app.py

import gradio as gr
from PIL import Image
import pdf2image
from transformers import (
    LayoutLMv3FeatureExtractor, 
    LayoutLMv3TokenizerFast, 
    LayoutLMv3Processor, 
    LayoutLMv3ForSequenceClassification
)
import torch
import logging
import traceback
import os
import tempfile

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

class DocumentClassifier:
    """Class to classify document types based on layout and text content"""
    def __init__(self):
        try:
            self.feature_extractor = LayoutLMv3FeatureExtractor()
            self.tokenizer = LayoutLMv3TokenizerFast.from_pretrained(
                "microsoft/layoutlmv3-base"
            )
            self.processor = LayoutLMv3Processor(self.feature_extractor, self.tokenizer)
            self.model = LayoutLMv3ForSequenceClassification.from_pretrained(
                "RavindiG/layoutlmv3-document-classification-v2"
            )
            self.model.eval()

            self.id2label = { 
                0: 'Financial Report', 
                1: 'Invoice or Receipt',
                2: 'Legal Document', 
                3: 'Medical Record',
                4: 'Research Paper'
            }

            logger.info("DocumentClassifier initialized successfully.")
        except Exception as e:
            logger.error(f"Error during initialization: {e}")
            logger.error(traceback.format_exc())
            raise

    def predict_document_class(self, image):
        try:
            logger.info(f"Image size: {image.size}, mode: {image.mode}")
            
            if image.mode != 'RGB':
                logger.info(f"Converting image from {image.mode} to RGB")
                image = image.convert('RGB')
            
            logger.info("Processing image...")
            encoded_inputs = self.processor(image, max_length=512, return_tensors="pt")

            device = next(self.model.parameters()).device
            
            for k, v in encoded_inputs.items():
                encoded_inputs[k] = v.to(device)

            logger.info("Running model inference...")
            with torch.no_grad():
                outputs = self.model(**encoded_inputs)

            logits = outputs.logits
            logger.info(f"Logits: {logits}")

            threshold = getattr(self, "confidence_threshold", 1.0)
            max_logit = logits.max().item()
            
            # Get probabilities for all classes
            probabilities = torch.nn.functional.softmax(logits, dim=-1)[0]
            
            if max_logit < threshold:
                logger.info("Max logit below threshold, returning 'Other'")
                predicted_label = "Other"
                confidence_dict = {"Other": 1.0}
            else:
                predicted_class_idx = logits.argmax(-1).item()
                predicted_label = self.id2label.get(predicted_class_idx, "Other")
                
                # Create confidence dictionary for all classes
                confidence_dict = {
                    self.id2label[i]: probabilities[i].item() 
                    for i in range(len(self.id2label))
                }
                
                logger.info(f"Predicted label: {predicted_label}")
            
            return predicted_label, confidence_dict
            
        except Exception as e:
            logger.error(f"Error in predict_document_class: {e}")
            logger.error(traceback.format_exc())
            raise

    def classify_document(self, file_path):
        """Classify document type based on layout and extracted text"""
        try:
            logger.info(f"Processing file: {file_path}")
            
            if file_path.lower().endswith('.pdf'):
                logger.info("Converting PDF to image...")
                images = pdf2image.convert_from_path(
                    file_path,
                    dpi=120,
                    fmt="RGB",
                    first_page=1,
                    last_page=1
                )
                if not images:
                    raise ValueError("No images extracted from PDF")
                image = images[0]
                logger.info(f"PDF converted successfully, image size: {image.size}")
            else:
                logger.info("Opening image file...")
                image = Image.open(file_path)
                logger.info(f"Image opened successfully, size: {image.size}")
            
            doc_type, confidence = self.predict_document_class(image)
            
            logger.info(f"Classification successful: {doc_type}")
            return doc_type, confidence, image
        
        except Exception as e:
            logger.error(f"Error classifying document {file_path}: {e}")
            logger.error(traceback.format_exc())
            return "Error", {"Error": 1.0}, None


# Initialize the classifier
logger.info("Initializing DocumentClassifier...")
classifier = DocumentClassifier()
logger.info("Classifier ready!")


def classify_upload(file):
    """Gradio interface function for file upload"""
    if file is None:
        return "No file uploaded", {}, None
    
    try:
        # Get the file path
        file_path = file.name
        
        # Classify the document
        doc_type, confidence, image = classifier.classify_document(file_path)
        
        return doc_type, confidence, image
    
    except Exception as e:
        logger.error(f"Error in classify_upload: {e}")
        logger.error(traceback.format_exc())
        return f"Error: {str(e)}", {}, None


# Create Gradio interface
with gr.Blocks(title="Document Classifier") as demo:
    gr.Markdown(
        """
        # 📄 Document Type Classifier
        
        Upload a document image or PDF to classify it into one of the following categories:
        - Financial Report
        - Invoice or Receipt
        - Legal Document
        - Medical Record
        - Research Paper
        
        **Note:** Only the first page of PDFs will be analyzed.
        """
    )
    
    with gr.Row():
        with gr.Column():
            file_input = gr.File(
                label="Upload Document (Image or PDF)",
                file_types=[".pdf", ".png", ".jpg", ".jpeg", ".tiff", ".bmp"]
            )
            classify_btn = gr.Button("Classify Document", variant="primary")
        
        with gr.Column():
            label_output = gr.Textbox(label="Document Type", lines=2)
            confidence_output = gr.Label(label="Confidence Scores", num_top_classes=5)
            image_output = gr.Image(label="Document Preview", type="pil")
    
    # Examples
    gr.Markdown("### Examples")
    gr.Markdown("Upload your document using the file uploader above.")
    
    # Connect the button to the function
    classify_btn.click(
        fn=classify_upload,
        inputs=[file_input],
        outputs=[label_output, confidence_output, image_output]
    )
    
    # Also trigger on file upload
    file_input.change(
        fn=classify_upload,
        inputs=[file_input],
        outputs=[label_output, confidence_output, image_output]
    )

# Launch the app
if __name__ == "__main__":
    demo.launch()