Update working_yolo_pipeline.py

working_yolo_pipeline.py

@@ -273,11 +273,66 @@ 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 = extracted_latex.replace('\\\\', '\\')
+        final_latex = re.sub(r'\s+', '', cleaned_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.
+        
+        
+        # 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.
+#     Decodes a Base64 image string, uses Pix2Text to recognize the formula,
+#     returns the LaTeX code stripped of all whitespace, and collapses unintended
+#     repeated backslashes into a single backslash.
 #     """
 #     if p2t is None:
 #         return "[P2T_ERROR: Model not initialized]"
@@ -286,40 +341,37 @@ except Exception as e:
 #         # 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)
-        
+#         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_parts = [result]
+
+#         # Join then strip
 #         extracted_latex = " ".join(extracted_latex_parts).strip()
-        
-#         # *** CORE CHANGE 1: Remove all spaces/line breaks ***
+
+#         # Remove all whitespace/newlines/tabs as requested
 #         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 "[P2T_WARNING: No formula found]"
+
+#         # COLLAPSE any run of 2 or more backslashes into a single backslash.
+#         # This handles inputs like '\\\\sqrt' or '\\\\\\frac' robustly.
+#         final_latex = re.sub(r'\\{2,}', r'\\', cleaned_latex)
+
 #         return final_latex
 
 #     except Exception as e:
-#         # Catch any unexpected errors
 #         print(f"  ❌ Pix2Text Recognition failed: {e}")
 #         return f"[P2T_ERROR: Recognition failed: {e}]"
 
@@ -327,57 +379,6 @@ except Exception as e:
 
 
 
-def get_latex_from_base64(base64_string: str) -> str:
-    """
-    Decodes a Base64 image string, uses Pix2Text to recognize the formula,
-    returns the LaTeX code stripped of all whitespace, and collapses unintended
-    repeated backslashes into a single backslash.
-    """
-    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
-        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:
-                text = item.text if hasattr(item, 'text') else str(item)
-                extracted_latex_parts.append(text)
-        elif isinstance(result, str):
-            extracted_latex_parts = [result]
-
-        # Join then strip
-        extracted_latex = " ".join(extracted_latex_parts).strip()
-
-        # Remove all whitespace/newlines/tabs as requested
-        cleaned_latex = re.sub(r'\s+', '', extracted_latex)
-
-        if not cleaned_latex:
-            return "[P2T_WARNING: No formula found]"
-
-        # COLLAPSE any run of 2 or more backslashes into a single backslash.
-        # This handles inputs like '\\\\sqrt' or '\\\\\\frac' robustly.
-        final_latex = re.sub(r'\\{2,}', r'\\', cleaned_latex)
-
-        return final_latex
-
-    except Exception as e:
-        print(f"  ❌ Pix2Text Recognition failed: {e}")
-        return f"[P2T_ERROR: Recognition failed: {e}]"
-
-
-
-
-
 
 
 # # Initialize the YOLO model