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 "plur": "plural",

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

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

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

44 "Deponente": "deponent",

45 "Ausiliare": "auxiliary",

46 "Passivo": "passive",

47 "Reciproco": "reciprocal",

48 "Riflessivo": "reflexive",

49 "Attivo": "active",

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

51}

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

54TERM_TEMPLATE_TOPICS = {

55 "abbigliamento": "clothing",

56 "aeronautica": "aeronautics",

57 "agricoltura": "agriculture",

58 "algebra": "algebra",

59 "ambito sportivo": "sports",

60 "anatomia": "anatomy",

61 # "animali": "",

62 "antropologia": "anthropology",

63 "araldica": "heraldry",

64 "archeologia": "archaeology",

65 "architettura": "architecture",

66 "aritmetica": "arithmetic",

67 "arm.": "weaponry",

68 "arma": "weaponry",

69 "armamento": "weaponry",

70 "armi": "weaponry",

71 "arte": "arts",

72 "astrologia": "astrology",

73 "astronomia": "astronomy",

74 "botanica": "botany",

75 "biochimica": "biochemistry",

76 "biologia": "biology",

77 "biotecnologia": "biotechnology",

78 # "burocrazia": "",

79 "calcio": "soccer",

80 "carte": "card-games",

81 "chimica": "chemistry",

82 "chimica generale": "chemistry",

83 "chimica inorganica": "chemistry",

84 "chimica organica": "chemistry",

85 "chimica analitica": "chemistry",

86 "chimica industriale": "chemistry",

87 "chirurgia": "surgery",

88 "cinematografia": "cinematography",

89 "colore": "color",

90 "commercio": "commerce",

91 # "composti organici": "",

92 # "composti inorganici": "",

93 "cristianesimo": "Christianity",

94 "danza": "dance",

95 # "diritto": "",

96 "ecclesiastico": "ecclesiastical",

97 "ecologia": "ecology",

98 "economia": "economics",

99 "edilizia": "construction",

100 "elementi chimici": "chemistry",

101 "elettronica": "electronics",

102 "elettrotecnica": "electrical-engineering",

103 "entomologia": "entomology",

104 "equitazione": "equitation",

105 "erpetologia": "herpetology",

106 # "esoterismo": "",

107 "etnologia": "ethnology",

108 "falegnameria": "carpentry",

109 # "familiare": "",

110 "farmacologia": "pharmacology",

111 "ferrovia": "railways",

112 "filosofia": "philosophy",

113 "finanza": "finance",

114 "fisica": "physics",

115 "fisiologia": "physiology",

116 "fonologia": "phonology",

117 # "forestierismo": "",

118 "fotografia": "photography",

119 # "gastronomia": "gastronomy",

120 "genetica": "genetics",

121 "geografia": "geography",

122 "geologia": "geology",

123 "geometria": "geometry",

124 "gioco": "games",

125 "giornalistico": "journalism",

126 "grammatica": "grammar",

127 "idraulica": "hydraulics",

128 "informatica": "informatics",

129 "ingegneria": "engineering",

130 "internet": "Internet",

131 "ittiologia": "ichthyology",

132 "legale": "law",

133 "letteratura": "literature",

134 "linguistica": "linguistics",

135 # "macelleria": "",

136 "malacologia": "malacology",

137 "mammalogia": "mammalogy",

138 "marina": "navy",

139 "matematica": "mathematics",

140 "meccanica": "mechanics",

141 "medicina": "medicine",

142 "metallurgia": "metallurgy",

143 "meteorologia": "meteorology",

144 # "Metrica": "",

145 # "Metrica classica": "",

146 # "Metrica contemporanea": "",

147 # "Metrica latina": "",

148 "militare": "military",

149 "minerale": "mineralogy",

150 "mineralogia": "mineralogy",

151 "mitologia": "mythology",

152 "moda": "fashion",

153 "musica": "music",

154 "numismatica": "numismatics",

155 "ornitologia": "ornithology",

156 # "pedagogia": "pedagogy",

157 # "pittura": "painting",

158 "poesia": "poetry",

159 # "polimeri": "",

160 "politica": "politics",

161 # "Popolare": "",

162 # "Professioni": "",

163 "psichiatria": "psychiatry",

164 "psicanalisi": "psychoanalysis",

165 "psicologia": "psychology",

166 "religione": "religion",

167 "topografia": "topography",

168 # "Toponimi": "",

169 "paleontologia": "paleontology",

170 "pianta": "botany",

171 "scacchi": "chess",

172 # "Scuola": "",

173 "sessualità": "sexuality",

174 "sociologia": "sociology",

175 "sport": "sports",

176 "sport invernali": "sports",

177 "statistica": "statistics",

178 "storia": "history",

179 # "strumenti musicali": "",

180 "teatro": "theater",

181 "tecnica": "technology",

182 "tecnologia": "technology",

183 "telecomunicazioni": "telecommunications",

184 "tessile": "textiles",

185 "tipografia": "typography",

186 "veterinaria": "veterinary",

187 "zoologia": "zoology",

188 # "zootecnica": "",

189}

190

191TERM_TEMPLATE_TAGS = {

192 "antico": "archaic",

193 "obsoleto": "obsolete",

194 "formale": "formal",

195 "gergale": "slang",

196 "informale": "informal",

197 "letterario": "literary",

198 "neologismo": "neologism",

199 "offensivo": "offensive",

200 "raro": "rare",

201 "regionale": "regional",

202 "volgare": "vulgar",

203}

204

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

206GLOSS_LIST_TEMPATE_TAGS = {

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

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

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

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

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

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

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

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

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

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

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

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

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

220}

221

222OTHER_TAGS = {

223 "Transitivo": "transitive",

224 "Intransitivo": "intransitive",

225 "toponimo": "toponymic",

226 "classe": "class",

227}

228

229

230TAGS = {

231 **TABLE_TAGS,

232 **FORM_LINE_TEMPLATE_TAGS,

233 **TERM_TEMPLATE_TAGS,

234 **GLOSS_LIST_TEMPATE_TAGS,

235 **OTHER_TAGS,

236}

237

238

239def translate_raw_tags(data: WordEntry) -> None:

240 raw_tags = []

241 for raw_tag in data.raw_tags:

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

243 tr_tag = TAGS[raw_tag]

244 if isinstance(tr_tag, str):

245 data.tags.append(tr_tag)

246 elif isinstance(tr_tag, list): 246 ↛ 241line 246 didn't jump to line 241 because the condition on line 246 was always true

247 data.tags.extend(tr_tag)

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

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

250 else:

251 raw_tags.append(raw_tag)

252 data.raw_tags = raw_tags