app.py

import os
os.environ["STREAMLIT_SERVER_PORT"] = "8501"
os.environ["STREAMLIT_SERVER_HEADLESS"] = "true"

# Must be first Streamlit command
import streamlit as st
st.set_page_config(page_title="DocAnalyzer Pro", layout="wide")

# Import libraries
from transformers import pipeline
from PyPDF2 import PdfReader
import docx
import time
import psutil
from pathlib import Path

# ======================
# CACHE SETUP
# ======================
def setup_environment():
    """Ensure cache directories exist"""
    cache_dir = Path("/app/models")
    cache_dir.mkdir(exist_ok=True, parents=True)
    return cache_dir

cache_dir = setup_environment()

# ======================
# MODEL LOADING
# ======================
@st.cache_resource(ttl=3600)  # Cache for 1 hour
def load_models():
    """Load optimized models for Hugging Face Spaces"""
    with st.spinner("🔄 Loading AI models (this may take 1-2 minutes)..."):
        return {
            'qa': pipeline(
                "question-answering",
                model="distilbert-base-cased-distilled-squad",
                device=-1  # Force CPU
            ),
            'summarizer': pipeline(
                "summarization",
                model="sshleifer/distilbart-cnn-12-6",
                device=-1  # Force CPU
            )
        }

models = load_models()

# ======================
# DOCUMENT PROCESSING
# ======================
def extract_text(file):
    """Extract text from PDF/DOCX files"""
    try:
        if file.type == "application/pdf":
            reader = PdfReader(file)
            return " ".join(page.extract_text() for page in reader.pages if page.extract_text())
        elif file.type == "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
            doc = docx.Document(file)
            return "\n".join(para.text for para in doc.paragraphs if para.text)
    except Exception as e:
        st.error(f"⚠️ Error processing document: {str(e)}")
        return ""

# ======================
# CORE FUNCTIONS
# ======================
def generate_summary(text, max_length=150):
    """Generate summary with chunking for large documents"""
    if not text:
        return ""
    
    try:
        if len(text) > 10000:  # Chunk large documents
            chunks = [text[i:i+3000] for i in range(0, len(text), 3000)]
            summaries = []
            for chunk in chunks:
                result = models['summarizer'](
                    chunk,
                    max_length=max_length//len(chunks),
                    min_length=30,
                    do_sample=False
                )
                summaries.append(result[0]['summary_text'])
            return " ".join(summaries)
        return models['summarizer'](text, max_length=max_length)[0]['summary_text']
    except Exception as e:
        st.error(f"❌ Summarization failed: {str(e)}")
        return ""

# ======================
# STREAMLIT UI
# ======================
st.title("📄 DocAnalyzer Pro")

# File Upload Section
with st.expander("📤 Upload Document", expanded=True):
    uploaded_file = st.file_uploader("Choose PDF/DOCX", type=["pdf", "docx"])
    manual_text = st.text_area("Or paste text here:", height=150)
    context = extract_text(uploaded_file) if uploaded_file else manual_text

# Main Features
tab1, tab2 = st.tabs(["🔍 Question Answering", "📝 Summarization"])

with tab1:
    if context:
        question = st.text_input("Ask about the document:")
        if question:
            with st.spinner("Analyzing..."):
                start_time = time.time()
                result = models['qa'](question=question, context=context[:100000])  # Limit context size
                st.success(f"Answered in {time.time()-start_time:.1f}s")
                st.markdown(f"**Answer:** {result['answer']}")
                st.progress(result['score'])
                st.caption(f"Confidence: {result['score']:.0%}")

with tab2:
    if context:
        with st.form("summary_form"):
            length = st.slider("Summary Length", 50, 300, 150)
            if st.form_submit_button("Generate Summary"):
                with st.spinner("Summarizing..."):
                    start_time = time.time()
                    summary = generate_summary(context, length)
                    st.success(f"Generated in {time.time()-start_time:.1f}s")
                    st.markdown(f"**Summary:**\n\n{summary}")

# System Info
with st.expander("⚙️ System Status"):
    st.code(f"""
    Models loaded: {', '.join(models.keys())}
    Device: {'GPU ✅' if torch.cuda.is_available() else 'CPU ⚠️'}
    Memory: {psutil.virtual_memory().percent}% used
    CPU: {psutil.cpu_percent()}% used
    """)