app.py

import gradio as gr
import os
from langchain.vectorstores import Chroma
from langchain.document_loaders import PyPDFLoader
from langchain.text_splitter import CharacterTextSplitter

from langchain.embeddings import HuggingFaceInferenceAPIEmbeddings

# Use Hugging Face Inference API embeddings
inference_api_key = os.environ['HF']
api_hf_embeddings = HuggingFaceInferenceAPIEmbeddings(
    api_key=inference_api_key,
    model_name="sentence-transformers/all-MiniLM-l6-v2"
)

# Load and process the PDF files
loader = PyPDFLoader("new_papers/ReACT.pdf")
loader
documents = loader.load()
print("-----------")
print(documents)
print("-----------")

# Load the document, split it into chunks, embed each chunk and load it into the vector store.
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
vdocuments = text_splitter.split_documents(documents)






# Create Chroma vector store for API embeddings
api_db = Chroma.from_documents(vdocuments, api_hf_embeddings, collection_name="api-collection")
#api_db = Chroma.from_texts(documents, api_hf_embeddings, collection_name="api-collection")

#Similarity search
query = "What did the president say about Ketanji Brown Jackson"
docs = db.similarity_search(query)
print(docs[0].page_content)


class PDFRetrievalTool:
    def __init__(self, retriever):
        self.retriever = retriever

    def __call__(self, query):
        # Run the query through the retriever
        response = self.retriever.run(query)
        return response['result']

# Create Gradio interface for the API retriever
api_tool = gr.Interface(
    PDFRetrievalTool(api_db.as_retriever(search_kwargs={"k": 1})),
    inputs=gr.Textbox(),
    outputs=gr.Textbox(),
    live=True,
    title="API PDF Retrieval Tool",
    description="This tool indexes PDF documents and retrieves relevant answers based on a given query (HF Inference API Embeddings).",
)

# Launch the Gradio interface
api_tool.launch()