generated-bill / backend /src /python /ocr_engine.py
narasimha01tm's picture
Deploy SmartSpend backend
4bf0aa2
Raw History Blame Contribute Delete
5.21 kB
import sys
import json
import base64
import cv2
import numpy as np
import easyocr
import logging
import traceback
# Configure logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
logger = logging.getLogger(__name__)
# Global reader instance
_reader = None
def get_reader():
global _reader
if _reader is None:
# Initialize with English
_reader = easyocr.Reader(['en'], gpu=False)
return _reader
def process_image(image_data):
try:
reader = get_reader()
# Decode base64 image
if ',' in image_data:
image_data = image_data.split(',')[1]
image_bytes = base64.b64decode(image_data)
nparr = np.frombuffer(image_bytes, np.uint8)
img = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
if img is None:
raise ValueError("Could not decode image")
# Run OCR
result = reader.readtext(img)
extracted_text = []
structured_data = []
# Process EasyOCR results
for item in result:
bbox, text, confidence = item
# Convert bbox points to standard list of lists
box = [[float(pt[0]), float(pt[1])] for pt in bbox]
extracted_text.append(text)
structured_data.append({
"text": text,
"confidence": float(confidence),
"box": box
})
# Reconstruct lines based on Y-coordinates
lines = []
if structured_data:
def get_y_min(item):
box = item['box']
return box[0][1]
structured_data.sort(key=get_y_min)
current_line = []
current_y = -1
y_threshold = 20 # Pixel threshold for same line
for item in structured_data:
box = item['box']
y_center = (box[0][1] + box[2][1]) / 2
if current_y == -1:
current_y = y_center
current_line.append(item)
elif abs(y_center - current_y) < y_threshold:
current_line.append(item)
else:
# Sort current line by X-coordinate
def get_x_min(item):
b = item['box']
return b[0][0]
current_line.sort(key=get_x_min)
lines.append(current_line)
current_line = [item]
current_y = y_center
if current_line:
def get_x_min(item):
b = item['box']
return b[0][0]
current_line.sort(key=get_x_min)
lines.append(current_line)
# Build full text from lines with enhancement (User request)
import re
final_lines = []
for line in lines:
line_text = " ".join([str(item['text']) for item in line])
# 1. Filter out "thank you" and "visit" phrases
junk_keywords = ["thank you", "visit again", "thanks", "welcome", "have a nice day", "visit us", "come again"]
if any(k in line_text.lower() for k in junk_keywords):
continue
# 2. Remove any numbers (User request: "dont add any number")
line_text = re.sub(r'\d+', '', line_text)
# Clean up extra whitespace and punctuation
line_text = re.sub(r'\s+', ' ', line_text).strip()
# Skip empty lines or lines with no alphanumeric characters
if line_text and any(c.isalnum() for c in line_text):
final_lines.append(line_text)
full_text = "\n".join(final_lines)
engine_type = "easyocr" # Define engine_type as requested
return {
"success": True,
"text": full_text,
"data": structured_data,
"engine": engine_type
}
except Exception as e:
logger.error(f"OCR processing failed: {str(e)}")
logger.error(traceback.format_exc())
return {
"success": False,
"error": str(e)
}
if __name__ == "__main__":
try:
# Read input from stdin
input_data = sys.stdin.read()
if not input_data:
print(json.dumps({"success": False, "error": "No input data received"}))
sys.exit(1)
try:
request = json.loads(input_data)
except json.JSONDecodeError:
print(json.dumps({"success": False, "error": "Invalid JSON input"}))
sys.exit(1)
image_base64 = request.get("imageBase64")
if not image_base64:
print(json.dumps({"success": False, "error": "imageBase64 is required"}))
sys.exit(1)
result = process_image(image_base64)
print(json.dumps(result))
except Exception as e:
print(json.dumps({"success": False, "error": f"Unexpected error: {str(e)}"}))
sys.exit(1)