app.py

import gradio as gr
import cv2
import numpy as np
from fpdf import FPDF
import tempfile
import os
from paddleocr import PaddleOCR
import time
from simple_salesforce import Salesforce
from datetime import datetime
import boto3

# Initialize PaddleOCR once with updated parameters
ocr_model = PaddleOCR(use_textline_orientation=True, lang='en')

def upload_image_and_get_url(image_path):
    """
    Upload the image to AWS S3 and return the public URL.
    """
    s3_client = boto3.client(
        's3',
        aws_access_key_id=os.environ['AWS_ACCESS_KEY_ID'],  # Set environment variable
        aws_secret_access_key=os.environ['AWS_SECRET_ACCESS_KEY'],  # Set environment variable
        region_name=os.environ['AWS_REGION']  # Set environment variable
    )

    # Define the S3 bucket name
    bucket_name = 'your-bucket-name'

    # Generate a unique key for the image (e.g., using the file name)
    image_key = f"images/{os.path.basename(image_path)}"

    # Upload the image to S3
    s3_client.upload_file(image_path, bucket_name, image_key)

    # Construct the public URL for the uploaded image
    image_url = f"https://{bucket_name}.s3.{os.environ['AWS_REGION']}.amazonaws.com/{image_key}"

    return image_url

def analyze_uv_coverage(img, brightness_threshold=150, kernel_size=5, apply_blur=True, adaptive_thresh=False):
    """
    Analyze UV sterilization coverage by thresholding the grayscale image.
    Optional adaptive thresholding and Gaussian blur for noise reduction.
    Morphological operations clean the mask for better accuracy.
    """
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    if apply_blur:
        gray = cv2.GaussianBlur(gray, (5, 5), 0)

    if adaptive_thresh:
        binary_mask = cv2.adaptiveThreshold(
            gray, 255,
            cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
            cv2.THRESH_BINARY,
            11, 2)
    else:
        _, binary_mask = cv2.threshold(gray, brightness_threshold, 255, cv2.THRESH_BINARY)

    # Morphological opening (erosion followed by dilation) to remove noise
    kernel = np.ones((kernel_size, kernel_size), np.uint8)
    binary_mask = cv2.morphologyEx(binary_mask, cv2.MORPH_OPEN, kernel, iterations=1)

    # Morphological closing (dilation followed by erosion) to close small holes inside foreground
    binary_mask = cv2.morphologyEx(binary_mask, cv2.MORPH_CLOSE, kernel, iterations=1)

    total_pixels = binary_mask.size
    sterilized_pixels = cv2.countNonZero(binary_mask)
    coverage_percent = (sterilized_pixels / total_pixels) * 100

    # Create overlay for visualization: Green = sterilized, Red = unsterilized
    overlay = img.copy()
    overlay[binary_mask == 255] = [0, 255, 0]  # Green
    overlay[binary_mask == 0] = [0, 0, 255]    # Red

    annotated_img = cv2.addWeighted(img, 0.6, overlay, 0.4, 0)

    return annotated_img, coverage_percent

def create_pdf_report(coverage_percent, extracted_texts, annotated_image_path, output_path):
    pdf = FPDF()
    pdf.add_page()

    pdf.set_font("Arial", 'B', 16)
    pdf.cell(200, 10, txt="UV Sterilization Report", ln=True, align='C')
    pdf.ln(10)

    pdf.set_font("Arial", size=12)
    pdf.cell(0, 10, f"Sterilization Coverage: {coverage_percent:.2f}%", ln=True)
    pdf.ln(5)

    pdf.cell(0, 10, "Extracted Text from Image (OCR):", ln=True)
    pdf.set_font("Arial", size=10)
    if extracted_texts:
        for text in extracted_texts:
            # Filter out very short or empty OCR texts to improve clarity
            if len(text.strip()) > 1:
                pdf.multi_cell(0, 8, f"- {text}")
    else:
        pdf.cell(0, 8, "No text detected.", ln=True)

    pdf.ln(10)
    pdf.cell(0, 10, "Annotated Image:", ln=True)
    pdf.image(annotated_image_path, x=10, y=pdf.get_y(), w=pdf.w - 20)

    pdf.output(output_path)

def save_record_to_salesforce(annotated_image_url, coverage_percent, original_image_pil, compliance_threshold=80):
    sf = Salesforce(
        username=os.environ['SF_USERNAME'],
        password=os.environ['SF_PASSWORD'],
        security_token=os.environ['SF_SECURITY_TOKEN'],
        domain=os.environ.get('SF_DOMAIN', 'login')  # 'test' for sandbox
    )
    
    # Save original image temporarily, upload it, get URL
    with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_orig_img_file:
        original_image_pil.save(temp_orig_img_file.name, format="JPEG")
        temp_orig_img_path = temp_orig_img_file.name
    
    original_image_url = upload_image_and_get_url(temp_orig_img_path)
    os.unlink(temp_orig_img_path)

    compliance_status = 'Pass' if coverage_percent >= compliance_threshold else 'Fail'
    technician_id = os.environ.get('SF_TECHNICIAN_ID')  # Salesforce UserId lookup
    
    record_name = f"UV Verification - {datetime.utcnow().strftime('%Y-%m-%d %H:%M:%S')}"

    sf.UV_Verification__c.create({
        'Name': record_name,
        'Annotated_Image__c': annotated_image_url,
        'Coverage_Percentage__c': round(coverage_percent, 2),
        'Original_Image__c': original_image_url,  # Correct field API name here
        'Compliance_Status__c': compliance_status,
        'Technician_ID__c': technician_id,
        'Verified_On__c': datetime.utcnow().isoformat()
    })

def process_image(input_img, brightness_threshold=150):
    img = cv2.cvtColor(np.array(input_img), cv2.COLOR_RGB2BGR)

    # Resize large images for faster processing, preserving aspect ratio
    max_dim = 640
    h, w = img.shape[:2]
    if max(h, w) > max_dim:
        scale = max_dim / max(h, w)
        img = cv2.resize(img, (int(w * scale), int(h * scale)))

    start_time = time.time()
    ocr_result = ocr_model.ocr(img)
    ocr_time = time.time() - start_time

    extracted_texts = []
    for line in ocr_result:
        if line:
            for word_info in line:
                text = word_info[1][0].strip()
                if len(text) > 1:
                    extracted_texts.append(text)

    annotated_img, coverage_percent = analyze_uv_coverage(img, brightness_threshold)

    with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as temp_img_file:
        cv2.imwrite(temp_img_file.name, annotated_img)
        annotated_img_path = temp_img_file.name

    temp_pdf_file = tempfile.NamedTemporaryFile(delete=False, suffix=".pdf")
    temp_pdf_file.close()
    create_pdf_report(coverage_percent, extracted_texts, annotated_img_path, temp_pdf_file.name)

    # Upload annotated image and get URL
    annotated_image_url = upload_image_and_get_url(annotated_img_path)

    # Save record in Salesforce
    save_record_to_salesforce(annotated_image_url, coverage_percent, input_img)

    annotated_img_rgb = cv2.cvtColor(annotated_img, cv2.COLOR_BGR2RGB)

    report_text = f"UV Sterilization Coverage: {coverage_percent:.2f}%"


    # Clean up temp image file after PDF generation
    os.unlink(annotated_img_path)

    return annotated_img_rgb, report_text, temp_pdf_file.name

iface = gr.Interface(
    fn=process_image,
    inputs=[ 
        gr.Image(type="pil", label="Upload Post-UV Sterilization Image"),
        gr.Slider(50, 255, value=150, step=1, label="Brightness Threshold")
    ],
    outputs=[
        gr.Image(type="numpy", label="Annotated Image"),
        gr.Textbox(label="UV Sterilization Report", lines=5),
        gr.File(label="Download PDF Report")
    ],
    title="UV Sterilization Coverage Analyzer",
    description="Upload a post-UV sterilization image to analyze surface coverage and generate a compliance report."
)

iface.queue()  # Enable request queuing to improve UX on heavy processing

if __name__ == "__main__":
    iface.launch()