Update app.py

Update for the 2026 hugging face compatibility

app.py

@@ -1,134 +1,82 @@
-
-# from ibm_watsonx_ai.foundation_models import ModelInference
-# from ibm_watsonx_ai.metanames import GenTextParamsMetaNames as GenParams
-# from ibm_watsonx_ai.metanames import EmbedTextParamsMetaNames
-# from ibm_watsonx_ai import Credentials
-# from langchain_ibm import WatsonxLLM, WatsonxEmbeddings
-
 import os
 import uuid
+import gradio as gr
 from langchain_huggingface import HuggingFaceEndpoint
-
 from langchain_text_splitters import RecursiveCharacterTextSplitter
 from langchain_community.vectorstores import Chroma
 from langchain_community.document_loaders import PyPDFLoader
-from langchain.chains import RetrievalQA
-#from huggingface_hub import HfFolder
 from langchain_community.embeddings import HuggingFaceEmbeddings
-from huggingface_hub import InferenceClient
-
-import gradio as gr
-## You can use this section to suppress warnings generated by your code:
-def warn(*args, **kwargs):
-    pass
-import warnings
-warnings.warn = warn
-warnings.filterwarnings('ignore')
-
-## LLM
-# def get_llm():
-#     model_id = 'ibm/granite-3-2-8b-instruct'
-#     parameters = {
-#         GenParams.MAX_NEW_TOKENS: 256, # This controls the maximum number of tokens in the generated output
-#         GenParams.TEMPERATURE: 0.5, # This controls randomness or creativity of the model's responses
-#     }
-#     project_id = 'skills-network'
-#     watsonx_llm = WatsonxLLM(
-#         model_id = model_id,
-#         url="https://us-south.ml.cloud.ibm.com",
-#         project_id = project_id,
-#         params = parameters,
-#     )
-#     return watsonx_llm
+from langchain.chains.combine_documents import create_stuff_documents_chain
+from langchain.chains import create_retrieval_chain
+from langchain_core.prompts import ChatPromptTemplate
 
+# 1. LLM SETUP (Using Granite via HF API)
 def get_llm():
-    # This uses Hugging Face's free Inference API to run the model
-    # You don't need Ollama for this!
-    llm = HuggingFaceEndpoint(
+    return HuggingFaceEndpoint(
         repo_id="ibm-granite/granite-3.1-8b-instruct", 
         task="text-generation",
         max_new_tokens=512,
         temperature=0.1,
-        huggingfacehub_api_token=os.getenv("HF_TOKEN") #
+        huggingfacehub_api_token=os.getenv("HF_TOKEN")
     )
-    return llm
 
-## Document loader
-def document_loader(file):
-    loader = PyPDFLoader(file)
-    loaded_document = loader.load()
-    return loaded_document
-## Text splitter
-def text_splitter(data):
-    text_splitter = RecursiveCharacterTextSplitter(
-        chunk_size = 1000,
-        chunk_overlap = 50,
-        length_function = len,
-    )
-    chuncks = text_splitter.split_documents(data)
-    return chuncks
-## Embedding model
-# def watsonx_embedding():
-#     embed_params = {
-#         EmbedTextParamsMetaNames.TRUNCATE_INPUT_TOKENS: 256,
-#         EmbedTextParamsMetaNames.RETURN_OPTIONS: {'input_text': True},
-#     }
+# 2. DOCUMENT PROCESSING
+def process_pdf(file_path):
+    loader = PyPDFLoader(file_path)
+    data = loader.load()
     
-#     watsonx_embedding = WatsonxEmbeddings(
-#         model_id = "ibm/slate-125m-english-rtrvr-v2",
-#         url="https://us-south.ml.cloud.ibm.com",
-#         project_id = "skills-network",
-#         params = embed_params,
-#     )
-#     return watsonx_embedding
-
-def watsonx_embedding():
-    return HuggingFaceEmbeddings(
-        model_name="sentence-transformers/all-MiniLM-L6-v2"
-    )
-
-## Vector db
-def vector_database(chunks):
+    text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=50)
+    chunks = text_splitter.split_documents(data)
+    
+    # Clean empty chunks
+    chunks = [c for c in chunks if len(c.page_content.strip()) > 10]
     
-    chunks = [
-        c for c in chunks
-        if c.page_content and c.page_content.strip() and len(c.page_content.strip()) > 10
-    ]
-    embedding_model = watsonx_embedding()
-    vectordb = Chroma.from_documents(chunks, embedding_model)
-    return vectordb
-## Retriever
-def retriever(file):
-    splits = document_loader(file)
-    chunks = text_splitter(splits)
-    vectordb = vector_database(chunks)
-    retriever = vectordb.as_retriever()
+    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
     
-    return retriever
-## QA Chain
+    # We use a unique collection name to prevent memory crossover
+    vectordb = Chroma.from_documents(
+        documents=chunks, 
+        embedding=embeddings,
+        collection_name=f"pdf_{uuid.uuid4().hex}"
+    )
+    return vectordb.as_retriever()
+
+# 3. QA CHAIN (The 2026 Modern Way)
 def retriever_qa(file, query):
+    if file is None:
+        return "Please upload a PDF first."
+        
     llm = get_llm()
-    retriever_obj = retriever(file)
-    qa = RetrievalQA.from_chain_type(
-        llm = llm, 
-        chain_type = 'stuff',
-        retriever = retriever_obj,
-        return_source_documents = False
+    retriever_obj = process_pdf(file)
+
+    # Define the Prompt
+    system_prompt = (
+        "Use the following pieces of retrieved context to answer the question. "
+        "If you don't know the answer, say you don't know. \n\n {context}"
     )
-    response = qa.invoke(query)
-    return response['result']
-    
-## Create Gradio interface
+    prompt = ChatPromptTemplate.from_messages([
+        ("system", system_prompt),
+        ("human", "{input}"),
+    ])
+
+    # Create the modern retrieval chain
+    combine_docs_chain = create_stuff_documents_chain(llm, prompt)
+    rag_chain = create_retrieval_chain(retriever_obj, combine_docs_chain)
+
+    # Invoke and return result
+    response = rag_chain.invoke({"input": query})
+    return response["answer"] # Note: 'answer' instead of 'result'
+
+# 4. GRADIO INTERFACE
 rag_application = gr.Interface(
-    fn = retriever_qa,
-    flagging_mode  = 'never',
-    inputs = [
-        gr.File(label = 'Upload PDF File', file_count = 'single', file_types = ['.pdf'], type ='filepath'), # Drag and drop file upload
-        gr.Textbox(label= 'Input Query', lines = 2, placeholder = 'Type your question here...')
+    fn=retriever_qa,
+    flagging_mode='never',
+    inputs=[
+        gr.File(label='Upload PDF', file_types=['.pdf'], type='filepath'),
+        gr.Textbox(label='Your Question', lines=2)
     ],
-    outputs = gr.Textbox(label = 'Ouput'),
-    title= 'PDF QA Chatbot',
-    description = 'Upload a PDF document and ask any question. The chatbot will try to answer using the provided document.'
+    outputs=gr.Textbox(label='AI Answer'),
+    title='Granite PDF Chatbot'
 )
-# Launch the app
-rag_application.launch()
+
+rag_application.launch()