michel/Philosophy-RAG-demo
1
0
Fork 0
forked from AI_team/Philosophy-RAG-demo
Code Pull Requests Activity

✨ Create single source aggregation definition

Browse Source
This commit is contained in:
Ruben Lucas 2025-04-09 16:03:35 +02:00
parent 3295bb8992
commit ab1235bd28
2 changed files with 15 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
generic_rag/app.py
View File

@ -93,15 +93,7 @@ async def on_message(message: cl.Message):
await process_response(message) await process_response(message)
async def process_response(message): async def add_sources(chainlit_response: cl.Message, pdf_sources: dict, web_sources: set | list):
config = {"configurable": {"thread_id": cl.user_session.get("id")}}
chainlit_response = cl.Message(content="")
async for response in graph.stream(message.content, config=config):
await chainlit_response.stream_token(response)
pdf_sources = graph.get_last_pdf_sources()
if len(pdf_sources) > 0: if len(pdf_sources) > 0:
await chainlit_response.stream_token("\nThe following PDF source were consulted:\n") await chainlit_response.stream_token("\nThe following PDF source were consulted:\n")
for source, page_numbers in pdf_sources.items(): for source, page_numbers in pdf_sources.items():
@ -111,13 +103,24 @@ async def process_response(message):
chainlit_response.elements.append(cl.Pdf(name="pdf", display="inline", path=source, page=page_numbers[0])) chainlit_response.elements.append(cl.Pdf(name="pdf", display="inline", path=source, page=page_numbers[0]))
await chainlit_response.update() await chainlit_response.update()
await chainlit_response.stream_token(f"- '{source}' on page(s): {page_numbers}\n") await chainlit_response.stream_token(f"- '{source}' on page(s): {page_numbers}\n")
web_sources = graph.get_last_web_sources()
if len(web_sources) > 0: if len(web_sources) > 0:
await chainlit_response.stream_token("\nThe following web sources were consulted:\n") await chainlit_response.stream_token("\nThe following web sources were consulted:\n")
for source in web_sources: for source in web_sources:
await chainlit_response.stream_token(f"- {source}\n") await chainlit_response.stream_token(f"- {source}\n")
async def process_response(message):
config = {"configurable": {"thread_id": cl.user_session.get("id")}}
chainlit_response = cl.Message(content="")
async for response in graph.stream(message.content, config=config):
await chainlit_response.stream_token(response)
pdf_sources = graph.get_last_pdf_sources()
web_sources = graph.get_last_web_sources()
await add_sources(chainlit_response, pdf_sources, web_sources)
await chainlit_response.send() await chainlit_response.send()
@ -129,20 +132,7 @@ async def process_cond_response(message):
for response in graph.stream(message.content, config=config): for response in graph.stream(message.content, config=config):
await chainlit_response.stream_token(response) await chainlit_response.stream_token(response)
if len(graph.last_retrieved_docs) > 0: await add_sources(chainlit_response, graph.last_retrieved_docs, graph.last_retrieved_sources)
await chainlit_response.stream_token("\nThe following PDF source were consulted:\n")
for source, page_numbers in graph.last_retrieved_docs.items():
page_numbers = list(page_numbers)
page_numbers.sort()
# display="side" seems to be not supported by chainlit for PDF's, so we use "inline" instead.
chainlit_response.elements.append(cl.Pdf(name="pdf", display="inline", path=source, page=page_numbers[0]))
await chainlit_response.update()
await chainlit_response.stream_token(f"- '{source}' on page(s): {page_numbers}\n")
if len(graph.last_retrieved_sources) > 0:
await chainlit_response.stream_token("\nThe following web sources were consulted:\n")
for source in graph.last_retrieved_sources:
await chainlit_response.stream_token(f"- {source}\n")
await chainlit_response.send() await chainlit_response.send()

1
generic_rag/graphs/cond_ret_gen.py
View File

@ -63,7 +63,6 @@ class CondRetGenLangGraph:
) # Use re.DOTALL if dicts might span newlines ) # Use re.DOTALL if dicts might span newlines
for dict_str in dictionary_strings: for dict_str in dictionary_strings:
parsed_dict = ast.literal_eval(dict_str) parsed_dict = ast.literal_eval(dict_str)
print(parsed_dict)
if "filetype" in parsed_dict and parsed_dict["filetype"] == "web": if "filetype" in parsed_dict and parsed_dict["filetype"] == "web":
self.last_retrieved_sources.add(parsed_dict["source"]) self.last_retrieved_sources.add(parsed_dict["source"])
elif Path(parsed_dict["source"]).suffix == ".pdf": elif Path(parsed_dict["source"]).suffix == ".pdf":