test_legacy_irregular_model.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253
  1. import sys
  2. import unittest
  3. from pathlib import Path
  4. sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
  5. from toolsvc.customs.engine.rules import Rules
  6. from toolsvc.customs.engine.structure import split_attributes
  7. ROOT = Path(__file__).resolve().parents[1] / 'toolsvc' / 'customs'
  8. class IrregularModelTests(unittest.TestCase):
  9. @classmethod
  10. def setUpClass(cls):
  11. cls.rules = Rules(ROOT / 'rules/数据映射规则.xlsx',
  12. ROOT / 'rules/进出口数据字段需求20260819.xlsx')
  13. def test_quoted_semicolon_and_sentence_usage(self):
  14. parsed = self.rules.parse('8532241000',
  15. '"品牌:MURATA;型号:TEST-1;用途:这句话说明用于计算机系统"')
  16. self.assertEqual(parsed['brand'], 'MURATA')
  17. self.assertEqual(parsed['model'], 'TEST-1')
  18. self.assertEqual(parsed['app_raw'], '这句话说明用于计算机系统')
  19. def test_placeholder_slash_between_semicolons(self):
  20. tokens, mode, _ = split_attributes('品牌:MURATA;/;型号:TEST-1')
  21. self.assertEqual(tokens, ['品牌:MURATA', '', '型号:TEST-1'])
  22. parsed = self.rules.parse('8532241000', '品牌:MURATA;/;型号:TEST-1')
  23. self.assertIn('斜杠占位字段', parsed['flags'])
  24. self.assertEqual(parsed['model'], 'TEST-1')
  25. def test_pure_slash_status_fields_and_fraction(self):
  26. text = '4/0/计算机用/MURATA/TEST-1/1/10W'
  27. tokens, mode, _ = split_attributes(text)
  28. self.assertEqual(mode, 'slash_inferred')
  29. self.assertEqual(tokens[-1], '1/10W')
  30. parsed = self.rules.parse('8532241000', text)
  31. self.assertEqual(parsed['brand'], 'MURATA')
  32. self.assertEqual(parsed['model'], 'TEST-1')
  33. self.assertIn('纯斜杠分段待核查', parsed['flags'])
  34. def test_pure_slash_brand_or_no_brand(self):
  35. text = 'MURATA/TEST-1/计算机用'
  36. parsed = self.rules.parse('8532241000', text)
  37. self.assertEqual(parsed['brand'], 'MURATA')
  38. self.assertEqual(parsed['model'], 'TEST-1')
  39. self.assertIn('纯斜杠分段待核查', parsed['flags'])
  40. parsed = self.rules.parse('8532241000', '无品牌/TEST-1/计算机用')
  41. self.assertEqual(parsed['brand'], '')
  42. self.assertEqual(parsed['model'], 'TEST-1')
  43. self.assertIn('无品牌标记', parsed['flags'])
  44. parsed = self.rules.parse('8532241000', '无中文品牌/MURATA/TEST-1')
  45. self.assertEqual(parsed['brand'], 'MURATA')
  46. self.assertEqual(parsed['model'], 'TEST-1')
  47. def test_fraction_or_slash_model_is_not_split(self):
  48. parsed = self.rules.parse('8532241000', '品牌:MURATA;型号:1/10W')
  49. self.assertEqual(parsed['model'], '1/10W')
  50. self.assertNotIn('多型号取首个', parsed['flags'])
  51. parsed = self.rules.parse('8532241000', '品牌:MURATA;型号:TEST-1/TEST-2')
  52. self.assertEqual(parsed['model'], 'TEST-1/TEST-2')
  53. def test_underscore_as_word_gap_without_changing_model_identifier(self):
  54. parsed = self.rules.parse('8532241000', 'MURATA_TEST-1')
  55. self.assertEqual(parsed['brand'], 'MURATA')
  56. self.assertEqual(parsed['model'], 'TEST-1')
  57. def test_repeated_brand_candidates(self):
  58. same = self.rules.parse('8532241000', '品牌:MURATA,MURATA;型号:TEST-1')
  59. self.assertEqual(same['brand'], 'MURATA')
  60. different = self.rules.parse('8532241000', '品牌:MURATA,SAMSUNG;型号:TEST-1')
  61. self.assertEqual(different['brand'], '')
  62. self.assertIn('品牌映射冲突', different['flags'])
  63. def test_chinese_status_with_known_schema_anchor(self):
  64. self.rules.schema['SYNTH_STATUS'] = ['品牌类型', '出口享惠情况', '品牌', '型号']
  65. parsed = self.rules.parse('SYNTH_STATUS', '同意|同意|MURATA|TEST-1')
  66. self.assertEqual(parsed['method'], '顺序提取')
  67. self.assertEqual(parsed['brand'], 'MURATA')
  68. self.assertEqual(parsed['model'], 'TEST-1')
  69. def test_missing_optional_tail_with_brand_anchor(self):
  70. self.rules.schema['SYNTH_TAIL'] = [
  71. '品牌类型', '出口享惠情况', '品牌', '型号', 'GTIN', 'CAS', '其他']
  72. parsed = self.rules.parse('SYNTH_TAIL', '4|0|MURATA|TEST-1')
  73. self.assertEqual(parsed['method'], '顺序提取')
  74. self.assertEqual(parsed['model'], 'TEST-1')
  75. self.assertIn('模板末尾字段缺失待核查', parsed['flags'])
  76. parsed = self.rules.parse('SYNTH_TAIL', '4|0|MURATA|TEST-1||||')
  77. self.assertEqual(parsed['method'], '顺序提取')
  78. def test_numbered_labels_do_not_invent_brand_or_model(self):
  79. parsed = self.rules.parse('8532241000', '3|0、品牌类型:xxx;1、品名:?xxx')
  80. self.assertEqual(parsed['brand'], '')
  81. self.assertEqual(parsed['model'], '')
  82. def test_empty_brand_slot_keeps_model_despite_placeholders(self):
  83. # 品牌位留空、GTIN/CAS 写“无”时,不能因为“无”在别名表里就把型号一起丢掉
  84. parsed = self.rules.parse('8532241000', '0|0|多层片式|瓷介||TEST-1|无|无|无')
  85. self.assertEqual(parsed['method'], '顺序提取')
  86. self.assertEqual(parsed['brand'], '')
  87. self.assertEqual(parsed['model_raw'], 'TEST-1')
  88. self.assertIn('品牌未识别', parsed['flags'])
  89. def test_position_brand_conflict_still_falls_back(self):
  90. # 品牌位是不认识的写法,但后面有已知品牌 → 仍应转缺项识别并取到 MURATA
  91. parsed = self.rules.parse('8532241000', '0|0|多层片式|瓷介|不存在的牌子|TEST-1|MURATA|无|无')
  92. self.assertEqual(parsed['method'], '缺项识别')
  93. self.assertIn('顺序位置与品牌线索冲突,转缺项识别', parsed['flags'])
  94. self.assertEqual(parsed['brand'], 'MURATA')
  95. def test_label_synonyms_lock_brand_and_model(self):
  96. parsed = self.rules.parse('8532241000', '型号规格:TEST-1|品牌名称:MURATA')
  97. self.assertEqual(parsed['brand'], 'MURATA')
  98. self.assertEqual(parsed['model_raw'], 'TEST-1')
  99. self.assertIn('品牌标签锁定', parsed['flags'])
  100. def test_no_model_marker_is_recognized(self):
  101. parsed = self.rules.parse('8532241000', '0|0|多层片式|瓷介|MURATA|无型号|无|无|无')
  102. self.assertEqual(parsed['brand'], 'MURATA')
  103. self.assertEqual(parsed['model_raw'], '')
  104. self.assertIn('无型号标记', parsed['flags'])
  105. self.assertNotIn('型号未识别', parsed['flags'])
  106. def test_no_brand_and_no_model_markers(self):
  107. parsed = self.rules.parse('8532241000', '型号:无;品牌:无中文品牌')
  108. self.assertIn('无品牌标记', parsed['flags'])
  109. self.assertIn('无型号标记', parsed['flags'])
  110. self.assertNotIn('品牌未识别', parsed['flags'])
  111. self.assertNotIn('型号未识别', parsed['flags'])
  112. def test_label_brand_inside_sentence(self):
  113. parsed = self.rules.parse('8532241000', '这是一颗贴片电容,品牌:TDK,用于手机')
  114. self.assertEqual(parsed['brand'], 'TDK')
  115. self.assertIn('品牌标签锁定', parsed['flags'])
  116. def test_multiple_model_labels_take_first(self):
  117. parsed = self.rules.parse('8542321010', '品牌:三星;型号:K4B4G1646D-BCK0;规格型号:DDR4 4GB')
  118. self.assertEqual(parsed['brand'], 'SAMSUNG')
  119. self.assertEqual(parsed['model_raw'], 'K4B4G1646D-BCK0')
  120. self.assertIn('多型号取首个', parsed['flags'])
  121. class LabelVariantTests(unittest.TestCase):
  122. """人眼一眼能读出、但旧逻辑取不到的写法(键值变体 / 位置变体)。"""
  123. @classmethod
  124. def setUpClass(cls):
  125. cls.rules = Rules(ROOT / 'rules/数据映射规则.xlsx',
  126. ROOT / 'rules/进出口数据字段需求20260819.xlsx')
  127. def pair(self, hs, raw):
  128. parsed = self.rules.parse(hs, raw)
  129. return parsed['brand'], parsed['model_raw']
  130. def test_key_and_value_in_separate_fields(self):
  131. self.assertEqual(self.pair('8532241000', '品牌|MURATA|型号|GRM188R71H104KA93D'),
  132. ('MURATA', 'GRM188R71H104KA93D'))
  133. self.assertEqual(self.pair('8542321010', '品牌|三星|型号|K4B4G1646D-BCK0'),
  134. ('SAMSUNG', 'K4B4G1646D-BCK0'))
  135. def test_label_inside_parentheses(self):
  136. self.assertEqual(self.pair('8532241000', '(品牌)MURATA(型号)GRM188R71H104KA93D'),
  137. ('MURATA', 'GRM188R71H104KA93D'))
  138. def test_value_before_label(self):
  139. self.assertEqual(self.pair('8532241000', 'MURATA(品牌)、GRM188R71H104KA93D(型号)'),
  140. ('MURATA', 'GRM188R71H104KA93D'))
  141. def test_connective_and_separator_variants(self):
  142. self.assertEqual(self.pair('8532241000', '品牌为MURATA,型号为GRM188R71H104KA93D'),
  143. ('MURATA', 'GRM188R71H104KA93D'))
  144. self.assertEqual(self.pair('8532241000', '品牌、村田、型号、GRM188R71H104KA93D'),
  145. ('MURATA', 'GRM188R71H104KA93D'))
  146. self.assertEqual(self.pair('8532241000', '型号"GRM188R71H104KA93D" 品牌"村田"'),
  147. ('MURATA', 'GRM188R71H104KA93D'))
  148. def test_model_written_before_brand(self):
  149. self.assertEqual(self.pair('8532241000', 'GRM188R71H104KA93D MURATA'),
  150. ('MURATA', 'GRM188R71H104KA93D'))
  151. self.assertEqual(self.pair('8542321010', 'K4B4G1646D-BCK0 三星'),
  152. ('SAMSUNG', 'K4B4G1646D-BCK0'))
  153. def test_brand_synonyms(self):
  154. self.assertEqual(self.pair('8532241000', '制造商:村田制作所 规格:GRM188R71H104KA93D'),
  155. ('MURATA', 'GRM188R71H104KA93D'))
  156. def test_parameters_are_never_taken_as_model(self):
  157. # 反向保护:包装尺寸、容值、用途不能被当成型号
  158. for raw in ('品牌:MURATA;规格:0402',
  159. '品牌:MURATA;规格:100nF',
  160. '品牌:MURATA;用途:5G基站'):
  161. brand, model = self.pair('8532241000', raw)
  162. self.assertEqual(brand, 'MURATA', raw)
  163. self.assertEqual(model, '', raw)
  164. def test_deng_suffix_is_not_part_of_brand_or_model(self):
  165. # 申报要素把“这个品牌/型号以及同类”写成 XXX等牌 / XXX等型
  166. parsed = self.rules.parse('8532241000', 'MURATA等牌|GRM188R71H104KA93D等型')
  167. self.assertEqual(parsed['brand'], 'MURATA')
  168. self.assertEqual(parsed['model_raw'], 'GRM188R71H104KA93D')
  169. self.assertIn('品牌带“等”后缀', parsed['flags'])
  170. self.assertIn('多型号取首个', parsed['flags'])
  171. parsed = self.rules.parse('8532241000', '4|3|村田等牌|GRM188R71H104KA93D等型')
  172. self.assertEqual(parsed['brand'], 'MURATA')
  173. self.assertEqual(parsed['model_raw'], 'GRM188R71H104KA93D')
  174. def test_missing_brand_is_named_in_flags(self):
  175. # 品牌表里没有 ABACUS,要让用户一眼看出“该补哪一行”
  176. parsed = self.rules.parse('8532241000', '4|3|ABACUS等牌|A5E30295495R2W等型')
  177. self.assertEqual(parsed['brand'], '')
  178. self.assertIn('疑似缺品牌:ABACUS', parsed['flags'])
  179. def test_model_survives_unresolved_brand_field(self):
  180. # 品牌认不出,但“品牌字段后面那一段”明显是型号,不能连型号一起丢
  181. parsed = self.rules.parse('8532241000', '4|3|225mW|PANJIT牌/强茂牌|MMBT3904_R1_000Z')
  182. self.assertEqual(parsed['brand'], '')
  183. self.assertEqual(parsed['model_raw'], 'MMBT3904_R1_000Z')
  184. self.assertIn('品牌字段后型号推断', parsed['flags'])
  185. self.assertIn('疑似缺品牌:PANJIT', parsed['flags'])
  186. def test_catch_all_label_is_not_reached_by_typo(self):
  187. # 规则表里有一行错写 PANJJIT → Others;PANJIT 不能靠拼写近似落到 Others 上
  188. parsed = self.rules.parse('8532241000', '4|3|225mW|PANJIT牌|MMBT3904_R1_000Z')
  189. self.assertNotEqual(parsed['brand'], 'Others')
  190. class ReverseContainmentTests(unittest.TestCase):
  191. """规则表的申报内容是一长串时,G_MODEL 里只写其中一段也要能认。"""
  192. @classmethod
  193. def setUpClass(cls):
  194. cls.rules = Rules(ROOT / 'rules/数据映射规则.xlsx',
  195. ROOT / 'rules/进出口数据字段需求20260819.xlsx')
  196. def test_long_rule_containing_the_name_matches(self):
  197. # 表里有 “村田制作所(MURATA)” 这类长写法,G_MODEL 写“村田制作所”要能对上
  198. self.assertEqual(self.rules.reverse_brand_labels('村田制作所'), frozenset({'MURATA'}))
  199. def test_multi_brand_rule_is_not_used_as_evidence(self):
  200. # “NEXPERIA、DIODES、PANJIT/无中文名称” 这种多品牌混写不能把 PANJIT 认成 Nexperia
  201. self.assertEqual(self.rules.reverse_brand_labels('PANJIT'), frozenset())
  202. parsed = self.rules.parse('8532241000', '4|3|225mW|PANJIT牌/强茂牌|MMBT3904_R1_000Z8')
  203. self.assertNotIn(parsed['brand'], ('Nexperia', 'Onsemi', 'Others'))
  204. def test_unknown_name_stays_unknown(self):
  205. self.assertEqual(self.rules.reverse_brand_labels('ABACUS'), frozenset())
  206. self.assertEqual(self.rules.reverse_brand_labels('MMBT3904'), frozenset())
  207. if __name__ == '__main__':
  208. unittest.main()