Skip to content

Instantly share code, notes, and snippets.

@MichelNivard
Created February 16, 2023 20:56
Show Gist options
  • Star 10 You must be signed in to star a gist
  • Fork 1 You must be signed in to fork a gist
  • Save MichelNivard/e8fd2efae5114a189f40d9eb6ea18dc2 to your computer and use it in GitHub Desktop.
Save MichelNivard/e8fd2efae5114a189f40d9eb6ea18dc2 to your computer and use it in GitHub Desktop.
import tkinter
import customtkinter
from bs4 import BeautifulSoup
# Langchain loads:
from langchain.document_loaders import DirectoryLoader,PagedPDFSplitter
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.text_splitter import CharacterTextSplitter
from langchain.vectorstores import ElasticVectorSearch, Pinecone, Weaviate, FAISS, Qdrant
from langchain.chains.qa_with_sources import load_qa_with_sources_chain
from langchain.llms import OpenAI
from langchain.prompts import PromptTemplate
# import filedialog module
from tkinter import filedialog
customtkinter.set_appearance_mode("System") # Modes: system (default), light, dark
customtkinter.set_default_color_theme("blue") # Themes: blue (default), dark-blue, green
app = customtkinter.CTk() # create CTk window like you do with the Tk window
app.geometry("1080x960")
OPENAI_API_KEY="YOUR API GOES HERE!"
embeddings = OpenAIEmbeddings(openai_api_key=OPENAI_API_KEY)
# Function for opening the
# file explorer window
def browseFiles():
filename = filedialog.askdirectory()
loader = DirectoryLoader(filename, glob="*.pdf")
pdfs = loader.load_and_split()
faiss_index = FAISS.from_documents(pdfs,embeddings)
faiss_index.save_local("faiss_index")
# Function for opening the
# file explorer window
def browseDocStore():
faiss_file = filedialog.askdirectory(initialdir = "/", title = "Select a Vectorstore Index")
faiss_index = FAISS.load_local(faiss_file , embeddings)
# Search Button
def button_function():
# clear previous search
textbox.delete("0.0", "end")
textbox1.delete("0.0", "end")
textbox2.delete("0.0", "end")
textbox3.delete("0.0", "end")
textbox4.delete("0.0", "end")
textbox1b.delete("0.0", "end")
textbox2b.delete("0.0", "end")
textbox3b.delete("0.0", "end")
textbox4b.delete("0.0", "end")
# get question:
query = entry.get()
faiss_index = FAISS.load_local("faiss_index" , embeddings)
docs = faiss_index.similarity_search(query=query, k=5)
# set up prompt:
question_prompt_template = """Summarize the following text in 250 tokens or less:
{context}"""
QUESTION_PROMPT = PromptTemplate(
template=question_prompt_template, input_variables=["context"]
)
combine_prompt_template = """Given the following extracted parts of a number of long documents and a question, create a final answer.
If you don't know the answer, just say that you don't know. Don't try to make up an answer.
QUESTION: {question}
=========
{summaries}
=========
FINAL ANSWER:"""
COMBINE_PROMPT = PromptTemplate(
template=combine_prompt_template, input_variables=["summaries", "question"]
)
chain = load_qa_with_sources_chain(OpenAI(temperature=0,openai_api_key=OPENAI_API_KEY), chain_type="map_reduce", return_intermediate_steps=True, question_prompt=QUESTION_PROMPT, combine_prompt=COMBINE_PROMPT)
# get answer and display the sources:
answer = chain({"input_documents": docs, "question": query}, return_only_outputs=False)
textbox.insert("0.0", answer["output_text"])
textbox1.insert("0.0", answer["intermediate_steps"][0])
textbox1b.insert("0.0", answer["input_documents"][0])
textbox2.insert("0.0", answer["intermediate_steps"][1])
textbox2b.insert("0.0", answer["input_documents"][1])
textbox3.insert("0.0", answer["intermediate_steps"][2])
textbox3b.insert("0.0", answer["input_documents"][2])
textbox4.insert("0.0", answer["intermediate_steps"][3])
textbox4b.insert("0.0", answer["input_documents"][3])
# seach field
entry = customtkinter.CTkEntry(master=app, placeholder_text="query",width = 600)
entry.pack(padx=20, pady=10)
entry.place(relx=0.5, rely=0.03, anchor=tkinter.CENTER)
# output textbox
textbox = customtkinter.CTkTextbox(master=app, width = 750, height=150)
textbox.pack(pady=20, padx=10)
textbox.place(relx=0.57, rely=0.15, anchor=tkinter.CENTER)
tabview = customtkinter.CTkTabview(master=app, width=750)
tabview.pack(pady=20, padx=10)
tabview.place(relx=0.5, rely=0.6, anchor=tkinter.CENTER)
tabview.add("Source and Summary 1")
tabview.add("Source and Summary 2")
tabview.add("Source and Summary 3")
tabview.add("Source and Summary 4")
# Sources textbox
textbox1 = customtkinter.CTkTextbox(tabview.tab("Source and Summary 1"), width = 950, height=150)
textbox1.pack(pady=20, padx=10)
textbox1b = customtkinter.CTkTextbox(tabview.tab("Source and Summary 1"), width = 950, height=350)
textbox1b.pack(pady=20, padx=10)
textbox2 = customtkinter.CTkTextbox(tabview.tab("Source and Summary 2"), width = 950, height=150)
textbox2.pack(pady=20, padx=10)
textbox2b = customtkinter.CTkTextbox(tabview.tab("Source and Summary 2"), width = 950, height=350)
textbox2b.pack(pady=20, padx=10)
textbox3 = customtkinter.CTkTextbox(tabview.tab("Source and Summary 3"), width = 950, height=150)
textbox3.pack(pady=20, padx=10)
textbox3b = customtkinter.CTkTextbox(tabview.tab("Source and Summary 3"), width = 950, height=350)
textbox3b.pack(pady=20, padx=10)
textbox4 = customtkinter.CTkTextbox(tabview.tab("Source and Summary 4"), width = 950, height=150)
textbox4.pack(pady=20, padx=10)
textbox4b = customtkinter.CTkTextbox(tabview.tab("Source and Summary 4"), width = 950, height=350)
textbox4b.pack(pady=20, padx=10)
# Browse for folder of ducuments:
button = customtkinter.CTkButton(master=app, text="Index files", command=browseFiles)
button.place(relx=0.1, rely=0.1, anchor=tkinter.CENTER)
# Browse for an existing faiss store:
button = customtkinter.CTkButton(master=app, text="Load existing Index", command=browseDocStore)
button.place(relx=0.1, rely=0.05, anchor=tkinter.CENTER)
# defien seach button:
button2 = customtkinter.CTkButton(master=app, text="Search", command=button_function)
button2.place(relx=0.9, rely=0.03, anchor=tkinter.CENTER)
app.mainloop()
@pvsundar
Copy link

pvsundar commented Mar 5, 2023

Hello, when I run this (after much trying and editing and installing) I get the following error:
... # get question:
... query = entry.get()
NameError: name 'entry' is not defined

faiss_index = FAISS.load_local("faiss_index" , embeddings)
RuntimeError: Error in __cdecl faiss::FileIOReader::FileIOReader(const char *) at D:\a\faiss-wheels\faiss-wheels\faiss\faiss\impl\io.cpp:68: Error: 'f' failed: could not open faiss_index\index.faiss for reading: No such file or directory
docs = faiss_index.similarity_search(query=query, k=5)
NameError: name 'faiss_index' is not defined

@pvsundar
Copy link

pvsundar commented Mar 5, 2023

I also added in:
pip install faiss-cpu

@pvsundar
Copy link

pvsundar commented Mar 5, 2023

image

Despite all that, I do get this dialog box!! But now, I dont know what I am supposed to select.

@pvsundar
Copy link

pvsundar commented Mar 5, 2023

i have now also added in
pip install unstructured
I now get the following error;
... query = entry.get()
TclError: invalid command name ".!ctkentry.!entry"

faiss_index = FAISS.load_local("faiss_index" , embeddings)
RuntimeError: Error in __cdecl faiss::FileIOReader::FileIOReader(const char *) at D:\a\faiss-wheels\faiss-wheels\faiss\faiss\impl\io.cpp:68: Error: 'f' failed: could not open faiss_index\index.faiss for reading: No such file or directory
docs = faiss_index.similarity_search(query=query, k=5)
NameError: name 'faiss_index' is not defined

@MichelNivard
Copy link
Author

Sorry for your trouble! Now you click “index files” and select a folder of files to index. When they done you can search the file with ChatGPT!

@pvsundar
Copy link

pvsundar commented Mar 6, 2023

Sorry to bother you. I am not able to get it to work. I still get the following error:
... # get question:
... query = entry.get()
NameError: name 'entry' is not defined

faiss_index = FAISS.load_local("faiss_index" , embeddings)
RuntimeError: Error in __cdecl faiss::FileIOReader::FileIOReader(const char *) at D:\a\faiss-wheels\faiss-wheels\faiss\faiss\impl\io.cpp:68: Error: 'f' failed: could not open faiss_index\index.faiss for reading: No such file or directory
docs = faiss_index.similarity_search(query=query, k=5)
NameError: name 'faiss_index' is not defined

@pvsundar
Copy link

pvsundar commented Mar 6, 2023

Would it be possible to start with something much smaller. Open one PDF file and get it to provide the key points in a bullet for summary? THanks for the consideration.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment