Update working_yolo_pipeline.py

def get_latex_from_base64(base64_string: str) -> str:
if p2t is None:
return "[P2T_ERROR: Model not initialized]"

try:
image_data = base64.b64decode(base64_string)
image = Image.open(io.BytesIO(image_data))

result = p2t.recognize(
image,
save_formula_images=False,
use_analyzer=True,
keep_original_image=False
)

extracted = []
if isinstance(result, list):
for item in result:
text = item.text if hasattr(item, "text") else str(item)
extracted.append(text)
else:
extracted = [result]

latex = " ".join(extracted).strip()
latex = re.sub(r"\s+", "", latex)

if not latex:
return "[P2T_WARNING: No formula found]"

# Correct way: fix ONLY real double-backslashes produced by Pix2Text
latex = latex.replace(r"\\", r"\")

return latex

except Exception as e:
print(f"❌ Pix2Text Recognition failed: {e}")
return f"[P2T_ERROR: Recognition failed: {e}]"

working_yolo_pipeline.py

@@ -273,54 +273,97 @@ except Exception as e:
 
 
 
+# def get_latex_from_base64(base64_string: str) -> str:
+#     """
+#     Decodes a Base64 image string, uses Pix2Text to recognize the formula, 
+#     and returns the LaTeX code, stripped of all whitespace, as requested, 
+#     and corrects unintended double backslashes.
+#     """
+#     if p2t is None:
+#         return "[P2T_ERROR: Model not initialized]"
+
+#     try:
+#         # 1. Decode Base64 to Image
+#         image_data = base64.b64decode(base64_string)
+#         image = Image.open(io.BytesIO(image_data))
+        
+#         # 2. Recognize text and formulas
+#         # Use keep_original_image=False to save memory
+#         result = p2t.recognize(image, save_formula_images=False, use_analyzer=True, keep_original_image=False)
+        
+#         # 3. Parse the result for LaTeX
+#         extracted_latex_parts = []
+#         if isinstance(result, list):
+#             for item in result:
+#                 # Use .text for structured output, item itself for string output
+#                 text = item.text if hasattr(item, 'text') else str(item)
+#                 extracted_latex_parts.append(text)
+#         elif isinstance(result, str):
+#              extracted_latex_parts = [result]
+        
+#         # Join with a space first, then clean all whitespace
+#         extracted_latex = " ".join(extracted_latex_parts).strip()
+        
+#         # *** CORE CHANGE 1: Remove all spaces/line breaks ***
+#         cleaned_latex = re.sub(r'\s+', '', extracted_latex)
+        
+#         if not cleaned_latex:
+#              return "[P2T_WARNING: No formula found]"
+        
+#         # *** CORE CHANGE 2: Fix unintended double backslashes for LaTeX rendering ***
+#         # This replaces every sequence of two literal backslashes ('\\') with one literal backslash ('\'),
+#         # ensuring LaTeX commands like '\frac' are correctly formed.
+#         final_latex = cleaned_latex.replace('\\\\', '\\')
+        
+#         # Return the clean and corrected LaTeX string.
+#         return final_latex
+
+#     except Exception as e:
+#         # Catch any unexpected errors
+#         print(f"  ❌ Pix2Text Recognition failed: {e}")
+#         return f"[P2T_ERROR: Recognition failed: {e}]"
+
+
+
+
+
+
 def get_latex_from_base64(base64_string: str) -> str:
-    """
-    Decodes a Base64 image string, uses Pix2Text to recognize the formula, 
-    and returns the LaTeX code, stripped of all whitespace, as requested, 
-    and corrects unintended double backslashes.
-    """
     if p2t is None:
         return "[P2T_ERROR: Model not initialized]"
 
     try:
-        # 1. Decode Base64 to Image
         image_data = base64.b64decode(base64_string)
         image = Image.open(io.BytesIO(image_data))
-        
-        # 2. Recognize text and formulas
-        # Use keep_original_image=False to save memory
-        result = p2t.recognize(image, save_formula_images=False, use_analyzer=True, keep_original_image=False)
-        
-        # 3. Parse the result for LaTeX
-        extracted_latex_parts = []
+
+        result = p2t.recognize(
+            image,
+            save_formula_images=False,
+            use_analyzer=True,
+            keep_original_image=False
+        )
+
+        extracted = []
         if isinstance(result, list):
             for item in result:
-                # Use .text for structured output, item itself for string output
-                text = item.text if hasattr(item, 'text') else str(item)
-                extracted_latex_parts.append(text)
-        elif isinstance(result, str):
-             extracted_latex_parts = [result]
-        
-        # Join with a space first, then clean all whitespace
-        extracted_latex = " ".join(extracted_latex_parts).strip()
-        
-        # *** CORE CHANGE 1: Remove all spaces/line breaks ***
-        cleaned_latex = re.sub(r'\s+', '', extracted_latex)
-        
-        if not cleaned_latex:
-             return "[P2T_WARNING: No formula found]"
-        
-        # *** CORE CHANGE 2: Fix unintended double backslashes for LaTeX rendering ***
-        # This replaces every sequence of two literal backslashes ('\\') with one literal backslash ('\'),
-        # ensuring LaTeX commands like '\frac' are correctly formed.
-        final_latex = cleaned_latex.replace('\\\\', '\\')
-        
-        # Return the clean and corrected LaTeX string.
-        return final_latex
+                text = item.text if hasattr(item, "text") else str(item)
+                extracted.append(text)
+        else:
+            extracted = [result]
+
+        latex = " ".join(extracted).strip()
+        latex = re.sub(r"\s+", "", latex)
+
+        if not latex:
+            return "[P2T_WARNING: No formula found]"
+
+        # Correct way: fix ONLY real double-backslashes produced by Pix2Text
+        latex = latex.replace(r"\\", r"\")
+
+        return latex
 
     except Exception as e:
-        # Catch any unexpected errors
-        print(f"  ❌ Pix2Text Recognition failed: {e}")
+        print(f"❌ Pix2Text Recognition failed: {e}")
         return f"[P2T_ERROR: Recognition failed: {e}]"
 
 
@@ -329,7 +372,6 @@ def get_latex_from_base64(base64_string: str) -> str:
 
 
 
-
 # # Initialize the YOLO model
 # model = YOLO(WEIGHTS_PATH)