Coverage for src/wiktextract/extractor/ku/models.py: 100%

103 statements  

« prev     ^ index     » next       coverage.py v7.10.3, created at 2025-08-15 05:18 +0000

1from pydantic import BaseModel, ConfigDict, Field 

2 

3 

4class KurdishBaseModel(BaseModel): 

5 model_config = ConfigDict( 

6 extra="forbid", 

7 strict=True, 

8 validate_assignment=True, 

9 validate_default=True, 

10 ) 

11 

12 

13class Sound(KurdishBaseModel): 

14 ipa: str = "" 

15 audio: str = Field(default="", description="Audio file name") 

16 wav_url: str = "" 

17 oga_url: str = "" 

18 ogg_url: str = "" 

19 mp3_url: str = "" 

20 opus_url: str = "" 

21 flac_url: str = "" 

22 tags: list[str] = [] 

23 raw_tags: list[str] = [] 

24 

25 

26class Example(KurdishBaseModel): 

27 text: str 

28 bold_text_offsets: list[tuple[int, int]] = [] 

29 translation: str = "" 

30 bold_translation_offsets: list[tuple[int, int]] = [] 

31 roman: str = Field( 

32 default="", description="Romanization of the example sentence" 

33 ) 

34 bold_roman_offsets: list[tuple[int, int]] = [] 

35 ref: str = Field( 

36 default="", 

37 description="Source of the sentence, like book title and page number", 

38 ) 

39 tags: list[str] = [] 

40 raw_tags: list[str] = [] 

41 sounds: list[Sound] = [] 

42 categories: list[str] = Field(default=[], exclude=True) 

43 

44 

45class AltForm(KurdishBaseModel): 

46 word: str 

47 

48 

49class Sense(KurdishBaseModel): 

50 glosses: list[str] = [] 

51 tags: list[str] = [] 

52 raw_tags: list[str] = [] 

53 categories: list[str] = [] 

54 examples: list[Example] = [] 

55 topics: list[str] = [] 

56 form_of: list[AltForm] = [] 

57 alt_of: list[AltForm] = [] 

58 

59 

60class Form(KurdishBaseModel): 

61 form: str 

62 tags: list[str] = [] 

63 raw_tags: list[str] = [] 

64 roman: str = "" 

65 translation: str = "" 

66 sense: str = "" 

67 source: str = "" 

68 

69 

70class Translation(KurdishBaseModel): 

71 lang_code: str = Field( 

72 description="Wiktionary language code of the translation term", 

73 ) 

74 lang: str = Field(description="Translation language name") 

75 word: str = Field(description="Translation term") 

76 sense: str = Field(default="", description="Translation gloss") 

77 sense_index: int = Field( 

78 default=0, ge=0, description="Number of the definition, start from 1" 

79 ) 

80 tags: list[str] = [] 

81 raw_tags: list[str] = [] 

82 roman: str = "" 

83 source: str = "" 

84 

85 

86class Linkage(KurdishBaseModel): 

87 word: str 

88 tags: list[str] = [] 

89 raw_tags: list[str] = [] 

90 roman: str = "" 

91 translation: str = "" 

92 sense: str = "" 

93 

94 

95class Descendant(KurdishBaseModel): 

96 lang_code: str = Field(description="Wiktionary language code") 

97 lang: str = Field(description="Language name") 

98 word: str 

99 roman: str = "" 

100 tags: list[str] = [] 

101 raw_tags: list[str] = [] 

102 descendants: list["Descendant"] = [] 

103 sense: str = "" 

104 

105 

106class Hyphenation(KurdishBaseModel): 

107 parts: list[str] = [] 

108 tags: list[str] = [] 

109 raw_tags: list[str] = [] 

110 

111 

112class WordEntry(KurdishBaseModel): 

113 model_config = ConfigDict(title="Kurdish Wiktionary") 

114 word: str = Field(description="Word string") 

115 lang_code: str = Field(description="Wiktionary language code") 

116 lang: str = Field(description="Localized language name") 

117 pos: str = Field(description="Part of speech type") 

118 pos_title: str = "" 

119 senses: list[Sense] = [] 

120 categories: list[str] = [] 

121 tags: list[str] = [] 

122 raw_tags: list[str] = [] 

123 forms: list[Form] = [] 

124 etymology_text: str = "" 

125 translations: list[Translation] = [] 

126 synonyms: list[Linkage] = [] 

127 antonyms: list[Linkage] = [] 

128 derived: list[Linkage] = [] 

129 related: list[Linkage] = [] 

130 hypernyms: list[Linkage] = [] 

131 hyponyms: list[Linkage] = [] 

132 anagrams: list[Linkage] = [] 

133 rhymes: list[Linkage] = [] 

134 sounds: list[Sound] = [] 

135 hyphenations: list[Hyphenation] = [] 

136 notes: list[str] = [] 

137 descendants: list[Descendant] = [] 

138 abbreviations: list[Linkage] = [] 

139 coordinate_terms: list[Linkage] = []