|
| 1 | +import warnings |
1 | 2 | from unittest.mock import MagicMock |
2 | 3 |
|
| 4 | +import pytest |
| 5 | + |
3 | 6 | import osw.wiki_tools as wt |
4 | 7 |
|
5 | 8 |
|
@@ -29,6 +32,12 @@ def _ask_result(*titles): |
29 | 32 | } |
30 | 33 |
|
31 | 34 |
|
| 35 | +def _ask_result_empty(): |
| 36 | + """Build the SMW ``ask`` API result dict for a zero-result query, mirroring |
| 37 | + SMW's behaviour of serialising an empty result set as a JSON array.""" |
| 38 | + return {"query": {"results": []}} |
| 39 | + |
| 40 | + |
32 | 41 | def test_semantic_search_return_json_single_query_returns_list_with_dict(): |
33 | 42 | result = _ask_result("Item:OSW1") |
34 | 43 | site = MagicMock() |
@@ -78,6 +87,126 @@ def test_semantic_search_returns_flat_list_of_titles(): |
78 | 87 | assert out == ["Item:OSW1", "Item:OSW2"] |
79 | 88 |
|
80 | 89 |
|
| 90 | +def test_semantic_search_zero_results_returns_empty_list(): |
| 91 | + site = MagicMock() |
| 92 | + site.api.return_value = _ask_result_empty() |
| 93 | + |
| 94 | + out = wt.semantic_search(site, "[[HasType::Category:Nonexistent]]") |
| 95 | + |
| 96 | + assert out == [] |
| 97 | + |
| 98 | + |
| 99 | +def test_semantic_search_zero_results_return_json_returns_list_with_dict(): |
| 100 | + result = _ask_result_empty() |
| 101 | + site = MagicMock() |
| 102 | + site.api.return_value = result |
| 103 | + |
| 104 | + out = wt.semantic_search( |
| 105 | + site, |
| 106 | + wt.SearchParam(query="[[HasType::Category:Nonexistent]]", return_json=True), |
| 107 | + ) |
| 108 | + |
| 109 | + assert out == [result] |
| 110 | + |
| 111 | + |
| 112 | +def test_semantic_search_batch_with_one_zero_result_query(): |
| 113 | + result_a = _ask_result("Item:OSW1") |
| 114 | + result_b = _ask_result_empty() |
| 115 | + result_c = _ask_result("Item:OSW3") |
| 116 | + site = MagicMock() |
| 117 | + site.api.side_effect = [result_a, result_b, result_c] |
| 118 | + |
| 119 | + out = wt.semantic_search( |
| 120 | + site, |
| 121 | + wt.SearchParam( |
| 122 | + query=[ |
| 123 | + "[[HasType::Category:A]]", |
| 124 | + "[[HasType::Category:B]]", |
| 125 | + "[[HasType::Category:C]]", |
| 126 | + ] |
| 127 | + ), |
| 128 | + ) |
| 129 | + |
| 130 | + assert out == ["Item:OSW1", "Item:OSW3"] |
| 131 | + |
| 132 | + |
| 133 | +def test_semantic_search_parallel_batch_with_one_zero_result_query(): |
| 134 | + results = [ |
| 135 | + _ask_result("Item:OSW1"), |
| 136 | + _ask_result("Item:OSW2"), |
| 137 | + _ask_result_empty(), |
| 138 | + _ask_result("Item:OSW4"), |
| 139 | + _ask_result("Item:OSW5"), |
| 140 | + _ask_result("Item:OSW6"), |
| 141 | + ] |
| 142 | + site = MagicMock() |
| 143 | + site.api.side_effect = results |
| 144 | + |
| 145 | + out = wt.semantic_search( |
| 146 | + site, |
| 147 | + wt.SearchParam( |
| 148 | + query=[ |
| 149 | + "[[HasType::Category:A]]", |
| 150 | + "[[HasType::Category:B]]", |
| 151 | + "[[HasType::Category:C]]", |
| 152 | + "[[HasType::Category:D]]", |
| 153 | + "[[HasType::Category:E]]", |
| 154 | + "[[HasType::Category:F]]", |
| 155 | + ] |
| 156 | + ), |
| 157 | + ) |
| 158 | + |
| 159 | + assert sorted(out) == [ |
| 160 | + "Item:OSW1", |
| 161 | + "Item:OSW2", |
| 162 | + "Item:OSW4", |
| 163 | + "Item:OSW5", |
| 164 | + "Item:OSW6", |
| 165 | + ] |
| 166 | + |
| 167 | + |
| 168 | +def test_semantic_search_truncation_warning(): |
| 169 | + titles = [f"Item:OSW{i}" for i in range(5)] |
| 170 | + result = _ask_result(*titles) |
| 171 | + site = MagicMock() |
| 172 | + site.api.return_value = result |
| 173 | + |
| 174 | + with pytest.warns(UserWarning, match="truncated"): |
| 175 | + out = wt.semantic_search( |
| 176 | + site, wt.SearchParam(query="[[HasType::Category:Item]]", limit=5) |
| 177 | + ) |
| 178 | + |
| 179 | + assert sorted(out) == sorted(titles) |
| 180 | + |
| 181 | + |
| 182 | +def test_semantic_search_no_truncation_warning_below_limit(): |
| 183 | + titles = [f"Item:OSW{i}" for i in range(5)] |
| 184 | + result = _ask_result(*titles) |
| 185 | + site = MagicMock() |
| 186 | + site.api.return_value = result |
| 187 | + |
| 188 | + with warnings.catch_warnings(record=True) as caught: |
| 189 | + warnings.simplefilter("always") |
| 190 | + out = wt.semantic_search( |
| 191 | + site, wt.SearchParam(query="[[HasType::Category:Item]]", limit=1000) |
| 192 | + ) |
| 193 | + |
| 194 | + assert not any("truncated" in str(w.message) for w in caught) |
| 195 | + assert sorted(out) == sorted(titles) |
| 196 | + |
| 197 | + |
| 198 | +def test_semantic_search_exists_drop_warning(): |
| 199 | + result = _ask_result("Item:OSW1", "Item:OSW2") |
| 200 | + result["query"]["results"]["Item:OSW2"]["exists"] = "" |
| 201 | + site = MagicMock() |
| 202 | + site.api.return_value = result |
| 203 | + |
| 204 | + with pytest.warns(UserWarning, match="non-existing"): |
| 205 | + out = wt.semantic_search(site, "[[HasType::Category:Item]]") |
| 206 | + |
| 207 | + assert out == ["Item:OSW1"] |
| 208 | + |
| 209 | + |
81 | 210 | def _prefixsearch_result(*titles): |
82 | 211 | """Build a minimal MediaWiki ``prefixsearch`` API result dict.""" |
83 | 212 | return { |
|
0 commit comments