Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -3,6 +3,7 @@ import streamlit as st
|
|
3 |
import os
|
4 |
import pickle
|
5 |
from PyPDF2 import PdfReader
|
|
|
6 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
7 |
from langchain.embeddings.openai import OpenAIEmbeddings
|
8 |
from langchain.vectorstores import FAISS
|
@@ -29,65 +30,42 @@ with st.sidebar:
|
|
29 |
def main():
|
30 |
st.header("Chat with PDF 💬")
|
31 |
|
32 |
-
#
|
33 |
-
|
34 |
-
|
35 |
-
if
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
|
|
42 |
text_splitter = RecursiveCharacterTextSplitter(
|
43 |
chunk_size=512,
|
44 |
chunk_overlap=128,
|
45 |
length_function=len
|
46 |
)
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
store_name
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
# PINECONE_API_ENV = os.environ.get('PINECONE_API_ENV',
|
68 |
-
# 'asia-southeast1-gcp-free') # You may need to switch with your env
|
69 |
-
# embeddings = OpenAIEmbeddings()
|
70 |
-
# # initialize pinecone
|
71 |
-
# pinecone.init(
|
72 |
-
# api_key=PINECONE_API_KEY, # find at app.pinecone.io
|
73 |
-
# environment=PINECONE_API_ENV # next to api key in console
|
74 |
-
# )
|
75 |
-
# index_name = "indexer" # put in the name of your pinecone index here
|
76 |
-
# VectorStore = Pinecone.from_texts(chunks, embeddings, index_name=index_name)
|
77 |
-
|
78 |
-
# Accept user questions/query
|
79 |
-
query = st.text_input("Ask questions about your PDF file:")
|
80 |
-
# st.write(query)
|
81 |
-
|
82 |
-
if query:
|
83 |
-
docs = VectorStore.similarity_search(query=query, k=3)
|
84 |
-
|
85 |
-
llm = OpenAI()
|
86 |
-
chain = load_qa_chain(llm=llm, chain_type="stuff")
|
87 |
-
with get_openai_callback() as cb:
|
88 |
-
response = chain.run(input_documents=docs, question=query)
|
89 |
-
print(cb)
|
90 |
-
st.write(response)
|
91 |
|
92 |
|
93 |
if __name__ == '__main__':
|
|
|
3 |
import os
|
4 |
import pickle
|
5 |
from PyPDF2 import PdfReader
|
6 |
+
from langchain.document_loaders import UnstructuredPDFLoader, OnlinePDFLoader, PyPDFLoader
|
7 |
from langchain.text_splitter import RecursiveCharacterTextSplitter
|
8 |
from langchain.embeddings.openai import OpenAIEmbeddings
|
9 |
from langchain.vectorstores import FAISS
|
|
|
30 |
def main():
|
31 |
st.header("Chat with PDF 💬")
|
32 |
|
33 |
+
# # embeddings
|
34 |
+
store_name = "coffee"
|
35 |
+
|
36 |
+
if os.path.exists(f"{store_name}.pkl"):
|
37 |
+
with open(f"{store_name}.pkl", "rb") as f:
|
38 |
+
VectorStore = pickle.load(f)
|
39 |
+
st.write('Embeddings Loaded from the Disk')
|
40 |
+
else:
|
41 |
+
st.write('Reading from prompt ...')
|
42 |
+
loader = PyPDFLoader("./咖啡语料.pdf")
|
43 |
+
data = loader.load()
|
44 |
text_splitter = RecursiveCharacterTextSplitter(
|
45 |
chunk_size=512,
|
46 |
chunk_overlap=128,
|
47 |
length_function=len
|
48 |
)
|
49 |
+
texts = text_splitter.split_documents(data)
|
50 |
+
embeddings = OpenAIEmbeddings()
|
51 |
+
VectorStore = FAISS.from_texts([t.page_content for t in texts], embedding=embeddings)
|
52 |
+
with open(f"{store_name}.pkl", "wb") as f:
|
53 |
+
pickle.dump(VectorStore, f)
|
54 |
+
|
55 |
+
|
56 |
+
|
57 |
+
query = st.text_input("Ask questions about Starbucks coffee:")
|
58 |
+
|
59 |
+
|
60 |
+
if query:
|
61 |
+
docs = VectorStore.similarity_search(query=query, k=3)
|
62 |
+
|
63 |
+
llm = OpenAI()
|
64 |
+
chain = load_qa_chain(llm=llm, chain_type="stuff")
|
65 |
+
with get_openai_callback() as cb:
|
66 |
+
response = chain.run(input_documents=docs, question=query)
|
67 |
+
print(cb)
|
68 |
+
st.write(response)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
69 |
|
70 |
|
71 |
if __name__ == '__main__':
|