jtcg_locale_detector 1.0.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (60) hide show
  1. checksums.yaml +7 -0
  2. data/CHANGELOG.md +37 -0
  3. data/PACKAGING_SUMMARY.md +195 -0
  4. data/README.md +226 -0
  5. data/bin/locale-detector +159 -0
  6. data/jtcg_locale_detector.gemspec +48 -0
  7. data/lib/locale_detector/client.rb +163 -0
  8. data/lib/locale_detector/detector.rb +46 -0
  9. data/lib/locale_detector/version.rb +3 -0
  10. data/lib/locale_detector.rb +25 -0
  11. data/locale_detector.gemspec +46 -0
  12. data/python/cli.py +220 -0
  13. data/python/requirements.txt +8 -0
  14. data/python/src/__init__.py +10 -0
  15. data/python/src/__pycache__/__init__.cpython-311.pyc +0 -0
  16. data/python/src/__pycache__/__init__.cpython-313.pyc +0 -0
  17. data/python/src/__pycache__/locale_data.cpython-311.pyc +0 -0
  18. data/python/src/__pycache__/locale_data.cpython-313.pyc +0 -0
  19. data/python/src/__pycache__/locale_detector.cpython-311.pyc +0 -0
  20. data/python/src/__pycache__/locale_detector.cpython-313.pyc +0 -0
  21. data/python/src/artifacts/fasttext/lid.176.bin +0 -0
  22. data/python/src/artifacts/fasttext/lid.176.ftz +0 -0
  23. data/python/src/download_fasttext.py +69 -0
  24. data/python/src/locale_data.py +178 -0
  25. data/python/src/locale_detector.py +534 -0
  26. data/python/src/locale_detector_c.c +403 -0
  27. data/python/src/locale_detector_c.h +37 -0
  28. data/python/src/locale_detector_cy.cpp +23126 -0
  29. data/python/src/locale_detector_cy.cpython-311-darwin.so +0 -0
  30. data/python/src/locale_detector_cy.cpython-313-darwin.so +0 -0
  31. data/python/src/locale_detector_cy.html +6460 -0
  32. data/python/src/locale_detector_cy.pyx +501 -0
  33. data/python/src/utils/__init__.py +1 -0
  34. data/python/src/utils/__pycache__/__init__.cpython-311.pyc +0 -0
  35. data/python/src/utils/__pycache__/__init__.cpython-313.pyc +0 -0
  36. data/python/src/utils/__pycache__/data_utils.cpython-311.pyc +0 -0
  37. data/python/src/utils/__pycache__/data_utils.cpython-313.pyc +0 -0
  38. data/python/src/utils/data_utils.py +50 -0
  39. data/python/src/utils/data_utils_cy.cpp +10086 -0
  40. data/python/src/utils/data_utils_cy.cpython-311-darwin.so +0 -0
  41. data/python/src/utils/data_utils_cy.cpython-313-darwin.so +0 -0
  42. data/python/src/utils/data_utils_cy.html +600 -0
  43. data/python/src/utils/data_utils_cy.pyx +94 -0
  44. data/python/src/zhon/__init__.py +7 -0
  45. data/python/src/zhon/__pycache__/__init__.cpython-311.pyc +0 -0
  46. data/python/src/zhon/__pycache__/hanzi.cpython-311.pyc +0 -0
  47. data/python/src/zhon/__pycache__/pinyin.cpython-311.pyc +0 -0
  48. data/python/src/zhon/__pycache__/zhuyin.cpython-311.pyc +0 -0
  49. data/python/src/zhon/cedict/__init__.py +14 -0
  50. data/python/src/zhon/cedict/__pycache__/__init__.cpython-311.pyc +0 -0
  51. data/python/src/zhon/cedict/__pycache__/all.cpython-311.pyc +0 -0
  52. data/python/src/zhon/cedict/__pycache__/simplified.cpython-311.pyc +0 -0
  53. data/python/src/zhon/cedict/__pycache__/traditional.cpython-311.pyc +0 -0
  54. data/python/src/zhon/cedict/all.py +4 -0
  55. data/python/src/zhon/cedict/simplified.py +4 -0
  56. data/python/src/zhon/cedict/traditional.py +4 -0
  57. data/python/src/zhon/hanzi.py +81 -0
  58. data/python/src/zhon/pinyin.py +187 -0
  59. data/python/src/zhon/zhuyin.py +46 -0
  60. metadata +198 -0
data/python/src/locale_data.py ADDED
@@ -0,0 +1,178 @@
1
+ LOCALE_MAP = {
2
+ "zh": {"i18n_code": "zh", "name": "Chinese"},
3
+ "af": {"i18n_code": "af", "name": "Afrikaans"},
4
+ "als": {"i18n_code": "not_supported", "name": "Tosk Albanian"},
5
+ "am": {"i18n_code": "am", "name": "Amharic"},
6
+ "an": {"i18n_code": "not_supported", "name": "Aragonese"},
7
+ "ar": {"i18n_code": "ar", "name": "Arabic"},
8
+ "arz": {"i18n_code": "not_supported", "name": "Egyptian Arabic"},
9
+ "as": {"i18n_code": "as", "name": "Assamese"},
10
+ "ast": {"i18n_code": "not_supported", "name": "Asturian"},
11
+ "av": {"i18n_code": "not_supported", "name": "Avaric"},
12
+ "az": {"i18n_code": "az", "name": "Azerbaijani"},
13
+ "azb": {"i18n_code": "not_supported", "name": "South Azerbaijani"},
14
+ "ba": {"i18n_code": "not_supported", "name": "Bashkir"},
15
+ "bar": {"i18n_code": "not_supported", "name": "Bavarian"},
16
+ "bcl": {"i18n_code": "not_supported", "name": "Central Bikol"},
17
+ "be": {"i18n_code": "be", "name": "Belarusian"},
18
+ "bg": {"i18n_code": "bg", "name": "Bulgarian"},
19
+ "bh": {"i18n_code": "not_supported", "name": "Bihari languages"},
20
+ "bn": {"i18n_code": "bn", "name": "Bengali"},
21
+ "bo": {"i18n_code": "bo", "name": "Tibetan"},
22
+ "bpy": {"i18n_code": "not_supported", "name": "Bishnupriya Manipuri"},
23
+ "br": {"i18n_code": "br", "name": "Breton"},
24
+ "bs": {"i18n_code": "bs", "name": "Bosnian"},
25
+ "bxr": {"i18n_code": "not_supported", "name": "Russia Buriat"},
26
+ "ca": {"i18n_code": "ca", "name": "Catalan"},
27
+ "cbk": {"i18n_code": "not_supported", "name": "Chavacano"},
28
+ "ce": {"i18n_code": "not_supported", "name": "Chechen"},
29
+ "ceb": {"i18n_code": "ceb", "name": "Cebuano"},
30
+ "ckb": {"i18n_code": "ckb", "name": "Sorani Kurdish"},
31
+ "co": {"i18n_code": "not_supported", "name": "Corsican"},
32
+ "cs": {"i18n_code": "cs", "name": "Czech"},
33
+ "cv": {"i18n_code": "not_supported", "name": "Chuvash"},
34
+ "cy": {"i18n_code": "cy", "name": "Welsh"},
35
+ "da": {"i18n_code": "da", "name": "Danish"},
36
+ "de": {"i18n_code": "de", "name": "German"},
37
+ "diq": {"i18n_code": "not_supported", "name": "Zazaki"},
38
+ "dsb": {"i18n_code": "dsb", "name": "Lower Sorbian"},
39
+ "dty": {"i18n_code": "not_supported", "name": "Doteli"},
40
+ "dv": {"i18n_code": "dv", "name": "Divehi"},
41
+ "el": {"i18n_code": "el", "name": "Greek"},
42
+ "eml": {"i18n_code": "not_supported", "name": "Emilian-Romagnol"},
43
+ "en": {"i18n_code": "en-US", "name": "English"},
44
+ "eo": {"i18n_code": "eo", "name": "Esperanto"},
45
+ "es": {"i18n_code": "es", "name": "Spanish"},
46
+ "et": {"i18n_code": "et", "name": "Estonian"},
47
+ "eu": {"i18n_code": "eu", "name": "Basque"},
48
+ "fa": {"i18n_code": "fa", "name": "Persian"},
49
+ "fi": {"i18n_code": "fi", "name": "Finnish"},
50
+ "fr": {"i18n_code": "fr", "name": "French"},
51
+ "frr": {"i18n_code": "not_supported", "name": "Northern Frisian"},
52
+ "fy": {"i18n_code": "fy", "name": "Western Frisian"},
53
+ "ga": {"i18n_code": "ga", "name": "Irish"},
54
+ "gd": {"i18n_code": "gd", "name": "Scottish Gaelic"},
55
+ "gl": {"i18n_code": "gl", "name": "Galician"},
56
+ "gn": {"i18n_code": "not_supported", "name": "Guarani"},
57
+ "gom": {"i18n_code": "not_supported", "name": "Goan Konkani"},
58
+ "gu": {"i18n_code": "gu", "name": "Gujarati"},
59
+ "gv": {"i18n_code": "gv", "name": "Manx"},
60
+ "he": {"i18n_code": "he", "name": "Hebrew"},
61
+ "hi": {"i18n_code": "hi", "name": "Hindi"},
62
+ "hif": {"i18n_code": "not_supported", "name": "Fiji Hindi"},
63
+ "hr": {"i18n_code": "hr", "name": "Croatian"},
64
+ "hsb": {"i18n_code": "hsb", "name": "Upper Sorbian"},
65
+ "ht": {"i18n_code": "not_supported", "name": "Haitian"},
66
+ "hu": {"i18n_code": "hu", "name": "Hungarian"},
67
+ "hy": {"i18n_code": "hy", "name": "Armenian"},
68
+ "ia": {"i18n_code": "not_supported", "name": "Interlingua"},
69
+ "id": {"i18n_code": "id", "name": "Indonesian"},
70
+ "ie": {"i18n_code": "not_supported", "name": "Interlingue"},
71
+ "ilo": {"i18n_code": "not_supported", "name": "Ilocano"},
72
+ "io": {"i18n_code": "not_supported", "name": "Ido"},
73
+ "is": {"i18n_code": "is", "name": "Icelandic"},
74
+ "it": {"i18n_code": "it", "name": "Italian"},
75
+ "ja": {"i18n_code": "ja", "name": "Japanese"},
76
+ "jbo": {"i18n_code": "not_supported", "name": "Lojban"},
77
+ "jv": {"i18n_code": "jv", "name": "Javanese"},
78
+ "ka": {"i18n_code": "ka", "name": "Georgian"},
79
+ "kk": {"i18n_code": "kk", "name": "Kazakh"},
80
+ "km": {"i18n_code": "km", "name": "Khmer"},
81
+ "kn": {"i18n_code": "kn", "name": "Kannada"},
82
+ "ko": {"i18n_code": "ko", "name": "Korean"},
83
+ "krc": {"i18n_code": "not_supported", "name": "Karachay-Balkar"},
84
+ "ku": {"i18n_code": "ku", "name": "Kurdish"},
85
+ "kv": {"i18n_code": "not_supported", "name": "Komi"},
86
+ "kw": {"i18n_code": "not_supported", "name": "Cornish"},
87
+ "ky": {"i18n_code": "ky", "name": "Kyrgyz"},
88
+ "la": {"i18n_code": "not_supported", "name": "Latin"},
89
+ "lb": {"i18n_code": "lb", "name": "Luxembourgish"},
90
+ "lez": {"i18n_code": "not_supported", "name": "Lezghian"},
91
+ "li": {"i18n_code": "not_supported", "name": "Limburgish"},
92
+ "lmo": {"i18n_code": "not_supported", "name": "Lombard"},
93
+ "lo": {"i18n_code": "lo", "name": "Lao"},
94
+ "lrc": {"i18n_code": "not_supported", "name": "Northern Luri"},
95
+ "lt": {"i18n_code": "lt", "name": "Lithuanian"},
96
+ "lv": {"i18n_code": "lv", "name": "Latvian"},
97
+ "mai": {"i18n_code": "not_supported", "name": "Maithili"},
98
+ "mg": {"i18n_code": "mg", "name": "Malagasy"},
99
+ "mhr": {"i18n_code": "not_supported", "name": "Meadow Mari"},
100
+ "min": {"i18n_code": "not_supported", "name": "Minangkabau"},
101
+ "mk": {"i18n_code": "mk", "name": "Macedonian"},
102
+ "ml": {"i18n_code": "ml", "name": "Malayalam"},
103
+ "mn": {"i18n_code": "mn", "name": "Mongolian"},
104
+ "mr": {"i18n_code": "mr", "name": "Marathi"},
105
+ "mrj": {"i18n_code": "not_supported", "name": "Hill Mari"},
106
+ "ms": {"i18n_code": "ms", "name": "Malay"},
107
+ "mt": {"i18n_code": "mt", "name": "Maltese"},
108
+ "mwl": {"i18n_code": "not_supported", "name": "Mirandese"},
109
+ "my": {"i18n_code": "my", "name": "Burmese"},
110
+ "myv": {"i18n_code": "not_supported", "name": "Erzya"},
111
+ "mzn": {"i18n_code": "not_supported", "name": "Mazanderani"},
112
+ "nah": {"i18n_code": "not_supported", "name": "Nahuatl"},
113
+ "nap": {"i18n_code": "not_supported", "name": "Neapolitan"},
114
+ "nds": {"i18n_code": "not_supported", "name": "Low German"},
115
+ "ne": {"i18n_code": "ne", "name": "Nepali"},
116
+ "new": {"i18n_code": "not_supported", "name": "Newar"},
117
+ "nl": {"i18n_code": "nl", "name": "Dutch"},
118
+ "nn": {"i18n_code": "not_supported", "name": "Norwegian Nynorsk"},
119
+ "no": {"i18n_code": "not_supported", "name": "Norwegian"},
120
+ "oc": {"i18n_code": "not_supported", "name": "Occitan"},
121
+ "or": {"i18n_code": "or", "name": "Oriya"},
122
+ "os": {"i18n_code": "os", "name": "Ossetian"},
123
+ "pa": {"i18n_code": "pa", "name": "Punjabi"},
124
+ "pam": {"i18n_code": "not_supported", "name": "Pampanga"},
125
+ "pfl": {"i18n_code": "not_supported", "name": "Palatine German"},
126
+ "pl": {"i18n_code": "pl", "name": "Polish"},
127
+ "pms": {"i18n_code": "not_supported", "name": "Piedmontese"},
128
+ "pnb": {"i18n_code": "not_supported", "name": "Western Punjabi"},
129
+ "ps": {"i18n_code": "ps", "name": "Pashto"},
130
+ "pt": {"i18n_code": "pt", "name": "Portuguese"},
131
+ "qu": {"i18n_code": "qu", "name": "Quechua"},
132
+ "rm": {"i18n_code": "rm", "name": "Romansh"},
133
+ "ro": {"i18n_code": "ro", "name": "Romanian"},
134
+ "ru": {"i18n_code": "ru", "name": "Russian"},
135
+ "rue": {"i18n_code": "not_supported", "name": "Rusyn"},
136
+ "sa": {"i18n_code": "sa", "name": "Sanskrit"},
137
+ "sah": {"i18n_code": "not_supported", "name": "Yakut"},
138
+ "sc": {"i18n_code": "not_supported", "name": "Sardinian"},
139
+ "scn": {"i18n_code": "not_supported", "name": "Sicilian"},
140
+ "sco": {"i18n_code": "not_supported", "name": "Scots"},
141
+ "sd": {"i18n_code": "not_supported", "name": "Sindhi"},
142
+ "sh": {"i18n_code": "not_supported", "name": "Serbo-Croatian"},
143
+ "si": {"i18n_code": "si", "name": "Sinhala"},
144
+ "sk": {"i18n_code": "sk", "name": "Slovak"},
145
+ "sl": {"i18n_code": "sl", "name": "Slovenian"},
146
+ "so": {"i18n_code": "so", "name": "Somali"},
147
+ "sq": {"i18n_code": "sq", "name": "Albanian"},
148
+ "sr": {"i18n_code": "sr", "name": "Serbian"},
149
+ "su": {"i18n_code": "not_supported", "name": "Sundanese"},
150
+ "sv": {"i18n_code": "sv", "name": "Swedish"},
151
+ "sw": {"i18n_code": "sw", "name": "Swahili"},
152
+ "ta": {"i18n_code": "ta", "name": "Tamil"},
153
+ "te": {"i18n_code": "te", "name": "Telugu"},
154
+ "tg": {"i18n_code": "tg", "name": "Tajik"},
155
+ "th": {"i18n_code": "th", "name": "Thai"},
156
+ "tk": {"i18n_code": "tk", "name": "Turkmen"},
157
+ "tl": {"i18n_code": "not_supported", "name": "Tagalog"},
158
+ "tr": {"i18n_code": "tr", "name": "Turkish"},
159
+ "tt": {"i18n_code": "not_supported", "name": "Tatar"},
160
+ "tyv": {"i18n_code": "not_supported", "name": "Tuvinian"},
161
+ "ug": {"i18n_code": "ug", "name": "Uyghur"},
162
+ "uk": {"i18n_code": "uk", "name": "Ukrainian"},
163
+ "ur": {"i18n_code": "ur", "name": "Urdu"},
164
+ "uz": {"i18n_code": "uz", "name": "Uzbek"},
165
+ "vec": {"i18n_code": "not_supported", "name": "Venetian"},
166
+ "vep": {"i18n_code": "not_supported", "name": "Veps"},
167
+ "vi": {"i18n_code": "vi", "name": "Vietnamese"},
168
+ "vls": {"i18n_code": "not_supported", "name": "West Flemish"},
169
+ "vo": {"i18n_code": "not_supported", "name": "Volapük"},
170
+ "wa": {"i18n_code": "not_supported", "name": "Walloon"},
171
+ "war": {"i18n_code": "not_supported", "name": "Waray-Waray"},
172
+ "wuu": {"i18n_code": "not_supported", "name": "Wu Chinese"},
173
+ "xal": {"i18n_code": "not_supported", "name": "Kalmyk"},
174
+ "xmf": {"i18n_code": "not_supported", "name": "Mingrelian"},
175
+ "yi": {"i18n_code": "yi", "name": "Yiddish"},
176
+ "yo": {"i18n_code": "yo", "name": "Yoruba"},
177
+ "yue": {"i18n_code": "not_supported", "name": "Cantonese"},
178
+ }