Coverage for src/wiktextract/extractor/fr/topics.py: 100%

1# https://fr.wiktionary.org/wiki/Annexe:Glossaire_grammatical

2# https://fr.wiktionary.org/wiki/Module:lexique/data

3# Catégorie:Modèles de thématique

5TOPIC_TAGS: dict[str, str] = {

6 "botanique": "botany",

7 "cuisine": "cuisine",

8 "informatique": "computing",

9 "marine": "nautical",

10 "droit": "law",

11 "politique": "politics",

12 "physique": "physics",

13 "sport": "sports",

14 "commerce": "commerce",

15 # from Module:lexique/data

16 "BDSM": "BDSM",

17 "Internet": "Internet",

18 "Scrabble": "Scrabble",

19 "agriculture": "agriculture",

20 "alchimie": "alchemy",

21 "algèbre": "algebra",

22 "alpinisme": "mountaineering",

23 "anarchisme": "anarchism",

24 "anatomie": "anatomy",

25 "anthropologie": "anthropology",

26 "apiculture": "beekeeping",

27 "arachnologie": "arachnology",

28 "architecture": "architecture",

29 "archéologie": "archeology",

30 "argent": "money",

31 "arithmétique": "arithmetic",

32 "art": "art",

33 "artillerie": "artillery",

34 "assurance": "insurance",

35 "astrologie": "astrology",

36 "astronautique": "astronautics",

37 "astronomie": "astronomy",

38 "astrophysique": "astrophysics",

39 "automobile": "automobile",

40 "aviation": "aviation",

41 "aéronautique": "aeronautics",

42 "bactériologie": "bacteriology",

43 "baseball": "baseball",

44 "basket-ball": "basketball",

45 "billard": "billiards",

46 "biochimie": "biochemistry",

47 "biogéographie": "biogeography",

48 "biologie": "biology",

49 "bouddhisme": "Buddhism",

50 "bowling": "bowling",

51 "boxe": "boxing",

52 "bridge": "bridge",

53 "bryologie": "bryology",

54 "calligraphie": "calligraphy",

55 "capitalisme": "capitalism",

56 "carcinologie": "oncology",

57 "cartographie": "cartography",

58 "catch": "wrestling",

59 "catholicisme": "Catholicism",

60 "charpenterie": "carpentry",

61 "charronnerie": "carpentry",

62 "chasse": "hunting",

63 "chimie": "chemistry",

64 "chirurgie": "surgery",

65 "christianisme": "Christianity",

66 "cinéma": "film",

67 "cirque": "circus",

68 "climatologie": "climatology",

69 "coiffure": "hairdressing",

70 "colorimétrie": "colorimetry",

71 "comptabilité": "accounting",

72 "conchyliologie": "conchology",

73 "construction": "construction",

74 "couture": "sewing",

75 "cricket": "cricket",

76 "cristallographie": "crystallography",

77 "cryptographie": "cryptography",

78 "cryptomonnaies": "cryptocurrencies",

79 "curling": "curling",

80 "cyclisme": "cycling",

81 "danse": "dance",

82 "dentisterie": "dentistry",

83 "dermatologie": "dermatology",

84 "diplomatie": "diplomacy",

85 "démographie": "demography",

86 "embryologie": "embryology",

87 "entomologie": "entomology",

88 "entreprise": "business",

89 "escrime": "fencing",

90 "ethnologie": "ethnology",

91 "fantasie": "fantasy",

92 "fascisme": "fascism",

93 "fauconnerie": "falconry",

94 "finance": "finance",

95 "fiscalité": "taxation",

96 "football": "soccer",

97 "foresterie": "forestry",

98 "fortification": "fortification",

99 "féminisme": "feminism",

100 "féodalité": "feudalism",

101 "glaciologie": "glaciology",

102 "golf": "golf",

103 "grammaire": "grammar",

104 "gravure": "engraving",

105 "géographie": "geography",

106 "géologie": "geology",

107 "géomorphologie": "geomorphology",

108 "géométrie": "geometry",

109 "géopolitique": "geopolitics",

110 "habillement": "clothing",

111 "handball": "handball",

112 "herpétologie": "herpetology",

113 "histoire": "history",

114 "histologie": "histology",

115 "horticulture": "horticulture",

116 "hydrographie": "hydrography",

117 "hydrologie": "hydrology",

118 "héraldique": "heraldry",

119 "ichtyologie": "ichthyology",

120 "iconographie": "iconography",

121 "immunologie": "immunology",

122 "islam": "Islam",

123 "jeux": "games",

124 "jeux d’argent": "gambling",

125 "joaillerie": "jewelry",

126 "jonglerie": "juggling",

127 "journalisme": "journalism",

128 "judo": "judo",

129 "lexicographie": "lexicography",

130 "lexicologie": "lexicology",

131 "linguistique": "linguistic",

132 "littérature": "literature",

133 "logique": "logic",

134 "loisirs": "hobbies",

135 "mah-jong": "mahjong",

136 "malacologie": "malacology",

137 "mammalogie": "mammalogy",

138 "management": "management",

139 "marketing": "marketing",

140 "maréchalerie": "farriery",

141 "mathématiques": "mathematics",

142 "maçonnerie": "masonry",

143 "menuiserie": "carpentry",

144 "microbiologie": "microbiology",

145 "militaire": "military",

146 "minéralogie": "mineralogy",

147 "monarchie": "monarchy",

148 "morphologie": "morphology",

149 "motocyclisme": "motorcycling",

150 "musculation": "bodybuilding",

151 "musique": "music",

152 "mycologie": "mycology",

153 "mythologie": "mythology",

154 "mécanique": "mechanical",

155 "médecine": "medicine",

156 "médias": "media",

157 "métallurgie": "metallurgy",

158 "métrologie": "metrology",

159 "météorologie": "meteorology",

160 "nanotechnologie": "nanotechnology",

161 "narratologie": "narratology",

162 "natation": "swimming",

163 "neurologie": "neurology",

164 "noblesse": "nobility",

165 "numismatique": "numismatics",

166 "obstétrique": "obstetrics",

167 "occultisme": "occultism",

168 "océanographie": "oceanography",

169 "oncologie": "oncology",

170 "ontologie": "ontology",

171 "ophtalmologie": "ophthalmology",

172 "ornithologie": "ornithology",

173 "outillage": "tools",

174 "paintball": "paintball",

175 "paléographie": "paleography",

176 "paléontologie": "paleontology",

177 "parachutisme": "skydiving",

178 "parasitologie": "parasitology",

179 "pathologie": "pathology",

180 "patinage": "skating",

181 "pharmacologie": "pharmacology",

182 "philatélie": "philately",

183 "philosophie": "philosophy",

184 "phonologie": "phonology",

185 "photographie": "photography",

186 "physiologie": "physiology",

187 "phytopathologie": "phytopathology",

188 "phénoménologie": "phenomenology",

189 "plomberie": "plumbing",

190 "plongée": "diving",

191 "poker": "poker",

192 "police": "police",

193 "poésie": "poetry",

194 "programmation": "programming",

195 "protestantisme": "Protestantism",

196 "psychanalyse": "psychoanalysis",

197 "psychiatrie": "psychiatry",

198 "psychologie": "psychology",

199 "psychopathologie": "psychopathology",

200 "pédologie": "pedology",

201 "pétrographie": "petrography",

202 "pétrologie": "petrology",

203 "pêche": "fishing",

204 "radiodiffusion": "broadcasting",

205 "religion": "religion",

206 "renseignement": "information",

207 "robotique": "robotics",

208 "rugby": "rugby",

209 "sciences": "science",

210 "sexualité": "sexuality",

211 "socialisme": "socialism",

212 "sociolinguistique": "sociolinguistics",

213 "sociologie": "sociology",

214 "soufisme": "Sufism",

215 "statistiques": "statistics",

216 "sylviculture": "forestry",

217 "syntaxe": "syntax",

218 "tauromachie": "bullfighting",

219 "taxinomie": "taxonomy",

220 "technique": "technical",

221 "technologie": "technology",

222 "temps": "time",

223 "tennis": "tennis",

224 "thermodynamique": "thermodynamics",

225 "théologie": "theology",

226 "théâtre": "theater",

227 "tir à l’arc": "archery",

228 "tissage": "weaving",

229 "topographie": "topography",

230 "topologie": "topology",

231 "tourisme": "tourism",

232 "trigonométrie": "trigonometry",

233 "typographie": "typography",

234 "télécommunications": "telecommunications",

235 "téléphonie": "telephony",

236 "télévision": "television",

237 "tératologie": "teratology",

238 "ufologie": "ufology",

239 "usinage": "machining",

240 "vexillologie": "vexillology",

241 "virologie": "virology",

242 "volcanologie": "volcanology",

243 "volley-ball": "volleyball",

244 "vénerie": "hunting",

245 "xiangqi": "xiangqi",

246 "zoologie": "zoology",

247 "échecs": "chess",

248 "écologie": "ecology",

249 "édition": "editing",

250 "éducation": "education",

251 "égyptologie": "Egyptology",

252 "électricité": "electricity",

253 "électromagnétisme": "electromagnetism",

254 "épidémiologie": "epidemiology",

255 "épistémologie": "epistemology",

256 "éthique": "ethics",

257 "œnologie": "oenology",

258 # manually translated

259 "CB": "radio", # Citizens band radio

260 "algèbre linéaire": "mathematics",

261 "alimentaire": "food",

262 "ameublement": "lifestyle",

263 "architecture des ordinateurs": "computing",

264 "art moderne": "arts",

265 "art urbain": "arts",

266 "artisanat": "crafts",

267 "arts appliqués": "arts",

268 "arts de la scène": "performing-arts",

269 "arts martiaux": "martial-arts",

270 "arts visuels": "visual-arts",

271 "athlétisme": "sports",

272 "baby-foot": "sports",

273 "badminton": "sports",

274 "bande dessinée": "comics",

275 "banque": "business",

276 "bases de données": "computing",

277 "beaux-arts": "arts",

278 "bijouterie": "jewelry",

279 "boulangerie": "cooking",

280 "bourse": "finance",

281 "brasserie": "beverages manufacturing",

282 "équitation": "equestrianism",

283 # Catégorie:Modèles de thématique

284 "phycologie": "phycology",

285 "bière": "beer",

286 "chronologie": "chronology",

287 "dialectologie": "dialectology",

288 "média": "media",

289 "état": "state",

290 "mobilier": "furniture",

291 "science": "science",

292 "boisson": "beverages",

293 "vêtement": "clothing",

294 "électronique": "electricity",

295 "carte à jouer": "card-games",

296 "lgbt": "LGBT",

297 "jeu vidéo, minecraft": "video-games",

298 "réseaux": "networking",

299 "science-fiction, univers de star wars": "science-fiction",

300 "textile": "textiles",

301 "transport": "transport",

302}

303

304SLANG_TOPICS = {

305 # https://fr.wiktionary.org/wiki/Modèle:argot

306 "argot scolaire": "school",

307 "argot polytechnicien": "polytechnicien",

308 "argot policier": "police",

309 "argot internet": "Internet",

310 "argot des typographes": "typography",

311 "argot militaire": "military",

312 "argot poilu": "poilu",

313 "argot des voleurs": "thieves",

314 "argot des gadz’arts": "Gadz'Arts",

315 "langage sms": "SMS",

316}