Tried vectorstore cache but failed.

app.py

@@ -34,6 +34,7 @@ async def on_chat_start():
     ).send()
 
     file = files[0]
+    print(f"filename is {file.name}")
 
     doc = pymupdf.Document(file.path)
     toc = doc.get_toc()
@@ -61,12 +62,18 @@ async def on_chat_start():
     # need a rect that will exclude headers and footers
     rect = pymupdf.Rect(0.0, 100.0, 612.0, 650.0)
 
-    #create the final text
+    #capture the first 2 page
     extracted_text = ""
+
+
     for page in doc.pages():
-        if page.number in range(start_page-1, end_page):
+        if page.number in [0, 1, 2]:
+            extracted_text += page.get_text()
+        elif page.number in range(start_page-1, end_page):
             # print(page.get_text(clip=rect))
             extracted_text += page.get_text(clip=rect)
+
+
     msg = cl.Message(
         content=f"""Processing selected file: `{file.name}`...
         Extraction beginning on page {start_page} and ending on page {end_page}.
@@ -102,29 +109,29 @@ async def on_chat_start():
 
     await msg.send()
 
+    qdrant_vectorstore = getVectorstore(document, file.name)
 
-    qdrant_vectorstore = getVectorstore(document, file.path)
+    protocol_retriever = qdrant_vectorstore.as_retriever(search_kwargs={"k":15})
+    # document_titles = [file.name]
 
-    document_titles = ["protocol.pdf", "consent.pdf"]
 
-    # protocol_retriever = qdrant_vectorstore.as_retriever()
-
-    # protocol_retriever = create_protocol_retriever(document_titles)
     protocol_retriever = qdrant_vectorstore.as_retriever(
         search_kwargs={
             'filter': rest.Filter(
-                must=[
+                 must=[
                     rest.FieldCondition(
                         key="metadata.document_title",
-                        match=rest.MatchAny(any=document_titles)
+                         match=rest.MatchAny(any=[file.name])
                     )
                 ]
             ),
-            'k':15,
+            'k': 15,                                       
         }
     )
+    # # protocol_retriever = qdrant_vectorstore.as_retriever()
 
-
+    # protocol_retriever = create_protocol_retriever(document_titles)
+ 
     # Create prompt
     rag_prompt = ChatPromptTemplate.from_template(prompts.rag_prompt_template)
 
@@ -134,3 +141,23 @@ async def on_chat_start():
         {"context": itemgetter("question") | protocol_retriever, "question": itemgetter("question")}
         | rag_prompt | llm | StrOutputParser()
     )
+
+    from datetime import date
+    # Heading for top of ICF document
+    protocol_title = rag_chain.invoke({"question": "What is the exact title of this protocol?  Only return the title itself without any other description."})
+    principal_investigator = rag_chain.invoke({"question":"What is the name of the principal investigator of the study?  Only return the name itself without any other description."})
+    support = rag_chain.invoke({"question":"What agency is funding the study?  Only return the name of the agency without any other description."})
+    version_date = date.today().strftime("%B %d, %Y")
+
+    msg = cl.Message(
+        content=f""" 
+        **Study Title:** {protocol_title}
+        **Principal Investigator:** {principal_investigator}
+        **Version Date:** {version_date}
+        **Source of Support:** {support}
+        ---
+        """
+    )
+
+    await msg.send()
+    

getVectorstore.py

@@ -19,29 +19,36 @@ This could also be useful if different versions of documents are in existence.
 recreate a large vectorstore.  But the user could select the most recent version.
 """
 
-
 def get_document_hash(doc_content):
     """Generate a unique hash for the document content."""
     return hashlib.md5(doc_content.encode()).hexdigest()
 
-def getVectorstore(document, file_path):
+def getVectorstore(document, file_name):
     # Add a unique hash to your documents
     for doc in document:
         doc.metadata['content_hash'] = get_document_hash(doc.page_content)
 
     # Add the document title
     for doc in document:
-        doc.metadata['document_title'] = file_path.split('/')[-1]
+        doc.metadata['document_title'] = file_name
+
+    # Add page to metadata
+    for i, doc in enumerate(document):
+        doc.metadata['source'] = f"source_{i}"
 
+    # collection_name = f"pdf_to_parse_{uuid.uuid4()}"
+    collection_name = "protocol_collection"
+    
     client = QdrantClient( url=qdrant_url)
-    # client = QdrantClient(":memory:")
+
+
     # If the collection exists, then we need to check to see if our document is already
     # present, in which case we would not want to store it again.
-    if client.collection_exists("protocol_collection"):
+    if client.collection_exists(collection_name):
         print("Collection exists")
         qdrant_vectorstore = QdrantVectorStore.from_existing_collection(
             embedding=embedding_model,
-            collection_name="protocol_collection",
+            collection_name=collection_name,
             url=qdrant_url
             # location = ":memory:"
         )
@@ -61,7 +68,7 @@ def getVectorstore(document, file_path):
         )
         
         scroll_results = client.scroll(
-            collection_name="protocol_collection",
+            collection_name=collection_name,
             scroll_filter=scroll_filter,
             limit=len(document)  # Adjust this if you have a large number of documents
         )
@@ -82,8 +89,8 @@ def getVectorstore(document, file_path):
         qdrant_vectorstore = QdrantVectorStore.from_documents(
             documents=document,
             embedding=embedding_model,
-            collection_name="protocol_collection",
+            collection_name=collection_name,
             # location = ":memory:"
             url=qdrant_url
         )
-    return qdrant_vectorstore
+    return qdrant_vectorstore

pyproject.toml

@@ -14,8 +14,8 @@ dependencies = [
     "langchain-openai>=0.2.14",
     "langgraph>=0.2.60",
     "pymupdf>=1.25.1",
-    "chainlit>=1.1.202",
     "websockets>=14.1",
+    "chainlit>=1.1.202",
 ]
 
 [dependency-groups]

uv.lock

@@ -258,7 +258,7 @@ dependencies = [
     { name = "pydantic", marker = "python_full_version >= '3.13'" },
     { name = "pyjwt", marker = "python_full_version >= '3.13'" },
     { name = "python-dotenv", marker = "python_full_version >= '3.13'" },
-    { name = "python-multipart", marker = "python_full_version >= '3.13'" },
+    { name = "python-multipart", version = "0.0.9", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.13'" },
     { name = "starlette", version = "0.37.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.13'" },
     { name = "syncer", marker = "python_full_version >= '3.13'" },
     { name = "tomli", marker = "python_full_version >= '3.13'" },
@@ -273,7 +273,7 @@ wheels = [
 
 [[package]]
 name = "chainlit"
-version = "1.3.2"
+version = "2.0.0"
 source = { registry = "https://pypi.org/simple" }
 resolution-markers = [
     "python_full_version < '3.12.4'",
@@ -295,7 +295,7 @@ dependencies = [
     { name = "pydantic", marker = "python_full_version < '3.13'" },
     { name = "pyjwt", marker = "python_full_version < '3.13'" },
     { name = "python-dotenv", marker = "python_full_version < '3.13'" },
-    { name = "python-multipart", marker = "python_full_version < '3.13'" },
+    { name = "python-multipart", version = "0.0.18", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.13'" },
     { name = "python-socketio", marker = "python_full_version < '3.13'" },
     { name = "starlette", version = "0.41.3", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.13'" },
     { name = "syncer", marker = "python_full_version < '3.13'" },
@@ -304,9 +304,9 @@ dependencies = [
     { name = "uvicorn", marker = "python_full_version < '3.13'" },
     { name = "watchfiles", marker = "python_full_version < '3.13'" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/6a/57/8c3c354b3e82fae0791b58304634b356aeea8791dd0b6f817a8d8787cad6/chainlit-1.3.2.tar.gz", hash = "sha256:5562246ca42c858a42e86efdd5a46c27fd93481706f6fe47416109c47ac41b77", size = 4271845 }
+sdist = { url = "https://files.pythonhosted.org/packages/45/24/424679b769664876093b3e42167911535d1739bc1bc88f3963c69affed9e/chainlit-2.0.0.tar.gz", hash = "sha256:47b3a274a20cefb443f356d69f1c6a48818d67eb4a11552c749bfa6f414423ed", size = 4637040 }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/ca/d8/2a0c00b2236c93f31df429642de6414c56510d39392101b1b102074520d0/chainlit-1.3.2-py3-none-any.whl", hash = "sha256:c0926e8b157a67aede511a8220307889d04e9458ff3f94176e004fe533662d74", size = 4332216 },
+    { url = "https://files.pythonhosted.org/packages/87/2a/e2bbb86fc3a34c7bf798644edb95bf14fd79a8b3f6c99e4b27e5df1e24f0/chainlit-2.0.0-py3-none-any.whl", hash = "sha256:2b58ac6b513d94aef0380d1d68b73f74718c0c844586b050ce8d5e0a82eb8133", size = 4703622 },
 ]
 
 [[package]]
@@ -783,7 +783,7 @@ version = "0.1.0"
 source = { virtual = "." }
 dependencies = [
     { name = "chainlit", version = "1.1.202", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version >= '3.13'" },
-    { name = "chainlit", version = "1.3.2", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.13'" },
+    { name = "chainlit", version = "2.0.0", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.13'" },
     { name = "langchain" },
     { name = "langchain-community" },
     { name = "langchain-openai" },
@@ -1808,11 +1808,27 @@ wheels = [
 name = "python-multipart"
 version = "0.0.9"
 source = { registry = "https://pypi.org/simple" }
+resolution-markers = [
+    "python_full_version >= '3.13'",
+]
 sdist = { url = "https://files.pythonhosted.org/packages/5c/0f/9c55ac6c84c0336e22a26fa84ca6c51d58d7ac3a2d78b0dfa8748826c883/python_multipart-0.0.9.tar.gz", hash = "sha256:03f54688c663f1b7977105f021043b0793151e4cb1c1a9d4a11fc13d622c4026", size = 31516 }
 wheels = [
     { url = "https://files.pythonhosted.org/packages/3d/47/444768600d9e0ebc82f8e347775d24aef8f6348cf00e9fa0e81910814e6d/python_multipart-0.0.9-py3-none-any.whl", hash = "sha256:97ca7b8ea7b05f977dc3849c3ba99d51689822fab725c3703af7c866a0c2b215", size = 22299 },
 ]
 
+[[package]]
+name = "python-multipart"
+version = "0.0.18"
+source = { registry = "https://pypi.org/simple" }
+resolution-markers = [
+    "python_full_version < '3.12.4'",
+    "python_full_version >= '3.12.4' and python_full_version < '3.13'",
+]
+sdist = { url = "https://files.pythonhosted.org/packages/b4/86/b6b38677dec2e2e7898fc5b6f7e42c2d011919a92d25339451892f27b89c/python_multipart-0.0.18.tar.gz", hash = "sha256:7a68db60c8bfb82e460637fa4750727b45af1d5e2ed215593f917f64694d34fe", size = 36622 }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/13/6b/b60f47101ba2cac66b4a83246630e68ae9bbe2e614cbae5f4465f46dee13/python_multipart-0.0.18-py3-none-any.whl", hash = "sha256:efe91480f485f6a361427a541db4796f9e1591afc0fb8e7a4ba06bfbc6708996", size = 24389 },
+]
+
 [[package]]
 name = "python-socketio"
 version = "5.12.1"