Update app.py

app.py

@@ -1,63 +1,141 @@
+import os
+import fitz
+import textwrap
+from dotenv import load_dotenv
+from langchain_community.document_loaders import UnstructuredPDFLoader
+from langchain_community.vectorstores import FAISS
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_text_splitters import CharacterTextSplitter
+from langchain_groq import ChatGroq
+from langchain.memory import ConversationBufferMemory
+from langchain.chains import ConversationalRetrievalChain
+from pdf2image import convert_from_path
+import pytesseract
+from gtts import gTTS
+import uuid
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
-
-
-if __name__ == "__main__":
-    demo.launch()
+
+# Load environment variables
+load_dotenv()
+os.environ["GROQ_API_KEY"] = "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"
+
+# File directories
+UPLOAD_FOLDER = 'uploads/'
+AUDIO_FOLDER = 'static/audio/'
+
+# Ensure directories exist
+for folder in [UPLOAD_FOLDER, AUDIO_FOLDER]:
+    if not os.path.exists(folder):
+        os.makedirs(folder)
+
+def load_pdf(file_path):
+    """
+    Load and preprocess Arabic text from a PDF file.
+    """
+    pages = convert_from_path(file_path, 500)
+    documents = []
+    for imgBlob in pages:
+        # Perform OCR on each image
+        text = pytesseract.image_to_string(imgBlob, lang="ara")
+        documents.append(text)
+    return documents
+
+def prepare_vectorstore(data):
+    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-mpnet-base-v2")
+    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=20, separator="\n")
+    texts = data
+    vectorstore = FAISS.from_texts(texts, embeddings)
+
+    # Save FAISS index to disk
+    vectorstore.save_local("faiss_index")
+
+    return vectorstore
+
+def load_vectorstore():
+    embeddings = HuggingFaceEmbeddings()
+    vectorstore = FAISS.load_local("faiss_index", embeddings, allow_dangerous_deserialization=True)
+    return vectorstore
+
+def create_chain(vectorstore):
+    llm = ChatGroq(model="gemma2-9b-it", temperature=0)
+    retriever = vectorstore.as_retriever()
+    memory = ConversationBufferMemory(llm=llm, output_key="answer", memory_key="chat_history", return_messages=True)
+    chain = ConversationalRetrievalChain.from_llm(
+        llm=llm,
+        retriever=retriever,
+        memory=memory,
+        verbose=False,
+        chain_type="map_reduce"
+    )
+    return chain
+
+def process_pdf(pdf_file):
+    if pdf_file is not None:
+        file_path = os.path.join(UPLOAD_FOLDER, pdf_file.name)
+        pdf_file.save(file_path)
+
+        # Load PDF, prepare vectorstore
+        data = load_pdf(file_path)
+        vectorstore = prepare_vectorstore(data)
+        chain = create_chain(vectorstore)
+        
+        return chain, f"تم تحميل الملف '{pdf_file.name}' بنجاح!"
+    return None, "الرجاء تحميل ملف PDF ."
+
+def chat_with_bot(user_input, chain):
+    if chain is None:
+        return "يرجى تحميل ملف PDF أولاً."
+
+    prompt=f"""
+        You are an expert Arabic-language assistant specialized in analyzing and responding to queries about Arabic PDF documents. Your responses should be precise, informative, and reflect the professional tone and structure expected in formal Arabic communication. Focus on extracting and presenting relevant information from the document clearly and systematically, while avoiding colloquial or informal language.
+
+        When responding, ensure the following:
+
+           - Your answer directly reflects the content of the document.
+           - If the requested information is not available in the document, clearly state that.
+           - Keep your response concise yet comprehensive, addressing the question fully.
+           - Always respond in formal Arabic, without using English.\n
+
+        Question: {user_input}\n
+        Helpful Answer:"""
+    
+    response = chain({"question": prompt})
+    assistant_response = response["answer"]
+    
+    # Generate and save audio response
+    audio_id = str(uuid.uuid4())
+    audio_file = f"{audio_id}.mp3"
+    tts = gTTS(text=assistant_response, lang='ar')
+    tts.save(os.path.join(AUDIO_FOLDER, audio_file))
+
+    return assistant_response, f"{AUDIO_FOLDER}/{audio_file}"
+
+# Gradio app interface
+def chatbot_interface(pdf_file, user_input):
+    chain, message = process_pdf(pdf_file)
+    
+    if user_input and chain:
+        response_text, audio_path = chat_with_bot(user_input, chain)
+        return response_text, audio_path
+    else:
+        return "يرجى إدخال السؤال.", None
+
+with gr.Blocks() as demo:
+    gr.Markdown("<h1 style='text-align:center;'>ديمو بوت للقاء مركز حضرموت</h1>")
+    
+    with gr.Row():
+        pdf_input = gr.File(label="اختر ملف 📑 PDF للدردشة", type="file")
+    
+    with gr.Row():
+        user_input = gr.Textbox(label="سؤالك")
+    
+    with gr.Row():
+        submit_button = gr.Button("رفع وبدء الدردشة")
+    
+    with gr.Row():
+        output_text = gr.Textbox(label="الرد")
+        audio_output = gr.Audio(label="الرد الصوتي")
+    
+    submit_button.click(chatbot_interface, inputs=[pdf_input, user_input], outputs=[output_text, audio_output])
+
+demo.launch()