utils/math_utils.py

import re

def sanitize_latex_for_sympy(latex_str: str) -> str:
    """
    מסיר פקודות עיצוב של LaTeX (כמו צבעים) כדי ש-SymPy יוכל לפענח את הביטוי.
    לדוגמה: הופך את \color{red}{5} ל- 5.
    """
    if not latex_str:
        return latex_str
        
    # שלב 1: מחלץ את התוכן מתוך פקודות \color{...}{content}
    cleaned = re.sub(r'\\color\{.*?\}\{(.*?)\}', r'\1', latex_str)
    
    # שלב 2: מוחק פקודות \color{...} גלמודות שלא עוטפות כלום
    cleaned = re.sub(r'\\color\{.*?\}', '', cleaned)
    
    # שלב 3: מחיקת פקודות \text{} (SymPy שונא טקסט חופשי)
    cleaned = re.sub(r'\\text\{.*?\}', '', cleaned)
    
    # שלב 4: ניקוי סוגריים מסוג \left( \right)
    cleaned = cleaned.replace(r'\left', '').replace(r'\right', '')
    
    return cleaned

from typing import List

def aggressive_sympy_sanitizer(latex_str: str) -> List[str]:
    """
    V9.0.1: Cleans dirty LaTeX generated by the LLM before feeding it to SymPy.
    Splits comma-separated equations into a list of pure math strings.
    """
    if not latex_str:
        return []
    
    s = latex_str
    
    # 1. Strip Hebrew Content including surrounding parentheses (e.g. (לכן C))
    # This prevents leaving behind empty shells like "( C)" which crash SymPy.
    s = re.sub(r'\([^)]*[\u0590-\u05FF]+[^)]*\)', '', s)
    s = re.sub(r'[\u0590-\u05FF]+', '', s)
    
    # 2. Strip formatting tags (\text{} and \color{})
    s = re.sub(r'\\text\{.*?\}', '', s)
    s = re.sub(r'\\color\{.*?\}', '', s)
    
    # 3. Replace Logical Arrows with Equals (to maintain equation structure)
    arrows = [r'\Rightarrow', r'\implies', r'\iff', r'\rightarrow']
    for arrow in arrows:
        s = s.replace(arrow, '=')

    # --- 🚀 BEGIN HOTFIX V9.0.3: LATEX MULTIPLICATION ---
    s = s.replace(r'\cdot', '*').replace(r'\times', '*')
    # --- END HOTFIX ---
        
    # 4. Split multiple equations
    # SymPy cannot parse "x=0, y=9" as a single AST node. 
    # We must split by comma and return a list of clean equations.
    # We also run the existing sanitize_latex_for_sympy on each part.
    expressions = [sanitize_latex_for_sympy(expr.strip()) for expr in s.split(',')]
    
    # Filter out empty strings that might have resulted from the split/regex
    return [e for e in expressions if e]