test_rsa_validator.py
1 """RSA validator テスト 2 3 _rsa_validator.pyの純粋な検証ロジックをテストする。 4 DB/API不要でテスト可能。 5 """ 6 7 from __future__ import annotations 8 9 import pytest 10 11 from mureo.google_ads._rsa_validator import ( 12 DESCRIPTION_MAX_WIDTH, 13 HEADLINE_MAX_WIDTH, 14 AdStrengthResult, 15 RSAValidationResult, 16 _bigram_similarity, 17 _check_headline_diversity, 18 _check_keyword_relevance, 19 _check_prohibited, 20 _has_synonym_overlap, 21 _sanitize_text, 22 _strip_match_type, 23 display_width, 24 predict_ad_strength, 25 validate_rsa_texts, 26 ) 27 28 29 # --------------------------------------------------------------------------- 30 # display_width 31 # --------------------------------------------------------------------------- 32 33 34 @pytest.mark.unit 35 class TestDisplayWidth: 36 def test_半角英数(self) -> None: 37 assert display_width("abc") == 3 38 39 def test_全角文字(self) -> None: 40 assert display_width("テスト") == 6 41 42 def test_混在(self) -> None: 43 assert display_width("ABCテスト") == 9 44 45 def test_空文字(self) -> None: 46 assert display_width("") == 0 47 48 def test_見出し上限(self) -> None: 49 assert HEADLINE_MAX_WIDTH == 30 50 51 def test_説明文上限(self) -> None: 52 assert DESCRIPTION_MAX_WIDTH == 90 53 54 55 # --------------------------------------------------------------------------- 56 # _sanitize_text 57 # --------------------------------------------------------------------------- 58 59 60 @pytest.mark.unit 61 class TestSanitizeText: 62 def test_連続感嘆符の縮約(self) -> None: 63 text, fixes = _sanitize_text("素晴らしい！！！") 64 assert text == "素晴らしい！" 65 assert any("punctuation" in f.lower() for f in fixes) 66 67 def test_連続疑問符の縮約(self) -> None: 68 text, fixes = _sanitize_text("本当？？？") 69 assert text == "本当？" 70 71 def test_装飾記号の縮約(self) -> None: 72 text, fixes = _sanitize_text("★★★限定") 73 assert text == "★限定" 74 assert any("decorative symbol" in f.lower() for f in fixes) 75 76 def test_全角スペースの正規化(self) -> None: 77 text, fixes = _sanitize_text("商品\u3000\u3000紹介") 78 assert text == "商品 紹介" 79 assert any("full-width space" in f.lower() for f in fixes) 80 81 def test_先頭末尾記号の除去(self) -> None: 82 text, fixes = _sanitize_text("！見出し。") 83 assert text == "見出し" 84 assert any("leading" in f.lower() for f in fixes) 85 86 def test_半角カタカナの変換(self) -> None: 87 text, fixes = _sanitize_text("ｷｰﾜｰﾄﾞ") 88 assert text == "キーワード" 89 assert any("half-width katakana" in f.lower() for f in fixes) 90 91 def test_絵文字の除去(self) -> None: 92 text, fixes = _sanitize_text("限定セール\U0001f525") 93 assert text == "限定セール" 94 assert any("emoji" in f.lower() for f in fixes) 95 96 def test_正常テキストは修正なし(self) -> None: 97 text, fixes = _sanitize_text("正常な広告テキスト") 98 assert text == "正常な広告テキスト" 99 assert fixes == [] 100 101 def test_超長テキストの切り詰め(self) -> None: 102 long_text = "あ" * 300 103 text, fixes = _sanitize_text(long_text) 104 assert len(text) <= 200 105 assert any("truncated" in f.lower() for f in fixes) 106 107 108 # --------------------------------------------------------------------------- 109 # _check_prohibited 110 # --------------------------------------------------------------------------- 111 112 113 @pytest.mark.unit 114 class TestCheckProhibited: 115 def test_最上級表現(self) -> None: 116 warnings = _check_prohibited("世界一の品質") 117 assert any("世界一" in w for w in warnings) 118 119 def test_価格系表現(self) -> None: 120 warnings = _check_prohibited("最安値で提供") 121 assert any("最安値" in w for w in warnings) 122 123 def test_効果保証表現(self) -> None: 124 warnings = _check_prohibited("効果保証付き") 125 assert any("効果保証" in w for w in warnings) 126 127 def test_医療系表現(self) -> None: 128 warnings = _check_prohibited("これで治る") 129 assert any("治る" in w for w in warnings) 130 131 def test_クリックベイト(self) -> None: 132 warnings = _check_prohibited("こちらをクリック") 133 assert any("clickbait" in w.lower() for w in warnings) 134 135 def test_正常テキスト(self) -> None: 136 warnings = _check_prohibited("高品質な商品をお届け") 137 assert warnings == [] 138 139 140 # --------------------------------------------------------------------------- 141 # validate_rsa_texts 142 # --------------------------------------------------------------------------- 143 144 145 @pytest.mark.unit 146 class TestValidateRsaTexts: 147 def test_正常なRSAテキスト(self) -> None: 148 result = validate_rsa_texts( 149 headlines=["見出し1", "見出し2", "見出し3"], 150 descriptions=["説明文1です", "説明文2です"], 151 final_url="https://example.com", 152 ) 153 154 assert isinstance(result, RSAValidationResult) 155 assert len(result.headlines) == 3 156 assert len(result.descriptions) == 2 157 158 def test_URLなしでValueError(self) -> None: 159 with pytest.raises(ValueError, match="final_url"): 160 validate_rsa_texts( 161 headlines=["見出し"], 162 descriptions=["説明文"], 163 final_url="", 164 ) 165 166 def test_不正URLでValueError(self) -> None: 167 with pytest.raises(ValueError, match="Invalid URL"): 168 validate_rsa_texts( 169 headlines=["見出し"], 170 descriptions=["説明文"], 171 final_url="invalid-url", 172 ) 173 174 def test_重複見出しの除去(self) -> None: 175 result = validate_rsa_texts( 176 headlines=["同じ見出し", "同じ見出し", "別の見出し"], 177 descriptions=["説明文"], 178 final_url="https://example.com", 179 ) 180 181 assert len(result.headlines) == 2 182 assert any("duplicate" in w for w in result.warnings) 183 184 def test_文字幅超過でValueError(self) -> None: 185 long_headline = "あ" * 20 # 全角20文字 = 幅40 > 30 186 with pytest.raises(ValueError, match="character limit"): 187 validate_rsa_texts( 188 headlines=[long_headline], 189 descriptions=["説明文"], 190 final_url="https://example.com", 191 ) 192 193 def test_禁止表現で警告(self) -> None: 194 result = validate_rsa_texts( 195 headlines=["世界一の品質"], 196 descriptions=["説明文です"], 197 final_url="https://example.com", 198 ) 199 200 assert any("世界一" in w for w in result.warnings) 201 202 def test_サニタイズ適用(self) -> None: 203 result = validate_rsa_texts( 204 headlines=["素晴らしい！！！"], 205 descriptions=["説明文です"], 206 final_url="https://example.com", 207 ) 208 209 assert result.headlines[0] == "素晴らしい！" 210 211 212 # --------------------------------------------------------------------------- 213 # _bigram_similarity 214 # --------------------------------------------------------------------------- 215 216 217 @pytest.mark.unit 218 class TestBigramSimilarity: 219 def test_同一文字列(self) -> None: 220 assert _bigram_similarity("テスト", "テスト") == 1.0 221 222 def test_完全に異なる文字列(self) -> None: 223 sim = _bigram_similarity("あいう", "かきく") 224 assert sim == 0.0 225 226 def test_短い文字列(self) -> None: 227 assert _bigram_similarity("a", "a") == 1.0 228 assert _bigram_similarity("a", "b") == 0.0 229 230 231 # --------------------------------------------------------------------------- 232 # _has_synonym_overlap 233 # --------------------------------------------------------------------------- 234 235 236 @pytest.mark.unit 237 class TestHasSynonymOverlap: 238 def test_同義語ペア検出(self) -> None: 239 assert _has_synonym_overlap("安い商品", "格安セール") is True 240 241 def test_同義語なし(self) -> None: 242 assert _has_synonym_overlap("高品質", "高速配送") is False 243 244 245 # --------------------------------------------------------------------------- 246 # _check_headline_diversity 247 # --------------------------------------------------------------------------- 248 249 250 @pytest.mark.unit 251 class TestCheckHeadlineDiversity: 252 def test_多様な見出し(self) -> None: 253 score, msgs = _check_headline_diversity(["商品紹介", "お客様の声", "無料体験"]) 254 assert score > 0.5 255 assert msgs == [] 256 257 def test_単一見出し(self) -> None: 258 score, msgs = _check_headline_diversity(["テスト"]) 259 assert score == 1.0 260 261 def test_同義語含む見出し(self) -> None: 262 score, msgs = _check_headline_diversity(["安い商品", "格安セール"]) 263 assert score < 1.0 264 assert len(msgs) > 0 265 266 267 # --------------------------------------------------------------------------- 268 # _check_keyword_relevance 269 # --------------------------------------------------------------------------- 270 271 272 @pytest.mark.unit 273 class TestCheckKeywordRelevance: 274 def test_全キーワード含有(self) -> None: 275 score, missing = _check_keyword_relevance( 276 ["テスト商品の紹介"], 277 ["テストの説明文"], 278 ["テスト", "商品"], 279 ) 280 assert score == 1.0 281 assert missing == [] 282 283 def test_一部未含有(self) -> None: 284 score, missing = _check_keyword_relevance( 285 ["テスト"], 286 ["説明文"], 287 ["テスト", "未知のKW"], 288 ) 289 assert score == 0.5 290 assert "未知のKW" in missing 291 292 def test_キーワードなし(self) -> None: 293 score, missing = _check_keyword_relevance( 294 ["テスト"], 295 ["説明文"], 296 [], 297 ) 298 assert score == 0.5 299 300 301 # --------------------------------------------------------------------------- 302 # _strip_match_type 303 # --------------------------------------------------------------------------- 304 305 306 @pytest.mark.unit 307 class TestStripMatchType: 308 def test_フレーズマッチ(self) -> None: 309 assert _strip_match_type('"テスト"') == "テスト" 310 311 def test_完全一致(self) -> None: 312 assert _strip_match_type("[テスト]") == "テスト" 313 314 def test_絞り込み部分一致(self) -> None: 315 assert _strip_match_type("+テスト") == "テスト" 316 317 def test_通常キーワード(self) -> None: 318 assert _strip_match_type("テスト") == "テスト" 319 320 321 # --------------------------------------------------------------------------- 322 # predict_ad_strength 323 # --------------------------------------------------------------------------- 324 325 326 @pytest.mark.unit 327 class TestPredictAdStrength: 328 def test_最小構成(self) -> None: 329 result = predict_ad_strength( 330 headlines=["見出し1", "見出し2", "見出し3"], 331 descriptions=["説明文1", "説明文2"], 332 ) 333 334 assert isinstance(result, AdStrengthResult) 335 assert result.level in ("POOR", "AVERAGE", "GOOD", "EXCELLENT") 336 assert 0.0 <= result.score <= 1.0 337 assert len(result.factors) == 6 338 339 def test_最大構成でEXCELLENT(self) -> None: 340 result = predict_ad_strength( 341 headlines=[f"見出し{i}" for i in range(15)], 342 descriptions=[f"説明文{i}" for i in range(4)], 343 keywords=["見出し0", "見出し1"], 344 has_sitelinks=True, 345 pinned_count=0, 346 ) 347 348 assert result.level in ("GOOD", "EXCELLENT") 349 assert result.score >= 0.65 350 351 def test_ピン留めでペナルティ(self) -> None: 352 base = predict_ad_strength( 353 headlines=["見出し1", "見出し2", "見出し3"], 354 descriptions=["説明文1", "説明文2"], 355 pinned_count=0, 356 ) 357 pinned = predict_ad_strength( 358 headlines=["見出し1", "見出し2", "見出し3"], 359 descriptions=["説明文1", "説明文2"], 360 pinned_count=3, 361 ) 362 363 assert pinned.score < base.score 364 365 def test_サジェスチョン生成(self) -> None: 366 result = predict_ad_strength( 367 headlines=["見出し1"], 368 descriptions=["説明文1"], 369 ) 370 371 assert len(result.suggestions) > 0