Upload 2 files

app.py

@@ -0,0 +1,73 @@
+import os
+import gradio as gr
+from langchain.document_loaders import PyPDFLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.vectorstores import FAISS
+from langchain.chains import RetrievalQA
+from langchain.llms import HuggingFaceHub
+
+# LLM for summarization and glossary extraction
+llm = HuggingFaceHub(repo_id="google/flan-t5-base", model_kwargs={"temperature": 0.7, "max_length": 512})
+
+summary_cache = ""
+glossary_cache = ""
+retriever_chain = None
+
+def extract_text_and_summary(file):
+    global retriever_chain, summary_cache, glossary_cache
+
+    loader = PyPDFLoader(file.name)
+    docs = loader.load()
+
+    splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
+    splits = splitter.split_documents(docs)
+    full_text = "\n".join([doc.page_content for doc in splits])
+
+    embeddings = HuggingFaceEmbeddings()
+    db = FAISS.from_documents(splits, embeddings)
+    retriever = db.as_retriever()
+    retriever_chain = RetrievalQA.from_chain_type(llm=llm, retriever=retriever)
+
+    summary_prompt = f"Summarize this legal document:\n{full_text[:1500]}"
+    glossary_prompt = f"Extract and define legal terms from the document:\n{full_text[:1500]}"
+
+    summary_cache = llm(summary_prompt)[0]['generated_text']
+    glossary_cache = llm(glossary_prompt)[0]['generated_text']
+
+    # Save downloadable summary file
+    filename = "summary_output.txt"
+    with open(filename, "w", encoding="utf-8") as f:
+        f.write("=== Summary ===\n")
+        f.write(summary_cache + "\n\n")
+        f.write("=== Glossary ===\n")
+        f.write(glossary_cache + "\n")
+
+    return full_text, summary_cache, glossary_cache, filename
+
+def answer_custom_question(question):
+    if retriever_chain:
+        return retriever_chain.run(question)
+    return "Please upload and process a document first."
+
+with gr.Blocks() as demo:
+    gr.Markdown("## 🧾 Legal Document Summarizer Using LangChain")
+
+    with gr.Row():
+        file = gr.File(label="📁 Upload Legal PDF", file_types=[".pdf"])
+        process_btn = gr.Button("🔍 Extract & Summarize")
+
+    extracted_text = gr.Textbox(label="📄 Extracted Text", lines=10)
+    summary_output = gr.Textbox(label="📝 Summary", lines=5)
+    glossary_output = gr.Textbox(label="📘 Glossary", lines=5)
+    download_link = gr.File(label="⬇️ Download Summary")
+
+    with gr.Row():
+        user_question = gr.Textbox(label="❓ Ask a Custom Question")
+        custom_answer = gr.Textbox(label="🤖 AI Answer")
+        ask_btn = gr.Button("🧠 Get Answer")
+
+    process_btn.click(fn=extract_text_and_summary, inputs=file, outputs=[extracted_text, summary_output, glossary_output, download_link])
+    ask_btn.click(fn=answer_custom_question, inputs=user_question, outputs=custom_answer)
+
+demo.launch()

requirements.txt

@@ -0,0 +1,9 @@
+gradio==4.14.0
+langchain==0.1.14
+huggingface_hub
+pdfplumber
+python-docx
+tiktoken
+faiss-cpu
+transformers==4.40.1
+torch