Update app.py

app.py

@@ -1,54 +1,111 @@
 import gradio as gr
-import fitz
+import fitz  # PyMuPDF
 import torch
 from transformers import pipeline
 import time, logging
+import re
+import tempfile
+import os
 
+# === Setup ===
 logging.basicConfig(level=logging.ERROR)
-device = -1  # CPU-only
-print("⚠️ CPU-only. Expect ~20–30s for 300,000 chars.")
+device = -1  # CPU
+print("⚠️ CPU-only mode. Expect ~20–30s for large documents.")
 
+# === Load summarization model ===
 try:
     summarizer = pipeline("summarization", model="t5-small", device=device, torch_dtype=torch.float32)
 except Exception as e:
-    print(f"❌ Model loading failed: {str(e)}")
+    print(f"❌ Model loading failed: {e}")
     exit(1)
 
-def summarize_file(file_bytes):
-    start = time.time()
-    print(f"File type: {type(file_bytes)}")
+# === Text Preprocessing ===
+def smart_chunk(text, max_chunk_len=2000):
+    sentences = re.split(r'(?<=[.!?]) +', text)
+    chunks, current_chunk = [], ""
+    for sentence in sentences:
+        if len(current_chunk) + len(sentence) < max_chunk_len:
+            current_chunk += sentence + " "
+        else:
+            chunks.append(current_chunk.strip())
+            current_chunk = sentence + " "
+    if current_chunk:
+        chunks.append(current_chunk.strip())
+    return chunks
+
+# === Summarization per file ===
+def summarize_file_bytes(file_bytes, filename):
+    start_time = time.time()
     try:
-        text = "".join(page.get_text("text", flags=16) for page in fitz.open(stream=file_bytes, filetype="pdf")) if file_bytes[:4].startswith(b'%PDF') else file_bytes.decode("utf-8", errors="ignore")
+        if file_bytes[:4].startswith(b'%PDF'):
+            text = "".join(page.get_text("text", flags=16) for page in fitz.open(stream=file_bytes, filetype="pdf"))
+        else:
+            text = file_bytes.decode("utf-8", errors="ignore")
     except Exception as e:
-        return f"❌ Text extraction failed: {str(e)}"
-    if not text.strip(): return "❌ No text found"
+        return f"{filename}: ❌ Text extraction failed: {e}", ""
+
+    text = text.strip()
+    if not text:
+        return f"{filename}: ❌ No text found.", ""
+
     text = text[:300000]
-    chunks = [text[i:i+2000] for i in range(0, len(text), 2000)]
-    print(f"Chunks created: {len(chunks)}")
-    if not chunks: return "❌ No chunks to summarize"
-    summaries = []
+    chunks = smart_chunk(text)
+    summaries, line_count = [], 0
+
     for i, chunk in enumerate(chunks):
-        if time.time() - start > 15: 
-            summaries.append("⚠️ Stopped early")
+        if time.time() - start_time > 20:
+            summaries.append("⚠️ Stopped early due to time limit.")
             break
         try:
-            summary = summarizer(chunk, max_length=60, min_length=10, do_sample=False)[0]['summary_text']
-            summaries.append(f"**Chunk {i+1}**:\n{summary}")
+            summary = summarizer(chunk, max_length=100, min_length=30, do_sample=False)[0]["summary_text"]
+            summaries.append(f"**Chunk {i+1}**:\n{summary.strip()}")
+            line_count += summary.count('\n') + 1
+            if line_count >= 15:
+                break
         except Exception as e:
-            summaries.append(f"**Chunk {i+1}**: ❌ Error: {str(e)}")
-    return f"**Chars**: {len(text)}\n**Time**: {time.time()-start:.2f}s\n\n" + "\n\n".join(summaries)
+            summaries.append(f"**Chunk {i+1}**: ❌ Error summarizing: {e}")
+
+    total_time = time.time() - start_time
+    summary_text = f"📄 **{filename}**\n**Characters**: {len(text)} | **Time**: {total_time:.2f}s\n\n" + "\n\n".join(summaries)
+    return summary_text, summary_text
 
+# === Gradio Wrapper ===
+def summarize_multiple_files(files):
+    all_summaries = []
+    combined_text = ""
+
+    for file_obj in files:
+        file_bytes = file_obj.read()
+        filename = file_obj.name.split("/")[-1]
+        summary, raw = summarize_file_bytes(file_bytes, filename)
+        all_summaries.append(summary)
+        combined_text += f"\n\n{raw}\n" + "="*60 + "\n"
+
+    # Write summary to temp .txt file
+    with tempfile.NamedTemporaryFile(delete=False, suffix=".txt", mode="w", encoding="utf-8") as f:
+        f.write(combined_text)
+        summary_file_path = f.name
+
+    return "\n\n".join(all_summaries), summary_file_path
+
+# === Gradio Interface ===
 demo = gr.Interface(
-    fn=summarize_file, inputs=gr.File(label="📄 PDF/TXT Notes", type="binary"), 
-    outputs=gr.Textbox(label="📝 Summary"), 
-    title="Fast Summarizer", description="300,000+ chars in ~20–30s (CPU)"
+    fn=summarize_multiple_files,
+    inputs=gr.File(label="📄 Upload PDF or TXT files", file_types=[".pdf", ".txt"], type="file", file_count="multiple"),
+    outputs=[
+        gr.Textbox(label="📝 Summary", lines=30, max_lines=100),
+        gr.File(label="📥 Download Summary as .txt")
+    ],
+    title="📚 Multi-File Summarizer",
+    description="Summarizes multiple PDFs or TXTs into at least 15 lines each. Download final output as .txt. CPU-optimized."
 )
 
 if __name__ == "__main__":
     try:
         demo.launch(share=False, server_port=7860)
     except Exception as e:
-        print(f"❌ Gradio launch failed: {str(e)}")
+        print(f"❌ Gradio launch failed: {e}")
+