pyWebLayout/tests/test_line_splitting_bug.py

#!/usr/bin/env python3
"""
Test to demonstrate and verify fix for the line splitting bug where 
text is lost at line breaks due to improper hyphenation handling.
"""

import unittest
from unittest.mock import patch, Mock
from pyWebLayout.concrete.text import Line
from pyWebLayout.abstract.inline import Word
from pyWebLayout.style import Font
from pyWebLayout.style.layout import Alignment


class TestLineSplittingBug(unittest.TestCase):
    """Test cases for the line splitting bug"""
    
    def setUp(self):
        """Set up test fixtures"""
        self.font = Font(
            font_path=None,
            font_size=12,
            colour=(0, 0, 0)
        )
        self.spacing = (5, 10)
        self.origin = (0, 0)
        self.size = (100, 20)  # Narrow line to force hyphenation
    
    @patch('pyWebLayout.abstract.inline.pyphen')
    def test_hyphenation_preserves_word_boundaries(self, mock_pyphen_module):
        """Test that hyphenation properly preserves word boundaries"""
        # Mock pyphen to return a multi-part hyphenated word
        mock_dic = Mock()
        mock_pyphen_module.Pyphen.return_value = mock_dic
        
        # Simulate hyphenating "supercalifragilisticexpialidocious"
        # into multiple parts: "super-", "cali-", "fragi-", "listic-", "expiali-", "docious"
        mock_dic.inserted.return_value = "super-cali-fragi-listic-expiali-docious"
        
        line = Line(self.spacing, self.origin, self.size, self.font)
        
        # Add the word that will be hyphenated
        overflow = line.add_word("supercalifragilisticexpialidocious")
        
        # The overflow should be the next part only, not all remaining parts joined
        # In the current buggy implementation, this would return "cali-fragi-listic-expiali-docious"
        # But it should return "cali-" (the next single part)
        print(f"Overflow returned: '{overflow}'")
        
        # Check that the first part was added to the line
        self.assertEqual(len(line.renderable_words), 1)
        first_word_text = line.renderable_words[0].word.text
        self.assertEqual(first_word_text, "super-")
        
        # The overflow should be just the next part, not all parts joined
        # This assertion will fail with the current bug, showing the issue
        self.assertEqual(overflow, "cali-")  # Should be next part only
        
        # NOT this (which is what the bug produces):
        # self.assertEqual(overflow, "cali-fragi-listic-expiali-docious")
    
    @patch('pyWebLayout.abstract.inline.pyphen')
    def test_single_word_overflow_behavior(self, mock_pyphen_module):
        """Test that overflow returns only the next part, not all remaining parts joined"""
        # Mock pyphen to return a simple two-part hyphenated word
        mock_dic = Mock()
        mock_pyphen_module.Pyphen.return_value = mock_dic
        mock_dic.inserted.return_value = "very-long"
        
        # Create a narrow line that will force hyphenation
        line = Line(self.spacing, (0, 0), (40, 20), self.font)
        
        # Add the word that will be hyphenated
        overflow = line.add_word("verylong")
        
        # Check that the first part was added to the line
        self.assertEqual(len(line.renderable_words), 1)
        first_word_text = line.renderable_words[0].word.text
        self.assertEqual(first_word_text, "very-")
        
        # The overflow should be just the next part ("long"), not multiple parts joined
        # This tests the core fix for the line splitting bug
        self.assertEqual(overflow, "long")
        
        print(f"First part in line: '{first_word_text}'")
        print(f"Overflow returned: '{overflow}'")
    
    def test_simple_overflow_case(self):
        """Test a simple word overflow without hyphenation to verify baseline behavior"""
        line = Line(self.spacing, self.origin, (50, 20), self.font)
        
        # Add a word that fits
        result1 = line.add_word("short")
        self.assertIsNone(result1)
        
        # Add a word that doesn't fit (should overflow)
        result2 = line.add_word("verylongword")
        self.assertEqual(result2, "verylongword")
        
        # Only the first word should be in the line
        self.assertEqual(len(line.renderable_words), 1)
        self.assertEqual(line.renderable_words[0].word.text, "short")


def demonstrate_bug():
    """Demonstrate the bug with a practical example"""
    print("=" * 60)
    print("DEMONSTRATING LINE SPLITTING BUG")
    print("=" * 60)
    
    font = Font(font_path=None, font_size=12, colour=(0, 0, 0))
    
    # Create a very narrow line that will force hyphenation
    line = Line((3, 6), (0, 0), (80, 20), font)
    
    # Try to add a long word that should be hyphenated
    with patch('pyWebLayout.abstract.inline.pyphen') as mock_pyphen_module:
        mock_dic = Mock()
        mock_pyphen_module.Pyphen.return_value = mock_dic
        mock_dic.inserted.return_value = "hyper-long-example-word"
        
        overflow = line.add_word("hyperlongexampleword")
        
        print(f"Original word: 'hyperlongexampleword'")
        print(f"Hyphenated to: 'hyper-long-example-word'")
        print(f"First part added to line: '{line.renderable_words[0].word.text if line.renderable_words else 'None'}'")
        print(f"Overflow returned: '{overflow}'")
        print()
        print("PROBLEM: The overflow should be 'long-' (next part only)")
        print("but instead it returns 'long-example-word' (all remaining parts joined)")
        print("This causes word boundary information to be lost!")


if __name__ == "__main__":
    # First demonstrate the bug
    demonstrate_bug()
    
    print("\n" + "=" * 60)
    print("RUNNING UNIT TESTS")
    print("=" * 60)
    
    # Run unit tests
    unittest.main()