Update app.py
Browse files
app.py
CHANGED
|
@@ -15,31 +15,67 @@ model = LatexOCR()
|
|
| 15 |
|
| 16 |
# Clean LaTeX output
|
| 17 |
def clean_latex(latex):
|
| 18 |
-
# π§
|
| 19 |
-
latex = latex.replace('\\ ', '')
|
| 20 |
-
latex = latex.replace('\\\\', '\\')
|
| 21 |
-
latex = re.sub(r'\\[ \t\n\r\f\v]*', '', latex)
|
| 22 |
|
| 23 |
# β
Fix invalid LaTeX escapes like \+ \- \=
|
| 24 |
latex = re.sub(r'\\([+\-=])', r'\1', latex)
|
| 25 |
|
| 26 |
-
# π Replace common misrecognized symbols
|
| 27 |
# π Replace common misrecognized symbols (with and without backslash)
|
| 28 |
replacements = {
|
| 29 |
-
r'\\chi': 'x',
|
| 30 |
-
r'
|
| 31 |
-
r'\\
|
| 32 |
-
r'
|
| 33 |
-
r'\\
|
| 34 |
-
r'alpha': 'a',
|
| 35 |
-
r'\\beta': 'b',
|
| 36 |
-
r'beta': 'b',
|
| 37 |
-
r'\\gamma': 'y',
|
| 38 |
-
r'gamma': 'y'
|
| 39 |
}
|
| 40 |
for wrong, correct in replacements.items():
|
| 41 |
latex = re.sub(wrong, correct, latex)
|
| 42 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
# Replace mathcal variables
|
| 45 |
latex = re.sub(r'\\(cal|mathcal)\s*X', 'x', latex)
|
|
|
|
| 15 |
|
| 16 |
# Clean LaTeX output
|
| 17 |
def clean_latex(latex):
|
| 18 |
+
# π§ Remove bad spacing escapes and redundant slashes
|
| 19 |
+
latex = latex.replace('\\ ', '')
|
| 20 |
+
latex = latex.replace('\\\\', '\\')
|
| 21 |
+
latex = re.sub(r'\\[ \t\n\r\f\v]*', '', latex)
|
| 22 |
|
| 23 |
# β
Fix invalid LaTeX escapes like \+ \- \=
|
| 24 |
latex = re.sub(r'\\([+\-=])', r'\1', latex)
|
| 25 |
|
|
|
|
| 26 |
# π Replace common misrecognized symbols (with and without backslash)
|
| 27 |
replacements = {
|
| 28 |
+
r'\\chi': 'x', r'chi': 'x',
|
| 29 |
+
r'\\xi': 'x', r'xi': 'x',
|
| 30 |
+
r'\\alpha': 'a', r'alpha': 'a',
|
| 31 |
+
r'\\beta': 'b', r'beta': 'b',
|
| 32 |
+
r'\\gamma': 'y', r'gamma': 'y'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
}
|
| 34 |
for wrong, correct in replacements.items():
|
| 35 |
latex = re.sub(wrong, correct, latex)
|
| 36 |
|
| 37 |
+
# β
Fix mathcal and cal variables in all formats
|
| 38 |
+
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*X\s*\}?', 'x', latex)
|
| 39 |
+
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*Y\s*\}?', 'y', latex)
|
| 40 |
+
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*Z\s*\}?', 'z', latex)
|
| 41 |
+
|
| 42 |
+
# β
Also clean up OCR leftovers like `cal X`, `mathcal x`
|
| 43 |
+
latex = latex.replace('cal x', 'x')
|
| 44 |
+
latex = latex.replace('cal X', 'x')
|
| 45 |
+
latex = latex.replace('mathcal x', 'x')
|
| 46 |
+
latex = latex.replace('mathcal X', 'x')
|
| 47 |
+
|
| 48 |
+
# π§Ή Remove curly braces and punctuation
|
| 49 |
+
latex = latex.replace('{', '').replace('}', '')
|
| 50 |
+
latex = latex.strip().rstrip(',.') # remove trailing punctuation
|
| 51 |
+
|
| 52 |
+
# β
Replace constants
|
| 53 |
+
latex = re.sub(r'(\d+)\s*\\pi', r'(\1*3.1416)', latex)
|
| 54 |
+
latex = latex.replace(r'\pi', '3.1416')
|
| 55 |
+
latex = re.sub(r'(\d+)\s*e', r'(\1*2.7183)', latex)
|
| 56 |
+
latex = re.sub(r'(?<![a-zA-Z0-9])e(?![a-zA-Z0-9])', '2.7183', latex)
|
| 57 |
+
|
| 58 |
+
# β
Add multiplication where missing
|
| 59 |
+
latex = re.sub(r'(\d)([a-zA-Z])', r'\1*\2', latex)
|
| 60 |
+
latex = re.sub(r'(\d+)\s*i', r'\1*I', latex)
|
| 61 |
+
latex = re.sub(r'(?<![a-zA-Z0-9])i(?![a-zA-Z0-9])', 'I', latex)
|
| 62 |
+
|
| 63 |
+
# β
(x+1)x β (x+1)*x
|
| 64 |
+
latex = re.sub(r'\(([^()]+?)\)\s*([a-zA-Z](\^\d+)?)', r'(\1)*\2', latex)
|
| 65 |
+
|
| 66 |
+
# β
Fix symbols
|
| 67 |
+
latex = latex.replace(r'\cdot', '*')
|
| 68 |
+
latex = latex.replace('β', '-') # Unicode minus
|
| 69 |
+
|
| 70 |
+
# β
Final cleanup: remove any stray garbage symbols
|
| 71 |
+
latex = re.sub(r'[^\w\s^=+*\-().]', '', latex)
|
| 72 |
+
|
| 73 |
+
# β
Ensure it's an equation
|
| 74 |
+
if '=' not in latex:
|
| 75 |
+
latex += '=0'
|
| 76 |
+
|
| 77 |
+
return latex
|
| 78 |
+
|
| 79 |
|
| 80 |
# Replace mathcal variables
|
| 81 |
latex = re.sub(r'\\(cal|mathcal)\s*X', 'x', latex)
|