Update app.py

app.py

@@ -53,39 +53,44 @@ def clean_latex(latex):
     return latex
 
 def clean_latex2(latex):
+    # Remove unnecessary LaTeX symbols and spaces
     latex = latex.replace('\\ ', '')
-    latex = latex.replace('\\\\', '\n')  # Convert LaTeX line breaks
-    latex = re.sub(r'\\(times)', '*', latex)  # Fix \times to *
-    latex = re.sub(r'\\(cdot)', '*', latex)
-    
-    # Remove wrappers like \mathbf{y}
+    latex = latex.replace('\\\\', '\n')
+    latex = latex.replace('~', '')
+    latex = latex.replace('−', '-').replace('–', '-')
+    latex = latex.replace('–', '-')
+
+    # Convert math symbols
+    latex = re.sub(r'\\times', '*', latex)
+    latex = re.sub(r'\\cdot', '*', latex)
+
+    # Fix multiple dashes (e.g., -- becomes -)
+    latex = re.sub(r'-{2,}', '-', latex)
+
+    # Remove commands like \mathbf{}, \mathcal{}, etc.
     latex = re.sub(r'\\(text|mathbf|mathrm|mathit|textbf|mathcal|cal)\s*\{([^{}]+)\}', r'\2', latex)
 
-    # Remove remaining commands (like \!\!\!)
+    # Remove leftover LaTeX commands
     latex = re.sub(r'\\[a-zA-Z]+', '', latex)
 
-    # Flatten array blocks
-    latex = latex.replace(r'\begin{array}', '').replace(r'\end{array}', '')
+    # Remove curly braces and ampersands
+    latex = latex.replace('{', '').replace('}', '').replace('&', '')
 
-    # Remove curly braces
-    latex = latex.replace('{', '').replace('}', '')
+    # Replace comma/semicolon with newlines to separate equations
+    latex = re.sub(r'[;,]', '\n', latex)
 
-    # Clean punctuation
-    latex = latex.replace('&', '').replace(';', '\n').replace(',', '\n')
-    latex = latex.replace('−', '-').replace('–', '-')
+    # Fix multiple equals (== becomes =)
+    latex = re.sub(r'=+', '=', latex)
 
-    # Normalize equals
-    latex = re.sub(r'(?<![=<>])=(?![=<>])', ' = ', latex)
+    # Normalize space
+    latex = re.sub(r'\s+', ' ', latex).strip()
 
-    # Replace Greek/misread symbols
+    # Symbol replacements (OCR common confusions)
     latex = re.sub(r'\bchi\b|\bxi\b|\balpha\b', 'x', latex)
     latex = re.sub(r'\bbeta\b', 'b', latex)
     latex = re.sub(r'\bgamma\b', 'y', latex)
 
-    # Remove extra whitespace
-    latex = re.sub(r'\s+', ' ', latex)
-
-    return latex.strip()
+    return latex
 
 def solve_polynomial(image):
     try: