Coverage for src/wiktextract/extractor/it/tags.py: 97%

1from .models import WordEntry

3TABLE_TAGS = {

4 # https://it.wiktionary.org/wiki/Template:It-decl-agg4

5 "singolare": "singular",

6 "plurale": "plural",

7 "positivo": "positive",

8 "superlativo assoluto": ["absolute", "superlative"],

9 "maschile": "masculine",

10 "femminile": "feminine",

11 # https://it.wiktionary.org/wiki/Template:It-decl-agg2

12 "m e f": ["masculine", "feminine"],

13 # https://it.wiktionary.org/wiki/Template:It-conj

14 "infinito": "infinitive",

15 "verbo ausiliare": "auxiliary",

16 "gerundio": "gerund",

17 "participio presente": ["present", "participle"],

18 "participio passato": ["past", "participle"],

19 "prima": "first-person",

20 "seconda": "second-person",

21 "terza": "third-person",

22 "presente": "present",

23 "imperfetto": "imperfect",

24 "passato remoto": "past-remote",

25 "futuro": "future",

26 "passato prossimo": ["past", "perfect"],

27 "trapassato prossimo": ["pluperfect", "past", "perfect"],

28 "trapassato remoto": ["historic", "past-remote"],

29 "futuro anteriore": ["future", "perfect"],

30 "passato": "past",

31 "trapassato": ["past", "perfect"],

32 "imperativo": "imperative",

33 "riflessivo pronominale": ["reflexive", "pronominal"],

34}

36FORM_LINE_TEMPLATE_TAGS = {

37 # https://it.wiktionary.org/wiki/Template:A_cmp

38 "comparativo": "comparative",

39 "superlativo": "superlative",

40}

42# https://it.wiktionary.org/wiki/Template:Term/d

43TERM_TEMPLATE_TOPICS = {

44 "abbigliamento": "clothing",

45 "aeronautica": "aeronautics",

46 "agricoltura": "agriculture",

47 "algebra": "algebra",

48 "ambito sportivo": "sports",

49 "anatomia": "anatomy",

50 # "animali": "",

51 "antropologia": "anthropology",

52 "araldica": "heraldry",

53 "archeologia": "archaeology",

54 "architettura": "architecture",

55 "aritmetica": "arithmetic",

56 "arm.": "weaponry",

57 "arma": "weaponry",

58 "armamento": "weaponry",

59 "armi": "weaponry",

60 "arte": "arts",

61 "astrologia": "astrology",

62 "astronomia": "astronomy",

63 "botanica": "botany",

64 "biochimica": "biochemistry",

65 "biologia": "biology",

66 "biotecnologia": "biotechnology",

67 # "burocrazia": "",

68 "calcio": "soccer",

69 "carte": "card-games",

70 "chimica": "chemistry",

71 "chimica generale": "chemistry",

72 "chimica inorganica": "chemistry",

73 "chimica organica": "chemistry",

74 "chimica analitica": "chemistry",

75 "chimica industriale": "chemistry",

76 "chirurgia": "surgery",

77 "cinematografia": "cinematography",

78 "colore": "color",

79 "commercio": "commerce",

80 # "composti organici": "",

81 # "composti inorganici": "",

82 "cristianesimo": "Christianity",

83 "danza": "dance",

84 # "diritto": "",

85 "ecclesiastico": "ecclesiastical",

86 "ecologia": "ecology",

87 "economia": "economics",

88 "edilizia": "construction",

89 "elementi chimici": "chemistry",

90 "elettronica": "electronics",

91 "elettrotecnica": "electrical-engineering",

92 "entomologia": "entomology",

93 "equitazione": "equitation",

94 "erpetologia": "herpetology",

95 # "esoterismo": "",

96 "etnologia": "ethnology",

97 "falegnameria": "carpentry",

98 # "familiare": "",

99 "farmacologia": "pharmacology",

100 "ferrovia": "railways",

101 "filosofia": "philosophy",

102 "finanza": "finance",

103 "fisica": "physics",

104 "fisiologia": "physiology",

105 "fonologia": "phonology",

106 # "forestierismo": "",

107 "fotografia": "photography",

108 # "gastronomia": "gastronomy",

109 "genetica": "genetics",

110 "geografia": "geography",

111 "geologia": "geology",

112 "geometria": "geometry",

113 "gioco": "games",

114 "giornalistico": "journalism",

115 "grammatica": "grammar",

116 "idraulica": "hydraulics",

117 "informatica": "informatics",

118 "ingegneria": "engineering",

119 "internet": "Internet",

120 "ittiologia": "ichthyology",

121 "legale": "law",

122 "letteratura": "literature",

123 "linguistica": "linguistics",

124 # "macelleria": "",

125 "malacologia": "malacology",

126 "mammalogia": "mammalogy",

127 "marina": "navy",

128 "matematica": "mathematics",

129 "meccanica": "mechanics",

130 "medicina": "medicine",

131 "metallurgia": "metallurgy",

132 "meteorologia": "meteorology",

133 # "Metrica": "",

134 # "Metrica classica": "",

135 # "Metrica contemporanea": "",

136 # "Metrica latina": "",

137 "militare": "military",

138 "minerale": "mineralogy",

139 "mineralogia": "mineralogy",

140 "mitologia": "mythology",

141 "moda": "fashion",

142 "musica": "music",

143 "numismatica": "numismatics",

144 "ornitologia": "ornithology",

145 # "pedagogia": "pedagogy",

146 # "pittura": "painting",

147 "poesia": "poetry",

148 # "polimeri": "",

149 "politica": "politics",

150 # "Popolare": "",

151 # "Professioni": "",

152 "psichiatria": "psychiatry",

153 "psicanalisi": "psychoanalysis",

154 "psicologia": "psychology",

155 "religione": "religion",

156 "topografia": "topography",

157 # "Toponimi": "",

158 "paleontologia": "paleontology",

159 "pianta": "botany",

160 "scacchi": "chess",

161 # "Scuola": "",

162 "sessualità": "sexuality",

163 "sociologia": "sociology",

164 "sport": "sports",

165 "sport invernali": "sports",

166 "statistica": "statistics",

167 "storia": "history",

168 # "strumenti musicali": "",

169 "teatro": "theater",

170 "tecnica": "technology",

171 "tecnologia": "technology",

172 "telecomunicazioni": "telecommunications",

173 "tessile": "textiles",

174 "tipografia": "typography",

175 "veterinaria": "veterinary",

176 "zoologia": "zoology",

177 # "zootecnica": "",

178}

179

180TERM_TEMPLATE_TAGS = {

181 "antico": "archaic",

182 "obsoleto": "obsolete",

183 "formale": "formal",

184 "gergale": "slang",

185 "informale": "informal",

186 "letterario": "literary",

187 "neologismo": "neologism",

188 "offensivo": "offensive",

189 "raro": "rare",

190 "regionale": "regional",

191 "volgare": "vulgar",

192}

193

194# https://it.wiktionary.org/wiki/Categoria:Template_ambito

195GLOSS_LIST_TEMPATE_TAGS = {

196 "accrescitivo": "augmentative", # Template:Accr

197 "colloquiale": "colloquial", # Template:Coll

198 "diminutivo": "diminutive", # Template:Dim

199 "per estensione": "broadly", # Template:Est

200 "senso figurato": "figuratively", # Template:Fig

201 "letteralmente": "literally", # Template:Lett

202 "peggiorativo": "pejorative", # Template:Pegg

203 "riferito solo a persone": "person", # Template:Pers

204 "per sineddoche": "synecdoche", # Template:Sndc

205 "specialmente al plurale": ["especially", "in-plural"], # Template:Spec pl

206 "spregiativo": "pejorative", # Template:Spreg

207 "vezzeggiativo": "endearing", # Template:Vezz

208 "volgare": "vulgar", # Template:Vulg

209}

210

211

212TAGS = {

213 **TABLE_TAGS,

214 **FORM_LINE_TEMPLATE_TAGS,

215 **TERM_TEMPLATE_TAGS,

216 **GLOSS_LIST_TEMPATE_TAGS,

217}

218

219

220def translate_raw_tags(data: WordEntry) -> None:

221 raw_tags = []

222 for raw_tag in data.raw_tags:

223 if raw_tag in TAGS and hasattr(data, "tags"):

224 tr_tag = TAGS[raw_tag]

225 if isinstance(tr_tag, str):

226 data.tags.append(tr_tag)

227 elif isinstance(tr_tag, list): 227 ↛ 222line 227 didn't jump to line 222 because the condition on line 227 was always true

228 data.tags.extend(tr_tag)

229 elif raw_tag in TERM_TEMPLATE_TOPICS and hasattr(data, "topics"):

230 data.topics.append(TERM_TEMPLATE_TOPICS[raw_tag])

231 else:

232 raw_tags.append(raw_tag)

233 data.raw_tags = raw_tags