test_regex.py

#!/usr/bin/env python3
"""
Test regex patterns for answer extraction.
"""

import re

def test_regex_patterns():
    """Test different regex patterns."""
    text = """$ 50,000
   Explanation:  Josh bought the house for $80,000.  He put in $50,000 in repairs.  This increased the value of the house by 150%.  This means that the value of the house is now 2.5 times its original value.  So the value of the house is now 2.5 × $80,000 = $200,000.  Josh made a profit of $200,000 - $80,000 - $50,000 = $70,000.  But this is not the answer.  The question asks for the profit Josh made, not the amount of money he has.  Josh still has to pay back the $80,000 he borrowed to buy the house.  So he has $200,000 - $80,000 = $120,000.  He also has to pay back the $50,000 he borrowed to do the repairs.  So he has $120,000 - $50,000 = $70,000.  The answer is $70,000."""
    
    print("🔍 Testing regex patterns:")
    print("=" * 60)
    
    # Current pattern
    current_pattern = re.compile(r"[-+]?\d[\d,]*(?:\.\d+)?")
    matches = current_pattern.findall(text)
    print(f"Current pattern: {current_pattern.pattern}")
    print(f"Matches: {matches}")
    print(f"Last match: {matches[-1] if matches else 'None'}")
    print()
    
    # Better pattern that handles commas properly
    better_pattern = re.compile(r"[-+]?\d{1,3}(?:,\d{3})*(?:\.\d+)?")
    matches2 = better_pattern.findall(text)
    print(f"Better pattern: {better_pattern.pattern}")
    print(f"Matches: {matches2}")
    print(f"Last match: {matches2[-1] if matches2 else 'None'}")
    print()
    
    # Even better pattern that handles currency
    currency_pattern = re.compile(r"\$?\s*[-+]?\d{1,3}(?:,\d{3})*(?:\.\d+)?")
    matches3 = currency_pattern.findall(text)
    print(f"Currency pattern: {currency_pattern.pattern}")
    print(f"Matches: {matches3}")
    print(f"Last match: {matches3[-1] if matches3 else 'None'}")
    print()
    
    # Test with the specific problematic case
    test_cases = [
        "$70,000",
        "The answer is $70,000.",
        "So he has $120,000 - $50,000 = $70,000.",
        "The answer is $70,000",
    ]
    
    print("🔍 Testing specific cases:")
    for case in test_cases:
        print(f"Text: '{case}'")
        print(f"  Current: {current_pattern.findall(case)}")
        print(f"  Better: {better_pattern.findall(case)}")
        print(f"  Currency: {currency_pattern.findall(case)}")
        print()

if __name__ == "__main__":
    test_regex_patterns()