Create ocr_module.py

import os
import json
import random
import re
from typing import Dict, List, Any
from PIL import Image

CRITICAL_OPERATORS = ["\\int", "\\sum", "=", "\\frac", "+", "-", "*", "\\times", "\\div"]
BRACKETS_LIMITS = ["(", ")", "[", "]", "\\{", "\\}", "^", "_"]
AMBIGUOUS_SYMBOLS = ["8", "B", "0", "O", "l", "1", "I", "S", "5", "Z", "2"]

def get_symbol_weight(symbol: str) -> float:
if symbol in CRITICAL_OPERATORS: return 1.5
elif symbol in BRACKETS_LIMITS: return 1.3
elif symbol in AMBIGUOUS_SYMBOLS: return 0.7
return 1.0

def calculate_weighted_confidence(latex_string: str, mock_logits: bool = True) -> float:
tokens = []
current_token = ""
for char in latex_string:
if char == '\\':
if current_token: tokens.append(current_token)
current_token = char
elif char.isalnum() and current_token.startswith('\\'):
current_token += char
else:
if current_token:
tokens.append(current_token)
current_token = ""
if char.strip(): tokens.append(char)
if current_token: tokens.append(current_token)
total_weighted_ci = 0.0
total_weights = 0.0
for token in tokens:
w_i = get_symbol_weight(token)
c_i = random.uniform(0.85, 0.99) if mock_logits else 0.95
total_weighted_ci += (w_i * c_i)
total_weights += w_i
if total_weights == 0: return 0.0
return round(total_weighted_ci / total_weights, 4)

class MVM2OCREngine:
def __init__(self):
try:
from pix2text import Pix2Text
self.p2t = Pix2Text.from_config()
self.model_loaded = True
except:
self.model_loaded = False

def clean_latex_output(self, text: str) -> str:
cjk_re = re.compile(r'[\u4e00-\u9fff\u3040-\u30ff\uac00-\ud7af]')
return cjk_re.sub('', text)

def process_image(self, image_path: str) -> Dict[str, Any]:
if not os.path.exists(image_path): return {"error": "File not found"}
if self.model_loaded:
try:
out = self.p2t.recognize(image_path)
if isinstance(out, str): raw_latex = out
else: raw_latex = "\n".join([item.get('text', '') for item in out])
except: raw_latex = "Error during OCR"
else: raw_latex = "No math detected (Simulated)."
raw_latex = self.clean_latex_output(raw_latex)
return {"latex_output": raw_latex, "weighted_confidence": calculate_weighted_confidence(raw_latex)}

ocr_module.py

@@ -0,0 +1,24 @@
+import os
+import json
+import random
+import re
+from typing import Dict, List, Any
+from PIL import Image
+CRITICAL_OPERATORS = ["\\int", "\\sum", "=", "\\frac", "+", "-", "*", "\\times", "\\div"]
+BRACKETS_LIMITS = ["(", ")", "[", "]", "\\{", "\\}", "^", "_"]
+AMBIGUOUS_SYMBOLS = ["8", "B", "0", "O", "l", "1", "I", "S", "5", "Z", "2"]
+
+def get_symbol_weight(symbol: str) -> float:
+        if symbol in CRITICAL_OPERATORS: return 1.5
+        elif symbol in BRACKETS_LIMITS: return 1.3
+        elif symbol in AMBIGUOUS_SYMBOLS: return 0.7
+        return 1.0
+
+    def calculate_weighted_confidence(latex_string: str, mock_logits: bool = True) -> float:
+            tokens = []
+            current_token = ""
+            for char in latex_string:
+                        if char == '\\':
+                                        if current_token: tokens.append(current_token)
+                                        current_token = char
+