Update app.py

app.py

@@ -1,15 +1,15 @@
 import os
 import re
-import uvicorn
 from fastapi import FastAPI, UploadFile, File
 from fastapi.middleware.cors import CORSMiddleware
 import pymupdf
 import torch
 from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+import gradio as gr
 
+# 1. FastAPI App Initialization
 app = FastAPI(title="CaseLoop PDF Simplifier Engine")
 
-# Allow frontend cross-origin requests
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -18,7 +18,7 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-# 1. Load Tokenizer & Model
+# 2. Load Model & Tokenizer
 MODEL_ID = "Ronak0/last-caseloop-pdf-simplifier"
 tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
 model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_ID)
@@ -35,7 +35,6 @@ def extract_clean_text(pdf_bytes: bytes) -> str:
     return " ".join(full_text)[:2000]
 
 def clean_and_split_bullets(raw_output: str) -> list[str]:
-    # Replace numbered prefixes and bullet symbols with newlines
     normalized = re.sub(r'(?:^\d+\.|\n\d+\.|\s*•\s*)', '\n', raw_output)
     lines = [line.strip() for line in normalized.split('\n') if line.strip()]
     
@@ -48,9 +47,9 @@ def clean_and_split_bullets(raw_output: str) -> list[str]:
             
     return cleaned_points
 
-@app.get("/")
-def health_check():
-    return {"status": "online", "model": MODEL_ID, "device": device}
+@app.get("/health")
+def health():
+    return {"status": "online", "model": MODEL_ID}
 
 @app.post("/simplify")
 async def simplify_pdf(file: UploadFile = File(...)):
@@ -78,7 +77,37 @@ async def simplify_pdf(file: UploadFile = File(...)):
         "simplified_points": formatted_bullets
     }
 
+# 3. Mount Gradio interface to satisfy HF Gradio-SDK health probes (/api/config)
+def gradio_simplify(pdf_file):
+    if not pdf_file:
+        return "Upload a PDF document to test."
+    with open(pdf_file.name, "rb") as f:
+        raw_text = extract_clean_text(f.read())
+    prompt = f"Simplify civic/RTI document:\n{raw_text}"
+    inputs = tokenizer(prompt, return_tensors="pt", max_length=512, truncation=True).to(device)
+    with torch.no_grad():
+        outputs = model.generate(
+            **inputs,
+            max_new_tokens=300,
+            no_repeat_ngram_size=3,
+            repetition_penalty=1.2,
+            temperature=0.2,
+            num_beams=2,
+            do_sample=False
+        )
+    gen = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return "\n".join(clean_and_split_bullets(gen))
+
+demo = gr.Interface(
+    fn=gradio_simplify,
+    inputs=gr.File(label="Upload PDF Document", file_types=[".pdf"]),
+    outputs=gr.Textbox(label="Simplified Takeaways"),
+    title="CaseLoop PDF Simplifier"
+)
+
+# Mounts Gradio on top of FastAPI and launches on 7860
+app = gr.mount_gradio_app(app, demo, path="/")
+
 if __name__ == "__main__":
-    # Hugging Face Spaces mandates port 7860
-    port = int(os.environ.get("PORT", 7860))
-    uvicorn.run(app, host="0.0.0.0", port=port)
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=7860)