Create app.py

app.py

@@ -0,0 +1,105 @@
+import streamlit as st
+import speech_recognition as sr
+from gtts import gTTS
+import requests
+import os
+import tempfile
+from langchain.chains import ConversationalRetrievalChain
+from langchain.vectorstores import FAISS
+from langchain.embeddings import HuggingFaceEmbeddings
+from langchain.text_splitter import CharacterTextSplitter
+from langchain.docstore.document import Document
+from langchain.prompts import PromptTemplate
+from langchain.llms import Anthropic
+from langchain.llms import AI21  # Use Claude or AI21 API
+
+# LLM API settings (choose between Claude or AI21)
+CLAUDE_API_KEY = st.secrets["claude_api_key"]
+# Initialize language model and embeddings
+llm = Anthropic(api_key=CLAUDE_API_KEY)  # Or AI21 for the API
+
+# Function for speech-to-text (Audio to Text)
+def record_audio():
+    recognizer = sr.Recognizer()
+    with sr.Microphone() as source:
+        st.info("Listening... Please speak now.")
+        audio = recognizer.listen(source)
+        try:
+            text_urdu = recognizer.recognize_google(audio, language="ur")
+            return text_urdu
+        except sr.UnknownValueError:
+            st.error("Sorry, I could not understand the audio.")
+            return None
+
+# Function for text-to-speech (Text to Audio)
+def text_to_speech_urdu(text):
+    tts = gTTS(text=text, lang='ur')
+    tts.save("response.mp3")
+    with open("response.mp3", "rb") as audio_file:
+        audio_bytes = audio_file.read()
+        st.audio(audio_bytes, format="audio/mp3")
+    if os.path.exists("response.mp3"):
+        os.remove("response.mp3")
+
+# Function to handle file uploads (process text content)
+def process_file_upload(file):
+    if file.type == "text/plain":
+        text_content = file.read().decode("utf-8")
+    elif file.type == "application/pdf":
+        # Process PDF files
+        text_content = extract_text_from_pdf(file)
+    else:
+        st.warning("Unsupported file format. Please upload a text or PDF file.")
+        return None
+    return text_content
+
+def extract_text_from_pdf(file):
+    # Extract text from uploaded PDF file (you can use PyPDF2 or other libraries)
+    pass
+
+# Function to create a Langchain Conversational Retrieval Chain (RAG)
+def create_chain(docs):
+    embeddings = HuggingFaceEmbeddings()
+    vectorstore = FAISS.from_documents(docs, embeddings)
+    chain = ConversationalRetrievalChain.from_llm(
+        llm=llm,
+        retriever=vectorstore.as_retriever(),
+    )
+    return chain
+
+# Streamlit app layout
+st.title("🤖 Urdu Voice & File Chatbot using Langchain (RAG)")
+st.write("Interact with the chatbot using Urdu voice input or upload any file in any language. The chatbot will respond with both text and audio in Urdu.")
+
+# Upload a file
+uploaded_file = st.file_uploader("Upload a file (text, PDF)", type=["txt", "pdf"])
+
+# Record and process audio input
+if st.button("Record Urdu Query"):
+    user_input_urdu = record_audio()
+    if user_input_urdu:
+        st.write(f"Your Query (Text): {user_input_urdu}")
+
+        # Simulate document processing with RAG
+        docs = [Document(page_content=user_input_urdu)]
+        chain = create_chain(docs)
+        response = chain.run(user_input_urdu)
+        
+        st.write(f"Chatbot Response (Text): {response}")
+        text_to_speech_urdu(response)
+    else:
+        st.error("No valid input provided. Please try again.")
+
+# Process uploaded file if any
+if uploaded_file:
+    file_text = process_file_upload(uploaded_file)
+    if file_text:
+        st.write(f"File Content (Text): {file_text}")
+
+        # Generate response based on file content
+        docs = [Document(page_content=file_text)]
+        chain = create_chain(docs)
+        response = chain.run(file_text)
+        
+        st.write(f"Chatbot Response (Text): {response}")
+        text_to_speech_urdu(response)