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 +1,62 @@
|
|
1
|
+
---
|
2
|
+
:transforms:
|
3
|
+
-
|
4
|
+
:direction: forward
|
5
|
+
:rules:
|
6
|
+
- ":: [ ְֱֲֳִֵֶַָֹֻּׁׂאבגדהוזחטיךכלםמןנסעףפץצקרשת׳] ;"
|
7
|
+
- ":: NFD (NFC) ;"
|
8
|
+
- "$alef = ’;"
|
9
|
+
- "$ayin = ‘;"
|
10
|
+
- "$wordBoundary = [^[:L:][:M:][:N:]] ;"
|
11
|
+
- "בּ > b ; # HEBREW LETTER BET + DAGESH"
|
12
|
+
- "פּ > P ; # HEBREW LETTER PE + DAGESH"
|
13
|
+
- "גּ > g ; # HEBREW LETTER GIMEL + DAGESH"
|
14
|
+
- "ג׳ > ǧ ; # HEBREW LETTER GIMEL + GERESH"
|
15
|
+
- "וּ > u ; # HEBREW LETTER VAV + POINT DAGESH"
|
16
|
+
- "וֹ > o ; # HEBREW LETTER VAV + POINT HOLAM"
|
17
|
+
- "צ׳ > č ; # HEBREW LETTER TSADI + GERESH"
|
18
|
+
- "ז׳ > ž ; # HEBREW LETTER ZAYIN + GERESH"
|
19
|
+
- "דּ > d ; # HEBREW LETTER DALET + DAGESH"
|
20
|
+
- "הּ > h ; # HEBREW LETTER HE + DAGESH"
|
21
|
+
- "ךּ > k ; # HEBREW LETTER FINAL KAF + DAGESH"
|
22
|
+
- "כּ > k ; # HEBREW LETTER KAF + DAGESH"
|
23
|
+
- "ךְ > kh ; # HEBREW LETTER FINAL KAF + SHEVA"
|
24
|
+
- "תּ > t ; # HEBREW LETTER TAV + DAGESH"
|
25
|
+
- "א > $alef ; # HEBREW LETTER ALEF"
|
26
|
+
- "ב > v ; # HEBREW LETTER BET"
|
27
|
+
- "ג > g ; # HEBREW LETTER GIMEL"
|
28
|
+
- "ד > d ; # HEBREW LETTER DALET"
|
29
|
+
- "ה > h ; # HEBREW LETTER HE"
|
30
|
+
- "ח > ẖ ; # HEBREW LETTER HET"
|
31
|
+
- "ו > w ; # HEBREW LETTER VAV"
|
32
|
+
- "ז > z ; # HEBREW LETTER ZAYIN"
|
33
|
+
- "[טת] > t ; # HEBREW LETTER TET"
|
34
|
+
- "י > y ; # HEBREW LETTER YOD"
|
35
|
+
- "[כך] > kh ; # HEBREW LETTER KAF and FINAL KAF"
|
36
|
+
- "ל > l ; # HEBREW LETTER LAMED"
|
37
|
+
- "[מם] > m ; # HEBREW LETTER MEM and FINAL MEM"
|
38
|
+
- "[נן] > n ; # HEBREW LETTER NUN and FINAL NUN"
|
39
|
+
- "ס > s ; # HEBREW LETTER SAMEKH"
|
40
|
+
- "ע > $ayin ; # HEBREW LETTER AYIN"
|
41
|
+
- "[פף] > f ; # HEBREW LETTER PE and FINAL PE"
|
42
|
+
- "[צץ] > ẕ ; # HEBREW LETTER TSADI and FINAL TSADI"
|
43
|
+
- "ק > q ; # HEBREW LETTER QOF"
|
44
|
+
- "ר > r ; # HEBREW LETTER RESH"
|
45
|
+
- "שׁ > sh ; # HEBREW LETTER SHIN"
|
46
|
+
- "שׂ > s ; # HEBREW LETTER SHIN"
|
47
|
+
- "ַ > a ; # HEBREW POINT PATAH"
|
48
|
+
- "ֲ > a ; # HEBREW POINT HATAF PATAH"
|
49
|
+
- "ָ > o ; # HEBREW POINT QAMATS"
|
50
|
+
- "ֶ > e ; # HEBREW POINT SEGOL"
|
51
|
+
- "ֱ > e ; # HEBREW POINT HATAF SEGOL"
|
52
|
+
- "ֵי > e ; # HEBREW POINT TSERE + LETTER YOD"
|
53
|
+
- "ֵ > e ; # HEBREW POINT TSERE"
|
54
|
+
- "ְ > e ; # HEBREW POINT SHEVA"
|
55
|
+
- "ִי > i ; # HEBREW POINT HIRIQ + LETTER YOD"
|
56
|
+
- "ִ > i ; # HEBREW POINT HIRIQ"
|
57
|
+
- "ֳ > o ; # HEBREW LETTER HATAF QAMATS"
|
58
|
+
- "ֹ > o ; # HEBREW POINT HOLAM"
|
59
|
+
- "ֻ > u ; # HEBREW POINT QUBUTS"
|
60
|
+
:source: he
|
61
|
+
:target: he_Latn
|
62
|
+
:variant: BGN
|
@@ -0,0 +1,56 @@
|
|
1
|
+
---
|
2
|
+
:transforms:
|
3
|
+
-
|
4
|
+
:direction: forward
|
5
|
+
:rules:
|
6
|
+
- "::lower();"
|
7
|
+
- "$wordBoundary = [^[:L:][:M:][:N:]];"
|
8
|
+
- "$vowel = [աեէըիոևօւ];"
|
9
|
+
- "ով > ov;"
|
10
|
+
- "մ > m;"
|
11
|
+
- "ն > n;"
|
12
|
+
- "պ > p;"
|
13
|
+
- "տ > t;"
|
14
|
+
- "կ > k;"
|
15
|
+
- "բ > b;"
|
16
|
+
- "դ > d;"
|
17
|
+
- "գ > ɡ;"
|
18
|
+
- "փ > pʰ;"
|
19
|
+
- "թ > tʰ;"
|
20
|
+
- "ք > kʰ;"
|
21
|
+
- "ծ > t͡s;"
|
22
|
+
- "ճ > t͡ʃ;"
|
23
|
+
- "ձ > d͡z;"
|
24
|
+
- "ջ > d͡ʒ;"
|
25
|
+
- "ց > t͡sʰ;"
|
26
|
+
- "չ > t͡ʃʰ;"
|
27
|
+
- "ֆ > f;"
|
28
|
+
- "ս > s;"
|
29
|
+
- "շ > ʃ;"
|
30
|
+
- "խ > x;"
|
31
|
+
- "հ > h;"
|
32
|
+
- "վ > v;"
|
33
|
+
- "ւ > v;"
|
34
|
+
- "զ > z;"
|
35
|
+
- "ժ > ʒ;"
|
36
|
+
- "ղ > ɣ;"
|
37
|
+
- "լ > l;"
|
38
|
+
- "յ > j;"
|
39
|
+
- "ռ > r;"
|
40
|
+
- "ր > ɾ;"
|
41
|
+
- "ի > i;"
|
42
|
+
- "ու > u;"
|
43
|
+
- "$wordBoundary {ո} > vo;"
|
44
|
+
- "ո > o;"
|
45
|
+
- "$vowel {ե} > jɛ;"
|
46
|
+
- "$wordBoundary {ե} > jɛ;"
|
47
|
+
- "ե > ɛ;"
|
48
|
+
- "$wordBoundary {և} > jɛv;"
|
49
|
+
- "և > ɛv;"
|
50
|
+
- "է > ɛ;"
|
51
|
+
- "ը > ə;"
|
52
|
+
- "օ > o;"
|
53
|
+
- "ա > a;"
|
54
|
+
:source: hy
|
55
|
+
:target: hy_FONIPA
|
56
|
+
:variant: ~
|
@@ -0,0 +1,133 @@
|
|
1
|
+
---
|
2
|
+
:transforms:
|
3
|
+
-
|
4
|
+
:direction: forward
|
5
|
+
:rules:
|
6
|
+
- ":: [ԱԲԳԴԵԶԷԸԹԺԻԼԽԾԿՀՁՂՃՄՅՆՇՈՉՊՋՌՍՎՏՐՑՒՓՔՕՖաբգդեզէըթժիլխծկհձղճմյնշոչպջռսվտրցւփքօֆև] ;"
|
7
|
+
- ":: NFD (NFC) ;"
|
8
|
+
- "$upperConsonants = [ԲԳԴԶԹԺԼԽԾԿՀՁՂՃՄՅՆՇՉՊՋՌՍՎՐՑՓՔՖ] ;"
|
9
|
+
- "$lowerConsonants = [բգդզթժլխծկհձղճմյնշչպջռսվտրցփքֆ] ;"
|
10
|
+
- "$consonants = [$upperConsonants $lowerConsonants] ;"
|
11
|
+
- "$upperVowels = [ԱԵԷԸԻՈՕՒ] ;"
|
12
|
+
- "$lowerVowels = [աեէըիոևօւ] ;"
|
13
|
+
- "$vowels = [$upperVowels $lowerVowels] ;"
|
14
|
+
- "$lower = [$lowerConsonants $lowerVowels] ;"
|
15
|
+
- "$aspirate = ’ ;"
|
16
|
+
- "$wordBoundary = [^[:L:][:M:][:N:]] ;"
|
17
|
+
- "$wordBoundary{ԵՒ}$wordBoundary > YEV ; # ARMENIAN CAPITAL ECH + CAPITAL YIWN"
|
18
|
+
- "$wordBoundary{ԵՒ > YEV ; # ARMENIAN CAPITAL ECH + CAPITAL YIWN"
|
19
|
+
- "$vowels{ԵՒ > YEV ; # ARMENIAN CAPITAL ECH + CAPITAL YIWN"
|
20
|
+
- "ԵՒ > EV ; # ARMENIAN CAPITAL ECH + CAPITAL YIWN"
|
21
|
+
- "$wordBoundary{Եւ}$wordBoundary > Yev ; # ARMENIAN CAPITAL ECH + SMALL YIWN"
|
22
|
+
- "$wordBoundary{Եւ > Yev ; # ARMENIAN CAPITAL ECH + SMALL YIWN"
|
23
|
+
- "$vowels{Եւ > Yev ; # ARMENIAN CAPITAL ECH + SMALL YIWN"
|
24
|
+
- "Եւ > Ev ; # ARMENIAN CAPITAL ECH + SMALL YIWN"
|
25
|
+
- "$wordBoundary{եւ}$wordBoundary > yev ; # ARMENIAN SMALL ECH + SMALL YIWN"
|
26
|
+
- "$wordBoundary{եւ > yev ; # ARMENIAN SMALL ECH + SMALL YIWN"
|
27
|
+
- "$vowels{եւ > yev ; # ARMENIAN SMALL ECH + SMALL YIWN"
|
28
|
+
- "եւ > ev ; # ARMENIAN SMALL ECH + SMALL YIWN"
|
29
|
+
- "$wordBoundary{և}$wordBoundary > yev ; # ARMENIAN SMALL LIGATURE ECH YIWN"
|
30
|
+
- "$wordBoundary{և > yev ; # ARMENIAN SMALL LIGATURE ECH YIWN"
|
31
|
+
- "$vowels{և > yev ; # ARMENIAN SMALL LIGATURE ECH YIWN"
|
32
|
+
- "և > ev ; # ARMENIAN SMALL LIGATURE ECH YIWN"
|
33
|
+
- "Ա > A ; # ARMENIAN CAPITAL LETTER AYB"
|
34
|
+
- "ա > a ; # ARMENIAN SMALL LETTER AYB"
|
35
|
+
- "Բ > B ; # ARMENIAN CAPITAL LETTER BEN"
|
36
|
+
- "բ > b ; # ARMENIAN SMALL LETTER BEN"
|
37
|
+
- "Գ > G ; # ARMENIAN CAPITAL LETTER GIM"
|
38
|
+
- "գ > g ; # ARMENIAN SMALL LETTER GIM"
|
39
|
+
- "Դ > D ; # ARMENIAN CAPITAL LETTER DA"
|
40
|
+
- "դ > d ; # ARMENIAN SMALL LETTER DA"
|
41
|
+
- "$upperVowels{Ե > YE ; # ARMENIAN CAPITAL LETTER ECH"
|
42
|
+
- "$lowerVowels{Ե > Ye ; # ARMENIAN CAPITAL LETTER ECH"
|
43
|
+
- "$wordBoundary{Ե > Ye ; # ARMENIAN CAPITAL LETTER ECH"
|
44
|
+
- "Ե > E ; # ARMENIAN CAPITAL LETTER ECH"
|
45
|
+
- "$vowels{ե > ye ; # ARMENIAN SMALL LETTER ECH"
|
46
|
+
- "$wordBoundary{ե > ye ; # ARMENIAN SMALL LETTER ECH"
|
47
|
+
- "ե > e ; # ARMENIAN SMALL LETTER ECH"
|
48
|
+
- "Զ > Z ; # ARMENIAN CAPITAL LETTER ZA"
|
49
|
+
- "զ > z ; # ARMENIAN SMALL LETTER ZA"
|
50
|
+
- "Է > E ; # ARMENIAN CAPITAL LETTER EH"
|
51
|
+
- "է > e ; # ARMENIAN SMALL LETTER EH"
|
52
|
+
- "Ը > Y ; # ARMENIAN CAPITAL LETTER ET"
|
53
|
+
- "ը > y ; # ARMENIAN SMALL LETTER ET"
|
54
|
+
- "Թ > T $aspirate ; # ARMENIAN CAPITAL LETTER TO"
|
55
|
+
- "թ > t $aspirate ; # ARMENIAN SMALL LETTER TO"
|
56
|
+
- "Ժ} $lower > Zh ; # ARMENIAN CAPITAL LETTER ZHE"
|
57
|
+
- "Ժ > ZH ; # ARMENIAN CAPITAL LETTER ZHE"
|
58
|
+
- "ժ > zh ; # ARMENIAN SMALL LETTER ZHE"
|
59
|
+
- "Ի > I ; # ARMENIAN CAPITAL LETTER INI"
|
60
|
+
- "ի > i ; # ARMENIAN SMALL LETTER INI"
|
61
|
+
- "Լ > L ; # ARMENIAN CAPITAL LETTER LIWN"
|
62
|
+
- "լ > l ; # ARMENIAN SMALL LETTER LIWN"
|
63
|
+
- "Խ} $lower > Kh ; # ARMENIAN CAPITAL LETTER XEH"
|
64
|
+
- "Խ > KH ; # ARMENIAN CAPITAL LETTER XEH"
|
65
|
+
- "խ > kh ; # ARMENIAN SMALL LETTER XEH"
|
66
|
+
- "Ծ} $lower > Ts ; # ARMENIAN CAPITAL LETTER CA"
|
67
|
+
- "Ծ > TS ; # ARMENIAN CAPITAL LETTER CA"
|
68
|
+
- "ծ > ts ; # ARMENIAN SMALL LETTER CA"
|
69
|
+
- "Կ > K ; # ARMENIAN CAPITAL LETTER KEN"
|
70
|
+
- "կ > k ; # ARMENIAN SMALL LETTER KEN"
|
71
|
+
- "Հ > H ; # ARMENIAN CAPITAL LETTER HO"
|
72
|
+
- "հ > h ; # ARMENIAN SMALL LETTER HO"
|
73
|
+
- "Ձ} $lower > Dz ; # ARMENIAN CAPITAL LETTER JA"
|
74
|
+
- "Ձ > DZ ; # ARMENIAN CAPITAL LETTER JA"
|
75
|
+
- "ձ > dz ; # ARMENIAN SMALL LETTER JA"
|
76
|
+
- "Ղ} $lower > Gh ; # ARMENIAN CAPITAL LETTER GHAD"
|
77
|
+
- "Ղ > GH ; # ARMENIAN CAPITAL LETTER GHAD"
|
78
|
+
- "ղ > gh ; # ARMENIAN SMALL LETTER GHAD"
|
79
|
+
- "Ճ} $lower > Ch ; # ARMENIAN CAPITAL LETTER CHEH"
|
80
|
+
- "Ճ > CH ; # ARMENIAN CAPITAL LETTER CHEH"
|
81
|
+
- "ճ > ch ; # ARMENIAN SMALL LETTER CHEH"
|
82
|
+
- "Մ > M ; # ARMENIAN CAPITAL LETTER MEN"
|
83
|
+
- "մ > m ; # ARMENIAN SMALL LETTER MEN"
|
84
|
+
- "Յ > Y ; # ARMENIAN CAPITAL LETTER YI"
|
85
|
+
- "յ > y ; # ARMENIAN SMALL LETTER YI"
|
86
|
+
- "Ն > N ; # ARMENIAN CAPITAL LETTER NOW"
|
87
|
+
- "ն > n ; # ARMENIAN SMALL LETTER NOW"
|
88
|
+
- "Շ} $lower > Sh ; # ARMENIAN CAPITAL LETTER SHA"
|
89
|
+
- "Շ > SH ; # ARMENIAN CAPITAL LETTER SHA"
|
90
|
+
- "շ > sh ; # ARMENIAN SMALL LETTER SHA"
|
91
|
+
- "Ու} $lower > U ; # ARMENIAN CAPITAL LETTER VO + SMALL YIWN"
|
92
|
+
- "ՈՒ > U ; # ARMENIAN CAPITAL LETTER VO + CAPITAL YIWN"
|
93
|
+
- "ու > u ; # ARMENIAN SMALL LETTER VO + SMALL YIWN"
|
94
|
+
- "ՈՎ > OV ;"
|
95
|
+
- "Ով > Ov ;"
|
96
|
+
- "ով > ov ;"
|
97
|
+
- "$wordBoundary{Ո}$lower > Vo ; # ARMENIAN CAPITAL LETTER VO"
|
98
|
+
- "$wordBoundary{Ո > VO ; # ARMENIAN CAPITAL LETTER VO"
|
99
|
+
- "Ո > O ; # ARMENIAN CAPITAL LETTER ECH"
|
100
|
+
- "$wordBoundary{ո > vo ; # ARMENIAN SMALL LETTER VO"
|
101
|
+
- "ո > o ; # ARMENIAN SMALL LETTER VO"
|
102
|
+
- "Չ} $lower > Ch $aspirate ; # ARMENIAN CAPITAL LETTER CHA"
|
103
|
+
- "Չ > CH $aspirate ; # ARMENIAN CAPITAL LETTER CHA"
|
104
|
+
- "չ > ch $aspirate ; # ARMENIAN SMALL LETTER CHA"
|
105
|
+
- "Պ > P ; # ARMENIAN CAPITAL LETTER PEH"
|
106
|
+
- "պ > p ; # ARMENIAN SMALL LETTER PEH"
|
107
|
+
- "Ջ > J ; # ARMENIAN CAPITAL LETTER JHEH"
|
108
|
+
- "ջ > j ; # ARMENIAN SMALL LETTER JHEH"
|
109
|
+
- "Ռ} $lower > Rr ; # ARMENIAN CAPITAL LETTER RA"
|
110
|
+
- "Ռ > RR ; # ARMENIAN CAPITAL LETTER RA"
|
111
|
+
- "ռ > rr ; # ARMENIAN SMALL LETTER RA"
|
112
|
+
- "Ս > S ; # ARMENIAN CAPITAL LETTER SEH"
|
113
|
+
- "ս > s ; # ARMENIAN SMALL LETTER SEH"
|
114
|
+
- "Վ > V ; # ARMENIAN CAPITAL LETTER VEW"
|
115
|
+
- "վ > v ; # ARMENIAN SMALL LETTER VEW"
|
116
|
+
- "Տ > T ; # ARMENIAN CAPITAL LETTER TIWN"
|
117
|
+
- "տ > t ; # ARMENIAN SMALL LETTER TIWN"
|
118
|
+
- "Ր > R ; # ARMENIAN CAPITAL LETTER REH"
|
119
|
+
- "ր > r ; # ARMENIAN SMALL LETTER REH"
|
120
|
+
- "Ց} $lower > Ts $aspirate ; # ARMENIAN CAPITAL LETTER CHEH"
|
121
|
+
- "Ց > TS $aspirate ; # ARMENIAN CAPITAL LETTER CO"
|
122
|
+
- "ց > ts $aspirate ; # ARMENIAN SMALL LETTER CO"
|
123
|
+
- "Փ > P $aspirate ; # ARMENIAN CAPITAL LETTER PIWR"
|
124
|
+
- "փ > p $aspirate ; # ARMENIAN SMALL LETTER PIWR"
|
125
|
+
- "Ք > K $aspirate ; # ARMENIAN CAPITAL LETTER KEH"
|
126
|
+
- "ք > k $aspirate ; # ARMENIAN SMALL LETTER KEH"
|
127
|
+
- "Օ > O ; # ARMENIAN CAPITAL LETTER OH"
|
128
|
+
- "օ > o ; # ARMENIAN SMALL LETTER OH"
|
129
|
+
- "Ֆ > F ; # ARMENIAN CAPITAL LETTER FEH"
|
130
|
+
- "ֆ > f ; # ARMENIAN SMALL LETTER FEH"
|
131
|
+
:source: hy
|
132
|
+
:target: hy_Latn
|
133
|
+
:variant: BGN
|
@@ -0,0 +1,82 @@
|
|
1
|
+
---
|
2
|
+
:transforms:
|
3
|
+
-
|
4
|
+
:direction: forward
|
5
|
+
:rules:
|
6
|
+
- "::lower();"
|
7
|
+
- "$wordBoundary = [^[:L:][:M:][:N:]];"
|
8
|
+
- "$vowel = [աեէըիոևօւ];"
|
9
|
+
- "'՚' > ; # կ՚ուտէ > /ɡudɛ/"
|
10
|
+
- "մ > m;"
|
11
|
+
- "ն > n;"
|
12
|
+
- "պ > b;"
|
13
|
+
- "տ > d;"
|
14
|
+
- "կ > ɡ;"
|
15
|
+
- "բ > pʰ;"
|
16
|
+
- "դ > tʰ;"
|
17
|
+
- "գ > kʰ;"
|
18
|
+
- "փ > pʰ;"
|
19
|
+
- "{թիւն} $wordBoundary > tʰjun; # միութիւն > /mijutʰjun/, գիտութիւն > /kʰidutʰjun/"
|
20
|
+
- "թ > tʰ;"
|
21
|
+
- "ք > kʰ;"
|
22
|
+
- "ծ > d͡z;"
|
23
|
+
- "ճ > d͡ʒ;"
|
24
|
+
- "ձ > t͡sʰ;"
|
25
|
+
- "ջ > t͡ʃʰ;"
|
26
|
+
- "ց > t͡sʰ;"
|
27
|
+
- "չ > t͡ʃʰ;"
|
28
|
+
- "ֆ > f;"
|
29
|
+
- "ս > s;"
|
30
|
+
- "շ > ʃ;"
|
31
|
+
- "խ > χ;"
|
32
|
+
- "հ > h;"
|
33
|
+
- "վ > v;"
|
34
|
+
- "ւ > v;"
|
35
|
+
- "զ > z;"
|
36
|
+
- "ժ > ʒ;"
|
37
|
+
- "ղ > ʁ;"
|
38
|
+
- "լ > l;"
|
39
|
+
- "$wordBoundary {յ} > h; # յետոյ > /hɛdo/, յատակ > /hɑdɑɡ/"
|
40
|
+
- "յ > j;"
|
41
|
+
- "ռ > ɾ;"
|
42
|
+
- "ր > ɾ;"
|
43
|
+
- "$wordBoundary {իւ} > ju; # իւղ > /juʁ/"
|
44
|
+
- "իու > iju; # միութիւն > /mijutʰjun/"
|
45
|
+
- "իւ > ʏ; # հիւր > /hʏɾ/"
|
46
|
+
- "{իայ} $wordBoundary > ja;"
|
47
|
+
- "իա > ijɑ; # միասին > /mijɑsin/"
|
48
|
+
- "ի > i;"
|
49
|
+
- "{եայ} $wordBoundary > jɑ; # առօրեայ > /ɑɾoɾjɑ/"
|
50
|
+
- "եա > jɑ; # Եանիքեան > /jɑnikʰjɑn/"
|
51
|
+
- "եօ > jo; # եօթը > /jotʰə/"
|
52
|
+
- "ով > ov; # երազով > /jɛɾɑzov/"
|
53
|
+
- "{ոյ} $wordBoundary > o; # երեկոյ > /jɛɾɛɡo/"
|
54
|
+
- "{ոյ} $vowel > oj; # գոյական > /kʰojɑɡɑn/"
|
55
|
+
- "ոյ > uj; # քոյր > /kʰujɾ/"
|
56
|
+
- "{ու} $vowel > v; # վաղուընէ > /vɑʁvənɛ/, պահուըտիլ > /bɑhvədil/"
|
57
|
+
- "ու > u; # մուկ > /muɡ/"
|
58
|
+
- "$wordBoundary {ո} > vo; # ոսկի > /vosɡi/"
|
59
|
+
- "ո > o; # ցորեն > /t͡sʰoɾɛn/"
|
60
|
+
- "$vowel {ե} > jɛ; # հայելի > /hɑjɛli/"
|
61
|
+
- "$wordBoundary {ե} > jɛ; # երազ > /jɛɾɑz/"
|
62
|
+
- "ե > ɛ;"
|
63
|
+
- "$wordBoundary {և} > jɛv;"
|
64
|
+
- "և > ɛv;"
|
65
|
+
- "{էայ} $wordBoundary > ɛjɑ;"
|
66
|
+
- "էա > ɛjɑ; # էակ > /ɛjɑɡ/"
|
67
|
+
- "էի > ɛji; # էի > /ɛji/, կուզէին > /ɡuzɛjin/"
|
68
|
+
- "էու > ɛju; # էութիւն > /ɛjutʰjun/"
|
69
|
+
- "էօ > œ; # Էօժենի > /œʒɛni/"
|
70
|
+
- "էյ > ej; # թէյ > /tʰej/"
|
71
|
+
- "է > ɛ;"
|
72
|
+
- "ը > ə;"
|
73
|
+
- "օ > o;"
|
74
|
+
- "{այ} $wordBoundary > ɑ; # ծառայ > /d͡zɑɾɑ/, կը դողայ > /ɡə tʰoʁɑ/"
|
75
|
+
- "ա > ɑ;"
|
76
|
+
- "::NULL;"
|
77
|
+
- "jj > j; # Գայեանէ > /kʰɑjjɑnɛ/ > /kʰɑjɑnɛ/"
|
78
|
+
- "nɡ > ŋɡ; # Քոնկօ > /kʰonɡo/ > /kʰoŋɡo/"
|
79
|
+
- "nk > ŋk; # օրէնք > /oɾɛnkʰ/ > /oɾɛŋkʰ/"
|
80
|
+
:source: hy_AREVMDA
|
81
|
+
:target: hy_AREVMDA_FONIPA
|
82
|
+
:variant: ~
|