twitter_cldr 4.0.0 → 4.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/Gemfile +3 -2
- data/README.md +18 -2
- data/Rakefile +39 -122
- data/lib/twitter_cldr.rb +3 -0
- data/lib/twitter_cldr/formatters/numbers/rbnf.rb +5 -1
- data/lib/twitter_cldr/resources.rb +86 -5
- data/lib/twitter_cldr/resources/bidi_test_importer.rb +50 -44
- data/lib/twitter_cldr/resources/casefolder_class_generator.rb +22 -13
- data/lib/twitter_cldr/resources/collation_tries_importer.rb +44 -0
- data/lib/twitter_cldr/resources/hyphenation_importer.rb +16 -42
- data/lib/twitter_cldr/resources/import_resolver.rb +71 -0
- data/lib/twitter_cldr/resources/importer.rb +107 -0
- data/lib/twitter_cldr/resources/language_codes_importer.rb +35 -38
- data/lib/twitter_cldr/resources/loader.rb +3 -3
- data/lib/twitter_cldr/resources/locales_resources_importer.rb +48 -35
- data/lib/twitter_cldr/resources/phone_codes_importer.rb +24 -23
- data/lib/twitter_cldr/resources/postal_codes_importer.rb +10 -11
- data/lib/twitter_cldr/resources/properties.rb +0 -4
- data/lib/twitter_cldr/resources/properties/age_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/arabic_shaping_property_importer.rb +9 -11
- data/lib/twitter_cldr/resources/properties/bidi_brackets_property_importer.rb +11 -9
- data/lib/twitter_cldr/resources/properties/blocks_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/derived_core_properties_importer.rb +9 -11
- data/lib/twitter_cldr/resources/properties/east_asian_width_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/grapheme_break_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/hangul_syllable_type_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/indic_positional_category_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/indic_syllabic_category_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/jamo_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/line_break_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/prop_list_importer.rb +9 -11
- data/lib/twitter_cldr/resources/properties/property_importer.rb +13 -22
- data/lib/twitter_cldr/resources/properties/script_extensions_property_importer.rb +12 -10
- data/lib/twitter_cldr/resources/properties/script_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/sentence_break_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/properties/unicode_data_properties_importer.rb +11 -9
- data/lib/twitter_cldr/resources/properties/word_break_property_importer.rb +13 -9
- data/lib/twitter_cldr/resources/rbnf_test_importer.rb +41 -38
- data/lib/twitter_cldr/resources/readme_renderer.rb +1 -2
- data/lib/twitter_cldr/resources/requirements.rb +18 -0
- data/lib/twitter_cldr/resources/requirements/cldr_requirement.rb +66 -0
- data/lib/twitter_cldr/resources/requirements/dependency_requirement.rb +23 -0
- data/lib/twitter_cldr/resources/requirements/git_requirement.rb +66 -0
- data/lib/twitter_cldr/resources/requirements/icu_requirement.rb +111 -0
- data/lib/twitter_cldr/resources/requirements/unicode_requirement.rb +51 -0
- data/lib/twitter_cldr/resources/segment_tests_importer.rb +15 -30
- data/lib/twitter_cldr/resources/tailoring_importer.rb +33 -26
- data/lib/twitter_cldr/resources/transform_test_importer.rb +15 -17
- data/lib/twitter_cldr/resources/uli/segment_exceptions_importer.rb +29 -17
- data/lib/twitter_cldr/resources/unicode_data_importer.rb +38 -31
- data/lib/twitter_cldr/resources/unicode_file_parser.rb +37 -0
- data/lib/twitter_cldr/resources/unicode_property_aliases_importer.rb +23 -27
- data/lib/twitter_cldr/shared/casefolder.rb +139 -115
- data/lib/twitter_cldr/version.rb +1 -1
- data/lib/twitter_cldr/versions.rb +0 -4
- data/resources/collation/tailoring/bo.yml +4 -0
- data/resources/collation/tries/bo.dump +0 -0
- data/resources/locales/bo/calendars.yml +247 -0
- data/resources/locales/bo/currencies.yml +208 -0
- data/resources/locales/bo/fields.yml +31 -0
- data/resources/locales/bo/languages.yml +24 -0
- data/resources/locales/bo/layout.yml +5 -0
- data/resources/locales/bo/lists.yml +12 -0
- data/resources/locales/bo/numbers.yml +111 -0
- data/resources/locales/bo/plural_rules.yml +6 -0
- data/resources/locales/bo/plurals.yml +12 -0
- data/resources/locales/bo/territories.yml +14 -0
- data/resources/locales/bo/units.yml +283 -0
- data/resources/shared/transforms/Arab-Latn.yml +109 -0
- data/resources/shared/transforms/Beng-Deva.yml +13 -0
- data/resources/shared/transforms/Beng-Gujr.yml +13 -0
- data/resources/shared/transforms/Beng-Guru.yml +13 -0
- data/resources/shared/transforms/Beng-Knda.yml +13 -0
- data/resources/shared/transforms/Beng-Latn.yml +13 -0
- data/resources/shared/transforms/Beng-Mlym.yml +13 -0
- data/resources/shared/transforms/Beng-Orya.yml +13 -0
- data/resources/shared/transforms/Beng-Taml.yml +13 -0
- data/resources/shared/transforms/Beng-Telu.yml +13 -0
- data/resources/shared/transforms/Cyrl-Latn.yml +128 -0
- data/resources/shared/transforms/Deva-Beng.yml +13 -0
- data/resources/shared/transforms/Deva-Gujr.yml +13 -0
- data/resources/shared/transforms/Deva-Guru.yml +13 -0
- data/resources/shared/transforms/Deva-Knda.yml +13 -0
- data/resources/shared/transforms/Deva-Latn.yml +13 -0
- data/resources/shared/transforms/Deva-Mlym.yml +13 -0
- data/resources/shared/transforms/Deva-Orya.yml +13 -0
- data/resources/shared/transforms/Deva-Taml.yml +13 -0
- data/resources/shared/transforms/Deva-Telu.yml +13 -0
- data/resources/shared/transforms/Geor-Latn.yml +43 -0
- data/resources/shared/transforms/Grek-Latn-UNGEGN.yml +160 -0
- data/resources/shared/transforms/Grek-Latn.yml +206 -0
- data/resources/shared/transforms/Gujr-Beng.yml +13 -0
- data/resources/shared/transforms/Gujr-Deva.yml +13 -0
- data/resources/shared/transforms/Gujr-Guru.yml +13 -0
- data/resources/shared/transforms/Gujr-Knda.yml +13 -0
- data/resources/shared/transforms/Gujr-Latn.yml +13 -0
- data/resources/shared/transforms/Gujr-Mlym.yml +13 -0
- data/resources/shared/transforms/Gujr-Orya.yml +13 -0
- data/resources/shared/transforms/Gujr-Taml.yml +13 -0
- data/resources/shared/transforms/Gujr-Telu.yml +13 -0
- data/resources/shared/transforms/Guru-Beng.yml +13 -0
- data/resources/shared/transforms/Guru-Deva.yml +13 -0
- data/resources/shared/transforms/Guru-Gujr.yml +13 -0
- data/resources/shared/transforms/Guru-Knda.yml +13 -0
- data/resources/shared/transforms/Guru-Latn.yml +13 -0
- data/resources/shared/transforms/Guru-Mlym.yml +13 -0
- data/resources/shared/transforms/Guru-Orya.yml +13 -0
- data/resources/shared/transforms/Guru-Taml.yml +13 -0
- data/resources/shared/transforms/Guru-Telu.yml +13 -0
- data/resources/shared/transforms/Han-Spacedhan.yml +1 -1
- data/resources/shared/transforms/Hang-Latn.yml +12 -0
- data/resources/shared/transforms/Hani-Latn.yml +1605 -0
- data/resources/shared/transforms/Hans-Hant.yml +3982 -0
- data/resources/shared/transforms/Hebr-Latn.yml +72 -0
- data/resources/shared/transforms/Hira-Kana.yml +114 -0
- data/resources/shared/transforms/Hira-Latn.yml +15 -0
- data/resources/shared/transforms/InterIndic-Latin.yml +2 -2
- data/resources/shared/transforms/Jamo-Latn.yml +12 -0
- data/resources/shared/transforms/Knda-Beng.yml +13 -0
- data/resources/shared/transforms/Knda-Deva.yml +13 -0
- data/resources/shared/transforms/Knda-Gujr.yml +13 -0
- data/resources/shared/transforms/Knda-Guru.yml +13 -0
- data/resources/shared/transforms/Knda-Latn.yml +13 -0
- data/resources/shared/transforms/Knda-Mlym.yml +13 -0
- data/resources/shared/transforms/Knda-Orya.yml +13 -0
- data/resources/shared/transforms/Knda-Taml.yml +13 -0
- data/resources/shared/transforms/Knda-Telu.yml +13 -0
- data/resources/shared/transforms/Latin-ASCII.yml +16 -1
- data/resources/shared/transforms/Latin-InterIndic.yml +2 -2
- data/resources/shared/transforms/Latn-Armn.yml +90 -0
- data/resources/shared/transforms/Latn-Beng.yml +14 -0
- data/resources/shared/transforms/Latn-Bopo.yml +1336 -0
- data/resources/shared/transforms/Latn-Cans.yml +190 -0
- data/resources/shared/transforms/Latn-Deva.yml +14 -0
- data/resources/shared/transforms/Latn-Ethi.yml +278 -0
- data/resources/shared/transforms/Latn-Gujr.yml +14 -0
- data/resources/shared/transforms/Latn-Guru.yml +14 -0
- data/resources/shared/transforms/Latn-Hang.yml +13 -0
- data/resources/shared/transforms/Latn-Jamo.yml +13 -0
- data/resources/shared/transforms/Latn-Kana.yml +274 -0
- data/resources/shared/transforms/Latn-Knda.yml +14 -0
- data/resources/shared/transforms/Latn-Mlym.yml +14 -0
- data/resources/shared/transforms/Latn-Orya.yml +14 -0
- data/resources/shared/transforms/Latn-Taml.yml +14 -0
- data/resources/shared/transforms/Latn-Telu.yml +14 -0
- data/resources/shared/transforms/Latn-Thaa.yml +439 -0
- data/resources/shared/transforms/Latn-Thai.yml +13 -0
- data/resources/shared/transforms/Mlym-Beng.yml +13 -0
- data/resources/shared/transforms/Mlym-Deva.yml +13 -0
- data/resources/shared/transforms/Mlym-Gujr.yml +13 -0
- data/resources/shared/transforms/Mlym-Guru.yml +13 -0
- data/resources/shared/transforms/Mlym-Knda.yml +13 -0
- data/resources/shared/transforms/Mlym-Latn.yml +13 -0
- data/resources/shared/transforms/Mlym-Orya.yml +13 -0
- data/resources/shared/transforms/Mlym-Taml.yml +13 -0
- data/resources/shared/transforms/Mlym-Telu.yml +13 -0
- data/resources/shared/transforms/Orya-Beng.yml +13 -0
- data/resources/shared/transforms/Orya-Deva.yml +13 -0
- data/resources/shared/transforms/Orya-Gujr.yml +13 -0
- data/resources/shared/transforms/Orya-Guru.yml +13 -0
- data/resources/shared/transforms/Orya-Knda.yml +13 -0
- data/resources/shared/transforms/Orya-Latn.yml +13 -0
- data/resources/shared/transforms/Orya-Mlym.yml +13 -0
- data/resources/shared/transforms/Orya-Taml.yml +13 -0
- data/resources/shared/transforms/Orya-Telu.yml +13 -0
- data/resources/shared/transforms/Syrc-Latn.yml +55 -0
- data/resources/shared/transforms/Taml-Beng.yml +13 -0
- data/resources/shared/transforms/Taml-Deva.yml +13 -0
- data/resources/shared/transforms/Taml-Gujr.yml +13 -0
- data/resources/shared/transforms/Taml-Guru.yml +13 -0
- data/resources/shared/transforms/Taml-Knda.yml +13 -0
- data/resources/shared/transforms/Taml-Latn.yml +13 -0
- data/resources/shared/transforms/Taml-Mlym.yml +13 -0
- data/resources/shared/transforms/Taml-Orya.yml +13 -0
- data/resources/shared/transforms/Taml-Telu.yml +13 -0
- data/resources/shared/transforms/Telu-Beng.yml +13 -0
- data/resources/shared/transforms/Telu-Deva.yml +13 -0
- data/resources/shared/transforms/Telu-Gujr.yml +13 -0
- data/resources/shared/transforms/Telu-Guru.yml +13 -0
- data/resources/shared/transforms/Telu-Knda.yml +13 -0
- data/resources/shared/transforms/Telu-Latn.yml +13 -0
- data/resources/shared/transforms/Telu-Mlym.yml +13 -0
- data/resources/shared/transforms/Telu-Orya.yml +13 -0
- data/resources/shared/transforms/Telu-Taml.yml +13 -0
- data/resources/shared/transforms/Thai-Latn.yml +15 -0
- data/resources/shared/transforms/am-am_FONIPA.yml +609 -0
- data/resources/shared/transforms/am-am_Latn-BGN.yml +336 -0
- data/resources/shared/transforms/am-ar.yml +11 -0
- data/resources/shared/transforms/am-fa.yml +10 -0
- data/resources/shared/transforms/ar-ar_Latn-BGN.yml +122 -0
- data/resources/shared/transforms/az_Cyrl-az-BGN.yml +93 -0
- data/resources/shared/transforms/be-be_Latn-BGN.yml +108 -0
- data/resources/shared/transforms/bg-bg_Latn-BGN.yml +99 -0
- data/resources/shared/transforms/ch-am.yml +10 -0
- data/resources/shared/transforms/ch-ar.yml +10 -0
- data/resources/shared/transforms/ch-ch_FONIPA.yml +0 -8
- data/resources/shared/transforms/ch-fa.yml +10 -0
- data/resources/shared/transforms/cs-am.yml +10 -0
- data/resources/shared/transforms/cs-ar.yml +10 -0
- data/resources/shared/transforms/cs-fa.yml +10 -0
- data/resources/shared/transforms/dsb-dsb_FONIPA.yml +0 -5
- data/resources/shared/transforms/dv-dv_Latn-BGN.yml +112 -0
- data/resources/shared/transforms/el-el_Latn-BGN.yml +208 -0
- data/resources/shared/transforms/eo-am.yml +10 -0
- data/resources/shared/transforms/eo-ar.yml +10 -0
- data/resources/shared/transforms/eo-eo_FONIPA.yml +52 -0
- data/resources/shared/transforms/eo-fa.yml +10 -0
- data/resources/shared/transforms/es-ar.yml +13 -0
- data/resources/shared/transforms/es-fa.yml +13 -0
- data/resources/shared/transforms/es_419-am.yml +11 -0
- data/resources/shared/transforms/es_419-ar.yml +14 -0
- data/resources/shared/transforms/es_419-fa.yml +14 -0
- data/resources/shared/transforms/fa-fa_Latn-BGN.yml +123 -0
- data/resources/shared/transforms/he-he_Latn-BGN.yml +62 -0
- data/resources/shared/transforms/hy-am.yml +10 -0
- data/resources/shared/transforms/hy-ar.yml +10 -0
- data/resources/shared/transforms/hy-fa.yml +10 -0
- data/resources/shared/transforms/hy-hy_FONIPA.yml +56 -0
- data/resources/shared/transforms/hy-hy_Latn-BGN.yml +133 -0
- data/resources/shared/transforms/hy_AREVMDA-am.yml +10 -0
- data/resources/shared/transforms/hy_AREVMDA-ar.yml +10 -0
- data/resources/shared/transforms/hy_AREVMDA-fa.yml +10 -0
- data/resources/shared/transforms/hy_AREVMDA-hy_AREVMDA_FONIPA.yml +82 -0
- data/resources/shared/transforms/ia-am.yml +10 -0
- data/resources/shared/transforms/ia-ar.yml +10 -0
- data/resources/shared/transforms/ia-fa.yml +10 -0
- data/resources/shared/transforms/ia-ia_FONIPA.yml +69 -0
- data/resources/shared/transforms/ja_Hrkt-ja_Latn-BGN.yml +310 -0
- data/resources/shared/transforms/ka-ka_Latn-BGN.yml +44 -0
- data/resources/shared/transforms/kk-am.yml +10 -0
- data/resources/shared/transforms/kk-ar.yml +10 -0
- data/resources/shared/transforms/kk-fa.yml +10 -0
- data/resources/shared/transforms/kk-kk_FONIPA.yml +53 -0
- data/resources/shared/transforms/kk-kk_Latn-BGN.yml +136 -0
- data/resources/shared/transforms/ko-ko_Latn-BGN.yml +282 -0
- data/resources/shared/transforms/ky-am.yml +10 -0
- data/resources/shared/transforms/ky-ar.yml +10 -0
- data/resources/shared/transforms/ky-fa.yml +10 -0
- data/resources/shared/transforms/ky-ky_FONIPA.yml +73 -0
- data/resources/shared/transforms/ky-ky_Latn-BGN.yml +107 -0
- data/resources/shared/transforms/la-la_FONIPA.yml +0 -8
- data/resources/shared/transforms/mk-mk_Latn-BGN.yml +89 -0
- data/resources/shared/transforms/mn-mn_Latn-BGN.yml +101 -0
- data/resources/shared/transforms/mn-mn_Latn-MNS.yml +89 -0
- data/resources/shared/transforms/my-am.yml +10 -0
- data/resources/shared/transforms/my-ar.yml +10 -0
- data/resources/shared/transforms/my-fa.yml +10 -0
- data/resources/shared/transforms/my-my_FONIPA.yml +260 -0
- data/resources/shared/transforms/pl-am.yml +10 -0
- data/resources/shared/transforms/pl-ar.yml +10 -0
- data/resources/shared/transforms/pl-fa.yml +10 -0
- data/resources/shared/transforms/ps-ps_Latn-BGN.yml +151 -0
- data/resources/shared/transforms/rm_SURSILV-am.yml +10 -0
- data/resources/shared/transforms/rm_SURSILV-ar.yml +10 -0
- data/resources/shared/transforms/rm_SURSILV-fa.yml +10 -0
- data/resources/shared/transforms/rm_SURSILV-rm_FONIPA_SURSILV.yml +84 -0
- data/resources/shared/transforms/ro-am.yml +10 -0
- data/resources/shared/transforms/ro-ar.yml +10 -0
- data/resources/shared/transforms/ro-fa.yml +10 -0
- data/resources/shared/transforms/ro-ro_FONIPA.yml +38 -6
- data/resources/shared/transforms/ro_FONIPA-ja.yml +1 -0
- data/resources/shared/transforms/ru-ru_Latn-BGN.yml +121 -0
- data/resources/shared/transforms/ru_Latn-ru-BGN.yml +101 -0
- data/resources/shared/transforms/sat-am.yml +10 -0
- data/resources/shared/transforms/sat-ar.yml +10 -0
- data/resources/shared/transforms/sat-fa.yml +10 -0
- data/resources/shared/transforms/sat_Olck-sat_FONIPA.yml +132 -0
- data/resources/shared/transforms/si-am.yml +10 -0
- data/resources/shared/transforms/si-ar.yml +10 -0
- data/resources/shared/transforms/si-fa.yml +10 -0
- data/resources/shared/transforms/si-si_FONIPA.yml +128 -0
- data/resources/shared/transforms/si-si_Latn.yml +96 -0
- data/resources/shared/transforms/sk-am.yml +10 -0
- data/resources/shared/transforms/sk-ar.yml +10 -0
- data/resources/shared/transforms/sk-fa.yml +10 -0
- data/resources/shared/transforms/sk-sk_FONIPA.yml +18 -2
- data/resources/shared/transforms/sk_FONIPA-ja.yml +2 -0
- data/resources/shared/transforms/sr-sr_Latn-BGN.yml +81 -0
- data/resources/shared/transforms/tk_Cyrl-tk-BGN.yml +122 -0
- data/resources/shared/transforms/tlh-am.yml +10 -0
- data/resources/shared/transforms/tlh-ar.yml +10 -0
- data/resources/shared/transforms/tlh-fa.yml +10 -0
- data/resources/shared/transforms/tlh-tlh_FONIPA.yml +0 -8
- data/resources/shared/transforms/uk-uk_Latn-BGN.yml +115 -0
- data/resources/shared/transforms/und_FONIPA-ar.yml +96 -0
- data/resources/shared/transforms/und_FONIPA-fa.yml +88 -0
- data/resources/shared/transforms/und_FONIPA-und_FONXSAMP.yml +198 -0
- data/resources/shared/transforms/uz_Cyrl-uz-BGN.yml +117 -0
- data/resources/shared/transforms/xh-am.yml +10 -0
- data/resources/shared/transforms/xh-ar.yml +10 -0
- data/resources/shared/transforms/xh-fa.yml +10 -0
- data/resources/shared/transforms/xh-xh_FONIPA.yml +71 -0
- data/resources/shared/transforms/zu-am.yml +10 -0
- data/resources/shared/transforms/zu-ar.yml +10 -0
- data/resources/shared/transforms/zu-fa.yml +10 -0
- data/resources/shared/transforms/zu-zu_FONIPA.yml +58 -0
- data/spec/formatters/numbers/rbnf/rbnf_spec.rb +3 -1
- data/spec/resources/loader_spec.rb +12 -5
- data/spec/spec_helper.rb +1 -1
- metadata +242 -10
- data/History.txt +0 -282
- data/lib/twitter_cldr/resources/collation_tries_dumper.rb +0 -43
- data/lib/twitter_cldr/resources/custom_locales_resources_importer.rb +0 -80
- data/lib/twitter_cldr/resources/download.rb +0 -64
- data/lib/twitter_cldr/resources/icu_based_importer.rb +0 -18
- data/lib/twitter_cldr/resources/properties/properties_importer.rb +0 -59
- data/lib/twitter_cldr/resources/unicode_importer.rb +0 -37
@@ -0,0 +1,84 @@
|
|
1
|
+
---
|
2
|
+
:transforms:
|
3
|
+
-
|
4
|
+
:direction: forward
|
5
|
+
:rules:
|
6
|
+
- "::Lower;"
|
7
|
+
- "ai > aɪ̯ ;"
|
8
|
+
- "au > aʊ̯ ;"
|
9
|
+
- "a > a ;"
|
10
|
+
- "b > b ;"
|
11
|
+
- "{c} [ei] > t͡s ;"
|
12
|
+
- "c > k ;"
|
13
|
+
- "d > d ;"
|
14
|
+
- "ei > ɛɪ̯ ; # can also be /aɪ̯/ or /ɔɪ̯/ in some regions"
|
15
|
+
- "eu > ɛʊ̯ ;"
|
16
|
+
- "e > e ; # can also be /ɛ/; needs a dictionary"
|
17
|
+
- "é > e ;"
|
18
|
+
- "è > ɛ ;"
|
19
|
+
- "f > f ;"
|
20
|
+
- "ge > ɟ ;"
|
21
|
+
- "gh > ɡ ;"
|
22
|
+
- "gi > ɟ ;"
|
23
|
+
- "{gl} [aeou] > ɡl ;"
|
24
|
+
- "{gl} > ʎ ;"
|
25
|
+
- "gn > ɲ ;"
|
26
|
+
- "g > ɡ ;"
|
27
|
+
- "h > ;"
|
28
|
+
- "iau > ɪa̯ʊ̯ ;"
|
29
|
+
- "ia > ɪa̯ ;"
|
30
|
+
- "ie > ɪɛ̯ ;"
|
31
|
+
- "iu > ɪʊ̯ ;"
|
32
|
+
- "i > i ;"
|
33
|
+
- "j > j ;"
|
34
|
+
- "k > k ;"
|
35
|
+
- "l > l ;"
|
36
|
+
- "m > m ;"
|
37
|
+
- "n > n ;"
|
38
|
+
- "o > ɔ ;"
|
39
|
+
- "p > p ;"
|
40
|
+
- "q > k ;"
|
41
|
+
- "r > r ;"
|
42
|
+
- "{sch} [aeiou] > ʒ ; # can also be /ʃ/; needs a dictionary"
|
43
|
+
- "{sch} > ʃ ; # can also be /ʒ/; needs a dictionary"
|
44
|
+
- "{s} [cptnm] > ʃ ;"
|
45
|
+
- "{s} [gbdv] > ʒ ;"
|
46
|
+
- "s > s ; # can also be /z/; needs a dictionary"
|
47
|
+
- "tg > c ;"
|
48
|
+
- "tsch > t͡ʃ ;"
|
49
|
+
- "t > t ;"
|
50
|
+
- "uau > ʊa̯ʊ̯ ;"
|
51
|
+
- "ua > ʊa̯ ;"
|
52
|
+
- "uei > ʊɛ̯ɪ̯ ;"
|
53
|
+
- "ue > ʊɛ̯ ;"
|
54
|
+
- "uo > ʊɔ̯ ;"
|
55
|
+
- "u > u ;"
|
56
|
+
- "v > v ;"
|
57
|
+
- "w > v ;"
|
58
|
+
- "x > ks ;"
|
59
|
+
- "y > i ;"
|
60
|
+
- "z > t͡s ;"
|
61
|
+
- "::NULL;"
|
62
|
+
- "mm+ > mː;"
|
63
|
+
- "nn+ > nː;"
|
64
|
+
- "ɲɲ+ > ɲː;"
|
65
|
+
- "pp+ > pː;"
|
66
|
+
- "bb+ > bː;"
|
67
|
+
- "tt+ > tː;"
|
68
|
+
- "dd+ > dː;"
|
69
|
+
- "cc+ > cː ;"
|
70
|
+
- "ɟɟ+ > ɟː ;"
|
71
|
+
- "kk+ > kː ;"
|
72
|
+
- "ɡɡ+ > ɡː ;"
|
73
|
+
- "ff+ > fː ;"
|
74
|
+
- "vv+ > vː ;"
|
75
|
+
- "ss+ > sː ;"
|
76
|
+
- "zz+ > zː ;"
|
77
|
+
- "ʃʃ+ > ʃː ;"
|
78
|
+
- "ʒʒ+ > ʒː ;"
|
79
|
+
- "rr+ > rː ;"
|
80
|
+
- "ll+ > lː ;"
|
81
|
+
- "jj+ > jː ;"
|
82
|
+
:source: rm_SURSILV
|
83
|
+
:target: rm_FONIPA_SURSILV
|
84
|
+
:variant: ~
|
@@ -3,6 +3,10 @@
|
|
3
3
|
-
|
4
4
|
:direction: forward
|
5
5
|
:rules:
|
6
|
+
- "$VowelEI = [e i î];"
|
7
|
+
- "$VowelAOU = [a â ă o u];"
|
8
|
+
- "$Vowel = [$VowelEI $VowelAOU];"
|
9
|
+
- "$Boundary = [^[:L:][:M:][:N:]];"
|
6
10
|
- ":: NFC () ;"
|
7
11
|
- ":: Lower () ;"
|
8
12
|
- "eoai > eo̯aj ; # eg. leoaică > /leo̯ajkə/, not /le̯o̯aikə/"
|
@@ -14,7 +18,7 @@
|
|
14
18
|
- "iau > jaw ;"
|
15
19
|
- "iei > jej ;"
|
16
20
|
- "ieu > jew ;"
|
17
|
-
- "
|
21
|
+
- "$Boundary {eu} > jew ;"
|
18
22
|
- "ioa > jo̯a ;"
|
19
23
|
- "ioi > joj ;"
|
20
24
|
- "i\\-oi > joj ;"
|
@@ -26,7 +30,7 @@
|
|
26
30
|
- "ai > aj ;"
|
27
31
|
- "âi > ɨj ;"
|
28
32
|
- "ăi > əj ;"
|
29
|
-
- "au
|
33
|
+
- "au} r > au ;"
|
30
34
|
- "au > aw ;"
|
31
35
|
- "âu > ɨw ;"
|
32
36
|
- "ău > əw ;"
|
@@ -41,7 +45,7 @@
|
|
41
45
|
- "ie > je ;"
|
42
46
|
- "ii > ij ;"
|
43
47
|
- "io > jo ;"
|
44
|
-
- "iu
|
48
|
+
- "iu} [$Vowel $Boundary] > iw ;"
|
45
49
|
- "iu > ju ;"
|
46
50
|
- "oa > o̯a ;"
|
47
51
|
- "oi > oj ;"
|
@@ -57,13 +61,15 @@
|
|
57
61
|
- "ă > ə ;"
|
58
62
|
- "b > b ;"
|
59
63
|
- "ch > k ;"
|
60
|
-
- "c
|
64
|
+
- "{c [ei]} $VowelAOU > t͡ʃ ;"
|
65
|
+
- "{c} [ei] > t͡ʃ ;"
|
61
66
|
- "c > k ;"
|
62
67
|
- "d > d ;"
|
63
68
|
- "e > e ;"
|
64
69
|
- "f > f ;"
|
65
70
|
- "gh > ɡ ;"
|
66
|
-
- "g
|
71
|
+
- "{g [ei]} $VowelAOU > d͡ʒ ;"
|
72
|
+
- "{g} [ei] > d͡ʒ ;"
|
67
73
|
- "g > ɡ ;"
|
68
74
|
- "h > h ;"
|
69
75
|
- "i > i ;"
|
@@ -72,17 +78,43 @@
|
|
72
78
|
- "k > k ;"
|
73
79
|
- "l > l ;"
|
74
80
|
- "m > m ;"
|
81
|
+
- "ng > ŋ ;"
|
75
82
|
- "n > n ;"
|
76
83
|
- "o > o ;"
|
77
84
|
- "p > p ;"
|
85
|
+
- "q > k ;"
|
78
86
|
- "r > r ;"
|
79
87
|
- "s > s ;"
|
80
88
|
- "ş > ʃ ;"
|
89
|
+
- "ș > ʃ ;"
|
81
90
|
- "t > t ;"
|
82
|
-
- "ţ > t
|
91
|
+
- "ţ > t͡s ;"
|
92
|
+
- "ț > t͡s ;"
|
83
93
|
- "u > u ;"
|
84
94
|
- "v > v ;"
|
95
|
+
- "x > ks ;"
|
96
|
+
- "y > i ;"
|
85
97
|
- "z > z ;"
|
98
|
+
- "[:P:]+ > ' ';"
|
99
|
+
- "::null;"
|
100
|
+
- "pp+ > p;"
|
101
|
+
- "bb+ > b;"
|
102
|
+
- "tt+ > t;"
|
103
|
+
- "dd+ > d;"
|
104
|
+
- "kk+ > k;"
|
105
|
+
- "dd+ > d;"
|
106
|
+
- "ɡɡ+ > ɡ;"
|
107
|
+
- "ff+ > f;"
|
108
|
+
- "vv+ > v;"
|
109
|
+
- "hh+ > h;"
|
110
|
+
- "ss+ > s;"
|
111
|
+
- "zz+ > z;"
|
112
|
+
- "ʃʃ+ > ʃ;"
|
113
|
+
- "ʒʒ+ > ʒ;"
|
114
|
+
- "rr+ > r;"
|
115
|
+
- "ll+ > l;"
|
116
|
+
- "jj+ > j;"
|
117
|
+
- "ww+ > w;"
|
86
118
|
:source: ro
|
87
119
|
:target: ro_FONIPA
|
88
120
|
:variant: ~
|
@@ -0,0 +1,121 @@
|
|
1
|
+
---
|
2
|
+
:transforms:
|
3
|
+
-
|
4
|
+
:direction: forward
|
5
|
+
:rules:
|
6
|
+
- "::[АБВГДЕЁЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯабвгдеёжзийклмнопрстуфхцчшщъыьэюя];"
|
7
|
+
- "$prime = ʹ ;"
|
8
|
+
- "$doublePrime = ʺ ;"
|
9
|
+
- "$wordBoundary = [^[:L:][:M:][:N:]] ;"
|
10
|
+
- "$upperConsonants = [БВГДЖЙКЛМНПРСТФХЦЧШЩЭ] ;"
|
11
|
+
- "$lowerConsonants = [бвгджйклмнпрстфхцчшщэ] ;"
|
12
|
+
- "$consonants = [$upperConsonants $lowerConsonants] ;"
|
13
|
+
- "$upperVowels = [АЕЁЭИОУЫЮЯ] ;"
|
14
|
+
- "$lowerVowels = [аеёэиоуыюя] ;"
|
15
|
+
- "$vowels = [$upperVowels $lowerVowels] ;"
|
16
|
+
- "$lower = [$lowerConsonants $lowerVowels] ;"
|
17
|
+
- "$upper = [$upperConsonants $upperVowels] ;"
|
18
|
+
- "$lowerVowels { ы > ·y ;"
|
19
|
+
- "$upperVowels { [Ыы] > ·Y ;"
|
20
|
+
- "[$consonants - [Йй]]{Э > ·E ;"
|
21
|
+
- "[$consonants - [Йй]]{э > ·e ;"
|
22
|
+
- "[$upperVowels [ЙЪЬ]] { Е } $upper > YE ; # CYRILLIC CAPITAL LETTER IE"
|
23
|
+
- "[$upperVowels [ЙЪЬ]] { Е > Ye ; # CYRILLIC CAPITAL LETTER IE"
|
24
|
+
- "[$upperVowels $lowerVowels [ЙйЪъЬь]] { е > ye ; # CYRILLIC SMALL LETTER IE"
|
25
|
+
- "[$upperVowels [ЙЪЬ]] { Ё } $upper > YË ; # CYRILLIC CAPITAL LETTER IO"
|
26
|
+
- "[$upperVowels [ЙЪЬ]] { Ё > Yë ; # CYRILLIC CAPITAL LETTER IO"
|
27
|
+
- "[$upperVowels $lowerVowels [ЙйЪъЬь]] { ё > yë ; # CYRILLIC SMALL LETTER IO"
|
28
|
+
- "::Null;"
|
29
|
+
- "А > A ; # CYRILLIC CAPITAL LETTER A"
|
30
|
+
- "а > a ; # CYRILLIC SMALL LETTER A"
|
31
|
+
- "Б > B ; # CYRILLIC CAPITAL LETTER BE"
|
32
|
+
- "б > b ; # CYRILLIC SMALL LETTER BE"
|
33
|
+
- "В > V ; # CYRILLIC CAPITAL LETTER VE"
|
34
|
+
- "в > v ; # CYRILLIC SMALL LETTER VE"
|
35
|
+
- "Г > G ; # CYRILLIC CAPITAL LETTER GHE"
|
36
|
+
- "г > g ; # CYRILLIC SMALL LETTER GHE"
|
37
|
+
- "Д > D ; # CYRILLIC CAPITAL LETTER DE"
|
38
|
+
- "д > d ; # CYRILLIC SMALL LETTER DE"
|
39
|
+
- "$wordBoundary{Е} $upper > YE ; # CYRILLIC CAPITAL LETTER IE"
|
40
|
+
- "$wordBoundary{Е > Ye ; # CYRILLIC CAPITAL LETTER IE"
|
41
|
+
- "Е > E ; # CYRILLIC CAPITAL LETTER IE"
|
42
|
+
- "$wordBoundary{е > ye ; # CYRILLIC SMALL LETTER IE"
|
43
|
+
- "е > e ; # CYRILLIC SMALL LETTER IE"
|
44
|
+
- "$wordBoundary {Ё} $upper > YË ; # CYRILLIC CAPITAL LETTER IO"
|
45
|
+
- "$wordBoundary {Ё} $lower > Yë ; # CYRILLIC CAPITAL LETTER IO"
|
46
|
+
- "Ё > Ë ; # CYRILLIC CAPITAL LETTER IO"
|
47
|
+
- "$wordBoundary{ё > yë ; # CYRILLIC SMALL LETTER IO"
|
48
|
+
- "ё > ë ; # CYRILLIC SMALL LETTER IO"
|
49
|
+
- "Ж} $lower > Zh ; # CYRILLIC CAPITAL LETTER ZHE"
|
50
|
+
- "Ж > ZH ; # CYRILLIC CAPITAL LETTER ZHE"
|
51
|
+
- "ж > zh ; # CYRILLIC SMALL LETTER ZHE"
|
52
|
+
- "З > Z ; # CYRILLIC CAPITAL LETTER ZE"
|
53
|
+
- "з > z ; # CYRILLIC SMALL LETTER ZE"
|
54
|
+
- "И > I ; # CYRILLIC CAPITAL LETTER I"
|
55
|
+
- "и > i ; # CYRILLIC SMALL LETTER I"
|
56
|
+
- "Й}[АаУуЫыЭэ] > Y· ; # CYRILLIC CAPITAL LETTER I"
|
57
|
+
- "й}[АаУуЫыЭэ] > y· ; # CYRILLIC SMALL LETTER I"
|
58
|
+
- "Й > Y ; # CYRILLIC CAPITAL LETTER I"
|
59
|
+
- "й > y ; # CYRILLIC SMALL LETTER I"
|
60
|
+
- "К > K ; # CYRILLIC CAPITAL LETTER KA"
|
61
|
+
- "к > k ; # CYRILLIC SMALL LETTER KA"
|
62
|
+
- "Л > L ; # CYRILLIC CAPITAL LETTER EL"
|
63
|
+
- "л > l ; # CYRILLIC SMALL LETTER EL"
|
64
|
+
- "М > M ; # CYRILLIC CAPITAL LETTER EM"
|
65
|
+
- "м > m ; # CYRILLIC SMALL LETTER EM"
|
66
|
+
- "Н > N ; # CYRILLIC CAPITAL LETTER EN"
|
67
|
+
- "н > n ; # CYRILLIC SMALL LETTER EN"
|
68
|
+
- "О > O ; # CYRILLIC CAPITAL LETTER O"
|
69
|
+
- "о > o ; # CYRILLIC SMALL LETTER O"
|
70
|
+
- "П > P ; # CYRILLIC CAPITAL LETTER PE"
|
71
|
+
- "п > p ; # CYRILLIC SMALL LETTER PE"
|
72
|
+
- "Р > R ; # CYRILLIC CAPITAL LETTER ER"
|
73
|
+
- "р > r ; # CYRILLIC SMALL LETTER ER"
|
74
|
+
- "С > S ; # CYRILLIC CAPITAL LETTER ES"
|
75
|
+
- "с > s ; # CYRILLIC SMALL LETTER ES"
|
76
|
+
- "ТС > T·S ; # CYRILLIC CAPITAL LETTER TE"
|
77
|
+
- "Тс > T·s ; # CYRILLIC CAPITAL LETTER TE"
|
78
|
+
- "тс > t·s ; # CYRILLIC SMALL LETTER TE"
|
79
|
+
- "Т > T ; # CYRILLIC CAPITAL LETTER TE"
|
80
|
+
- "т > t ; # CYRILLIC SMALL LETTER TE"
|
81
|
+
- "У > U ; # CYRILLIC CAPITAL LETTER U"
|
82
|
+
- "у > u ; # CYRILLIC SMALL LETTER U"
|
83
|
+
- "Ф > F ; # CYRILLIC CAPITAL LETTER EF"
|
84
|
+
- "ф > f ; # CYRILLIC SMALL LETTER EF"
|
85
|
+
- "Х} $lower > Kh ; # CYRILLIC CAPITAL LETTER HA"
|
86
|
+
- "Х > KH ; # CYRILLIC CAPITAL LETTER HA"
|
87
|
+
- "х > kh ; # CYRILLIC SMALL LETTER HA"
|
88
|
+
- "Ц} $lower > Ts ; # CYRILLIC CAPITAL LETTER TSE"
|
89
|
+
- "Ц > TS ; # CYRILLIC CAPITAL LETTER TSE"
|
90
|
+
- "ц > ts ; # CYRILLIC SMALL LETTER TSE"
|
91
|
+
- "Ч} $lower > Ch ; # CYRILLIC CAPITAL LETTER CHE"
|
92
|
+
- "Ч > CH ; # CYRILLIC CAPITAL LETTER CHE"
|
93
|
+
- "ч > ch ; # CYRILLIC SMALL LETTER CHE"
|
94
|
+
- "ШЧ > SH·CH ; # CYRILLIC CAPITAL LETTER SHA"
|
95
|
+
- "Шч > Sh·ch ; # CYRILLIC CAPITAL LETTER SHA"
|
96
|
+
- "шч > sh·ch ; # CYRILLIC SMALL LETTER SHA"
|
97
|
+
- "Ш} $lower > Sh ; # CYRILLIC CAPITAL LETTER SHA"
|
98
|
+
- "Ш > SH ; # CYRILLIC CAPITAL LETTER SHA"
|
99
|
+
- "ш > sh ; # CYRILLIC SMALL LETTER SHA"
|
100
|
+
- "Щ} $lower > Shch ; # CYRILLIC CAPITAL LETTER SHCHA"
|
101
|
+
- "Щ > SHCH ; # CYRILLIC CAPITAL LETTER SHCHA"
|
102
|
+
- "щ > shch ; # CYRILLIC SMALL LETTER SHCHA"
|
103
|
+
- "Ъ > $doublePrime ; # CYRILLIC CAPITAL LETTER HARD SIGN"
|
104
|
+
- "ъ > $doublePrime ; # CYRILLIC SMALL LETTER HARD SIGN"
|
105
|
+
- "Ы}[АаУуЫыЭэ] > Y· ; # CYRILLIC CAPITAL LETTER YERU"
|
106
|
+
- "ы}[ауыэ] > y· ; # CYRILLIC SMALL LETTER YERU"
|
107
|
+
- "Ы > Y ; # CYRILLIC CAPITAL LETTER YERU"
|
108
|
+
- "ы > y ; # CYRILLIC SMALL LETTER YERU"
|
109
|
+
- "Ь > $prime ; # CYRILLIC CAPITAL LETTER SOFT SIGN"
|
110
|
+
- "ь > $prime ; # CYRILLIC SMALL LETTER SOFT SIGN"
|
111
|
+
- "Э > E ; # CYRILLIC CAPITAL LETTER E"
|
112
|
+
- "э > e ; # CYRILLIC SMALL LETTER E"
|
113
|
+
- "Ю} $lower > Yu ; # CYRILLIC CAPITAL LETTER YU"
|
114
|
+
- "Ю > YU ; # CYRILLIC CAPITAL LETTER YU"
|
115
|
+
- "ю > yu ; # CYRILLIC SMALL LETTER YU"
|
116
|
+
- "Я} $lower > Ya ; # CYRILLIC CAPITAL LETTER YA"
|
117
|
+
- "Я > YA ; # CYRILLIC CAPITAL LETTER YA"
|
118
|
+
- "я > ya ; # CYRILLIC SMALL LETTER YA"
|
119
|
+
:source: ru
|
120
|
+
:target: ru_Latn
|
121
|
+
:variant: BGN
|
@@ -0,0 +1,101 @@
|
|
1
|
+
---
|
2
|
+
:transforms:
|
3
|
+
-
|
4
|
+
:direction: forward
|
5
|
+
:rules:
|
6
|
+
- "$prime = ʹ;"
|
7
|
+
- "$doublePrime = ʺ;"
|
8
|
+
- "$wordBoundary = [^[:L:][:M:][:N:]];"
|
9
|
+
- "$upperConsonant = [БВГДЖЙКЛМНПРСТФХЦЧШЩЭ];"
|
10
|
+
- "$lowerConsonant = [бвгджйклмнпрстфхцчшщэ];"
|
11
|
+
- "$consonant = [$upperConsonant $lowerConsonant];"
|
12
|
+
- "::NFC;"
|
13
|
+
- "[:Upper:] {$prime} [^[:Lower:]] > Ь;"
|
14
|
+
- "$prime > ь;"
|
15
|
+
- "[:Upper:] {$doublePrime} [^[:Lower:]] > Ъ;"
|
16
|
+
- "$doublePrime > ъ;"
|
17
|
+
- "K[Hh] > Х;"
|
18
|
+
- "k[Hh] > х;"
|
19
|
+
- "T·S > ТС;"
|
20
|
+
- "T·s > Тс;"
|
21
|
+
- "t·S > тС;"
|
22
|
+
- "t·s > тс;"
|
23
|
+
- "T[Ss] > Ц;"
|
24
|
+
- "t[Ss] > ц;"
|
25
|
+
- "C[Hh] > Ч;"
|
26
|
+
- "c[Hh] > ч;"
|
27
|
+
- "S[Hh]·C[Hh] > ШЧ;"
|
28
|
+
- "S[Hh]·c[Hh] > Шч;"
|
29
|
+
- "s[Hh]·C[Hh] > шЧ;"
|
30
|
+
- "s[Hh]·c[Hh] > шч;"
|
31
|
+
- "S[Hh][Cc][Hh] > Щ;"
|
32
|
+
- "s[Hh][Cc][Hh] > щ;"
|
33
|
+
- "S[Hh] > Ш;"
|
34
|
+
- "s[Hh] > ш;"
|
35
|
+
- "Y[Ee] > Е;"
|
36
|
+
- "y[Ee] > е;"
|
37
|
+
- "Y[Ëë] > Ё;"
|
38
|
+
- "y[Ëë] > ё;"
|
39
|
+
- "Y[Uu] > Ю;"
|
40
|
+
- "y[Uu] > ю;"
|
41
|
+
- "Y[Aa] > Я;"
|
42
|
+
- "y[Aa] > я;"
|
43
|
+
- "{yy} $wordBoundary > ый;"
|
44
|
+
- "$wordBoundary {Y} [^aeëiouyAEËIOUY] > Ы;"
|
45
|
+
- "$wordBoundary {y} [^aeëiouyAEËIOUY] > ы;"
|
46
|
+
- "$consonant {Y} > Ы;"
|
47
|
+
- "$consonant {y} > ы;"
|
48
|
+
- "Y > Й;"
|
49
|
+
- "y > й;"
|
50
|
+
- "$wordBoundary {E} > Э;"
|
51
|
+
- "$wordBoundary {e} > э;"
|
52
|
+
- "·E > Э;"
|
53
|
+
- "·e > э;"
|
54
|
+
- "E > Е;"
|
55
|
+
- "e > е;"
|
56
|
+
- "A > А;"
|
57
|
+
- "a > а;"
|
58
|
+
- "B > Б;"
|
59
|
+
- "b > б;"
|
60
|
+
- "V > В;"
|
61
|
+
- "v > в;"
|
62
|
+
- "G > Г;"
|
63
|
+
- "g > г;"
|
64
|
+
- "D > Д;"
|
65
|
+
- "d > д;"
|
66
|
+
- "Ë > Ё;"
|
67
|
+
- "ë > ё;"
|
68
|
+
- "Z[Hh] > Ж;"
|
69
|
+
- "z[Hh] > ж;"
|
70
|
+
- "Z > З;"
|
71
|
+
- "z > з;"
|
72
|
+
- "I > И;"
|
73
|
+
- "i > и;"
|
74
|
+
- "K > К;"
|
75
|
+
- "k > к;"
|
76
|
+
- "L > Л;"
|
77
|
+
- "l > л;"
|
78
|
+
- "M > М;"
|
79
|
+
- "m > м;"
|
80
|
+
- "N > Н;"
|
81
|
+
- "n > н;"
|
82
|
+
- "O > О;"
|
83
|
+
- "o > о;"
|
84
|
+
- "P > П;"
|
85
|
+
- "p > п;"
|
86
|
+
- "R > Р;"
|
87
|
+
- "r > р;"
|
88
|
+
- "S > С;"
|
89
|
+
- "s > с;"
|
90
|
+
- "T > Т;"
|
91
|
+
- "t > т;"
|
92
|
+
- "U > У;"
|
93
|
+
- "u > у;"
|
94
|
+
- "F > Ф;"
|
95
|
+
- "f > ф;"
|
96
|
+
- "·Y > Ы;"
|
97
|
+
- "·y > ы;"
|
98
|
+
- "· > ;"
|
99
|
+
:source: ru_Latn
|
100
|
+
:target: ru
|
101
|
+
:variant: BGN
|