",
-        "tar": "^6.2.0",
+        "tar": "^6.2.1",
         "text-table": "~0.2.0",
         "tiny-relative-date": "^1.3.0",
         "treeverse": "^3.0.0",
@@ -8339,8 +8341,6 @@
     },
     "node_modules/npm/node_modules/@isaacs/cliui": {
       "version": "8.0.2",
-      "resolved": "https://registry.npmjs.org/@isaacs/cliui/-/cliui-8.0.2.tgz",
-      "integrity": "sha512-O8jcjabXaleOG9DQ0+ARXWZBTfnP4WNAqzuiJK7ll44AmxGKv/J2M4TPjxjY3znBCfvBXFzucm1twdyFybFqEA==",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -8357,8 +8357,6 @@
     },
     "node_modules/npm/node_modules/@isaacs/cliui/node_modules/ansi-regex": {
       "version": "6.0.1",
-      "resolved": "https://registry.npmjs.org/ansi-regex/-/ansi-regex-6.0.1.tgz",
-      "integrity": "sha512-n5M855fKb2SsfMIiFFoVrABHJC8QtHwVx+mHWP3QcEqBHYienj5dHSgjbxtC0WEZXYt4wcD6zrQElDPhFuZgfA==",
       "inBundle": true,
       "license": "MIT",
       "engines": {
@@ -8370,8 +8368,6 @@
     },
     "node_modules/npm/node_modules/@isaacs/cliui/node_modules/emoji-regex": {
       "version": "9.2.2",
-      "resolved": "https://registry.npmjs.org/emoji-regex/-/emoji-regex-9.2.2.tgz",
-      "integrity": "sha512-L18DaJsXSUk2+42pv8mLs5jJT2hqFkFE4j21wOmgbUqsZ2hL72NsUU785g9RXgo3s0ZNgVl42TiHp3ZtOv/Vyg==",
       "inBundle": true,
       "license": "MIT"
     },
@@ -8393,8 +8389,6 @@
     },
     "node_modules/npm/node_modules/@isaacs/cliui/node_modules/strip-ansi": {
       "version": "7.1.0",
-      "resolved": "https://registry.npmjs.org/strip-ansi/-/strip-ansi-7.1.0.tgz",
-      "integrity": "sha512-iq6eVVI64nQQTRYq2KtEg2d2uU7LElhTJwsH4YzIHZshxlgZms/wIc4VoDQTlG/IvVIrBKG06CrZnp0qv7hkcQ==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -8428,7 +8422,7 @@
       }
     },
     "node_modules/npm/node_modules/@npmcli/arborist": {
-      "version": "7.4.0",
+      "version": "7.4.1",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -8448,12 +8442,12 @@
         "hosted-git-info": "^7.0.1",
         "json-parse-even-better-errors": "^3.0.0",
         "json-stringify-nice": "^1.1.4",
-        "minimatch": "^9.0.0",
+        "minimatch": "^9.0.4",
         "nopt": "^7.0.0",
         "npm-install-checks": "^6.2.0",
         "npm-package-arg": "^11.0.1",
         "npm-pick-manifest": "^9.0.0",
-        "npm-registry-fetch": "^16.0.0",
+        "npm-registry-fetch": "^16.2.0",
         "npmlog": "^7.0.1",
         "pacote": "^17.0.4",
         "parse-conflict-json": "^3.0.0",
@@ -8474,13 +8468,13 @@
       }
     },
     "node_modules/npm/node_modules/@npmcli/config": {
-      "version": "8.2.0",
+      "version": "8.2.1",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "@npmcli/map-workspaces": "^3.0.2",
         "ci-info": "^4.0.0",
-        "ini": "^4.1.0",
+        "ini": "^4.1.2",
         "nopt": "^7.0.0",
         "proc-log": "^3.0.0",
         "read-package-json-fast": "^3.0.2",
@@ -8643,6 +8637,14 @@
         "node": "^14.17.0 || ^16.13.0 || >=18.0.0"
       }
     },
+    "node_modules/npm/node_modules/@npmcli/redact": {
+      "version": "1.1.0",
+      "inBundle": true,
+      "license": "ISC",
+      "engines": {
+        "node": "^16.14.0 || >=18.0.0"
+      }
+    },
     "node_modules/npm/node_modules/@npmcli/run-script": {
       "version": "7.0.4",
       "inBundle": true,
@@ -8660,8 +8662,6 @@
     },
     "node_modules/npm/node_modules/@pkgjs/parseargs": {
       "version": "0.11.0",
-      "resolved": "https://registry.npmjs.org/@pkgjs/parseargs/-/parseargs-0.11.0.tgz",
-      "integrity": "sha512-+1VkjdD0QBLPodGrJUeqarH8VAIvQODIbwh9XpP5Syisf7YoQgsJKPNFoqqLQlu+VQ/tVSshMR6loPMn8U+dPg==",
       "inBundle": true,
       "license": "MIT",
       "optional": true,
@@ -8711,7 +8711,7 @@
       }
     },
     "node_modules/npm/node_modules/@sigstore/tuf": {
-      "version": "2.3.1",
+      "version": "2.3.2",
       "inBundle": true,
       "license": "Apache-2.0",
       "dependencies": {
@@ -8764,7 +8764,7 @@
       }
     },
     "node_modules/npm/node_modules/agent-base": {
-      "version": "7.1.0",
+      "version": "7.1.1",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -8788,8 +8788,6 @@
     },
     "node_modules/npm/node_modules/ansi-regex": {
       "version": "5.0.1",
-      "resolved": "https://registry.npmjs.org/ansi-regex/-/ansi-regex-5.0.1.tgz",
-      "integrity": "sha512-quJQXlTSUGL2LH9SUXo8VwsY4soanhgo6LNSm84E1LBcE8s3O0wpdiRzyR9z/ZZJMlMWv37qOOb9pdJlMUEKFQ==",
       "inBundle": true,
       "license": "MIT",
       "engines": {
@@ -8798,8 +8796,6 @@
     },
     "node_modules/npm/node_modules/ansi-styles": {
       "version": "6.2.1",
-      "resolved": "https://registry.npmjs.org/ansi-styles/-/ansi-styles-6.2.1.tgz",
-      "integrity": "sha512-bN798gFfQX+viw3R7yrGWRqnrN2oRkEkUjjl4JNn4E8GxxbjtG3FbrEIIY3l8/hrwUwIeCZvi4QuOTP4MErVug==",
       "inBundle": true,
       "license": "MIT",
       "engines": {
@@ -8829,8 +8825,6 @@
     },
     "node_modules/npm/node_modules/balanced-match": {
       "version": "1.0.2",
-      "resolved": "https://registry.npmjs.org/balanced-match/-/balanced-match-1.0.2.tgz",
-      "integrity": "sha512-3oSeUO0TMV67hN1AmbXsK4yaqU7tjiHlbxRDZOpH0KW9+CeX4bRAaX0Anxt0tx2MrpRpWwQaPwIlISEJhYU5Pw==",
       "inBundle": true,
       "license": "MIT"
     },
@@ -8849,17 +8843,18 @@
       }
     },
     "node_modules/npm/node_modules/binary-extensions": {
-      "version": "2.2.0",
+      "version": "2.3.0",
       "inBundle": true,
       "license": "MIT",
       "engines": {
         "node": ">=8"
+      },
+      "funding": {
+        "url": "https://github.com/sponsors/sindresorhus"
       }
     },
     "node_modules/npm/node_modules/brace-expansion": {
       "version": "2.0.1",
-      "resolved": "https://registry.npmjs.org/brace-expansion/-/brace-expansion-2.0.1.tgz",
-      "integrity": "sha512-XnAIvQ8eM+kC6aULx6wuQiwVsnzsi9d3WxzV3FpWTGA19F621kwdbsAcFKXgKUHZWsy+mY6iL1sHTxWEFCytDA==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -8961,7 +8956,7 @@
       }
     },
     "node_modules/npm/node_modules/cli-table3": {
-      "version": "0.6.3",
+      "version": "0.6.4",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -8992,8 +8987,6 @@
     },
     "node_modules/npm/node_modules/color-convert": {
       "version": "2.0.1",
-      "resolved": "https://registry.npmjs.org/color-convert/-/color-convert-2.0.1.tgz",
-      "integrity": "sha512-RRECPsj7iu/xb5oKYcsFHSppFNnsj/52OVTRKb4zP5onXwVF3zVmmToNcOfGC+CRDpfK/U584fMg38ZHCaElKQ==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -9005,8 +8998,6 @@
     },
     "node_modules/npm/node_modules/color-name": {
       "version": "1.1.4",
-      "resolved": "https://registry.npmjs.org/color-name/-/color-name-1.1.4.tgz",
-      "integrity": "sha512-dOy+3AuW3a2wNbZHIuMZpTcgjGuLU/uBL/ubcZF9OXbDo8ff4O8yVp5Bf0efS8uEoYo5q4Fx7dY9OgQGXgAsQA==",
       "inBundle": true,
       "license": "MIT"
     },
@@ -9042,8 +9033,6 @@
     },
     "node_modules/npm/node_modules/cross-spawn": {
       "version": "7.0.3",
-      "resolved": "https://registry.npmjs.org/cross-spawn/-/cross-spawn-7.0.3.tgz",
-      "integrity": "sha512-iRDPJKUPVEND7dHPO8rkbOnPpyDygcDFtWjpeWNCgy8WP2rXcxXL8TskReQl6OrB2G7+UJrags1q15Fudc7G6w==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -9071,8 +9060,6 @@
     },
     "node_modules/npm/node_modules/cssesc": {
       "version": "3.0.0",
-      "resolved": "https://registry.npmjs.org/cssesc/-/cssesc-3.0.0.tgz",
-      "integrity": "sha512-/Tb/JcjK111nNScGob5MNtsntNM1aCNUDipB/TkwZFhyDrrE47SOx/18wF2bbjgc3ZzCSKW1T5nt5EbFoAz/Vg==",
       "inBundle": true,
       "license": "MIT",
       "bin": {
@@ -9124,15 +9111,11 @@
     },
     "node_modules/npm/node_modules/eastasianwidth": {
       "version": "0.2.0",
-      "resolved": "https://registry.npmjs.org/eastasianwidth/-/eastasianwidth-0.2.0.tgz",
-      "integrity": "sha512-I88TYZWc9XiYHRQ4/3c5rjjfgkjhLyW2luGIheGERbNQ6OY7yTybanSpDXZa8y7VUP9YmDcYa+eyq4ca7iLqWA==",
       "inBundle": true,
       "license": "MIT"
     },
     "node_modules/npm/node_modules/emoji-regex": {
       "version": "8.0.0",
-      "resolved": "https://registry.npmjs.org/emoji-regex/-/emoji-regex-8.0.0.tgz",
-      "integrity": "sha512-MSjYzcWNOA0ewAHpz0MxpYFvwg6yjy1NG3xteoqz644VCo/RPgnr1/GGt+ic3iJTzQ8Eu3TdM14SawnVUmGE6A==",
       "inBundle": true,
       "license": "MIT"
     },
@@ -9173,8 +9156,6 @@
     },
     "node_modules/npm/node_modules/foreground-child": {
       "version": "3.1.1",
-      "resolved": "https://registry.npmjs.org/foreground-child/-/foreground-child-3.1.1.tgz",
-      "integrity": "sha512-TMKDUnIte6bfb5nWv7V/caI169OHgvwjb7V4WkeUvbQQdjr5rWKqHFiKWb/fcOwB+CzBT+qbWjvj+DVwRskpIg==",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -9201,8 +9182,6 @@
     },
     "node_modules/npm/node_modules/function-bind": {
       "version": "1.1.2",
-      "resolved": "https://registry.npmjs.org/function-bind/-/function-bind-1.1.2.tgz",
-      "integrity": "sha512-7XHNxH7qX9xG5mIwxkhumTox/MIRNcOgDrxWsMt2pAr23WHp6MrRlN7FBSFpCpr+oVO0F744iUgR82nJMfG2SA==",
       "inBundle": true,
       "license": "MIT",
       "funding": {
@@ -9228,17 +9207,15 @@
       }
     },
     "node_modules/npm/node_modules/glob": {
-      "version": "10.3.10",
-      "resolved": "https://registry.npmjs.org/glob/-/glob-10.3.10.tgz",
-      "integrity": "sha512-fa46+tv1Ak0UPK1TOy/pZrIybNNt4HCv7SDzwyfiOZkvZLEbjsZkJBPtDHVshZjbecAoAGSC20MjLDG/qr679g==",
+      "version": "10.3.12",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "foreground-child": "^3.1.0",
-        "jackspeak": "^2.3.5",
+        "jackspeak": "^2.3.6",
         "minimatch": "^9.0.1",
-        "minipass": "^5.0.0 || ^6.0.2 || ^7.0.0",
-        "path-scurry": "^1.10.1"
+        "minipass": "^7.0.4",
+        "path-scurry": "^1.10.2"
       },
       "bin": {
         "glob": "dist/esm/bin.mjs"
@@ -9252,8 +9229,6 @@
     },
     "node_modules/npm/node_modules/graceful-fs": {
       "version": "4.2.11",
-      "resolved": "https://registry.npmjs.org/graceful-fs/-/graceful-fs-4.2.11.tgz",
-      "integrity": "sha512-RbJ5/jmFcNNCcDV5o9eTnBLJ/HszWV0P73bc+Ff4nS/rJj+YaS6IGyiOL0VoBYX+l1Wrl3k63h/KrH+nhJ0XvQ==",
       "inBundle": true,
       "license": "ISC"
     },
@@ -9353,7 +9328,7 @@
       }
     },
     "node_modules/npm/node_modules/ini": {
-      "version": "4.1.1",
+      "version": "4.1.2",
       "inBundle": true,
       "license": "ISC",
       "engines": {
@@ -9361,14 +9336,14 @@
       }
     },
     "node_modules/npm/node_modules/init-package-json": {
-      "version": "6.0.0",
+      "version": "6.0.2",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
+        "@npmcli/package-json": "^5.0.0",
         "npm-package-arg": "^11.0.0",
         "promzard": "^1.0.0",
-        "read": "^2.0.0",
-        "read-package-json": "^7.0.0",
+        "read": "^3.0.1",
         "semver": "^7.3.5",
         "validate-npm-package-license": "^3.0.4",
         "validate-npm-package-name": "^5.0.0"
@@ -9429,8 +9404,6 @@
     },
     "node_modules/npm/node_modules/is-fullwidth-code-point": {
       "version": "3.0.0",
-      "resolved": "https://registry.npmjs.org/is-fullwidth-code-point/-/is-fullwidth-code-point-3.0.0.tgz",
-      "integrity": "sha512-zymm5+u+sCsSWyD9qNaejV3DFvhCKclKdizYaJUuHA83RLjb7nSuGnddCHGv0hk+KY7BMAlsWeK4Ueg6EV6XQg==",
       "inBundle": true,
       "license": "MIT",
       "engines": {
@@ -9444,15 +9417,11 @@
     },
     "node_modules/npm/node_modules/isexe": {
       "version": "2.0.0",
-      "resolved": "https://registry.npmjs.org/isexe/-/isexe-2.0.0.tgz",
-      "integrity": "sha512-RHxMLp9lnKHGHRng9QFhRCMbYAcVpn69smSGcq3f36xjgVVWThj4qqLbTLlq7Ssj8B+fIQ1EuCEGI2lKsyQeIw==",
       "inBundle": true,
       "license": "ISC"
     },
     "node_modules/npm/node_modules/jackspeak": {
       "version": "2.3.6",
-      "resolved": "https://registry.npmjs.org/jackspeak/-/jackspeak-2.3.6.tgz",
-      "integrity": "sha512-N3yCS/NegsOBokc8GAdM8UcmfsKiSS8cipheD/nivzr700H+nsMOxJjQnvwOcRYVuFkdH0wGUvW2WbXGmrZGbQ==",
       "inBundle": true,
       "license": "BlueOak-1.0.0",
       "dependencies": {
@@ -9508,38 +9477,38 @@
       "license": "MIT"
     },
     "node_modules/npm/node_modules/libnpmaccess": {
-      "version": "8.0.2",
+      "version": "8.0.3",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "npm-package-arg": "^11.0.1",
-        "npm-registry-fetch": "^16.0.0"
+        "npm-registry-fetch": "^16.2.0"
       },
       "engines": {
         "node": "^16.14.0 || >=18.0.0"
       }
     },
     "node_modules/npm/node_modules/libnpmdiff": {
-      "version": "6.0.7",
+      "version": "6.0.8",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "@npmcli/arborist": "^7.2.1",
         "@npmcli/disparity-colors": "^3.0.0",
         "@npmcli/installed-package-contents": "^2.0.2",
-        "binary-extensions": "^2.2.0",
+        "binary-extensions": "^2.3.0",
         "diff": "^5.1.0",
-        "minimatch": "^9.0.0",
+        "minimatch": "^9.0.4",
         "npm-package-arg": "^11.0.1",
         "pacote": "^17.0.4",
-        "tar": "^6.2.0"
+        "tar": "^6.2.1"
       },
       "engines": {
         "node": "^16.14.0 || >=18.0.0"
       }
     },
     "node_modules/npm/node_modules/libnpmexec": {
-      "version": "7.0.8",
+      "version": "7.0.9",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -9550,7 +9519,7 @@
         "npmlog": "^7.0.1",
         "pacote": "^17.0.4",
         "proc-log": "^3.0.0",
-        "read": "^2.0.0",
+        "read": "^3.0.1",
         "read-package-json-fast": "^3.0.2",
         "semver": "^7.3.7",
         "walk-up-path": "^3.0.1"
@@ -9560,7 +9529,7 @@
       }
     },
     "node_modules/npm/node_modules/libnpmfund": {
-      "version": "5.0.5",
+      "version": "5.0.6",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -9571,31 +9540,31 @@
       }
     },
     "node_modules/npm/node_modules/libnpmhook": {
-      "version": "10.0.1",
+      "version": "10.0.2",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "aproba": "^2.0.0",
-        "npm-registry-fetch": "^16.0.0"
+        "npm-registry-fetch": "^16.2.0"
       },
       "engines": {
         "node": "^16.14.0 || >=18.0.0"
       }
     },
     "node_modules/npm/node_modules/libnpmorg": {
-      "version": "6.0.2",
+      "version": "6.0.3",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "aproba": "^2.0.0",
-        "npm-registry-fetch": "^16.0.0"
+        "npm-registry-fetch": "^16.2.0"
       },
       "engines": {
         "node": "^16.14.0 || >=18.0.0"
       }
     },
     "node_modules/npm/node_modules/libnpmpack": {
-      "version": "6.0.7",
+      "version": "6.0.8",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -9609,14 +9578,14 @@
       }
     },
     "node_modules/npm/node_modules/libnpmpublish": {
-      "version": "9.0.4",
+      "version": "9.0.5",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "ci-info": "^4.0.0",
         "normalize-package-data": "^6.0.0",
         "npm-package-arg": "^11.0.1",
-        "npm-registry-fetch": "^16.0.0",
+        "npm-registry-fetch": "^16.2.0",
         "proc-log": "^3.0.0",
         "semver": "^7.3.7",
         "sigstore": "^2.2.0",
@@ -9627,23 +9596,23 @@
       }
     },
     "node_modules/npm/node_modules/libnpmsearch": {
-      "version": "7.0.1",
+      "version": "7.0.2",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
-        "npm-registry-fetch": "^16.0.0"
+        "npm-registry-fetch": "^16.2.0"
       },
       "engines": {
         "node": "^16.14.0 || >=18.0.0"
       }
     },
     "node_modules/npm/node_modules/libnpmteam": {
-      "version": "6.0.1",
+      "version": "6.0.2",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
         "aproba": "^2.0.0",
-        "npm-registry-fetch": "^16.0.0"
+        "npm-registry-fetch": "^16.2.0"
       },
       "engines": {
         "node": "^16.14.0 || >=18.0.0"
@@ -9694,9 +9663,7 @@
       }
     },
     "node_modules/npm/node_modules/minimatch": {
-      "version": "9.0.3",
-      "resolved": "https://registry.npmjs.org/minimatch/-/minimatch-9.0.3.tgz",
-      "integrity": "sha512-RHiac9mvaRw0x3AYRgDC1CxAP7HTcNrrECeA8YYJeWnpo+2Q5CegtZjaotWTWxDG3UeGA1coE05iH1mPjT/2mg==",
+      "version": "9.0.4",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -9711,8 +9678,6 @@
     },
     "node_modules/npm/node_modules/minipass": {
       "version": "7.0.4",
-      "resolved": "https://registry.npmjs.org/minipass/-/minipass-7.0.4.tgz",
-      "integrity": "sha512-jYofLM5Dam9279rdkWzqHozUo4ybjdZmCsDHePy5V/PbBcVMiSZR97gmAy45aqi8CK1lG2ECd356FU86avfwUQ==",
       "inBundle": true,
       "license": "ISC",
       "engines": {
@@ -9888,7 +9853,7 @@
       }
     },
     "node_modules/npm/node_modules/node-gyp": {
-      "version": "10.0.1",
+      "version": "10.1.0",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10028,10 +9993,11 @@
       }
     },
     "node_modules/npm/node_modules/npm-registry-fetch": {
-      "version": "16.1.0",
+      "version": "16.2.0",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
+        "@npmcli/redact": "^1.1.0",
         "make-fetch-happen": "^13.0.0",
         "minipass": "^7.0.2",
         "minipass-fetch": "^3.0.0",
@@ -10126,8 +10092,6 @@
     },
     "node_modules/npm/node_modules/path-key": {
       "version": "3.1.1",
-      "resolved": "https://registry.npmjs.org/path-key/-/path-key-3.1.1.tgz",
-      "integrity": "sha512-ojmeN0qd+y0jszEtoY48r0Peq5dwMEkIlCOu6Q5f41lfkswXuKtYrhgoTpLnyIcHm24Uhqx+5Tqm2InSwLhE6Q==",
       "inBundle": true,
       "license": "MIT",
       "engines": {
@@ -10135,13 +10099,11 @@
       }
     },
     "node_modules/npm/node_modules/path-scurry": {
-      "version": "1.10.1",
-      "resolved": "https://registry.npmjs.org/path-scurry/-/path-scurry-1.10.1.tgz",
-      "integrity": "sha512-MkhCqzzBEpPvxxQ71Md0b1Kk51W01lrYvlMzSUaIzNsODdd7mqhiimSZlr+VegAz5Z6Vzt9Xg2ttE//XBhH3EQ==",
+      "version": "1.10.2",
       "inBundle": true,
       "license": "BlueOak-1.0.0",
       "dependencies": {
-        "lru-cache": "^9.1.1 || ^10.0.0",
+        "lru-cache": "^10.2.0",
         "minipass": "^5.0.0 || ^6.0.2 || ^7.0.0"
       },
       "engines": {
@@ -10205,11 +10167,11 @@
       }
     },
     "node_modules/npm/node_modules/promzard": {
-      "version": "1.0.0",
+      "version": "1.0.1",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
-        "read": "^2.0.0"
+        "read": "^3.0.1"
       },
       "engines": {
         "node": "^14.17.0 || ^16.13.0 || >=18.0.0"
@@ -10223,11 +10185,11 @@
       }
     },
     "node_modules/npm/node_modules/read": {
-      "version": "2.1.0",
+      "version": "3.0.1",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
-        "mute-stream": "~1.0.0"
+        "mute-stream": "^1.0.0"
       },
       "engines": {
         "node": "^14.17.0 || ^16.13.0 || >=18.0.0"
@@ -10277,8 +10239,6 @@
     },
     "node_modules/npm/node_modules/safer-buffer": {
       "version": "2.1.2",
-      "resolved": "https://registry.npmjs.org/safer-buffer/-/safer-buffer-2.1.2.tgz",
-      "integrity": "sha512-YZo3K82SD7Riyi0E1EQPojLz7kpepnSQI9IyPbHHg1XXXevb5dJI7tpyN2ADxGcQbHG7vcyRHk0cbwqcQriUtg==",
       "inBundle": true,
       "license": "MIT",
       "optional": true
@@ -10315,8 +10275,6 @@
     },
     "node_modules/npm/node_modules/shebang-command": {
       "version": "2.0.0",
-      "resolved": "https://registry.npmjs.org/shebang-command/-/shebang-command-2.0.0.tgz",
-      "integrity": "sha512-kHxr2zZpYtdmrN1qDjrrX/Z1rR1kG8Dx+gkpK1G4eXmvXswmcE1hTWBWYUzlraYw1/yZp6YuDY77YtvbN0dmDA==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10328,8 +10286,6 @@
     },
     "node_modules/npm/node_modules/shebang-regex": {
       "version": "3.0.0",
-      "resolved": "https://registry.npmjs.org/shebang-regex/-/shebang-regex-3.0.0.tgz",
-      "integrity": "sha512-7++dFhtcx3353uBaq8DDR4NuxBetBzC7ZQOhmTQInHEd6bSrXdiEyzCvG07Z44UYdLShWUyXt5M/yhz8ekcb1A==",
       "inBundle": true,
       "license": "MIT",
       "engines": {
@@ -10338,8 +10294,6 @@
     },
     "node_modules/npm/node_modules/signal-exit": {
       "version": "4.1.0",
-      "resolved": "https://registry.npmjs.org/signal-exit/-/signal-exit-4.1.0.tgz",
-      "integrity": "sha512-bzyZ1e88w9O1iNJbKnOlvYTrWPDl46O1bG0D3XInv+9tkPrxrN8jUUTiFlDkkmKWgn1M6CfIA13SuGqOa9Korw==",
       "inBundle": true,
       "license": "ISC",
       "engines": {
@@ -10441,8 +10395,6 @@
     },
     "node_modules/npm/node_modules/string-width": {
       "version": "4.2.3",
-      "resolved": "https://registry.npmjs.org/string-width/-/string-width-4.2.3.tgz",
-      "integrity": "sha512-wKyQRQpjJ0sIp62ErSZdGsjMJWsap5oRNihHhu6G7JVO/9jIB6UyevL+tXuOqrng8j/cxKTWyWUwvSTriiZz/g==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10457,8 +10409,6 @@
     "node_modules/npm/node_modules/string-width-cjs": {
       "name": "string-width",
       "version": "4.2.3",
-      "resolved": "https://registry.npmjs.org/string-width/-/string-width-4.2.3.tgz",
-      "integrity": "sha512-wKyQRQpjJ0sIp62ErSZdGsjMJWsap5oRNihHhu6G7JVO/9jIB6UyevL+tXuOqrng8j/cxKTWyWUwvSTriiZz/g==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10472,8 +10422,6 @@
     },
     "node_modules/npm/node_modules/strip-ansi": {
       "version": "6.0.1",
-      "resolved": "https://registry.npmjs.org/strip-ansi/-/strip-ansi-6.0.1.tgz",
-      "integrity": "sha512-Y38VPSHcqkFrCpFnQ9vuSXmquuv5oXOKpGeT6aGrr3o3Gc9AlVa6JBfUSOCnbxGGZF+/0ooI7KrPuUSztUdU5A==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10486,8 +10434,6 @@
     "node_modules/npm/node_modules/strip-ansi-cjs": {
       "name": "strip-ansi",
       "version": "6.0.1",
-      "resolved": "https://registry.npmjs.org/strip-ansi/-/strip-ansi-6.0.1.tgz",
-      "integrity": "sha512-Y38VPSHcqkFrCpFnQ9vuSXmquuv5oXOKpGeT6aGrr3o3Gc9AlVa6JBfUSOCnbxGGZF+/0ooI7KrPuUSztUdU5A==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10509,7 +10455,7 @@
       }
     },
     "node_modules/npm/node_modules/tar": {
-      "version": "6.2.0",
+      "version": "6.2.1",
       "inBundle": true,
       "license": "ISC",
       "dependencies": {
@@ -10609,8 +10555,6 @@
     },
     "node_modules/npm/node_modules/util-deprecate": {
       "version": "1.0.2",
-      "resolved": "https://registry.npmjs.org/util-deprecate/-/util-deprecate-1.0.2.tgz",
-      "integrity": "sha512-EPD5q1uXyFxJpCrLnCc1nHnq3gOa6DZBocAIiI2TaSCA7VCJ1UJDMagCzIkXNsUYfD1daK//LTEQ8xiIbrHtcw==",
       "inBundle": true,
       "license": "MIT"
     },
@@ -10679,8 +10623,6 @@
     },
     "node_modules/npm/node_modules/wrap-ansi": {
       "version": "8.1.0",
-      "resolved": "https://registry.npmjs.org/wrap-ansi/-/wrap-ansi-8.1.0.tgz",
-      "integrity": "sha512-si7QWI6zUMq56bESFvagtmzMdGOtoxfR+Sez11Mobfc7tm+VkUckk9bW2UeffTGVUbOksxmSw0AA2gs8g71NCQ==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10698,8 +10640,6 @@
     "node_modules/npm/node_modules/wrap-ansi-cjs": {
       "name": "wrap-ansi",
       "version": "7.0.0",
-      "resolved": "https://registry.npmjs.org/wrap-ansi/-/wrap-ansi-7.0.0.tgz",
-      "integrity": "sha512-YVGIj2kamLSTxw6NsZjoBxfSwsn0ycdesmc4p+Q21c5zPuZ1pl+NfxVdxPtdHvmNVOQ6XSYG4AUtyt/Fi7D16Q==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {
@@ -10730,8 +10670,6 @@
     },
     "node_modules/npm/node_modules/wrap-ansi/node_modules/ansi-regex": {
       "version": "6.0.1",
-      "resolved": "https://registry.npmjs.org/ansi-regex/-/ansi-regex-6.0.1.tgz",
-      "integrity": "sha512-n5M855fKb2SsfMIiFFoVrABHJC8QtHwVx+mHWP3QcEqBHYienj5dHSgjbxtC0WEZXYt4wcD6zrQElDPhFuZgfA==",
       "inBundle": true,
       "license": "MIT",
       "engines": {
@@ -10743,8 +10681,6 @@
     },
     "node_modules/npm/node_modules/wrap-ansi/node_modules/emoji-regex": {
       "version": "9.2.2",
-      "resolved": "https://registry.npmjs.org/emoji-regex/-/emoji-regex-9.2.2.tgz",
-      "integrity": "sha512-L18DaJsXSUk2+42pv8mLs5jJT2hqFkFE4j21wOmgbUqsZ2hL72NsUU785g9RXgo3s0ZNgVl42TiHp3ZtOv/Vyg==",
       "inBundle": true,
       "license": "MIT"
     },
@@ -10766,8 +10702,6 @@
     },
     "node_modules/npm/node_modules/wrap-ansi/node_modules/strip-ansi": {
       "version": "7.1.0",
-      "resolved": "https://registry.npmjs.org/strip-ansi/-/strip-ansi-7.1.0.tgz",
-      "integrity": "sha512-iq6eVVI64nQQTRYq2KtEg2d2uU7LElhTJwsH4YzIHZshxlgZms/wIc4VoDQTlG/IvVIrBKG06CrZnp0qv7hkcQ==",
       "inBundle": true,
       "license": "MIT",
       "dependencies": {

extensions/react-widget/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "docsgpt",
-  "version": "0.3.6",
+  "version": "0.3.7",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "docsgpt",
-      "version": "0.3.6",
+      "version": "0.3.7",
       "license": "Apache-2.0",
       "dependencies": {
         "@babel/plugin-transform-flow-strip-types": "^7.23.3",

extensions/react-widget/package.json

@@ -19,7 +19,7 @@
   },
   "scripts": {
     "build": "parcel build src/index.ts",
-    "dev": "parcel",
+    "dev": "parcel src/index.html -p 3000",
     "test": "jest",
     "lint": "eslint",
     "check": "tsc --noEmit",

extensions/react-widget/src/components/DocsGPTWidget.tsx

@@ -1,8 +1,7 @@
 "use client";
 import { Fragment, useEffect, useRef, useState } from 'react'
 import { PaperPlaneIcon, RocketIcon, ExclamationTriangleIcon, Cross2Icon } from '@radix-ui/react-icons';
-import { MESSAGE_TYPE } from '../models/types';
-import { Query, Status } from '../models/types';
+import { MESSAGE_TYPE, Query, Status } from '../types/index';
 import MessageIcon from '../assets/message.svg'
 import { fetchAnswerStreaming } from '../requests/streamingApi';
 import styled, { keyframes, createGlobalStyle } from 'styled-components';

extensions/react-widget/src/types/index.ts

@@ -0,0 +1,13 @@
+export type MESSAGE_TYPE = 'QUESTION' | 'ANSWER' | 'ERROR';
+export type Status = 'idle' | 'loading' | 'failed';
+export type FEEDBACK = 'LIKE' | 'DISLIKE';
+
+export interface Query {
+  prompt: string;
+  response?: string;
+  feedback?: FEEDBACK;
+  error?: string;
+  sources?: { title: string; text: string }[];
+  conversationId?: string | null;
+  title?: string | null;
+}

frontend/src/App.tsx

@@ -6,7 +6,7 @@ import PageNotFound from './PageNotFound';
 import { inject } from '@vercel/analytics';
 import { useMediaQuery } from './hooks';
 import { useState } from 'react';
-import Setting from './Setting';
+import Setting from './settings';
 
 inject();
 

frontend/src/components/Dropdown.tsx

@@ -1,4 +1,4 @@
-import { useState } from 'react';
+import React from 'react';
 import Arrow2 from '../assets/dropdown-arrow.svg';
 import Edit from '../assets/edit.svg';
 import Trash from '../assets/trash.svg';
@@ -14,8 +14,6 @@ function Dropdown({
   showDelete,
   onDelete,
   placeholder,
-  fullWidth,
-  alignMidddle,
 }: {
   options:
     | string[]
@@ -33,10 +31,8 @@ function Dropdown({
   showDelete?: boolean;
   onDelete?: (value: string) => void;
   placeholder?: string;
-  fullWidth?: boolean;
-  alignMidddle?: boolean;
 }) {
-  const [isOpen, setIsOpen] = useState(false);
+  const [isOpen, setIsOpen] = React.useState(false);
   return (
     <div
       className={[
@@ -58,9 +54,7 @@ function Dropdown({
           </span>
         ) : (
           <span
-            className={`${
-              alignMidddle && 'flex-1'
-            } overflow-hidden text-ellipsis dark:text-bright-gray ${
+            className={`overflow-hidden text-ellipsis dark:text-bright-gray ${
               !selectedValue && 'text-silver dark:text-gray-400'
             }`}
           >
@@ -80,7 +74,7 @@ function Dropdown({
         />
       </button>
       {isOpen && (
-        <div className="absolute left-0 right-0 z-20 -mt-1 max-h-40 overflow-y-auto rounded-b-xl border-2 bg-white shadow-lg dark:border-chinese-silver dark:bg-dark-charcoal">
+        <div className="absolute left-0 right-0 z-20 -mt-1 max-h-40 overflow-y-auto rounded-b-xl border-2 border-silver bg-white shadow-lg dark:border-chinese-silver dark:bg-dark-charcoal">
           {options.map((option: any, index) => (
             <div
               key={index}
@@ -91,7 +85,7 @@ function Dropdown({
                   onSelect(option);
                   setIsOpen(false);
                 }}
-                className="ml-2 flex-1 overflow-hidden overflow-ellipsis whitespace-nowrap py-3 dark:text-light-gray"
+                className="ml-5 flex-1 overflow-hidden overflow-ellipsis whitespace-nowrap py-3 dark:text-light-gray"
               >
                 {typeof option === 'string'
                   ? option

frontend/src/conversation/conversationApi.ts

@@ -3,6 +3,33 @@ import { Doc } from '../preferences/preferenceApi';
 
 const apiHost = import.meta.env.VITE_API_HOST || 'https://docsapi.arc53.com';
 
+function getDocPath(selectedDocs: Doc | null): string {
+  let docPath = 'default';
+
+  if (selectedDocs) {
+    let namePath = selectedDocs.name;
+    if (selectedDocs.language === namePath) {
+      namePath = '.project';
+    }
+    if (selectedDocs.location === 'local') {
+      docPath = 'local' + '/' + selectedDocs.name + '/';
+    } else if (selectedDocs.location === 'remote') {
+      docPath =
+        selectedDocs.language +
+        '/' +
+        namePath +
+        '/' +
+        selectedDocs.version +
+        '/' +
+        selectedDocs.model +
+        '/';
+    } else if (selectedDocs.location === 'custom') {
+      docPath = selectedDocs.docLink;
+    }
+  }
+
+  return docPath;
+}
 export function fetchAnswerApi(
   question: string,
   signal: AbortSignal,
@@ -28,27 +55,7 @@ export function fetchAnswerApi(
       title: any;
     }
 > {
-  let docPath = 'default';
-
-  if (selectedDocs) {
-    let namePath = selectedDocs.name;
-    if (selectedDocs.language === namePath) {
-      namePath = '.project';
-    }
-    if (selectedDocs.location === 'local') {
-      docPath = 'local' + '/' + selectedDocs.name + '/';
-    } else if (selectedDocs.location === 'remote') {
-      docPath =
-        selectedDocs.language +
-        '/' +
-        namePath +
-        '/' +
-        selectedDocs.version +
-        '/' +
-        selectedDocs.model +
-        '/';
-    }
-  }
+  const docPath = getDocPath(selectedDocs);
   //in history array remove all keys except prompt and response
   history = history.map((item) => {
     return { prompt: item.prompt, response: item.response };
@@ -98,27 +105,7 @@ export function fetchAnswerSteaming(
   chunks: string,
   onEvent: (event: MessageEvent) => void,
 ): Promise<Answer> {
-  let docPath = 'default';
-
-  if (selectedDocs) {
-    let namePath = selectedDocs.name;
-    if (selectedDocs.language === namePath) {
-      namePath = '.project';
-    }
-    if (selectedDocs.location === 'local') {
-      docPath = 'local' + '/' + selectedDocs.name + '/';
-    } else if (selectedDocs.location === 'remote') {
-      docPath =
-        selectedDocs.language +
-        '/' +
-        namePath +
-        '/' +
-        selectedDocs.version +
-        '/' +
-        selectedDocs.model +
-        '/';
-    }
-  }
+  const docPath = getDocPath(selectedDocs);
 
   history = history.map((item) => {
     return { prompt: item.prompt, response: item.response };
@@ -195,31 +182,7 @@ export function searchEndpoint(
   history: Array<any> = [],
   chunks: string,
 ) {
-  /*
-  "active_docs": "default",
-  "question": "Summarise",
-  "conversation_id": null,
-  "history": "[]" */
-  let docPath = 'default';
-  if (selectedDocs) {
-    let namePath = selectedDocs.name;
-    if (selectedDocs.language === namePath) {
-      namePath = '.project';
-    }
-    if (selectedDocs.location === 'local') {
-      docPath = 'local' + '/' + selectedDocs.name + '/';
-    } else if (selectedDocs.location === 'remote') {
-      docPath =
-        selectedDocs.language +
-        '/' +
-        namePath +
-        '/' +
-        selectedDocs.version +
-        '/' +
-        selectedDocs.model +
-        '/';
-    }
-  }
+  const docPath = getDocPath(selectedDocs);
 
   const body = {
     question: question,

frontend/src/models/misc.ts

@@ -3,3 +3,42 @@ export type ActiveState = 'ACTIVE' | 'INACTIVE';
 export type User = {
   avatar: string;
 };
+export type Doc = {
+  location: string;
+  name: string;
+  language: string;
+  version: string;
+  description: string;
+  fullName: string;
+  date: string;
+  docLink: string;
+  model: string;
+};
+
+export type PromptProps = {
+  prompts: { name: string; id: string; type: string }[];
+  selectedPrompt: { name: string; id: string; type: string };
+  onSelectPrompt: (name: string, id: string, type: string) => void;
+  setPrompts: (prompts: { name: string; id: string; type: string }[]) => void;
+  apiHost: string;
+};
+
+export type DocumentsProps = {
+  documents: Doc[] | null;
+  handleDeleteDocument: (index: number, document: Doc) => void;
+};
+
+export type CreateAPIKeyModalProps = {
+  close: () => void;
+  createAPIKey: (payload: {
+    name: string;
+    source: string;
+    prompt_id: string;
+    chunks: string;
+  }) => void;
+};
+
+export type SaveAPIKeyModalProps = {
+  apiKey: string;
+  close: () => void;
+};

frontend/src/preferences/PromptsModal.tsx

@@ -1,4 +1,5 @@
 import { ActiveState } from '../models/misc';
+import Exit from '../assets/exit.svg';
 
 function AddPrompt({
   setModalState,
@@ -16,50 +17,54 @@ function AddPrompt({
   setNewPromptContent: (content: string) => void;
 }) {
   return (
-    <div className="rounded-3xl px-4 py-2">
-      <p className="mb-1 text-xl text-jet dark:text-bright-gray">Add Prompt</p>
-      <p className="mb-7 text-xs text-[#747474] dark:text-[#7F7F82]">
-        Add your custom prompt and save it to DocsGPT
-      </p>
-      <div>
-        <input
-          placeholder="Prompt Name"
-          type="text"
-          className="h-10 w-full rounded-lg border-2 border-silver px-3 outline-none dark:bg-transparent dark:text-silver"
-          value={newPromptName}
-          onChange={(e) => setNewPromptName(e.target.value)}
-        ></input>
-        <div className="relative bottom-12 left-3 mt-[-3.00px]">
-          <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
-            Prompt Name
-          </span>
+    <div className="relative">
+      <button
+        className="absolute top-3 right-4 m-2 w-3"
+        onClick={() => {
+          setModalState('INACTIVE');
+        }}
+      >
+        <img className="filter dark:invert" src={Exit} />
+      </button>
+      <div className="p-8">
+        <p className="mb-1 text-xl text-jet dark:text-bright-gray">
+          Add Prompt
+        </p>
+        <p className="mb-7 text-xs text-[#747474] dark:text-[#7F7F82]">
+          Add your custom prompt and save it to DocsGPT
+        </p>
+        <div>
+          <input
+            placeholder="Prompt Name"
+            type="text"
+            className="h-10 w-full rounded-lg border-2 border-silver px-3 outline-none dark:bg-transparent dark:text-silver"
+            value={newPromptName}
+            onChange={(e) => setNewPromptName(e.target.value)}
+          ></input>
+          <div className="relative bottom-12 left-3 mt-[-3.00px]">
+            <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
+              Prompt Name
+            </span>
+          </div>
+          <div className="relative top-[7px] left-3">
+            <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
+              Prompt Text
+            </span>
+          </div>
+          <textarea
+            className="h-56 w-full rounded-lg border-2 border-silver px-3 py-2 outline-none dark:bg-transparent dark:text-silver"
+            value={newPromptContent}
+            onChange={(e) => setNewPromptContent(e.target.value)}
+          ></textarea>
         </div>
-        <div className="relative top-[7px] left-3">
-          <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
-            Prompt Text
-          </span>
+        <div className="mt-6 flex flex-row-reverse">
+          <button
+            onClick={handleAddPrompt}
+            className="rounded-3xl bg-purple-30 px-5 py-2 text-sm text-white transition-all hover:opacity-90"
+          >
+            Save
+          </button>
         </div>
-        <textarea
-          className="h-56 w-full rounded-lg border-2 border-silver px-3 py-2 outline-none dark:bg-transparent dark:text-silver"
-          value={newPromptContent}
-          onChange={(e) => setNewPromptContent(e.target.value)}
-        ></textarea>
-      </div>
-      <div className="mt-6 flex flex-row-reverse gap-4">
-        <button
-          onClick={handleAddPrompt}
-          className="rounded-3xl bg-purple-30 px-5 py-2 text-white transition-all hover:opacity-90"
-        >
-          Save
-        </button>
-        <button
-          onClick={() => {
-            setModalState('INACTIVE');
-          }}
-          className="cursor-pointer font-medium dark:text-light-gray"
-        >
-          Cancel
-        </button>
       </div>
     </div>
   );
@@ -83,58 +88,62 @@ function EditPrompt({
   currentPromptEdit: { name: string; id: string; type: string };
 }) {
   return (
-    <div className="rounded-3xl px-4 py-2">
-      <p className="mb-1 text-xl text-jet dark:text-bright-gray">Edit Prompt</p>
-      <p className="mb-7 text-xs text-[#747474] dark:text-[#7F7F82]">
-        Edit your custom prompt and save it to DocsGPT
-      </p>
-      <div>
-        <input
-          placeholder="Prompt Name"
-          type="text"
-          className="h-10 w-full rounded-lg border-2 border-silver px-3 outline-none dark:bg-transparent dark:text-silver"
-          value={editPromptName}
-          onChange={(e) => setEditPromptName(e.target.value)}
-        ></input>
-        <div className="relative bottom-12 left-3 mt-[-3.00px]">
-          <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
-            Prompt Name
-          </span>
+    <div className="relative">
+      <button
+        className="absolute top-3 right-4 m-2 w-3"
+        onClick={() => {
+          setModalState('INACTIVE');
+        }}
+      >
+        <img className="filter dark:invert" src={Exit} />
+      </button>
+      <div className="p-8">
+        <p className="mb-1 text-xl text-jet dark:text-bright-gray">
+          Edit Prompt
+        </p>
+        <p className="mb-7 text-xs text-[#747474] dark:text-[#7F7F82]">
+          Edit your custom prompt and save it to DocsGPT
+        </p>
+        <div>
+          <input
+            placeholder="Prompt Name"
+            type="text"
+            className="h-10 w-full rounded-lg border-2 border-silver px-3 outline-none dark:bg-transparent dark:text-silver"
+            value={editPromptName}
+            onChange={(e) => setEditPromptName(e.target.value)}
+          ></input>
+          <div className="relative bottom-12 left-3 mt-[-3.00px]">
+            <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
+              Prompt Name
+            </span>
+          </div>
+          <div className="relative top-[7px] left-3">
+            <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
+              Prompt Text
+            </span>
+          </div>
+          <textarea
+            className="h-56 w-full rounded-lg border-2 border-silver px-3 py-2 outline-none dark:bg-transparent dark:text-silver"
+            value={editPromptContent}
+            onChange={(e) => setEditPromptContent(e.target.value)}
+          ></textarea>
         </div>
-        <div className="relative top-[7px] left-3">
-          <span className="bg-white px-1 text-xs text-silver dark:bg-outer-space dark:text-silver">
-            Prompt Text
-          </span>
+        <div className="mt-6 flex flex-row-reverse gap-4">
+          <button
+            className={`rounded-3xl bg-purple-30 px-5 py-2 text-sm text-white transition-all ${
+              currentPromptEdit.type === 'public'
+                ? 'cursor-not-allowed opacity-50'
+                : 'hover:opacity-90'
+            }`}
+            onClick={() => {
+              handleEditPrompt &&
+                handleEditPrompt(currentPromptEdit.id, currentPromptEdit.type);
+            }}
+            disabled={currentPromptEdit.type === 'public'}
+          >
+            Save
+          </button>
         </div>
-        <textarea
-          className="h-56 w-full rounded-lg border-2 border-silver px-3 py-2 outline-none dark:bg-transparent dark:text-silver"
-          value={editPromptContent}
-          onChange={(e) => setEditPromptContent(e.target.value)}
-        ></textarea>
-      </div>
-      <div className="mt-6 flex flex-row-reverse gap-4">
-        <button
-          className={`rounded-3xl bg-purple-30 px-5 py-2 text-white transition-all ${
-            currentPromptEdit.type === 'public'
-              ? 'cursor-not-allowed opacity-50'
-              : 'hover:opacity-90'
-          }`}
-          onClick={() => {
-            handleEditPrompt &&
-              handleEditPrompt(currentPromptEdit.id, currentPromptEdit.type);
-          }}
-          disabled={currentPromptEdit.type === 'public'}
-        >
-          Save
-        </button>
-        <button
-          onClick={() => {
-            setModalState('INACTIVE');
-          }}
-          className="cursor-pointer font-medium dark:text-light-gray"
-        >
-          Cancel
-        </button>
       </div>
     </div>
   );
@@ -205,7 +214,7 @@ export default function PromptsModal({
         modalState === 'ACTIVE' ? 'visible' : 'hidden'
       } fixed top-0 left-0 z-30  h-screen w-screen  bg-gray-alpha`}
     >
-      <article className="mx-auto mt-24 flex w-[90vw] max-w-lg  flex-col gap-4 rounded-lg bg-white p-6 shadow-lg dark:bg-outer-space">
+      <article className="mx-auto mt-24 flex w-[90vw] max-w-lg  flex-col gap-4 rounded-2xl bg-white shadow-lg dark:bg-outer-space">
         {view}
       </article>
     </article>

frontend/src/settings/APIKeys.tsx

@@ -0,0 +1,336 @@
+import React from 'react';
+import { useSelector } from 'react-redux';
+import Dropdown from '../components/Dropdown';
+import {
+  Doc,
+  CreateAPIKeyModalProps,
+  SaveAPIKeyModalProps,
+} from '../models/misc';
+import { selectSourceDocs } from '../preferences/preferenceSlice';
+import Exit from '../assets/exit.svg';
+import Trash from '../assets/trash.svg';
+
+const apiHost = import.meta.env.VITE_API_HOST || 'https://docsapi.arc53.com';
+const embeddingsName =
+  import.meta.env.VITE_EMBEDDINGS_NAME ||
+  'huggingface_sentence-transformers/all-mpnet-base-v2';
+
+const APIKeys: React.FC = () => {
+  const [isCreateModalOpen, setCreateModal] = React.useState(false);
+  const [isSaveKeyModalOpen, setSaveKeyModal] = React.useState(false);
+  const [newKey, setNewKey] = React.useState('');
+  const [apiKeys, setApiKeys] = React.useState<
+    { name: string; key: string; source: string; id: string }[]
+  >([]);
+  const handleDeleteKey = (id: string) => {
+    fetch(`${apiHost}/api/delete_api_key`, {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+      },
+      body: JSON.stringify({ id }),
+    })
+      .then((response) => {
+        if (!response.ok) {
+          throw new Error('Failed to delete API Key');
+        }
+        return response.json();
+      })
+      .then((data) => {
+        data.status === 'ok' &&
+          setApiKeys((previous) => previous.filter((elem) => elem.id !== id));
+      })
+      .catch((error) => {
+        console.error(error);
+      });
+  };
+  React.useEffect(() => {
+    fetchAPIKeys();
+  }, []);
+  const fetchAPIKeys = async () => {
+    try {
+      const response = await fetch(`${apiHost}/api/get_api_keys`);
+      if (!response.ok) {
+        throw new Error('Failed to fetch API Keys');
+      }
+      const apiKeys = await response.json();
+      setApiKeys(apiKeys);
+    } catch (error) {
+      console.log(error);
+    }
+  };
+  const createAPIKey = (payload: {
+    name: string;
+    source: string;
+    prompt_id: string;
+    chunks: string;
+  }) => {
+    fetch(`${apiHost}/api/create_api_key`, {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+      },
+      body: JSON.stringify(payload),
+    })
+      .then((response) => {
+        if (!response.ok) {
+          throw new Error('Failed to create API Key');
+        }
+        return response.json();
+      })
+      .then((data) => {
+        setApiKeys([...apiKeys, data]);
+        setCreateModal(false);
+        setNewKey(data.key);
+        setSaveKeyModal(true);
+        fetchAPIKeys();
+      })
+      .catch((error) => {
+        console.error(error);
+      });
+  };
+  return (
+    <div className="mt-8">
+      <div className="flex w-full flex-col lg:w-max">
+        <div className="flex justify-end">
+          <button
+            onClick={() => setCreateModal(true)}
+            className="rounded-full bg-purple-30 px-4 py-3 text-sm text-white hover:opacity-90"
+          >
+            Create new
+          </button>
+        </div>
+        {isCreateModalOpen && (
+          <CreateAPIKeyModal
+            close={() => setCreateModal(false)}
+            createAPIKey={createAPIKey}
+          />
+        )}
+        {isSaveKeyModalOpen && (
+          <SaveAPIKeyModal
+            apiKey={newKey}
+            close={() => setSaveKeyModal(false)}
+          />
+        )}
+        <div className="mt-[27px] w-full">
+          <div className="w-full overflow-x-auto">
+            <table className="block w-max table-auto content-center justify-center rounded-xl border text-center dark:border-chinese-silver dark:text-bright-gray">
+              <thead>
+                <tr>
+                  <th className="border-r p-4 md:w-[244px]">Name</th>
+                  <th className="w-[244px] border-r px-4 py-2">
+                    Source document
+                  </th>
+                  <th className="w-[244px] border-r px-4 py-2">API Key</th>
+                  <th className="px-4 py-2"></th>
+                </tr>
+              </thead>
+              <tbody>
+                {apiKeys?.map((element, index) => (
+                  <tr key={index}>
+                    <td className="border-r border-t p-4">{element.name}</td>
+                    <td className="border-r border-t p-4">{element.source}</td>
+                    <td className="border-r border-t p-4">{element.key}</td>
+                    <td className="border-t p-4">
+                      <img
+                        src={Trash}
+                        alt="Delete"
+                        className="h-4 w-4 cursor-pointer hover:opacity-50"
+                        id={`img-${index}`}
+                        onClick={() => handleDeleteKey(element.id)}
+                      />
+                    </td>
+                  </tr>
+                ))}
+              </tbody>
+            </table>
+          </div>
+        </div>
+      </div>
+    </div>
+  );
+};
+
+const CreateAPIKeyModal: React.FC<CreateAPIKeyModalProps> = ({
+  close,
+  createAPIKey,
+}) => {
+  const [APIKeyName, setAPIKeyName] = React.useState<string>('');
+  const [sourcePath, setSourcePath] = React.useState<{
+    label: string;
+    value: string;
+  } | null>(null);
+
+  const chunkOptions = ['0', '2', '4', '6', '8', '10'];
+  const [chunk, setChunk] = React.useState<string>('2');
+  const [activePrompts, setActivePrompts] = React.useState<
+    { name: string; id: string; type: string }[]
+  >([]);
+  const [prompt, setPrompt] = React.useState<{
+    name: string;
+    id: string;
+    type: string;
+  } | null>(null);
+  const docs = useSelector(selectSourceDocs);
+  React.useEffect(() => {
+    const fetchPrompts = async () => {
+      try {
+        const response = await fetch(`${apiHost}/api/get_prompts`);
+        if (!response.ok) {
+          throw new Error('Failed to fetch prompts');
+        }
+        const promptsData = await response.json();
+        setActivePrompts(promptsData);
+      } catch (error) {
+        console.error(error);
+      }
+    };
+    fetchPrompts();
+  }, []);
+  const extractDocPaths = () =>
+    docs
+      ? docs
+          .filter((doc) => doc.model === embeddingsName)
+          .map((doc: Doc) => {
+            let namePath = doc.name;
+            if (doc.language === namePath) {
+              namePath = '.project';
+            }
+            let docPath = 'default';
+            if (doc.location === 'local') {
+              docPath = 'local' + '/' + doc.name + '/';
+            } else if (doc.location === 'remote') {
+              docPath =
+                doc.language +
+                '/' +
+                namePath +
+                '/' +
+                doc.version +
+                '/' +
+                doc.model +
+                '/';
+            }
+            return {
+              label: doc.name,
+              value: docPath,
+            };
+          })
+      : [];
+
+  return (
+    <div className="fixed top-0 left-0 z-30 flex h-screen w-screen items-center justify-center bg-gray-alpha bg-opacity-50">
+      <div className="relative w-11/12 rounded-2xl bg-white p-10 dark:bg-outer-space sm:w-[512px]">
+        <button className="absolute top-3 right-4 m-2 w-3" onClick={close}>
+          <img className="filter dark:invert" src={Exit} />
+        </button>
+        <div className="mb-6">
+          <span className="text-xl text-jet dark:text-bright-gray">
+            Create New API Key
+          </span>
+        </div>
+        <div className="relative mt-5 mb-4">
+          <span className="absolute left-2 -top-2 bg-white px-2 text-xs text-gray-4000 dark:bg-outer-space dark:text-silver">
+            API Key Name
+          </span>
+          <input
+            type="text"
+            className="h-10 w-full rounded-md border-2 border-silver px-3 outline-none dark:bg-transparent dark:text-silver"
+            value={APIKeyName}
+            onChange={(e) => setAPIKeyName(e.target.value)}
+          />
+        </div>
+        <div className="my-4">
+          <Dropdown
+            placeholder="Source document"
+            selectedValue={sourcePath}
+            onSelect={(selection: { label: string; value: string }) =>
+              setSourcePath(selection)
+            }
+            options={extractDocPaths()}
+            size="w-full"
+            rounded="xl"
+          />
+        </div>
+        <div className="my-4">
+          <Dropdown
+            options={activePrompts}
+            selectedValue={prompt ? prompt.name : null}
+            placeholder="Select active prompt"
+            onSelect={(value: { name: string; id: string; type: string }) =>
+              setPrompt(value)
+            }
+            size="w-full"
+          />
+        </div>
+        <div className="my-4">
+          <p className="mb-2 ml-2 font-bold text-jet dark:text-bright-gray">
+            Chunks processed per query
+          </p>
+          <Dropdown
+            options={chunkOptions}
+            selectedValue={chunk}
+            onSelect={(value: string) => setChunk(value)}
+            size="w-full"
+          />
+        </div>
+        <button
+          disabled={!sourcePath || APIKeyName.length === 0 || !prompt}
+          onClick={() =>
+            sourcePath &&
+            prompt &&
+            createAPIKey({
+              name: APIKeyName,
+              source: sourcePath.value,
+              prompt_id: prompt.id,
+              chunks: chunk,
+            })
+          }
+          className="float-right mt-4 rounded-full bg-purple-30 px-4 py-3 text-white disabled:opacity-50"
+        >
+          Create
+        </button>
+      </div>
+    </div>
+  );
+};
+
+const SaveAPIKeyModal: React.FC<SaveAPIKeyModalProps> = ({ apiKey, close }) => {
+  const [isCopied, setIsCopied] = React.useState(false);
+  const handleCopyKey = () => {
+    navigator.clipboard.writeText(apiKey);
+    setIsCopied(true);
+  };
+  return (
+    <div className="fixed top-0 left-0 z-30 flex h-screen w-screen items-center justify-center bg-gray-alpha bg-opacity-50">
+      <div className="relative w-11/12 rounded-3xl bg-white px-6 py-8 dark:bg-outer-space dark:text-bright-gray sm:w-[512px]">
+        <button className="absolute top-3 right-4 m-2 w-3" onClick={close}>
+          <img className="filter dark:invert" src={Exit} />
+        </button>
+        <h1 className="my-0 text-xl font-medium">Please save your Key</h1>
+        <h3 className="text-sm font-normal text-outer-space">
+          This is the only time your key will be shown.
+        </h3>
+        <div className="flex justify-between py-2">
+          <div>
+            <h2 className="text-base font-semibold">API Key</h2>
+            <span className="text-sm font-normal leading-7 ">{apiKey}</span>
+          </div>
+          <button
+            className="my-1 h-10 w-20 rounded-full border border-purple-30 p-2 text-sm text-purple-30 hover:bg-purple-30 hover:text-white dark:border-purple-500 dark:text-purple-500"
+            onClick={handleCopyKey}
+          >
+            {isCopied ? 'Copied' : 'Copy'}
+          </button>
+        </div>
+        <button
+          onClick={close}
+          className="rounded-full bg-philippine-yellow px-4 py-3 font-medium text-black hover:bg-[#E6B91A]"
+        >
+          I saved the Key
+        </button>
+      </div>
+    </div>
+  );
+};
+
+export default APIKeys;

frontend/src/settings/Documents.tsx

@@ -0,0 +1,60 @@
+import { DocumentsProps } from '../models/misc';
+import Trash from '../assets/trash.svg';
+
+const Documents: React.FC<DocumentsProps> = ({
+  documents,
+  handleDeleteDocument,
+}) => {
+  return (
+    <div className="mt-8">
+      <div className="flex flex-col">
+        <div className="mt-[27px] w-max overflow-x-auto rounded-xl border dark:border-chinese-silver">
+          <table className="block w-full table-auto content-center justify-center text-center dark:text-bright-gray">
+            <thead>
+              <tr>
+                <th className="border-r p-4 md:w-[244px]">Document Name</th>
+                <th className="w-[244px] border-r px-4 py-2">Vector Date</th>
+                <th className="w-[244px] border-r px-4 py-2">Type</th>
+                <th className="px-4 py-2"></th>
+              </tr>
+            </thead>
+            <tbody>
+              {documents &&
+                documents.map((document, index) => (
+                  <tr key={index}>
+                    <td className="border-r border-t px-4 py-2">
+                      {document.name}
+                    </td>
+                    <td className="border-r border-t px-4 py-2">
+                      {document.date}
+                    </td>
+                    <td className="border-r border-t px-4 py-2">
+                      {document.location === 'remote'
+                        ? 'Pre-loaded'
+                        : 'Private'}
+                    </td>
+                    <td className="border-t px-4 py-2">
+                      {document.location !== 'remote' && (
+                        <img
+                          src={Trash}
+                          alt="Delete"
+                          className="h-4 w-4 cursor-pointer hover:opacity-50"
+                          id={`img-${index}`}
+                          onClick={(event) => {
+                            event.stopPropagation();
+                            handleDeleteDocument(index, document);
+                          }}
+                        />
+                      )}
+                    </td>
+                  </tr>
+                ))}
+            </tbody>
+          </table>
+        </div>
+      </div>
+    </div>
+  );
+};
+
+export default Documents;

frontend/src/settings/General.tsx

@@ -0,0 +1,100 @@
+import React from 'react';
+import { useSelector, useDispatch } from 'react-redux';
+import Prompts from './Prompts';
+import { useDarkTheme } from '../hooks';
+import Dropdown from '../components/Dropdown';
+import {
+  selectPrompt,
+  setPrompt,
+  setChunks,
+  selectChunks,
+} from '../preferences/preferenceSlice';
+
+const apiHost = import.meta.env.VITE_API_HOST || 'https://docsapi.arc53.com';
+
+const General: React.FC = () => {
+  const themes = ['Light', 'Dark'];
+  const languages = ['English'];
+  const chunks = ['0', '2', '4', '6', '8', '10'];
+  const [prompts, setPrompts] = React.useState<
+    { name: string; id: string; type: string }[]
+  >([]);
+  const selectedChunks = useSelector(selectChunks);
+  const [isDarkTheme, toggleTheme] = useDarkTheme();
+  const [selectedTheme, setSelectedTheme] = React.useState(
+    isDarkTheme ? 'Dark' : 'Light',
+  );
+  const dispatch = useDispatch();
+  const [selectedLanguage, setSelectedLanguage] = React.useState(languages[0]);
+  const selectedPrompt = useSelector(selectPrompt);
+
+  React.useEffect(() => {
+    const fetchPrompts = async () => {
+      try {
+        const response = await fetch(`${apiHost}/api/get_prompts`);
+        if (!response.ok) {
+          throw new Error('Failed to fetch prompts');
+        }
+        const promptsData = await response.json();
+        setPrompts(promptsData);
+      } catch (error) {
+        console.error(error);
+      }
+    };
+    fetchPrompts();
+  }, []);
+  return (
+    <div className="mt-[59px]">
+      <div className="mb-4">
+        <p className="font-bold text-jet dark:text-bright-gray">Select Theme</p>
+        <Dropdown
+          options={themes}
+          selectedValue={selectedTheme}
+          onSelect={(option: string) => {
+            setSelectedTheme(option);
+            option !== selectedTheme && toggleTheme();
+          }}
+          size="w-56"
+          rounded="3xl"
+        />
+      </div>
+      <div className="mb-4">
+        <p className="font-bold text-jet dark:text-bright-gray">
+          Select Language
+        </p>
+        <Dropdown
+          options={languages}
+          selectedValue={selectedLanguage}
+          onSelect={setSelectedLanguage}
+          size="w-56"
+          rounded="3xl"
+        />
+      </div>
+      <div className="mb-4">
+        <p className="font-bold text-jet dark:text-bright-gray">
+          Chunks processed per query
+        </p>
+        <Dropdown
+          options={chunks}
+          selectedValue={selectedChunks}
+          onSelect={(value: string) => dispatch(setChunks(value))}
+          size="w-56"
+          rounded="3xl"
+        />
+      </div>
+      <div>
+        <Prompts
+          prompts={prompts}
+          selectedPrompt={selectedPrompt}
+          onSelectPrompt={(name, id, type) =>
+            dispatch(setPrompt({ name: name, id: id, type: type }))
+          }
+          setPrompts={setPrompts}
+          apiHost={apiHost}
+        />
+      </div>
+    </div>
+  );
+};
+
+export default General;

frontend/src/settings/Prompts.tsx

@@ -0,0 +1,219 @@
+import React from 'react';
+import { PromptProps, ActiveState } from '../models/misc';
+import Dropdown from '../components/Dropdown';
+import PromptsModal from '../preferences/PromptsModal';
+
+const apiHost = import.meta.env.VITE_API_HOST || 'https://docsapi.arc53.com';
+
+const Prompts: React.FC<PromptProps> = ({
+  prompts,
+  selectedPrompt,
+  onSelectPrompt,
+  setPrompts,
+}) => {
+  const handleSelectPrompt = ({
+    name,
+    id,
+    type,
+  }: {
+    name: string;
+    id: string;
+    type: string;
+  }) => {
+    setEditPromptName(name);
+    onSelectPrompt(name, id, type);
+  };
+  const [newPromptName, setNewPromptName] = React.useState('');
+  const [newPromptContent, setNewPromptContent] = React.useState('');
+  const [editPromptName, setEditPromptName] = React.useState('');
+  const [editPromptContent, setEditPromptContent] = React.useState('');
+  const [currentPromptEdit, setCurrentPromptEdit] = React.useState({
+    id: '',
+    name: '',
+    type: '',
+  });
+  const [modalType, setModalType] = React.useState<'ADD' | 'EDIT'>('ADD');
+  const [modalState, setModalState] = React.useState<ActiveState>('INACTIVE');
+
+  const handleAddPrompt = async () => {
+    try {
+      const response = await fetch(`${apiHost}/api/create_prompt`, {
+        method: 'POST',
+        headers: {
+          'Content-Type': 'application/json',
+        },
+        body: JSON.stringify({
+          name: newPromptName,
+          content: newPromptContent,
+        }),
+      });
+      if (!response.ok) {
+        throw new Error('Failed to add prompt');
+      }
+      const newPrompt = await response.json();
+      if (setPrompts) {
+        setPrompts([
+          ...prompts,
+          { name: newPromptName, id: newPrompt.id, type: 'private' },
+        ]);
+      }
+      setModalState('INACTIVE');
+      onSelectPrompt(newPromptName, newPrompt.id, newPromptContent);
+      setNewPromptName(newPromptName);
+    } catch (error) {
+      console.error(error);
+    }
+  };
+
+  const handleDeletePrompt = (id: string) => {
+    setPrompts(prompts.filter((prompt) => prompt.id !== id));
+    fetch(`${apiHost}/api/delete_prompt`, {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+      },
+      body: JSON.stringify({ id: id }),
+    })
+      .then((response) => {
+        if (!response.ok) {
+          throw new Error('Failed to delete prompt');
+        }
+        // get 1st prompt and set it as selected
+        if (prompts.length > 0) {
+          onSelectPrompt(prompts[0].name, prompts[0].id, prompts[0].type);
+        }
+      })
+      .catch((error) => {
+        console.error(error);
+      });
+  };
+
+  const fetchPromptContent = async (id: string) => {
+    console.log('fetching prompt content');
+    try {
+      const response = await fetch(
+        `${apiHost}/api/get_single_prompt?id=${id}`,
+        {
+          method: 'GET',
+          headers: {
+            'Content-Type': 'application/json',
+          },
+        },
+      );
+      if (!response.ok) {
+        throw new Error('Failed to fetch prompt content');
+      }
+      const promptContent = await response.json();
+      setEditPromptContent(promptContent.content);
+    } catch (error) {
+      console.error(error);
+    }
+  };
+
+  const handleSaveChanges = (id: string, type: string) => {
+    fetch(`${apiHost}/api/update_prompt`, {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+      },
+      body: JSON.stringify({
+        id: id,
+        name: editPromptName,
+        content: editPromptContent,
+      }),
+    })
+      .then((response) => {
+        if (!response.ok) {
+          throw new Error('Failed to update prompt');
+        }
+        if (setPrompts) {
+          const existingPromptIndex = prompts.findIndex(
+            (prompt) => prompt.id === id,
+          );
+          if (existingPromptIndex === -1) {
+            setPrompts([
+              ...prompts,
+              { name: editPromptName, id: id, type: type },
+            ]);
+          } else {
+            const updatedPrompts = [...prompts];
+            updatedPrompts[existingPromptIndex] = {
+              name: editPromptName,
+              id: id,
+              type: type,
+            };
+            setPrompts(updatedPrompts);
+          }
+        }
+        setModalState('INACTIVE');
+        onSelectPrompt(editPromptName, id, type);
+      })
+      .catch((error) => {
+        console.error(error);
+      });
+  };
+
+  return (
+    <>
+      <div>
+        <div className="mb-4 flex flex-row items-center gap-8">
+          <div>
+            <p className="font-semibold dark:text-bright-gray">Active Prompt</p>
+            <Dropdown
+              options={prompts}
+              selectedValue={selectedPrompt.name}
+              onSelect={handleSelectPrompt}
+              size="w-56"
+              rounded="3xl"
+              showEdit
+              showDelete
+              onEdit={({
+                id,
+                name,
+                type,
+              }: {
+                id: string;
+                name: string;
+                type: string;
+              }) => {
+                setModalType('EDIT');
+                setEditPromptName(name);
+                fetchPromptContent(id);
+                setCurrentPromptEdit({ id: id, name: name, type: type });
+                setModalState('ACTIVE');
+              }}
+              onDelete={handleDeletePrompt}
+            />
+          </div>
+          <button
+            className="mt-[24px] rounded-3xl border-2 border-solid border-purple-30 px-5 py-3 text-purple-30 hover:bg-purple-30 hover:text-white"
+            onClick={() => {
+              setModalType('ADD');
+              setModalState('ACTIVE');
+            }}
+          >
+            Add new
+          </button>
+        </div>
+      </div>
+      <PromptsModal
+        type={modalType}
+        modalState={modalState}
+        setModalState={setModalState}
+        newPromptName={newPromptName}
+        setNewPromptName={setNewPromptName}
+        newPromptContent={newPromptContent}
+        setNewPromptContent={setNewPromptContent}
+        editPromptName={editPromptName}
+        setEditPromptName={setEditPromptName}
+        editPromptContent={editPromptContent}
+        setEditPromptContent={setEditPromptContent}
+        currentPromptEdit={currentPromptEdit}
+        handleAddPrompt={handleAddPrompt}
+        handleEditPrompt={handleSaveChanges}
+      />
+    </>
+  );
+};
+
+export default Prompts;

frontend/src/settings/Widgets.tsx

@@ -0,0 +1,112 @@
+import React from 'react';
+import Dropdown from '../components/Dropdown';
+
+const Widgets: React.FC<{
+  widgetScreenshot: File | null;
+  onWidgetScreenshotChange: (screenshot: File | null) => void;
+}> = ({ widgetScreenshot, onWidgetScreenshotChange }) => {
+  const widgetSources = ['Source 1', 'Source 2', 'Source 3'];
+  const widgetMethods = ['Method 1', 'Method 2', 'Method 3'];
+  const widgetTypes = ['Type 1', 'Type 2', 'Type 3'];
+
+  const [selectedWidgetSource, setSelectedWidgetSource] = React.useState(
+    widgetSources[0],
+  );
+  const [selectedWidgetMethod, setSelectedWidgetMethod] = React.useState(
+    widgetMethods[0],
+  );
+  const [selectedWidgetType, setSelectedWidgetType] = React.useState(
+    widgetTypes[0],
+  );
+
+  // const [widgetScreenshot, setWidgetScreenshot] = useState<File | null>(null);
+  const [widgetCode, setWidgetCode] = React.useState<string>(''); // Your widget code state
+
+  const handleScreenshotChange = (
+    event: React.ChangeEvent<HTMLInputElement>,
+  ) => {
+    const files = event.target.files;
+
+    if (files && files.length > 0) {
+      const selectedScreenshot = files[0];
+      onWidgetScreenshotChange(selectedScreenshot); // Update the screenshot in the parent component
+    }
+  };
+
+  const handleCopyToClipboard = () => {
+    // Create a new textarea element to select the text
+    const textArea = document.createElement('textarea');
+    textArea.value = widgetCode;
+    document.body.appendChild(textArea);
+
+    // Select and copy the text
+    textArea.select();
+    document.execCommand('copy');
+
+    // Clean up the textarea element
+    document.body.removeChild(textArea);
+  };
+
+  return (
+    <div>
+      <div className="mt-[59px]">
+        <p className="font-bold text-jet">Widget Source</p>
+        <Dropdown
+          options={widgetSources}
+          selectedValue={selectedWidgetSource}
+          onSelect={setSelectedWidgetSource}
+        />
+      </div>
+      <div className="mt-5">
+        <p className="font-bold text-jet">Widget Method</p>
+        <Dropdown
+          options={widgetMethods}
+          selectedValue={selectedWidgetMethod}
+          onSelect={setSelectedWidgetMethod}
+        />
+      </div>
+      <div className="mt-5">
+        <p className="font-bold text-jet">Widget Type</p>
+        <Dropdown
+          options={widgetTypes}
+          selectedValue={selectedWidgetType}
+          onSelect={setSelectedWidgetType}
+        />
+      </div>
+      <div className="mt-6">
+        <p className="font-bold text-jet">Widget Code Snippet</p>
+        <textarea
+          rows={4}
+          value={widgetCode}
+          onChange={(e) => setWidgetCode(e.target.value)}
+          className="mt-3 w-full rounded-lg border-2 p-2"
+        />
+      </div>
+      <div className="mt-1">
+        <button
+          onClick={handleCopyToClipboard}
+          className="rounded-lg bg-blue-400 px-2 py-2 font-bold text-white transition-all hover:bg-blue-600"
+        >
+          Copy
+        </button>
+      </div>
+
+      <div className="mt-4">
+        <p className="text-lg font-semibold">Widget Screenshot</p>
+        <input type="file" accept="image/*" onChange={handleScreenshotChange} />
+      </div>
+
+      {widgetScreenshot && (
+        <div className="mt-4">
+          <img
+            src={URL.createObjectURL(widgetScreenshot)}
+            alt="Widget Screenshot"
+            className="max-w-full rounded-lg border border-gray-300"
+          />
+        </div>
+      )}
+    </div>
+  );
+};
+
+export default Widgets;

frontend/src/settings/index.tsx

@@ -0,0 +1,127 @@
+import React from 'react';
+import { useSelector, useDispatch } from 'react-redux';
+import General from './General';
+import Documents from './Documents';
+import APIKeys from './APIKeys';
+import Widgets from './Widgets';
+import {
+  selectSourceDocs,
+  setSourceDocs,
+} from '../preferences/preferenceSlice';
+import { Doc } from '../preferences/preferenceApi';
+import ArrowLeft from '../assets/arrow-left.svg';
+import ArrowRight from '../assets/arrow-right.svg';
+
+const apiHost = import.meta.env.VITE_API_HOST || 'https://docsapi.arc53.com';
+
+const Settings: React.FC = () => {
+  const dispatch = useDispatch();
+  const tabs = ['General', 'Documents', 'API Keys'];
+  const [activeTab, setActiveTab] = React.useState('General');
+  const [widgetScreenshot, setWidgetScreenshot] = React.useState<File | null>(
+    null,
+  );
+
+  const documents = useSelector(selectSourceDocs);
+  const updateWidgetScreenshot = (screenshot: File | null) => {
+    setWidgetScreenshot(screenshot);
+  };
+  const handleDeleteClick = (index: number, doc: Doc) => {
+    const docPath = 'indexes/' + 'local' + '/' + doc.name;
+    fetch(`${apiHost}/api/delete_old?path=${docPath}`, {
+      method: 'GET',
+    })
+      .then((response) => {
+        if (response.ok && documents) {
+          const updatedDocuments = [
+            ...documents.slice(0, index),
+            ...documents.slice(index + 1),
+          ];
+          dispatch(setSourceDocs(updatedDocuments));
+        }
+      })
+      .catch((error) => console.error(error));
+  };
+  return (
+    <div className="wa p-4 pt-20 md:p-12">
+      <p className="text-2xl font-bold text-eerie-black dark:text-bright-gray">
+        Settings
+      </p>
+      <div className="mt-6 flex flex-row items-center space-x-4 overflow-x-auto md:space-x-8 ">
+        <div className="md:hidden">
+          <button
+            onClick={() => scrollTabs(-1)}
+            className="flex h-8 w-8 items-center justify-center rounded-full border-2 border-purple-30 transition-all hover:bg-gray-100"
+          >
+            <img src={ArrowLeft} alt="left-arrow" className="h-6 w-6" />
+          </button>
+        </div>
+        <div className="flex flex-nowrap space-x-4 overflow-x-auto md:space-x-8">
+          {tabs.map((tab, index) => (
+            <button
+              key={index}
+              onClick={() => setActiveTab(tab)}
+              className={`h-9 rounded-3xl px-4 font-bold ${
+                activeTab === tab
+                  ? 'bg-purple-3000 text-purple-30 dark:bg-dark-charcoal'
+                  : 'text-gray-6000'
+              }`}
+            >
+              {tab}
+            </button>
+          ))}
+        </div>
+        <div className="md:hidden">
+          <button
+            onClick={() => scrollTabs(1)}
+            className="flex h-8 w-8 items-center justify-center rounded-full border-2 border-purple-30 hover:bg-gray-100"
+          >
+            <img src={ArrowRight} alt="right-arrow" className="h-6 w-6" />
+          </button>
+        </div>
+      </div>
+      {renderActiveTab()}
+
+      {/* {activeTab === 'Widgets' && (
+        <Widgets
+          widgetScreenshot={widgetScreenshot}
+          onWidgetScreenshotChange={updateWidgetScreenshot}
+        />
+      )} */}
+    </div>
+  );
+
+  function scrollTabs(direction: number) {
+    const container = document.querySelector('.flex-nowrap');
+    if (container) {
+      container.scrollLeft += direction * 100; // Adjust the scroll amount as needed
+    }
+  }
+
+  function renderActiveTab() {
+    switch (activeTab) {
+      case 'General':
+        return <General />;
+      case 'Documents':
+        return (
+          <Documents
+            documents={documents}
+            handleDeleteDocument={handleDeleteClick}
+          />
+        );
+      case 'Widgets':
+        return (
+          <Widgets
+            widgetScreenshot={widgetScreenshot} // Add this line
+            onWidgetScreenshotChange={updateWidgetScreenshot} // Add this line
+          />
+        );
+      case 'API Keys':
+        return <APIKeys />;
+      default:
+        return null;
+    }
+  }
+};
+
+export default Settings;

frontend/src/upload/Upload.tsx

@@ -201,7 +201,7 @@ export default function Upload({
 
   const { getRootProps, getInputProps, isDragActive } = useDropzone({
     onDrop,
-    multiple: false,
+    multiple: true,
     onDragEnter: doNothing,
     onDragOver: doNothing,
     onDragLeave: doNothing,
@@ -302,7 +302,6 @@ export default function Upload({
         {activeTab === 'remote' && (
           <>
             <Dropdown
-              fullWidth
               options={urlOptions}
               selectedValue={urlType}
               onSelect={(value: { label: string; value: string }) =>
@@ -420,7 +419,7 @@ export default function Upload({
             disabled={
               (files.length === 0 || docName.trim().length === 0) &&
               activeTab === 'file'
-            } // Disable the button if no file is selected or docName is empty
+            }
           >
             Train
           </button>
@@ -451,4 +450,3 @@ export default function Upload({
     </article>
   );
 }
-// TODO: sanitize all inputs

tests/llm/test_sagemaker.py

@@ -54,7 +54,7 @@ class TestSagemakerAPILLM(unittest.TestCase):
     def test_gen(self):
         with patch.object(self.sagemaker.runtime, 'invoke_endpoint', 
                           return_value=self.response) as mock_invoke_endpoint:
-            output = self.sagemaker.gen(None, None, self.messages)
+            output = self.sagemaker.gen(None, self.messages)
             mock_invoke_endpoint.assert_called_once_with(
                 EndpointName=self.sagemaker.endpoint,
                 ContentType='application/json',
@@ -66,7 +66,7 @@ class TestSagemakerAPILLM(unittest.TestCase):
     def test_gen_stream(self):
         with patch.object(self.sagemaker.runtime, 'invoke_endpoint_with_response_stream', 
                           return_value=self.response) as mock_invoke_endpoint:
-            output = list(self.sagemaker.gen_stream(None, None, self.messages))
+            output = list(self.sagemaker.gen_stream(None, self.messages))
             mock_invoke_endpoint.assert_called_once_with(
                 EndpointName=self.sagemaker.endpoint,
                 ContentType='application/json',