|
| 1 | +# docker-compose -f local.yml run --rm django pytest sde_collections/tests/test_title_resolution.py |
| 2 | + |
| 3 | +from unittest.mock import Mock, patch |
| 4 | + |
| 5 | +import pytest |
| 6 | + |
| 7 | +from ..utils.title_resolver import ( |
| 8 | + clean_text, |
| 9 | + is_valid_xpath, |
| 10 | + parse_title, |
| 11 | + resolve_brace, |
| 12 | + resolve_title, |
| 13 | + resolve_xpath, |
| 14 | + validate_fstring, |
| 15 | +) |
| 16 | + |
| 17 | + |
| 18 | +def test_parse_title(): |
| 19 | + # Test basic string |
| 20 | + assert parse_title("Simple Title") == [("str", "Simple Title")] |
| 21 | + |
| 22 | + # Test f-string |
| 23 | + assert parse_title("Hello {title}") == [("str", "Hello "), ("brace", "{title}")] |
| 24 | + |
| 25 | + # Test xpath |
| 26 | + assert parse_title("xpath://h1") == [("xpath", "//h1")] |
| 27 | + |
| 28 | + # Test complex pattern |
| 29 | + result = parse_title("xpath://h1 | {title} - {collection}") |
| 30 | + assert result == [ |
| 31 | + ("xpath", "//h1"), |
| 32 | + ("str", " | "), |
| 33 | + ("brace", "{title}"), |
| 34 | + ("str", " - "), |
| 35 | + ("brace", "{collection}"), |
| 36 | + ] |
| 37 | + |
| 38 | + |
| 39 | +def test_is_valid_xpath(): |
| 40 | + assert is_valid_xpath("//h1") is True |
| 41 | + assert is_valid_xpath("//div[@class='title']") is True |
| 42 | + assert is_valid_xpath("invalid xpath") is False |
| 43 | + assert is_valid_xpath("//h1[") is False |
| 44 | + |
| 45 | + |
| 46 | +def test_validate_fstring(): |
| 47 | + # Valid cases - should not raise |
| 48 | + validate_fstring("{title}") |
| 49 | + validate_fstring("{url}") |
| 50 | + validate_fstring("{collection}") |
| 51 | + |
| 52 | + # Invalid cases |
| 53 | + with pytest.raises(ValueError): |
| 54 | + validate_fstring("{invalid_var}") |
| 55 | + with pytest.raises(ValueError): |
| 56 | + validate_fstring("{title.upper()}") |
| 57 | + with pytest.raises(ValueError): |
| 58 | + validate_fstring("{len(title)}") |
| 59 | + |
| 60 | + |
| 61 | +def test_resolve_brace(): |
| 62 | + context = {"title": "Test Title", "url": "https://example.com", "collection": "Test Collection"} |
| 63 | + |
| 64 | + assert resolve_brace("{title}", context) == "Test Title" |
| 65 | + assert resolve_brace("{title} - {collection}", context) == "Test Title - Test Collection" |
| 66 | + |
| 67 | + with pytest.raises(ValueError): |
| 68 | + resolve_brace("{invalid}", context) |
| 69 | + |
| 70 | + |
| 71 | +def test_clean_text(): |
| 72 | + # Test whitespace handling |
| 73 | + assert clean_text(" Title \n With\tSpaces ") == "Title With Spaces" |
| 74 | + |
| 75 | + # Test HTML entities |
| 76 | + assert clean_text("Title & More") == "Title & More" |
| 77 | + |
| 78 | + # Test unicode normalization |
| 79 | + assert clean_text("Café") == "Cafe" |
| 80 | + |
| 81 | + |
| 82 | +@patch("requests.get") |
| 83 | +def test_resolve_xpath(mock_get): |
| 84 | + mock_response = Mock() |
| 85 | + mock_response.ok = True |
| 86 | + mock_response.content = b""" |
| 87 | + <html> |
| 88 | + <body> |
| 89 | + <h1>Test Title</h1> |
| 90 | + <div class="content">Inner Content</div> |
| 91 | + </body> |
| 92 | + </html> |
| 93 | + """ |
| 94 | + mock_get.return_value = mock_response |
| 95 | + |
| 96 | + # Test basic xpath |
| 97 | + assert resolve_xpath("//h1", "https://example.com") == "Test Title" |
| 98 | + assert resolve_xpath("//div[@class='content']", "https://example.com") == "Inner Content" |
| 99 | + |
| 100 | + # Test error cases |
| 101 | + mock_response.ok = False |
| 102 | + with pytest.raises(ValueError): |
| 103 | + resolve_xpath("//h1", "https://example.com") |
| 104 | + |
| 105 | + mock_response.ok = True |
| 106 | + with pytest.raises(ValueError): |
| 107 | + resolve_xpath("//nonexistent", "https://example.com") |
| 108 | + |
| 109 | + |
| 110 | +@patch("requests.get") |
| 111 | +def test_resolve_title(mock_get): |
| 112 | + mock_response = Mock() |
| 113 | + mock_response.ok = True |
| 114 | + mock_response.content = b""" |
| 115 | + <html> |
| 116 | + <body> |
| 117 | + <h1>Dynamic Content</h1> |
| 118 | + </body> |
| 119 | + </html> |
| 120 | + """ |
| 121 | + mock_get.return_value = mock_response |
| 122 | + |
| 123 | + context = {"title": "Original Title", "url": "https://example.com", "collection": "Test Collection"} |
| 124 | + |
| 125 | + # Test combination of xpath and f-string |
| 126 | + pattern = "xpath://h1 | {title} - {collection}" |
| 127 | + assert resolve_title(pattern, context) == "Dynamic Content | Original Title - Test Collection" |
| 128 | + |
| 129 | + # Test simple f-string |
| 130 | + assert resolve_title("{title} ({collection})", context) == "Original Title (Test Collection)" |
| 131 | + |
| 132 | + # Test plain string |
| 133 | + assert resolve_title("Static Title", context) == "Static Title" |
0 commit comments