Add application file

.env.example

@@ -0,0 +1 @@
+APP_NAME = "ITI-RAG"

Dockerfile

@@ -0,0 +1,17 @@
+FROM python:3.10-slim
+
+WORKDIR /app
+
+RUN apt-get update && apt-get install -y \
+    build-essential \
+    libmagic-dev \
+    && rm -rf /var/lib/apt/lists/*
+
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY . .
+
+EXPOSE 8000
+
+CMD ["python", "app/main.py"]

app/core/constants.py

@@ -0,0 +1,3 @@
+CHUNK_SIZE = 1500
+CHUNK_OVERLAP = 200
+VECTOR_DB_PATH = "data/vector_db"

app/services/ingestion.py

@@ -0,0 +1,17 @@
+from langchain_community.document_loaders import PyMuPDFLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from app.core.constants import CHUNK_SIZE, CHUNK_OVERLAP
+
+def load_and_split_book(file_path):
+    loader = PyMuPDFLoader(file_path)
+    documents = loader.load()
+    
+    text_splitter = RecursiveCharacterTextSplitter(
+        chunk_size=CHUNK_SIZE,
+        chunk_overlap=CHUNK_OVERLAP,
+        length_function=len
+    )
+    chunks = text_splitter.split_documents(documents)
+    
+    print(f"Book splited to {len(chunks)}")
+    return chunks

app/services/llm_service.py

@@ -0,0 +1,36 @@
+import os
+from langchain_google_genai import ChatGoogleGenerativeAI
+from langchain.prompts import PromptTemplate
+from langchain.chains import RetrievalQA
+from app.services.vector_store import load_vector_store
+
+llm = ChatGoogleGenerativeAI(
+    model="gemini-1.5-flash",
+    google_api_key="YOUR_GEMINI_API_KEY",
+    temperature=0 
+)
+
+def get_book_assistant_chain(mode="Explain"):
+    vector_db = load_vector_store()
+    retriever = vector_db.as_retriever(search_kwargs={"k": 5})
+
+    if mode == "Explain":
+        template = """You are an expert tutor. Use the provided context to explain the concept simply.
+        Context: {context}
+        Question: {question}
+        Explanation:"""
+    else:
+        template = """Based on the following text, generate 3 multiple-choice questions with answers.
+        Context: {context}
+        Question: {question}
+        Questions:"""
+
+    QA_PROMPT = PromptTemplate(template=template, input_variables=["context", "question"])
+
+    return RetrievalQA.from_chain_type(
+        llm=llm,
+        chain_type="stuff",
+        retriever=retriever,
+        return_source_documents=True,
+        chain_type_kwargs={"prompt": QA_PROMPT}
+    )

app/services/vector_store.py

@@ -0,0 +1,18 @@
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_community.vectorstores import FAISS
+from app.core.constants import VECTOR_DB_PATH
+
+# اختيارنا لـ all-MiniLM-L6-v2 لأنه خفيف على جهازك Dell G12 [cite: 51]
+embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
+
+def create_vector_store(chunks):
+    # تحويل النص لأرقام وتخزينه [cite: 29, 30]
+    vector_db = FAISS.from_documents(chunks, embeddings)
+    
+    # حفظ القاعدة محلياً (Local Storage) 
+    vector_db.save_local(VECTOR_DB_PATH)
+    return vector_db
+
+def load_vector_store():
+    # تحميل القاعدة عند الحاجة [cite: 32]
+    return FAISS.load_local(VECTOR_DB_PATH, embeddings, allow_dangerous_deserialization=True)

docker-compose.yml

@@ -0,0 +1,12 @@
+version: '3.8'
+
+services:
+  smart-assistant:
+    build: .
+    ports:
+      - "8000:8000"
+    environment:
+      - GROQ_API_KEY=${GROQ_API_KEY}
+    volumes:
+      - ./data:/app/data  
+

main.py

@@ -0,0 +1,11 @@
+from fastapi import FastAPI
+import uvicorn
+from ui.gradio_app import demo
+import gradio as gr
+
+app = FastAPI(title="Smart Contract/Book Assistant API")
+
+app = gr.mount_gradio_app(app, demo, path="/")
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=8000)

requirements.txt

@@ -0,0 +1,23 @@
+# Core Web Framework
+fastapi==0.111.0
+uvicorn[standard]==0.30.1
+
+# AI & RAG Engine
+langchain>=0.2.0
+langchain-community>=0.2.0
+langchain-google-genai>=1.0.0
+sentence-transformers>=2.7.0
+faiss-cpu>=1.8.0
+
+# Document Processing
+pypdf>=4.2.0
+python-docx>=1.1.0
+
+# UI & Local Execution
+gradio>=4.36.1
+transformers>=4.41.0
+torch>=2.3.0
+
+# Critical Dependencies (To fix the Numpy/Scipy error)
+numpy<2.0.0
+scipy>=1.13.0

ui/gradio_app.py

@@ -0,0 +1,49 @@
+import gradio as gr
+from app.services.ingestion import load_and_split_book
+from app.services.vector_store import create_vector_store
+from app.services.llm_service import get_book_assistant_chain
+
+def handle_upload(file):
+    if file is None:
+        return "Please upload a file first."
+    
+    chunks = load_and_split_book(file.name)
+    
+    create_vector_store(chunks)
+    
+    return "✅ Document processed successfully! You can now go to the Chat tab." 
+
+def handle_chat(user_question, mode):
+    qa_chain = get_book_assistant_chain(mode=mode)
+    
+    result = qa_chain({"query": user_question})
+    
+    answer = result["result"]
+    sources = "\n\n📌 Sources:\n" + "\n".join([doc.metadata.get('source', 'Unknown') for doc in result["source_documents"]])
+    
+    return answer + sources
+
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 📚 Smart Book & Contract Assistant")
+    
+    with gr.Tabs():
+        with gr.TabItem("⬆️ Upload Document"):
+            file_input = gr.File(label="Upload PDF or DOCX", file_types=[".pdf", ".docx"])
+            upload_btn = gr.Button("Process Document", variant="primary")
+            status_output = gr.Textbox(label="Status", interactive=False)
+            upload_btn.click(handle_upload, inputs=file_input, outputs=status_output)
+            
+        with gr.TabItem("💬 Explanation & Q&A"):
+            mode_radio = gr.Radio(
+                choices=["Explain", "Generate Questions"], 
+                label="Interaction Mode", 
+                value="Explain"
+            )
+            chat_input = gr.Textbox(label="Ask a question or request a summary/questions")
+            chat_btn = gr.Button("Submit", variant="primary")
+            chat_output = gr.Textbox(label="AI Assistant Response", lines=12) 
+            
+            chat_btn.click(handle_chat, inputs=[chat_input, mode_radio], outputs=chat_output)
+
+if __name__ == "__main__":
+    demo.launch()