Update app.py
Browse files
app.py
CHANGED
|
@@ -8,6 +8,14 @@ from langchain.chains import ConversationalRetrievalChain, ConversationChain
|
|
| 8 |
from langchain.memory import ConversationBufferMemory
|
| 9 |
from langchain.document_loaders import PyPDFLoader
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
def create_sidebar():
|
| 12 |
with st.sidebar:
|
| 13 |
st.title("PDF Chat")
|
|
@@ -36,7 +44,6 @@ def save_uploaded_file(uploaded_file, path='./uploads/'):
|
|
| 36 |
f.write(uploaded_file.getbuffer())
|
| 37 |
return file_path
|
| 38 |
|
| 39 |
-
@st.cache_data
|
| 40 |
def load_texts_from_papers(papers):
|
| 41 |
all_texts = []
|
| 42 |
for paper in papers:
|
|
@@ -57,77 +64,77 @@ def load_texts_from_papers(papers):
|
|
| 57 |
st.error(f"Error processing {paper.name}: {str(e)}")
|
| 58 |
return all_texts
|
| 59 |
|
| 60 |
-
|
| 61 |
-
def initialize_vectorstore(api_key): # Added api_key parameter
|
| 62 |
embedding = OpenAIEmbeddings(openai_api_key=api_key)
|
| 63 |
vectorstore = Chroma(embedding_function=embedding, persist_directory="db")
|
| 64 |
return vectorstore
|
| 65 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
def main():
|
| 67 |
st.set_page_config(page_title="PDF Chat", layout="wide")
|
| 68 |
-
|
| 69 |
-
# Get API key from sidebar
|
| 70 |
api_key = create_sidebar()
|
| 71 |
-
|
| 72 |
-
st.title("Chat with PDF")
|
| 73 |
-
papers = st.file_uploader("Upload PDFs", type=["pdf"], accept_multiple_files=True)
|
| 74 |
-
|
| 75 |
-
if "messages" not in st.session_state:
|
| 76 |
-
st.session_state.messages = []
|
| 77 |
-
|
| 78 |
if not api_key:
|
| 79 |
st.warning("Please enter your OpenAI API key")
|
| 80 |
return
|
| 81 |
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
)
|
| 96 |
-
st.success("PDF processed successfully!")
|
| 97 |
-
else:
|
| 98 |
-
memory = ConversationBufferMemory(memory_key="chat_history")
|
| 99 |
-
qa_chain = ConversationChain(
|
| 100 |
-
llm=ChatOpenAI(temperature=0, model_name="gpt-3.5-turbo", openai_api_key=api_key), # Added api_key here
|
| 101 |
-
memory=memory
|
| 102 |
-
)
|
| 103 |
-
|
| 104 |
-
# Chat interface
|
| 105 |
for message in st.session_state.messages:
|
| 106 |
with st.chat_message(message["role"]):
|
| 107 |
st.markdown(message["content"])
|
| 108 |
-
|
|
|
|
| 109 |
if prompt := st.chat_input("Ask about your PDFs"):
|
| 110 |
-
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
|
| 114 |
-
with st.chat_message("assistant"):
|
| 115 |
-
try:
|
| 116 |
-
if texts:
|
| 117 |
-
result = qa_chain({"question": prompt})
|
| 118 |
-
response = result["answer"]
|
| 119 |
-
else:
|
| 120 |
-
result = qa_chain.predict(input=prompt)
|
| 121 |
-
response = result
|
| 122 |
-
|
| 123 |
-
st.session_state.messages.append({"role": "assistant", "content": response})
|
| 124 |
-
st.markdown(response)
|
| 125 |
-
|
| 126 |
-
except Exception as e:
|
| 127 |
-
st.error(f"Error: {str(e)}")
|
| 128 |
-
|
| 129 |
-
except Exception as e:
|
| 130 |
-
st.error(f"Error: {str(e)}")
|
| 131 |
|
| 132 |
if __name__ == "__main__":
|
| 133 |
main()
|
|
|
|
| 8 |
from langchain.memory import ConversationBufferMemory
|
| 9 |
from langchain.document_loaders import PyPDFLoader
|
| 10 |
|
| 11 |
+
# Initialize session state variables
|
| 12 |
+
if "messages" not in st.session_state:
|
| 13 |
+
st.session_state.messages = []
|
| 14 |
+
if "chain" not in st.session_state:
|
| 15 |
+
st.session_state.chain = None
|
| 16 |
+
if "processed_pdfs" not in st.session_state:
|
| 17 |
+
st.session_state.processed_pdfs = False
|
| 18 |
+
|
| 19 |
def create_sidebar():
|
| 20 |
with st.sidebar:
|
| 21 |
st.title("PDF Chat")
|
|
|
|
| 44 |
f.write(uploaded_file.getbuffer())
|
| 45 |
return file_path
|
| 46 |
|
|
|
|
| 47 |
def load_texts_from_papers(papers):
|
| 48 |
all_texts = []
|
| 49 |
for paper in papers:
|
|
|
|
| 64 |
st.error(f"Error processing {paper.name}: {str(e)}")
|
| 65 |
return all_texts
|
| 66 |
|
| 67 |
+
def initialize_vectorstore(api_key):
|
|
|
|
| 68 |
embedding = OpenAIEmbeddings(openai_api_key=api_key)
|
| 69 |
vectorstore = Chroma(embedding_function=embedding, persist_directory="db")
|
| 70 |
return vectorstore
|
| 71 |
|
| 72 |
+
def process_pdfs(papers, api_key):
|
| 73 |
+
if papers and not st.session_state.processed_pdfs:
|
| 74 |
+
with st.spinner("Processing PDFs..."):
|
| 75 |
+
texts = load_texts_from_papers(papers)
|
| 76 |
+
if texts:
|
| 77 |
+
vectorstore = initialize_vectorstore(api_key)
|
| 78 |
+
vectorstore.add_documents(texts)
|
| 79 |
+
st.session_state.chain = ConversationalRetrievalChain.from_llm(
|
| 80 |
+
ChatOpenAI(temperature=0, model_name="gpt-3.5-turbo", openai_api_key=api_key),
|
| 81 |
+
vectorstore.as_retriever(),
|
| 82 |
+
memory=ConversationBufferMemory(
|
| 83 |
+
memory_key="chat_history",
|
| 84 |
+
return_messages=True
|
| 85 |
+
)
|
| 86 |
+
)
|
| 87 |
+
st.session_state.processed_pdfs = True
|
| 88 |
+
st.success("PDFs processed successfully!")
|
| 89 |
+
return texts
|
| 90 |
+
return []
|
| 91 |
+
|
| 92 |
+
def handle_chat(prompt, texts):
|
| 93 |
+
st.session_state.messages.append({"role": "user", "content": prompt})
|
| 94 |
+
|
| 95 |
+
try:
|
| 96 |
+
if texts or st.session_state.processed_pdfs:
|
| 97 |
+
result = st.session_state.chain({"question": prompt})
|
| 98 |
+
response = result["answer"]
|
| 99 |
+
else:
|
| 100 |
+
response = "Please upload a PDF first."
|
| 101 |
+
|
| 102 |
+
st.session_state.messages.append({"role": "assistant", "content": response})
|
| 103 |
+
|
| 104 |
+
except Exception as e:
|
| 105 |
+
st.error(f"Error: {str(e)}")
|
| 106 |
+
|
| 107 |
def main():
|
| 108 |
st.set_page_config(page_title="PDF Chat", layout="wide")
|
| 109 |
+
|
|
|
|
| 110 |
api_key = create_sidebar()
|
| 111 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 112 |
if not api_key:
|
| 113 |
st.warning("Please enter your OpenAI API key")
|
| 114 |
return
|
| 115 |
|
| 116 |
+
st.title("Chat with PDF")
|
| 117 |
+
|
| 118 |
+
# File uploader
|
| 119 |
+
papers = st.file_uploader("Upload PDFs", type=["pdf"], accept_multiple_files=True)
|
| 120 |
+
|
| 121 |
+
# Process PDFs
|
| 122 |
+
texts = process_pdfs(papers, api_key)
|
| 123 |
+
|
| 124 |
+
# Chat interface container
|
| 125 |
+
chat_container = st.container()
|
| 126 |
+
|
| 127 |
+
with chat_container:
|
| 128 |
+
# Display chat messages
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 129 |
for message in st.session_state.messages:
|
| 130 |
with st.chat_message(message["role"]):
|
| 131 |
st.markdown(message["content"])
|
| 132 |
+
|
| 133 |
+
# Chat input
|
| 134 |
if prompt := st.chat_input("Ask about your PDFs"):
|
| 135 |
+
handle_chat(prompt, texts)
|
| 136 |
+
# Force a rerun to display the new message
|
| 137 |
+
st.rerun()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
|
| 139 |
if __name__ == "__main__":
|
| 140 |
main()
|