Update app.py
Browse files
app.py
CHANGED
|
@@ -15,15 +15,14 @@ model = LatexOCR()
|
|
| 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 |
-
#
|
| 24 |
latex = re.sub(r'\\([+\-=])', r'\1', latex)
|
| 25 |
|
| 26 |
-
#
|
| 27 |
replacements = {
|
| 28 |
r'\\chi': 'x', r'chi': 'x',
|
| 29 |
r'\\xi': 'x', r'xi': 'x',
|
|
@@ -32,42 +31,32 @@ def clean_latex(latex):
|
|
| 32 |
r'\\gamma': 'y', r'gamma': 'y',
|
| 33 |
r'\\vartheta': '3', r'vartheta': '3',
|
| 34 |
r'\\mathcalW': 'x', r'mathcalW': 'x',
|
|
|
|
|
|
|
| 35 |
}
|
| 36 |
for wrong, correct in replacements.items():
|
| 37 |
latex = re.sub(wrong, correct, latex)
|
| 38 |
|
| 39 |
-
# β
Fix mathcal and cal variables
|
| 40 |
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*[Xx]\s*\}?', 'x', latex)
|
| 41 |
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*[Yy]\s*\}?', 'y', latex)
|
| 42 |
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*[Zz]\s*\}?', 'z', latex)
|
|
|
|
| 43 |
latex = latex.replace('cal x', 'x').replace('cal X', 'x')
|
| 44 |
latex = latex.replace('mathcal x', 'x').replace('mathcal X', 'x')
|
| 45 |
|
| 46 |
-
# π§Ή Remove curly braces and trailing punctuation
|
| 47 |
latex = latex.replace('{', '').replace('}', '')
|
| 48 |
-
latex = latex.strip().rstrip(',.')
|
| 49 |
-
|
| 50 |
-
# β
Replace symbolic constants and protect them
|
| 51 |
-
latex = latex.replace(r'\pi', 'SYMPY_PI') # protect pi
|
| 52 |
latex = re.sub(r'(?<![a-zA-Z0-9])e(?![a-zA-Z0-9])', 'E', latex)
|
| 53 |
|
| 54 |
-
|
| 55 |
-
latex = re.sub(r'(\d)([a-zA-Z])', r'\1*\2', latex) # 4x β 4*x
|
| 56 |
latex = re.sub(r'(\d+)\s*i', r'\1*I', latex)
|
| 57 |
latex = re.sub(r'(?<![a-zA-Z0-9])i(?![a-zA-Z0-9])', 'I', latex)
|
| 58 |
latex = re.sub(r'\(([^()]+?)\)\s*([a-zA-Z](\^\d+)?)', r'(\1)*\2', latex)
|
| 59 |
|
| 60 |
-
# β
Fix symbols
|
| 61 |
latex = latex.replace(r'\cdot', '*')
|
| 62 |
latex = latex.replace('β', '-') # Unicode minus
|
| 63 |
-
|
| 64 |
-
# β
Final cleanup
|
| 65 |
latex = re.sub(r'[^\w\s^=+*\-().]', '', latex)
|
| 66 |
|
| 67 |
-
# β
Restore pi
|
| 68 |
-
latex = latex.replace('SYMPY_PI', 'pi')
|
| 69 |
-
|
| 70 |
-
# β
Ensure equation
|
| 71 |
if '=' not in latex:
|
| 72 |
latex += '=0'
|
| 73 |
|
|
@@ -75,7 +64,9 @@ def clean_latex(latex):
|
|
| 75 |
|
| 76 |
# Fallback to external LLM microservice
|
| 77 |
def request_llm_fallback(bad_latex, llm_url):
|
| 78 |
-
pre_cleaned = re.sub(
|
|
|
|
|
|
|
| 79 |
try:
|
| 80 |
response = requests.post(f"{llm_url}/clean", json={"prompt": pre_cleaned})
|
| 81 |
if response.status_code == 200:
|
|
@@ -84,7 +75,7 @@ def request_llm_fallback(bad_latex, llm_url):
|
|
| 84 |
print(f"LLM fallback failed: {e}")
|
| 85 |
return pre_cleaned
|
| 86 |
|
| 87 |
-
#
|
| 88 |
def request_llm_explanation(latex, llm_url):
|
| 89 |
try:
|
| 90 |
response = requests.post(f"{llm_url}/explain", json={"prompt": latex})
|
|
@@ -110,12 +101,15 @@ def solve_polynomial(image, llm_url):
|
|
| 110 |
expr = parse_latex(cleaned_latex)
|
| 111 |
if not isinstance(expr, sp.Equality):
|
| 112 |
raise ValueError("Expression is not an equation.")
|
|
|
|
| 113 |
lhs_minus_rhs = expr.lhs - expr.rhs
|
| 114 |
if not lhs_minus_rhs.is_polynomial():
|
| 115 |
-
raise ValueError("
|
|
|
|
| 116 |
junk_symbols = {"pm", "cdot", "mathrm", "boldmath", "bar", "L"}
|
| 117 |
-
|
| 118 |
-
|
|
|
|
| 119 |
except Exception:
|
| 120 |
fixed_latex = request_llm_fallback(cleaned_latex, llm_url)
|
| 121 |
cleaned_latex = clean_latex(fixed_latex)
|
|
@@ -125,19 +119,24 @@ def solve_polynomial(image, llm_url):
|
|
| 125 |
raise ValueError("Expression is not an equation.")
|
| 126 |
lhs_minus_rhs = expr.lhs - expr.rhs
|
| 127 |
if not lhs_minus_rhs.is_polynomial():
|
| 128 |
-
raise ValueError("
|
| 129 |
-
|
| 130 |
-
|
|
|
|
| 131 |
except Exception:
|
| 132 |
expr = None
|
| 133 |
|
| 134 |
if expr is None:
|
| 135 |
-
return
|
|
|
|
|
|
|
|
|
|
|
|
|
| 136 |
|
| 137 |
output = (
|
| 138 |
-
f"## π Extracted LaTeX\n```latex\n{latex_result}\n```\n"
|
| 139 |
-
f"\n---\n## π§Ή Cleaned LaTeX Used\n```latex\n{cleaned_latex}\n```\n"
|
| 140 |
-
f"\n---\n## π§ Parsed Expression\n$$ {sp.latex(expr)} $$\n---\n"
|
| 141 |
)
|
| 142 |
|
| 143 |
if isinstance(expr, sp.Equality):
|
|
@@ -151,9 +150,10 @@ def solve_polynomial(image, llm_url):
|
|
| 151 |
x_sym = sym
|
| 152 |
break
|
| 153 |
if x_sym is None:
|
| 154 |
-
raise ValueError("β Could not find variable 'x'.")
|
| 155 |
|
| 156 |
roots = sp.solve(sp.Eq(lhs, 0), x_sym, dict=True)
|
|
|
|
| 157 |
output += "## β
Step 3: Solve Roots\n"
|
| 158 |
if roots:
|
| 159 |
output += "$$\n\\begin{aligned}\n"
|
|
@@ -173,8 +173,8 @@ def solve_polynomial(image, llm_url):
|
|
| 173 |
|
| 174 |
# Gradio UI
|
| 175 |
with gr.Blocks() as demo:
|
| 176 |
-
llm_url = gr.Textbox(label="π Enter LLM Microservice URL", placeholder="https://xxxx.ngrok-free.app")
|
| 177 |
-
image_input = gr.Image(type="pil", label="π· Upload Polynomial
|
| 178 |
output_box = gr.Markdown(label="π Step-by-step Solution")
|
| 179 |
explain_box = gr.Markdown(label="π£οΈ Human-style Explanation")
|
| 180 |
hidden_latex = gr.Textbox(visible=False)
|
|
@@ -190,7 +190,7 @@ with gr.Blocks() as demo:
|
|
| 190 |
explain_btn.click(fn=request_llm_explanation, inputs=[hidden_latex, llm_url], outputs=explain_box)
|
| 191 |
|
| 192 |
demo.title = "π§ Polynomial Solver from Image"
|
| 193 |
-
demo.description = "Upload a typed or handwritten
|
| 194 |
|
| 195 |
if __name__ == "__main__":
|
| 196 |
demo.launch()
|
|
|
|
| 15 |
|
| 16 |
# Clean LaTeX output
|
| 17 |
def clean_latex(latex):
|
|
|
|
| 18 |
latex = latex.replace('\\ ', '')
|
| 19 |
latex = latex.replace('\\\\', '\\')
|
| 20 |
latex = re.sub(r'\\[ \t\n\r\f\v]*', '', latex)
|
| 21 |
|
| 22 |
+
# Fix invalid LaTeX escapes like \+ \- \=
|
| 23 |
latex = re.sub(r'\\([+\-=])', r'\1', latex)
|
| 24 |
|
| 25 |
+
# Replace common misrecognized symbols
|
| 26 |
replacements = {
|
| 27 |
r'\\chi': 'x', r'chi': 'x',
|
| 28 |
r'\\xi': 'x', r'xi': 'x',
|
|
|
|
| 31 |
r'\\gamma': 'y', r'gamma': 'y',
|
| 32 |
r'\\vartheta': '3', r'vartheta': '3',
|
| 33 |
r'\\mathcalW': 'x', r'mathcalW': 'x',
|
| 34 |
+
r'\\pi': 'pi', r'pi': 'pi', # β
prevent pi β ip
|
| 35 |
+
r'\\mathrm': '', r'mathrm': '',
|
| 36 |
}
|
| 37 |
for wrong, correct in replacements.items():
|
| 38 |
latex = re.sub(wrong, correct, latex)
|
| 39 |
|
|
|
|
| 40 |
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*[Xx]\s*\}?', 'x', latex)
|
| 41 |
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*[Yy]\s*\}?', 'y', latex)
|
| 42 |
latex = re.sub(r'\\(cal|mathcal)\s*\{?\s*[Zz]\s*\}?', 'z', latex)
|
| 43 |
+
|
| 44 |
latex = latex.replace('cal x', 'x').replace('cal X', 'x')
|
| 45 |
latex = latex.replace('mathcal x', 'x').replace('mathcal X', 'x')
|
| 46 |
|
|
|
|
| 47 |
latex = latex.replace('{', '').replace('}', '')
|
| 48 |
+
latex = latex.strip().rstrip(',.')
|
|
|
|
|
|
|
|
|
|
| 49 |
latex = re.sub(r'(?<![a-zA-Z0-9])e(?![a-zA-Z0-9])', 'E', latex)
|
| 50 |
|
| 51 |
+
latex = re.sub(r'(\d)([a-zA-Z])', r'\1*\2', latex)
|
|
|
|
| 52 |
latex = re.sub(r'(\d+)\s*i', r'\1*I', latex)
|
| 53 |
latex = re.sub(r'(?<![a-zA-Z0-9])i(?![a-zA-Z0-9])', 'I', latex)
|
| 54 |
latex = re.sub(r'\(([^()]+?)\)\s*([a-zA-Z](\^\d+)?)', r'(\1)*\2', latex)
|
| 55 |
|
|
|
|
| 56 |
latex = latex.replace(r'\cdot', '*')
|
| 57 |
latex = latex.replace('β', '-') # Unicode minus
|
|
|
|
|
|
|
| 58 |
latex = re.sub(r'[^\w\s^=+*\-().]', '', latex)
|
| 59 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
if '=' not in latex:
|
| 61 |
latex += '=0'
|
| 62 |
|
|
|
|
| 64 |
|
| 65 |
# Fallback to external LLM microservice
|
| 66 |
def request_llm_fallback(bad_latex, llm_url):
|
| 67 |
+
pre_cleaned = re.sub(
|
| 68 |
+
r'(\\!?pm|\\not=?|\\!|\\L|\\perp|\\bar|\\Sigma|\\boldmath|G|L(?=\^))', '', bad_latex
|
| 69 |
+
)
|
| 70 |
try:
|
| 71 |
response = requests.post(f"{llm_url}/clean", json={"prompt": pre_cleaned})
|
| 72 |
if response.status_code == 200:
|
|
|
|
| 75 |
print(f"LLM fallback failed: {e}")
|
| 76 |
return pre_cleaned
|
| 77 |
|
| 78 |
+
# Request human-style explanation
|
| 79 |
def request_llm_explanation(latex, llm_url):
|
| 80 |
try:
|
| 81 |
response = requests.post(f"{llm_url}/explain", json={"prompt": latex})
|
|
|
|
| 101 |
expr = parse_latex(cleaned_latex)
|
| 102 |
if not isinstance(expr, sp.Equality):
|
| 103 |
raise ValueError("Expression is not an equation.")
|
| 104 |
+
|
| 105 |
lhs_minus_rhs = expr.lhs - expr.rhs
|
| 106 |
if not lhs_minus_rhs.is_polynomial():
|
| 107 |
+
raise ValueError("Expression is not a polynomial.")
|
| 108 |
+
|
| 109 |
junk_symbols = {"pm", "cdot", "mathrm", "boldmath", "bar", "L"}
|
| 110 |
+
symbols = {str(s) for s in expr.free_symbols}
|
| 111 |
+
if symbols & junk_symbols:
|
| 112 |
+
raise ValueError("Expression contains junk symbols.")
|
| 113 |
except Exception:
|
| 114 |
fixed_latex = request_llm_fallback(cleaned_latex, llm_url)
|
| 115 |
cleaned_latex = clean_latex(fixed_latex)
|
|
|
|
| 119 |
raise ValueError("Expression is not an equation.")
|
| 120 |
lhs_minus_rhs = expr.lhs - expr.rhs
|
| 121 |
if not lhs_minus_rhs.is_polynomial():
|
| 122 |
+
raise ValueError("Expression is not a polynomial.")
|
| 123 |
+
symbols = {str(s) for s in expr.free_symbols}
|
| 124 |
+
if symbols & junk_symbols:
|
| 125 |
+
raise ValueError("Expression contains junk symbols.")
|
| 126 |
except Exception:
|
| 127 |
expr = None
|
| 128 |
|
| 129 |
if expr is None:
|
| 130 |
+
return (
|
| 131 |
+
f"β Could not parse expression even after fallback:\n\n```latex\n{cleaned_latex}\n```",
|
| 132 |
+
cleaned_latex,
|
| 133 |
+
""
|
| 134 |
+
)
|
| 135 |
|
| 136 |
output = (
|
| 137 |
+
f"## π Extracted LaTeX\n\n```latex\n{latex_result}\n```\n"
|
| 138 |
+
f"\n---\n## π§Ή Cleaned LaTeX Used\n\n```latex\n{cleaned_latex}\n```\n"
|
| 139 |
+
f"\n---\n## π§ Parsed Expression\n\n$$ {sp.latex(expr)} $$\n\n---\n"
|
| 140 |
)
|
| 141 |
|
| 142 |
if isinstance(expr, sp.Equality):
|
|
|
|
| 150 |
x_sym = sym
|
| 151 |
break
|
| 152 |
if x_sym is None:
|
| 153 |
+
raise ValueError("β Could not find variable 'x' in the equation.")
|
| 154 |
|
| 155 |
roots = sp.solve(sp.Eq(lhs, 0), x_sym, dict=True)
|
| 156 |
+
|
| 157 |
output += "## β
Step 3: Solve Roots\n"
|
| 158 |
if roots:
|
| 159 |
output += "$$\n\\begin{aligned}\n"
|
|
|
|
| 173 |
|
| 174 |
# Gradio UI
|
| 175 |
with gr.Blocks() as demo:
|
| 176 |
+
llm_url = gr.Textbox(label="π Enter LLM Microservice URL (from Colab)", placeholder="https://xxxx.ngrok-free.app")
|
| 177 |
+
image_input = gr.Image(type="pil", label="π· Upload Image of Polynomial")
|
| 178 |
output_box = gr.Markdown(label="π Step-by-step Solution")
|
| 179 |
explain_box = gr.Markdown(label="π£οΈ Human-style Explanation")
|
| 180 |
hidden_latex = gr.Textbox(visible=False)
|
|
|
|
| 190 |
explain_btn.click(fn=request_llm_explanation, inputs=[hidden_latex, llm_url], outputs=explain_box)
|
| 191 |
|
| 192 |
demo.title = "π§ Polynomial Solver from Image"
|
| 193 |
+
demo.description = "Upload a polynomial image (typed or handwritten). View symbolic steps and human-style explanation."
|
| 194 |
|
| 195 |
if __name__ == "__main__":
|
| 196 |
demo.launch()
|