Update app.py

app.py

@@ -1,28 +1,18 @@
 import gradio as gr
 from transformers import pipeline
-from PIL import Image
-import pytesseract
-import shutil, os
 
-print("Checking tesseract path:", pytesseract.pytesseract.tesseract_cmd)
-print("Does it exist?", os.path.exists(pytesseract.pytesseract.tesseract_cmd))
+# Load your trained event extractor
+event_model_id = "King-8/flyer-event-extractor"
+event_extractor = pipeline("text2text-generation", model=event_model_id)
 
-cache_dir = os.path.expanduser("~/.cache/huggingface")
-if os.path.exists(cache_dir):
-    shutil.rmtree(cache_dir)
+# Load an OCR model (image → text)
+ocr_model = pipeline("image-to-text", model="microsoft/trocr-base-handwritten")
 
-# explicitly set path to tesseract
-pytesseract.pytesseract.tesseract_cmd = "/usr/bin/tesseract"
+def extract_event_details(text):
+    """Extract event details from flyer text using your fine-tuned model"""
+    output = event_extractor(f"Flyer text: {text}", max_length=128)[0]["generated_text"]
 
-# Load your trained model from Hugging Face Hub
-model_id = "King-8/flyer-event-extractor"  # replace with your actual model repo
-extractor = pipeline("text2text-generation", model=model_id)
-
-def extract_event_details(flyer_text):
-    """Extract structured details from flyer text using fine-tuned model"""
-    output = extractor(f"Flyer text: {flyer_text}", max_length=128)[0]["generated_text"]
-
-    # Try to parse into Event, Date, Location
+    # Try parsing into fields
     event, date, location = "", "", ""
     if "Event:" in output and "Date:" in output and "Location:" in output:
         try:
@@ -33,20 +23,20 @@ def extract_event_details(flyer_text):
         except:
             event = output
     else:
-        event = output  # fallback
+        event = output  # fallback if parsing fails
 
     return event, date, location, output
 
 def process_image(image):
-    """Run OCR on uploaded flyer image, then extract details with model"""
-    flyer_text = pytesseract.image_to_string(Image.open(image))
-    return extract_event_details(flyer_text)
+    """OCR via Hugging Face model, then event extraction"""
+    ocr_text = ocr_model(image)[0]["generated_text"]
+    return extract_event_details(ocr_text)
 
 with gr.Blocks() as demo:
     gr.Markdown("## 📅 Flyer Event Extractor\nUpload a flyer image or paste text to extract event details.")
 
     with gr.Tab("Upload Flyer Image"):
-        img_input = gr.Image(type="filepath", label="Upload Flyer Image")
+        img_input = gr.Image(type="pil", label="Upload Flyer Image")
         btn_img = gr.Button("Extract from Image")
         event_out_img = gr.Textbox(label="Event Name")
         date_out_img = gr.Textbox(label="Date & Time")