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}

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

85TERM_TEMPLATE_TOPICS = {

86 "abbigliamento": "clothing",

87 "aeronautica": "aeronautics",

88 "agricoltura": "agriculture",

89 "algebra": "algebra",

90 "ambito sportivo": "sports",

91 "anatomia": "anatomy",

92 # "animali": "",

93 "antropologia": "anthropology",

94 "araldica": "heraldry",

95 "archeologia": "archaeology",

96 "architettura": "architecture",

97 "aritmetica": "arithmetic",

98 "arm.": "weaponry",

99 "arma": "weaponry",

100 "armamento": "weaponry",

101 "armi": "weaponry",

102 "arte": "arts",

103 "astrologia": "astrology",

104 "astronomia": "astronomy",

105 "botanica": "botany",

106 "biochimica": "biochemistry",

107 "biologia": "biology",

108 "biotecnologia": "biotechnology",

109 # "burocrazia": "",

110 "calcio": "soccer",

111 "carte": "card-games",

112 "chimica": "chemistry",

113 "chimica generale": "chemistry",

114 "chimica inorganica": "chemistry",

115 "chimica organica": "chemistry",

116 "chimica analitica": "chemistry",

117 "chimica industriale": "chemistry",

118 "chirurgia": "surgery",

119 "cinematografia": "cinematography",

120 "colore": "color",

121 "commercio": "commerce",

122 # "composti organici": "",

123 # "composti inorganici": "",

124 "cristianesimo": "Christianity",

125 "danza": "dance",

126 # "diritto": "",

127 "ecclesiastico": "ecclesiastical",

128 "ecologia": "ecology",

129 "economia": "economics",

130 "edilizia": "construction",

131 "elementi chimici": "chemistry",

132 "elettronica": "electronics",

133 "elettrotecnica": "electrical-engineering",

134 "entomologia": "entomology",

135 "equitazione": "equitation",

136 "erpetologia": "herpetology",

137 # "esoterismo": "",

138 "etnologia": "ethnology",

139 "falegnameria": "carpentry",

140 # "familiare": "",

141 "farmacologia": "pharmacology",

142 "ferrovia": "railways",

143 "filosofia": "philosophy",

144 "finanza": "finance",

145 "fisica": "physics",

146 "fisiologia": "physiology",

147 "fonologia": "phonology",

148 # "forestierismo": "",

149 "fotografia": "photography",

150 # "gastronomia": "gastronomy",

151 "genetica": "genetics",

152 "geografia": "geography",

153 "geologia": "geology",

154 "geometria": "geometry",

155 "gioco": "games",

156 "giornalistico": "journalism",

157 "grammatica": "grammar",

158 "idraulica": "hydraulics",

159 "informatica": "informatics",

160 "ingegneria": "engineering",

161 "internet": "Internet",

162 "ittiologia": "ichthyology",

163 "legale": "law",

164 "letteratura": "literature",

165 "linguistica": "linguistics",

166 # "macelleria": "",

167 "malacologia": "malacology",

168 "mammalogia": "mammalogy",

169 "marina": "navy",

170 "matematica": "mathematics",

171 "meccanica": "mechanics",

172 "medicina": "medicine",

173 "metallurgia": "metallurgy",

174 "meteorologia": "meteorology",

175 # "Metrica": "",

176 # "Metrica classica": "",

177 # "Metrica contemporanea": "",

178 # "Metrica latina": "",

179 "militare": "military",

180 "minerale": "mineralogy",

181 "mineralogia": "mineralogy",

182 "mitologia": "mythology",

183 "moda": "fashion",

184 "musica": "music",

185 "numismatica": "numismatics",

186 "ornitologia": "ornithology",

187 # "pedagogia": "pedagogy",

188 # "pittura": "painting",

189 "poesia": "poetry",

190 # "polimeri": "",

191 "politica": "politics",

192 # "Popolare": "",

193 # "Professioni": "",

194 "psichiatria": "psychiatry",

195 "psicanalisi": "psychoanalysis",

196 "psicologia": "psychology",

197 "religione": "religion",

198 "topografia": "topography",

199 # "Toponimi": "",

200 "paleontologia": "paleontology",

201 "pianta": "botany",

202 "scacchi": "chess",

203 # "Scuola": "",

204 "sessualità": "sexuality",

205 "sociologia": "sociology",

206 "sport": "sports",

207 "sport invernali": "sports",

208 "statistica": "statistics",

209 "storia": "history",

210 # "strumenti musicali": "",

211 "teatro": "theater",

212 "tecnica": "technology",

213 "tecnologia": "technology",

214 "telecomunicazioni": "telecommunications",

215 "tessile": "textiles",

216 "tipografia": "typography",

217 "veterinaria": "veterinary",

218 "zoologia": "zoology",

219 # "zootecnica": "",

220}

221

222TERM_TEMPLATE_TAGS = {

223 "antico": "archaic",

224 "obsoleto": "obsolete",

225 "formale": "formal",

226 "gergale": "slang",

227 "informale": "informal",

228 "letterario": "literary",

229 "neologismo": "neologism",

230 "offensivo": "offensive",

231 "raro": "rare",

232 "regionale": "regional",

233 "volgare": "vulgar",

234}

235

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

237GLOSS_LIST_TEMPATE_TAGS = {

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

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

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

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

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

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

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

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

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

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

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

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

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

251}

252

253OTHER_TAGS = {

254 "Transitivo": "transitive",

255 "Intransitivo": "intransitive",

256 "toponimo": "toponymic",

257 "classe": "class",

258}

259

260

261TAGS = {

262 **TABLE_TAGS,

263 **FORM_LINE_TEMPLATE_TAGS,

264 **TERM_TEMPLATE_TAGS,

265 **GLOSS_LIST_TEMPATE_TAGS,

266 **OTHER_TAGS,

267}

268

269

270def translate_raw_tags(data: WordEntry) -> None:

271 raw_tags = []

272 for raw_tag in data.raw_tags:

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

274 tr_tag = TAGS[raw_tag]

275 if isinstance(tr_tag, str):

276 data.tags.append(tr_tag)

277 elif isinstance(tr_tag, list): 277 ↛ 272line 277 didn't jump to line 272 because the condition on line 277 was always true

278 data.tags.extend(tr_tag)

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

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

281 else:

282 raw_tags.append(raw_tag)

283 data.raw_tags = raw_tags