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 # Template:Fr-conj

35 "verbo ausiliare\nauxiliaire": "auxiliary",

36 "gerundio\ngérondif": "gerund",

37 "participio presente\nparticipe présent": ["present", "participle"],

38 "participio passato\nparticipe passé": ["past", "participle"],

39 "presente\nprésent": "present",

40 "imperfetto\nimparfait": "imperfect",

41 "passato remoto\npassé simple": "past-remote",

42 "futuro\nfutur simple": "future",

43 "passato prossimo\npassé composé": ["past", "perfect"],

44 "trapassato prossimo\nplus-que-parfait": ["pluperfect", "past", "perfect"],

45 "trapassato remoto\npassé anterieur": ["historic", "past-remote"],

46 "futuro anteriore\nfutur antérieur": ["future", "perfect"],

47 "passato\npassé": "past",

48 "trapassato\nplus-que-parfait": ["past", "perfect"],

49 "imperativo\nimpératif": "imperative",

50 # Template:La-conj

51 "paradigma": "paradigm",

52 "attivo": "active",

53 "passivo": "passive",

54 "perfetto": "perfect",

55 "participio": "participle",

56 "nominativo": "nominative",

57 "genitivo": "genitive",

58 "dativo": "dative",

59 "accusativo": "accusative",

60 "ablativo": "ablative",

61 "gerundivo": "gerund",

62 "supino": "supine",

63 "futuro semplice": "future",

64 "piuccheperfetto": "pluperfect",

65}

67FORM_LINE_TEMPLATE_TAGS = {

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

69 "comparativo": "comparative",

70 "superlativo": "superlative",

71 "plur": "plural",

72 "Participio presente": ["present", "participle"],

73 "Participio passato": ["past", "participle"],

74 "Intransitivo pronominale": ["intransitive", "pronominal"], # Template:inpr

75 "Deponente": "deponent",

76 "Ausiliare": "auxiliary",

77 "Passivo": "passive",

78 "Reciproco": "reciprocal",

79 "Riflessivo": "reflexive",

80 "Attivo": "active",

81 "Riflessivo pronominale": ["reflexive", "pronominal"],

82 # Template:en-verb

83 "3ª persona sing. presente": ["third-person", "singular", "present"],

84 "passato semplice": "past",

85 "passato semplice e participio passato": ["past", "participle"],

86}

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

89TERM_TEMPLATE_TOPICS = {

90 "abbigliamento": "clothing",

91 "aeronautica": "aeronautics",

92 "agricoltura": "agriculture",

93 "algebra": "algebra",

94 "ambito sportivo": "sports",

95 "anatomia": "anatomy",

96 # "animali": "",

97 "antropologia": "anthropology",

98 "araldica": "heraldry",

99 "archeologia": "archaeology",

100 "architettura": "architecture",

101 "aritmetica": "arithmetic",

102 "arm.": "weaponry",

103 "arma": "weaponry",

104 "armamento": "weaponry",

105 "armi": "weaponry",

106 "arte": "arts",

107 "astrologia": "astrology",

108 "astronomia": "astronomy",

109 "botanica": "botany",

110 "biochimica": "biochemistry",

111 "biologia": "biology",

112 "biotecnologia": "biotechnology",

113 # "burocrazia": "",

114 "calcio": "soccer",

115 "carte": "card-games",

116 "chimica": "chemistry",

117 "chimica generale": "chemistry",

118 "chimica inorganica": "chemistry",

119 "chimica organica": "chemistry",

120 "chimica analitica": "chemistry",

121 "chimica industriale": "chemistry",

122 "chirurgia": "surgery",

123 "cinematografia": "cinematography",

124 "colore": "color",

125 "commercio": "commerce",

126 # "composti organici": "",

127 # "composti inorganici": "",

128 "cristianesimo": "Christianity",

129 "danza": "dance",

130 # "diritto": "",

131 "ecclesiastico": "ecclesiastical",

132 "ecologia": "ecology",

133 "economia": "economics",

134 "edilizia": "construction",

135 "elementi chimici": "chemistry",

136 "elettronica": "electronics",

137 "elettrotecnica": "electrical-engineering",

138 "entomologia": "entomology",

139 "equitazione": "equitation",

140 "erpetologia": "herpetology",

141 # "esoterismo": "",

142 "etnologia": "ethnology",

143 "falegnameria": "carpentry",

144 # "familiare": "",

145 "farmacologia": "pharmacology",

146 "ferrovia": "railways",

147 "filosofia": "philosophy",

148 "finanza": "finance",

149 "fisica": "physics",

150 "fisiologia": "physiology",

151 "fonologia": "phonology",

152 # "forestierismo": "",

153 "fotografia": "photography",

154 # "gastronomia": "gastronomy",

155 "genetica": "genetics",

156 "geografia": "geography",

157 "geologia": "geology",

158 "geometria": "geometry",

159 "gioco": "games",

160 "giornalistico": "journalism",

161 "grammatica": "grammar",

162 "idraulica": "hydraulics",

163 "informatica": "informatics",

164 "ingegneria": "engineering",

165 "internet": "Internet",

166 "ittiologia": "ichthyology",

167 "legale": "law",

168 "letteratura": "literature",

169 "linguistica": "linguistics",

170 # "macelleria": "",

171 "malacologia": "malacology",

172 "mammalogia": "mammalogy",

173 "marina": "navy",

174 "matematica": "mathematics",

175 "meccanica": "mechanics",

176 "medicina": "medicine",

177 "metallurgia": "metallurgy",

178 "meteorologia": "meteorology",

179 # "Metrica": "",

180 # "Metrica classica": "",

181 # "Metrica contemporanea": "",

182 # "Metrica latina": "",

183 "militare": "military",

184 "minerale": "mineralogy",

185 "mineralogia": "mineralogy",

186 "mitologia": "mythology",

187 "moda": "fashion",

188 "musica": "music",

189 "numismatica": "numismatics",

190 "ornitologia": "ornithology",

191 # "pedagogia": "pedagogy",

192 # "pittura": "painting",

193 "poesia": "poetry",

194 # "polimeri": "",

195 "politica": "politics",

196 # "Popolare": "",

197 # "Professioni": "",

198 "psichiatria": "psychiatry",

199 "psicanalisi": "psychoanalysis",

200 "psicologia": "psychology",

201 "religione": "religion",

202 "topografia": "topography",

203 # "Toponimi": "",

204 "paleontologia": "paleontology",

205 "pianta": "botany",

206 "scacchi": "chess",

207 # "Scuola": "",

208 "sessualità": "sexuality",

209 "sociologia": "sociology",

210 "sport": "sports",

211 "sport invernali": "sports",

212 "statistica": "statistics",

213 "storia": "history",

214 # "strumenti musicali": "",

215 "teatro": "theater",

216 "tecnica": "technology",

217 "tecnologia": "technology",

218 "telecomunicazioni": "telecommunications",

219 "tessile": "textiles",

220 "tipografia": "typography",

221 "veterinaria": "veterinary",

222 "zoologia": "zoology",

223 # "zootecnica": "",

224}

225

226TERM_TEMPLATE_TAGS = {

227 "antico": "archaic",

228 "obsoleto": "obsolete",

229 "formale": "formal",

230 "gergale": "slang",

231 "informale": "informal",

232 "letterario": "literary",

233 "neologismo": "neologism",

234 "offensivo": "offensive",

235 "raro": "rare",

236 "regionale": "regional",

237 "volgare": "vulgar",

238}

239

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

241GLOSS_LIST_TEMPLATE_TAGS = {

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

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

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

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

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

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

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

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

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

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

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

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

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

255}

256

257OTHER_TAGS = {

258 "Transitivo": "transitive",

259 "Intransitivo": "intransitive",

260 "toponimo": "toponymic",

261 "classe": "class",

262}

263

264

265TAGS = {

266 **TABLE_TAGS,

267 **FORM_LINE_TEMPLATE_TAGS,

268 **TERM_TEMPLATE_TAGS,

269 **GLOSS_LIST_TEMPLATE_TAGS,

270 **OTHER_TAGS,

271}

272

273

274def translate_raw_tags(data: WordEntry) -> None:

275 raw_tags = []

276 for raw_tag in data.raw_tags:

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

278 tr_tag = TAGS[raw_tag]

279 if isinstance(tr_tag, str):

280 data.tags.append(tr_tag)

281 elif isinstance(tr_tag, list): 281 ↛ 276line 281 didn't jump to line 276 because the condition on line 281 was always true

282 data.tags.extend(tr_tag)

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

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

285 else:

286 raw_tags.append(raw_tag)

287 data.raw_tags = raw_tags

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

21 statements