interscript 0.1.9 → 2.0.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.gitignore +11 -0
- data/.rspec +3 -0
- data/Gemfile +29 -0
- data/LICENSE.adoc +31 -0
- data/README.md +3 -0
- data/Rakefile +53 -0
- data/bin/console +14 -0
- data/bin/interscript +5 -0
- data/bin/maps_analyze_staging +168 -0
- data/bin/maps_debug_compilers +58 -0
- data/bin/maps_debug_ordering +88 -0
- data/bin/maps_debug_ruby_compile +24 -0
- data/bin/maps_debug_step_by_step +44 -0
- data/bin/maps_optimize_order +112 -0
- data/bin/maps_v1_analyze_regexps +45 -0
- data/bin/maps_v1_to_v2 +426 -0
- data/bin/setup +8 -0
- data/exe/interscript +6 -0
- data/interscript.gemspec +31 -0
- data/lib/interscript.rb +80 -135
- data/lib/interscript/command.rb +5 -5
- data/lib/interscript/compiler.rb +22 -0
- data/lib/interscript/compiler/javascript.rb +292 -0
- data/lib/interscript/compiler/ruby.rb +262 -0
- data/lib/interscript/dsl.rb +67 -0
- data/lib/interscript/dsl/aliases.rb +23 -0
- data/lib/interscript/dsl/document.rb +46 -0
- data/lib/interscript/dsl/group.rb +45 -0
- data/lib/interscript/dsl/group/parallel.rb +6 -0
- data/lib/interscript/dsl/items.rb +89 -0
- data/lib/interscript/dsl/metadata.rb +26 -0
- data/lib/interscript/dsl/stage.rb +6 -0
- data/lib/interscript/dsl/symbol_mm.rb +11 -0
- data/lib/interscript/dsl/tests.rb +12 -0
- data/lib/interscript/interpreter.rb +251 -0
- data/lib/interscript/node.rb +25 -0
- data/lib/interscript/node/alias_def.rb +15 -0
- data/lib/interscript/node/dependency.rb +13 -0
- data/lib/interscript/node/document.rb +45 -0
- data/lib/interscript/node/group.rb +34 -0
- data/lib/interscript/node/group/parallel.rb +9 -0
- data/lib/interscript/node/group/sequential.rb +2 -0
- data/lib/interscript/node/item.rb +52 -0
- data/lib/interscript/node/item/alias.rb +42 -0
- data/lib/interscript/node/item/any.rb +61 -0
- data/lib/interscript/node/item/capture.rb +50 -0
- data/lib/interscript/node/item/group.rb +51 -0
- data/lib/interscript/node/item/repeat.rb +40 -0
- data/lib/interscript/node/item/stage.rb +23 -0
- data/lib/interscript/node/item/string.rb +51 -0
- data/lib/interscript/node/metadata.rb +18 -0
- data/lib/interscript/node/rule.rb +6 -0
- data/lib/interscript/node/rule/funcall.rb +18 -0
- data/lib/interscript/node/rule/run.rb +15 -0
- data/lib/interscript/node/rule/sub.rb +65 -0
- data/lib/interscript/node/stage.rb +19 -0
- data/lib/interscript/node/tests.rb +15 -0
- data/lib/interscript/stdlib.rb +211 -0
- data/lib/interscript/utils/regexp_converter.rb +283 -0
- data/lib/interscript/version.rb +1 -1
- data/requirements.txt +1 -0
- metadata +73 -458
- data/README.adoc +0 -296
- data/aliases.json +0 -1
- data/lib/g2pwrapper.py +0 -34
- data/lib/interscript/fs.rb +0 -96
- data/lib/interscript/mapping.rb +0 -144
- data/lib/interscript/opal.rb +0 -196
- data/lib/interscript/opal/entrypoint.rb +0 -20
- data/lib/interscript/opal/exports.rb +0 -11
- data/lib/interscript/opal/maps.js.erb +0 -8
- data/lib/model-7 +0 -0
- data/lib/tha-pt-b-7 +0 -0
- data/maps/acadsin-zho-Hani-Latn-2002.yaml +0 -38916
- data/maps/alalc-amh-Ethi-Latn-1997.yaml +0 -513
- data/maps/alalc-amh-Ethi-Latn-2011.yaml +0 -138
- data/maps/alalc-ara-Arab-Latn-1997.yaml +0 -1287
- data/maps/alalc-asm-Deva-Latn-1997.yaml +0 -259
- data/maps/alalc-asm-Deva-Latn-2012.yaml +0 -55
- data/maps/alalc-aze-Arab-Latn-1997.yaml +0 -376
- data/maps/alalc-aze-Cyrl-Latn-1997.yaml +0 -145
- data/maps/alalc-bel-Cyrl-Latn-1997.yaml +0 -129
- data/maps/alalc-ben-Beng-Latn-1997.yaml +0 -291
- data/maps/alalc-ben-Beng-Latn-2017.yaml +0 -130
- data/maps/alalc-bul-Cyrl-Latn-1997.yaml +0 -98
- data/maps/alalc-div-Thaa-Latn-1997.yaml +0 -211
- data/maps/alalc-ell-Grek-Latn-1997.yaml +0 -628
- data/maps/alalc-ell-Grek-Latn-2010.yaml +0 -626
- data/maps/alalc-guj-Gujr-Latn-1997.yaml +0 -266
- data/maps/alalc-guj-Gujr-Latn-2011.yaml +0 -64
- data/maps/alalc-hin-Deva-Latn-1997.yaml +0 -303
- data/maps/alalc-hin-Deva-Latn-2011.yaml +0 -65
- data/maps/alalc-kan-Kana-Latn-1997.yaml +0 -274
- data/maps/alalc-kan-Kana-Latn-2011.yaml +0 -63
- data/maps/alalc-kat-Geok-Latn-1997.yaml +0 -111
- data/maps/alalc-kat-Geor-Latn-1997.yaml +0 -150
- data/maps/alalc-kor-Hang-Latn-1997.yaml +0 -98
- data/maps/alalc-mal-Mlym-Latn-1997.yaml +0 -303
- data/maps/alalc-mal-Mlym-Latn-2012.yaml +0 -73
- data/maps/alalc-mar-Deva-Latn-1997.yaml +0 -189
- data/maps/alalc-mar-Deva-Latn-2011.yaml +0 -45
- data/maps/alalc-mkd-Cyrl-Latn-1997.yaml +0 -114
- data/maps/alalc-mkd-Cyrl-Latn-2013.yaml +0 -103
- data/maps/alalc-mon-Cyrl-Latn-1997.yaml +0 -220
- data/maps/alalc-ori-Orya-Latn-1997.yaml +0 -284
- data/maps/alalc-ori-Orya-Latn-2011.yaml +0 -67
- data/maps/alalc-pan-Guru-Latn-1997.yaml +0 -256
- data/maps/alalc-pan-Guru-Latn-2011.yaml +0 -78
- data/maps/alalc-per-Arab-Latn-1997.yaml +0 -375
- data/maps/alalc-pli-Deva-Latn-2012.yaml +0 -144
- data/maps/alalc-pra-Deva-Latn-2012.yaml +0 -47
- data/maps/alalc-rus-Cyrl-Latn-1997.yaml +0 -225
- data/maps/alalc-rus-Cyrl-Latn-2012.yaml +0 -162
- data/maps/alalc-san-Deva-Latn-2012.yaml +0 -241
- data/maps/alalc-sin-Sinh-Latn-1997.yaml +0 -292
- data/maps/alalc-sin-Sinh-Latn-2011.yaml +0 -71
- data/maps/alalc-srp-Cyrl-Latn-1997.yaml +0 -118
- data/maps/alalc-srp-Cyrl-Latn-2013.yaml +0 -135
- data/maps/alalc-tam-Taml-Latn-1997.yaml +0 -62
- data/maps/alalc-tam-Taml-Latn-2011.yaml +0 -58
- data/maps/alalc-tel-Telu-Latn-1997.yaml +0 -284
- data/maps/alalc-tel-Telu-Latn-2011.yaml +0 -64
- data/maps/alalc-ukr-Cyrl-Latn-1997.yaml +0 -145
- data/maps/alalc-ukr-Cyrl-Latn-2011.yaml +0 -16
- data/maps/apcbg-bul-Cyrl-Latn-1995.yaml +0 -283
- data/maps/az-aze-Cyrl-Latn-1939.yaml +0 -105
- data/maps/az-aze-Cyrl-Latn-1958.yaml +0 -45
- data/maps/bas-rus-Cyrl-Latn-2017-bss.yaml +0 -174
- data/maps/bas-rus-Cyrl-Latn-2017-oss.yaml +0 -169
- data/maps/bgn-jpn-Hrkt-Latn-1962.yaml +0 -292
- data/maps/bgn-kor-Hang-Latn-1943.yaml +0 -35
- data/maps/bgn-kor-Kore-Latn-1943.yaml +0 -31
- data/maps/bgna-bul-Cyrl-Latn-2006.yaml +0 -208
- data/maps/bgna-bul-Cyrl-Latn-2009.yaml +0 -208
- data/maps/bgnpcgn-amh-Ethi-Latn-1967.yaml +0 -532
- data/maps/bgnpcgn-ara-Arab-Latn-1956.yaml +0 -598
- data/maps/bgnpcgn-arm-Armn-Latn-1981.yaml +0 -108
- data/maps/bgnpcgn-aze-Cyrl-Latn-1993.yaml +0 -111
- data/maps/bgnpcgn-bak-Cyrl-Latn-2007.yaml +0 -188
- data/maps/bgnpcgn-bal-Arab-Latn-2008.yaml +0 -329
- data/maps/bgnpcgn-bel-Cyrl-Latn-1979.yaml +0 -289
- data/maps/bgnpcgn-bul-Cyrl-Latn-1952.yaml +0 -119
- data/maps/bgnpcgn-bul-Cyrl-Latn-2013.yaml +0 -42
- data/maps/bgnpcgn-che-Cyrl-Latn-2008.yaml +0 -184
- data/maps/bgnpcgn-div-Thaa-Latn-1988.yaml +0 -75
- data/maps/bgnpcgn-ell-Grek-Latn-1962.yaml +0 -705
- data/maps/bgnpcgn-ell-Grek-Latn-1996.yaml +0 -23
- data/maps/bgnpcgn-far-Latn-Latn-1964.yaml +0 -28
- data/maps/bgnpcgn-fas-Arab-Latn-1956.yaml +0 -96
- data/maps/bgnpcgn-isl-Latn-Latn-1964.yaml +0 -37
- data/maps/bgnpcgn-jpn-Hrkt-Latn-1976.yaml +0 -257
- data/maps/bgnpcgn-kat-Geor-Latn-1981.yaml +0 -131
- data/maps/bgnpcgn-kat-Geor-Latn-2009.yaml +0 -42
- data/maps/bgnpcgn-kaz-Cyrl-Latn-1979.yaml +0 -247
- data/maps/bgnpcgn-kir-Cyrl-Latn-1979.yaml +0 -218
- data/maps/bgnpcgn-kor-Hang-Latn-kn-1945.yaml +0 -253
- data/maps/bgnpcgn-kor-Hang-Latn-rok-2011.yaml +0 -48
- data/maps/bgnpcgn-kor-Kore-Latn-rok-2011.yaml +0 -48
- data/maps/bgnpcgn-kur-Arab-Latn-2007.yaml +0 -249
- data/maps/bgnpcgn-mkd-Cyrl-Latn-1981.yaml +0 -163
- data/maps/bgnpcgn-mkd-Cyrl-Latn-2013.yaml +0 -190
- data/maps/bgnpcgn-mon-Cyrl-Latn-1964.yaml +0 -223
- data/maps/bgnpcgn-nep-Deva-Latn-2011.yaml +0 -230
- data/maps/bgnpcgn-per-Arab-Latn-1958.yaml +0 -338
- data/maps/bgnpcgn-prs-Arab-Latn-2007.yaml +0 -673
- data/maps/bgnpcgn-prs-Arab-Latn-yaghoubi.yaml +0 -459
- data/maps/bgnpcgn-pus-Arab-Latn-1968.yaml +0 -377
- data/maps/bgnpcgn-rue-Cyrl-Latn-2016.yaml +0 -168
- data/maps/bgnpcgn-rus-Cyrl-Latn-1947.yaml +0 -318
- data/maps/bgnpcgn-srp-Cyrl-Latn-1962.yaml +0 -73
- data/maps/bgnpcgn-srp-Cyrl-Latn-2005.yaml +0 -170
- data/maps/bgnpcgn-tat-Cyrl-Latn-2007.yaml +0 -220
- data/maps/bgnpcgn-tgk-Cyrl-Latn-1994.yaml +0 -240
- data/maps/bgnpcgn-ukr-Cyrl-Latn-1965.yaml +0 -166
- data/maps/bgnpcgn-ukr-Cyrl-Latn-2019.yaml +0 -119
- data/maps/bgnpcgn-urd-Arab-Latn-2007.yaml +0 -459
- data/maps/bgnpcgn-uzb-Cyrl-Latn-1979.yaml +0 -127
- data/maps/bgnpcgn-uzb-Cyrl-Latn-2000.yaml +0 -82
- data/maps/bgnpcgn-zho-Hans-Latn-1979.yaml +0 -7456
- data/maps/bis-asm-Beng-Latn-13194-1991.yaml +0 -159
- data/maps/bis-ben-Beng-Latn-13194-1991.yaml +0 -156
- data/maps/bis-dev-Deva-Latn-13194-1991.yaml +0 -184
- data/maps/bis-guj-Gujr-Latn-13194-1991.yaml +0 -181
- data/maps/bis-kan-Kana-Latn-13194-1991.yaml +0 -173
- data/maps/bis-mlm-Mlym-Latn-13194-1991.yaml +0 -176
- data/maps/bis-ori-Orya-Latn-13194-1991.yaml +0 -175
- data/maps/bis-pnj-Guru-Latn-13194-1991.yaml +0 -175
- data/maps/bis-tel-Telu-Latn-13194-1991.yaml +0 -170
- data/maps/bis-tml-Taml-Latn-13194-1991.yaml +0 -155
- data/maps/by-bel-Cyrl-Latn-1998.yaml +0 -172
- data/maps/by-bel-Cyrl-Latn-2007.yaml +0 -115
- data/maps/din-grc-Grek-Latn-31634-2011-t1.yaml +0 -899
- data/maps/din-hin-Deva-Latn-33904-2018.yaml +0 -100
- data/maps/din-kat-Geor-Latn-32707-2010.yaml +0 -145
- data/maps/din-mar-Deva-Latn-33904-2018.yaml +0 -84
- data/maps/din-nep-Deva-Latn-33904-2018.yaml +0 -119
- data/maps/din-pli-Deva-Latn-33904-2018.yaml +0 -75
- data/maps/din-pra-Deva-Latn-33904-2018.yaml +0 -63
- data/maps/din-san-Deva-Latn-33904-2018.yaml +0 -338
- data/maps/din-tam-Taml-Latn-33903-2016.yaml +0 -213
- data/maps/dos-nep-Deva-Latn-1997.yaml +0 -47
- data/maps/elot-ell-Grek-Latn-743-1982-tl.yaml +0 -684
- data/maps/elot-ell-Grek-Latn-743-1982-ts.yaml +0 -680
- data/maps/elot-ell-Grek-Latn-743-2001-tl.yaml +0 -19
- data/maps/elot-ell-Grek-Latn-743-2001-ts.yaml +0 -31
- data/maps/ggg-kat-Geor-Latn-2002.yaml +0 -92
- data/maps/gki-bel-Cyrl-Latn-1992.yaml +0 -33
- data/maps/gki-bel-Cyrl-Latn-2000.yaml +0 -201
- data/maps/gost-rus-Cyrl-Latn-16876-71-1983.yaml +0 -190
- data/maps/gost-rus-Cyrl-Latn-7.79-2000-2002.yaml +0 -157
- data/maps/hk-yue-Hani-Latn-1888.yaml +0 -38497
- data/maps/icao-bel-Cyrl-Latn-9303.yaml +0 -136
- data/maps/icao-bul-Cyrl-Latn-9303.yaml +0 -118
- data/maps/icao-fas-Arab-Latn-9303.yaml +0 -103
- data/maps/icao-heb-Hebr-Latn-9303.yaml +0 -151
- data/maps/icao-mkd-Cyrl-Latn-9303.yaml +0 -117
- data/maps/icao-rus-Cyrl-Latn-9303.yaml +0 -117
- data/maps/icao-srp-Cyrl-Latn-9303.yaml +0 -117
- data/maps/icao-ukr-Cyrl-Latn-9303.yaml +0 -119
- data/maps/iso-ara-Arab-Latn-233-1984.yaml +0 -323
- data/maps/iso-asm-Beng-Latn-15919-2001.yaml +0 -75
- data/maps/iso-ben-Beng-Latn-15919-2001.yaml +0 -175
- data/maps/iso-ell-Grek-Latn-843-1997-t1.yaml +0 -613
- data/maps/iso-ell-Grek-Latn-843-1997-t2.yaml +0 -44
- data/maps/iso-guj-Gujr-Latn-15919-2001.yaml +0 -220
- data/maps/iso-hin-Deva-Latn-15919-2001.yaml +0 -87
- data/maps/iso-inc-Deva-Latn-15919-2001.yaml +0 -61
- data/maps/iso-jpn-Hrkt-Latn-3602-1989.yaml +0 -66
- data/maps/iso-kan-Kana-Latn-15919-2001.yaml +0 -220
- data/maps/iso-kat-Geor-Latn-9984-1996.yaml +0 -145
- data/maps/iso-kor-Hang-Latn-1996-method1.yaml +0 -240
- data/maps/iso-kor-Hang-Latn-1996-method2.yaml +0 -226
- data/maps/iso-mal-Mlym-Latn-15919-2001.yaml +0 -281
- data/maps/iso-mar-Deva-Latn-15919-2001.yaml +0 -75
- data/maps/iso-nep-Deva-Latn-15919-2001.yaml +0 -87
- data/maps/iso-ori-Orya-Latn-15919-2001.yaml +0 -193
- data/maps/iso-pan-Guru-Latn-15919-2001.yaml +0 -222
- data/maps/iso-pli-Beng-Latn-15919-2001.yaml +0 -73
- data/maps/iso-pli-Deva-Latn-15919-2001.yaml +0 -74
- data/maps/iso-pli-Sinh-Latn-15919-2001.yaml +0 -219
- data/maps/iso-pli-Thai-Latn-15919-2001.yaml +0 -55
- data/maps/iso-pra-Deva-Latn-15919-2001.yaml +0 -59
- data/maps/iso-prs-Arab-Latn-233-3-1999.yaml +0 -366
- data/maps/iso-rus-Cyrl-Latn-9-1995.yaml +0 -271
- data/maps/iso-san-Deva-Latn-15919-2001.yaml +0 -220
- data/maps/iso-tam-Taml-Latn-15919-2001.yaml +0 -159
- data/maps/iso-tel-Telu-Latn-15919-2001.yaml +0 -220
- data/maps/iso-tha-Thai-Latn-11940-1998.yaml +0 -109
- data/maps/kp-kor-Hang-Latn-2002.yaml +0 -909
- data/maps/lshk-yue-Hani-Latn-jyutping-1993.yaml +0 -44820
- data/maps/masm-mon-Cyrl-Latn-5217-2012.yaml +0 -163
- data/maps/masm-mon-Latn-Cyrl-5217-2012.yaml +0 -200
- data/maps/mext-jpn-Hrkt-Latn-1954.yaml +0 -411
- data/maps/moct-kor-Hang-Latn-2000.yaml +0 -807
- data/maps/mofa-jpn-Hrkt-Latn-1989.yaml +0 -541
- data/maps/mv-div-Thaa-Latn-1987.yaml +0 -200
- data/maps/mvd-bel-Cyrl-Latn-2008.yaml +0 -225
- data/maps/mvd-bel-Cyrl-Latn-2010.yaml +0 -63
- data/maps/mvd-rus-Cyrl-Latn-2008.yaml +0 -109
- data/maps/mvd-rus-Cyrl-Latn-2010.yaml +0 -37
- data/maps/odni-ara-Arab-Latn-2004.yaml +0 -137
- data/maps/odni-ara-Arab-Latn-2015.yaml +0 -315
- data/maps/odni-aze-Cyrl-Latn-2015.yaml +0 -144
- data/maps/odni-bel-Cyrl-Latn-2015.yaml +0 -148
- data/maps/odni-bul-Cyrl-Latn-2005.yaml +0 -90
- data/maps/odni-bul-Cyrl-Latn-2015.yaml +0 -96
- data/maps/odni-che-Cyrl-Latn-2015.yaml +0 -169
- data/maps/odni-fas-Arab-Latn-2004.yaml +0 -276
- data/maps/odni-fas-Arab-Latn-2015.yaml +0 -406
- data/maps/odni-hin-Deva-Latn-2004.yaml +0 -182
- data/maps/odni-hin-Deva-Latn-2015.yaml +0 -258
- data/maps/odni-kat-Geor-Latn-2015.yaml +0 -87
- data/maps/odni-kaz-Cyrl-Latn-2015.yaml +0 -148
- data/maps/odni-kir-Cyrl-Latn-2015.yaml +0 -136
- data/maps/odni-kor-Hang-Latn-2015.yaml +0 -375
- data/maps/odni-mkd-Cyrl-Latn-2005.yaml +0 -21
- data/maps/odni-mkd-Cyrl-Latn-2015.yaml +0 -122
- data/maps/odni-prs-Arab-Latn-2004.yaml +0 -123
- data/maps/odni-prs-Arab-Latn-2015.yaml +0 -228
- data/maps/odni-rus-Cyrl-Latn-2015.yaml +0 -77
- data/maps/odni-srp-Cyrl-Latn-2005.yaml +0 -36
- data/maps/odni-srp-Cyrl-Latn-2015.yaml +0 -129
- data/maps/odni-tat-Cyrl-Latn-2015.yaml +0 -142
- data/maps/odni-tgk-Cyrl-Latn-2015.yaml +0 -148
- data/maps/odni-tuk-Cyrl-Latn-2015.yaml +0 -170
- data/maps/odni-uig-Cyrl-Latn-2015.yaml +0 -138
- data/maps/odni-ukr-Cyrl-Latn-2015.yaml +0 -161
- data/maps/odni-urd-Arab-Latn-2015.yaml +0 -221
- data/maps/odni-uzb-Cyrl-Latn-2015.yaml +0 -166
- data/maps/royin-tha-Thai-Latn-1939-generic.yaml +0 -90
- data/maps/royin-tha-Thai-Latn-1968.yaml +0 -183
- data/maps/royin-tha-Thai-Latn-1999-chained.yaml +0 -180
- data/maps/royin-tha-Thai-Latn-1999.yaml +0 -80
- data/maps/sac-zho-Hans-Latn-1979.yaml +0 -24763
- data/maps/sasm-mon-Mong-Latn-general-1978.yaml +0 -389
- data/maps/sasm-mon-Mong-Latn-phonetic-1978.yaml +0 -354
- data/maps/ses-ara-Arab-Latn-1930.yaml +0 -283
- data/maps/stategeocadastre-ukr-Cyrl-Latn-1993.yaml +0 -222
- data/maps/ua-ukr-Cyrl-Latn-1996.yaml +0 -197
- data/maps/ua-ukr-Cyrl-Latn-2007.yaml +0 -75
- data/maps/ua-ukr-Cyrl-Latn-2010.yaml +0 -192
- data/maps/un-amh-Ethi-Latn-2016.yaml +0 -602
- data/maps/un-ara-Arab-Latn-1971.yaml +0 -139
- data/maps/un-ara-Arab-Latn-1972.yaml +0 -159
- data/maps/un-ara-Arab-Latn-2017.yaml +0 -420
- data/maps/un-asm-Beng-Latn-1972.yaml +0 -223
- data/maps/un-bel-Cyrl-Latn-2007.yaml +0 -114
- data/maps/un-ben-Beng-Latn-2016.yaml +0 -534
- data/maps/un-ell-Grek-Latn-1987-phonetic.yaml +0 -780
- data/maps/un-ell-Grek-Latn-1987-tl.yaml +0 -31
- data/maps/un-ell-Grek-Latn-1987-ts.yaml +0 -19
- data/maps/un-guj-Gujr-Latn-1972.yaml +0 -229
- data/maps/un-hin-Deva-Latn-2016.yaml +0 -316
- data/maps/un-kan-Kana-Latn-2016.yaml +0 -254
- data/maps/un-mal-Mlym-Latn-1972.yaml +0 -251
- data/maps/un-mar-Deva-Latn-2016.yaml +0 -102
- data/maps/un-mon-Mong-Latn-general-2013.yaml +0 -264
- data/maps/un-mon-Mong-Latn-phonetic-2013.yaml +0 -264
- data/maps/un-nep-Deva-Latn-1972.yaml +0 -269
- data/maps/un-nep-Deva-Latn-2013.yaml +0 -74
- data/maps/un-ori-Orya-Latn-1972.yaml +0 -247
- data/maps/un-pan-Guru-Latn-1972.yaml +0 -402
- data/maps/un-prs-Arab-Latn-1967.yaml +0 -236
- data/maps/un-rus-Cyrl-Latn-1987.yaml +0 -166
- data/maps/un-tam-Taml-Latn-1972.yaml +0 -194
- data/maps/un-tel-Telu-Latn-1972.yaml +0 -270
- data/maps/un-ukr-Cyrl-Latn-1998.yaml +0 -53
- data/maps/un-ukr-Cyrl-Latn-2012.yaml +0 -162
- data/maps/un-urd-Arab-Latn-1972.yaml +0 -405
- data/maps/var-amh-Ethi-Latn-eae-2003.yaml +0 -466
- data/maps/var-gez-Ethi-Latn-eae-2003.yaml +0 -76
- data/maps/var-hin-Deva-Latn-hunterian-1872.yaml +0 -221
- data/maps/var-jpn-Hrkt-Latn-hepburn-1886.yaml +0 -406
- data/maps/var-jpn-Hrkt-Latn-hepburn-1954.yaml +0 -386
- data/maps/var-kor-Hang-Hang-jamo.yaml +0 -11193
- data/maps/var-kor-Hang-Latn-mr-1939.yaml +0 -1054
- data/maps/var-kor-Kore-Hang-2013.yaml +0 -59754
- data/maps/var-kor-Kore-Latn-mr-1939.yaml +0 -36
- data/maps/var-mar-Deva-Latn-hunterian-1872.yaml +0 -43
- data/maps/var-mon-Mong-Latn-1930.yaml +0 -102
- data/maps/var-mon-Mong-Latn-lessing.yaml +0 -272
- data/maps/var-mon-Mong-Latn-vpmc.yaml +0 -274
- data/maps/var-pra-Deva-Latn-iast-1912.yaml +0 -30
- data/maps/var-san-Deva-Latn-iast-1912.yaml +0 -149
- data/maps/var-tha-Thai-Thai-phonemic.yaml +0 -59
- data/maps/var-tha-Thai-Zsym-ipa.yaml +0 -301
- data/maps/var-zho-Hani-Latn-wd-1979.yaml +0 -38912
- data/spec/interscript/filenames_spec.rb +0 -21
- data/spec/interscript/mapping_spec.rb +0 -42
- data/spec/interscript_spec.rb +0 -37
- data/spec/spec_helper.rb +0 -3
@@ -1,53 +0,0 @@
|
|
1
|
-
---
|
2
|
-
authority_id: un
|
3
|
-
id: 1998
|
4
|
-
language: iso-639-2:ukr
|
5
|
-
source_script: Cyrl
|
6
|
-
destination_script: Latn
|
7
|
-
name: ROMANIZATION SYSTEM FOR BELARUSIAN, RUSSIAN AND UKRAINIAN CYRILLIC
|
8
|
-
url: https://unstats.un.org/unsd/geoinfo/UNGEGN/docs/7th-uncsgn-docs/econf/7th_UNCSGN_econf.91_3_Add1.pdf
|
9
|
-
creation_date: 1998
|
10
|
-
|
11
|
-
tests:
|
12
|
-
|
13
|
-
map:
|
14
|
-
inherit: gost-rus-Cyrl-Latn-16876-71-1983
|
15
|
-
|
16
|
-
characters:
|
17
|
-
'\u0413' : 'H' # Г
|
18
|
-
'\u0490' : 'G' # Ґ
|
19
|
-
'\u0404' : 'Ye' # Є
|
20
|
-
'\u0416' : 'Zh' # Ж
|
21
|
-
'\u0418' : 'Y' # И
|
22
|
-
'\u0419' : 'Y' # Й
|
23
|
-
'\u0406' : 'I' # І
|
24
|
-
'\u0407' : 'Yi' # Ї
|
25
|
-
'\u0425' : 'Kh' # Х
|
26
|
-
'\u0426' : 'Ts' # Ц
|
27
|
-
'\u0427' : 'Ch' # Ч
|
28
|
-
'\u0428' : 'Sh' # Ш
|
29
|
-
'\u0429' : 'Shch' # Щ
|
30
|
-
'\u042C' : '' # Ь
|
31
|
-
'\u042E' : 'yu' # Ю
|
32
|
-
'\u042F' : 'ya' # Я
|
33
|
-
'\u042D' : 'E' # Э
|
34
|
-
|
35
|
-
'\u0433' : 'h' # г
|
36
|
-
'\u0491' : 'g' # ґ
|
37
|
-
'\u0454' : 'ye' # є
|
38
|
-
'\u0436' : 'zh' # ж
|
39
|
-
'\u0438' : 'y' # и
|
40
|
-
'\u0439' : 'y' # й
|
41
|
-
'\u0456' : 'i' # і
|
42
|
-
'\u0457' : 'yi' # ї
|
43
|
-
'\u0445' : 'kh' # х
|
44
|
-
'\u0446' : 'ts' # ц
|
45
|
-
'\u0447' : 'ch' # ч
|
46
|
-
'\u0448' : 'sh' # ш
|
47
|
-
'\u0449' : 'shch' # щ
|
48
|
-
'\u044C' : '' # ь
|
49
|
-
'\u044E' : 'yu' # ю
|
50
|
-
'\u044F' : 'ya' # я
|
51
|
-
'\u044D' : 'e' # э
|
52
|
-
|
53
|
-
'\u0027' : '"' # '
|
@@ -1,162 +0,0 @@
|
|
1
|
-
---
|
2
|
-
authority_id: un
|
3
|
-
id: 2012
|
4
|
-
language: iso-639-2:ukr
|
5
|
-
source_script: Cyrl
|
6
|
-
destination_script: Latn
|
7
|
-
name: ROMANIZATION SYSTEM FOR BELARUSIAN, RUSSIAN AND UKRAINIAN CYRILLIC
|
8
|
-
url: http://www.eki.ee/wgrs/rom1_uk.htm
|
9
|
-
creation_date: 2010
|
10
|
-
|
11
|
-
tests:
|
12
|
-
- source: Алушта
|
13
|
-
expected: Alushta
|
14
|
-
- source: Андрій
|
15
|
-
expected: Andrii
|
16
|
-
- source: Борщагівка
|
17
|
-
expected: Borshchahivka
|
18
|
-
- source: Борисенко
|
19
|
-
expected: Borysenko
|
20
|
-
- source: Вінниця
|
21
|
-
expected: Vinnytsia
|
22
|
-
- source: Володимир
|
23
|
-
expected: Volodymyr
|
24
|
-
- source: Гадяч
|
25
|
-
expected: Hadiach
|
26
|
-
- source: Богдан
|
27
|
-
expected: Bohdan
|
28
|
-
- source: Згурський
|
29
|
-
expected: Zghurskyi
|
30
|
-
- source: Ґалаґан
|
31
|
-
expected: Galagan
|
32
|
-
- source: Ґорґани
|
33
|
-
expected: Gorgany
|
34
|
-
- source: Донецьк
|
35
|
-
expected: Donetsk
|
36
|
-
- source: Дмитро
|
37
|
-
expected: Dmytro
|
38
|
-
- source: Рівне
|
39
|
-
expected: Rivne
|
40
|
-
- source: Олег
|
41
|
-
expected: Oleh
|
42
|
-
- source: Есмань
|
43
|
-
expected: Esman
|
44
|
-
- source: Єнакієве
|
45
|
-
expected: Yenakiieve
|
46
|
-
- source: Гаєвич
|
47
|
-
expected: Haievych
|
48
|
-
- source: Короп'є
|
49
|
-
expected: Koropie
|
50
|
-
- source: Житомир
|
51
|
-
expected: Zhytomyr
|
52
|
-
- source: Жанна
|
53
|
-
expected: Zhanna
|
54
|
-
- source: Жежелів
|
55
|
-
expected: Zhezheliv
|
56
|
-
- source: Закарпаття
|
57
|
-
expected: Zakarpattia
|
58
|
-
- source: Казимирчук
|
59
|
-
expected: Kazymyrchuk
|
60
|
-
- source: Медвин
|
61
|
-
expected: Medvyn
|
62
|
-
- source: Михайленко
|
63
|
-
expected: Mykhailenko
|
64
|
-
- source: Іванків
|
65
|
-
expected: Ivankiv
|
66
|
-
- source: Іващенко
|
67
|
-
expected: Ivashchenko
|
68
|
-
- source: Їжакевич
|
69
|
-
expected: Yizhakevych
|
70
|
-
- source: Кадиївка
|
71
|
-
expected: Kadyivka
|
72
|
-
- source: Мар'їне
|
73
|
-
expected: Marine
|
74
|
-
- source: Йосипівка
|
75
|
-
expected: Yosypivka
|
76
|
-
- source: Стрий
|
77
|
-
expected: Stryi
|
78
|
-
- source: Олексій
|
79
|
-
expected: Oleksii
|
80
|
-
- source: Київ
|
81
|
-
expected: Kyiv
|
82
|
-
- source: Коваленко
|
83
|
-
expected: Kovalenko
|
84
|
-
- source: Лебедин
|
85
|
-
expected: Lebedyn
|
86
|
-
- source: Леонід
|
87
|
-
expected: Leonid
|
88
|
-
- source: Миколаїв
|
89
|
-
expected: Mykolaiv
|
90
|
-
- source: Маринич
|
91
|
-
expected: Marynych
|
92
|
-
- source: Ніжин
|
93
|
-
expected: Nizhyn
|
94
|
-
- source: Наталія
|
95
|
-
expected: Nataliia
|
96
|
-
- source: Одеса
|
97
|
-
expected: Odesa
|
98
|
-
- source: Онищенко
|
99
|
-
expected: Onyshchenko
|
100
|
-
- source: Полтава
|
101
|
-
expected: Poltava
|
102
|
-
- source: Петро
|
103
|
-
expected: Petro
|
104
|
-
- source: Решетилівка
|
105
|
-
expected: Reshetylivka
|
106
|
-
- source: Рибчинський
|
107
|
-
expected: Rybchynskyi
|
108
|
-
- source: Суми
|
109
|
-
expected: Sumy
|
110
|
-
- source: Соломія
|
111
|
-
expected: Solomiia
|
112
|
-
- source: Тернопіль
|
113
|
-
expected: Ternopil
|
114
|
-
- source: Троць
|
115
|
-
expected: Trots
|
116
|
-
- source: Ужгород
|
117
|
-
expected: Uzhhorod
|
118
|
-
- source: Уляна
|
119
|
-
expected: Uliana
|
120
|
-
- source: Фастів
|
121
|
-
expected: Fastiv
|
122
|
-
- source: Філіпчук
|
123
|
-
expected: Filipchuk
|
124
|
-
- source: Харків
|
125
|
-
expected: Kharkiv
|
126
|
-
- source: Христина
|
127
|
-
expected: Khrystyna
|
128
|
-
- source: Біла Церква
|
129
|
-
expected: Bila Tserkva
|
130
|
-
- source: Стеценко
|
131
|
-
expected: Stetsenko
|
132
|
-
- source: Чернівці
|
133
|
-
expected: Chernivtsi
|
134
|
-
- source: Шевченко
|
135
|
-
expected: Shevchenko
|
136
|
-
- source: Шостка
|
137
|
-
expected: Shostka
|
138
|
-
- source: Кишеньки
|
139
|
-
expected: Kyshenky
|
140
|
-
- source: Щербухи
|
141
|
-
expected: Shcherbukhy
|
142
|
-
- source: Гоща
|
143
|
-
expected: Hoshcha
|
144
|
-
- source: Гаращенко
|
145
|
-
expected: Harashchenko
|
146
|
-
- source: Яготин
|
147
|
-
expected: Yahotyn
|
148
|
-
- source: Ярошенко
|
149
|
-
expected: Yaroshenko
|
150
|
-
- source: Костянтин
|
151
|
-
expected: Kostiantyn
|
152
|
-
- source: Знам'янка
|
153
|
-
expected: Znamianka
|
154
|
-
- source: Феодосія
|
155
|
-
expected: Feodosiia
|
156
|
-
- source: Згорани
|
157
|
-
expected: Zghorany
|
158
|
-
- source: Розгон
|
159
|
-
expected: Rozghon
|
160
|
-
|
161
|
-
map:
|
162
|
-
inherit: ua-ukr-Cyrl-Latn-2010
|
@@ -1,405 +0,0 @@
|
|
1
|
-
---
|
2
|
-
authority_id: ungegn
|
3
|
-
id: 2017
|
4
|
-
language: iso-639-2:ara
|
5
|
-
source_script: Arab
|
6
|
-
destination_script: Latn
|
7
|
-
name: ROMANIZATION OF ARABIC -- UNGEGN 2017 System
|
8
|
-
url: https://www.eki.ee/wgrs/rom1_ur.htm
|
9
|
-
creation_date: 2017
|
10
|
-
confirmation date: 2018-06
|
11
|
-
description: |
|
12
|
-
The United Nations recommended system was approved in 1972 (
|
13
|
-
II/11) and amended in 1977 (III/12), based on a report
|
14
|
-
prepared by D. N. Sharma. The tables and their corrections
|
15
|
-
were published in volume II of the conference reports1,2.
|
16
|
-
|
17
|
-
There is no evidence of the use of the system either in
|
18
|
-
Pakistan, India or in international cartographic products.
|
19
|
-
Instead, in Pakistan the Hunterian system is officially
|
20
|
-
used3. The resolutions III/12 (1977) and IV/17 (1982)
|
21
|
-
recommended association, inter alia, with Pakistan, in
|
22
|
-
carrying out further studies on the system.
|
23
|
-
|
24
|
-
Urdu (Urdū) uses the Perso-Arabic script which is written
|
25
|
-
from right to left. In the script vowel points are usually
|
26
|
-
omitted which makes it difficult to obtain uniform
|
27
|
-
romanizations. Some of the Arabic consonants are
|
28
|
-
undifferentiated in romanization which means that the
|
29
|
-
system is not fully reversible.
|
30
|
-
notes:
|
31
|
-
- A If preceded by short a, it is romanized ‘ā, e.g. مَعمُل M‘āmul.
|
32
|
-
- B When و is imperceptible, e.g. in a few words of Persian origin when preceded by خ (ḳh).
|
33
|
-
- C Word-finally after a short vowel.
|
34
|
-
- D Marks aspiration of consonants.
|
35
|
-
- E The character ے is used only word-finally.
|
36
|
-
tests:
|
37
|
-
- source: بوغدِی
|
38
|
-
expected: Bvghdī
|
39
|
-
|
40
|
-
- source: مَعمُل
|
41
|
-
expected: M‘āmul
|
42
|
-
|
43
|
-
- source: پَالِير
|
44
|
-
expected: Pālīr
|
45
|
-
|
46
|
-
- source: بیزوت كَلے
|
47
|
-
expected: Byzvt Kale
|
48
|
-
|
49
|
-
- source: عَمَل كوٹ
|
50
|
-
expected: ‘Amal Kvṭ
|
51
|
-
|
52
|
-
- source: ثَابِر
|
53
|
-
expected: Sābir
|
54
|
-
|
55
|
-
- source: شَاه نَثَار ميلة
|
56
|
-
expected: Shāh Nasār Mylah
|
57
|
-
|
58
|
-
- source: چَپرِی
|
59
|
-
expected: Chaprī
|
60
|
-
|
61
|
-
- source: أَحمَد خَان كَلے
|
62
|
-
expected: Ahmad Ḳhān Kale
|
63
|
-
|
64
|
-
- source: دُرَانِي
|
65
|
-
expected: Durānī
|
66
|
-
|
67
|
-
- source: ڈَنگِیلا
|
68
|
-
expected: Ḍangīlā
|
69
|
-
|
70
|
-
- source: ذَرَانِی
|
71
|
-
expected: Zarānī
|
72
|
-
|
73
|
-
- source: بُركِي
|
74
|
-
expected: Burkī
|
75
|
-
|
76
|
-
- source: گِیدَڑَه
|
77
|
-
expected: Gīdaṙah
|
78
|
-
|
79
|
-
- source: عَلِي زَائِي
|
80
|
-
expected: ‘Alī Zā-ī
|
81
|
-
|
82
|
-
- source: ژوب
|
83
|
-
expected: Ỵvb
|
84
|
-
|
85
|
-
- source: بِسَاتُو
|
86
|
-
expected: Bisātū
|
87
|
-
|
88
|
-
- source: أَحمَدِي شَامَا
|
89
|
-
expected: Ahmadī Shāmā
|
90
|
-
|
91
|
-
- source: اَصَالَت كَلے
|
92
|
-
expected: Asālat Kale
|
93
|
-
|
94
|
-
- source: خَضَر خَان
|
95
|
-
expected: Ḳhazar Ḳhān
|
96
|
-
|
97
|
-
- source: سُلْطَان
|
98
|
-
expected: Sultān
|
99
|
-
|
100
|
-
- source: عَزَم سَيِّد نُور كَلے
|
101
|
-
expected: ‘Azam Sayyid Nūr Kale
|
102
|
-
|
103
|
-
- source: بغَاكِي
|
104
|
-
expected: Bghākī
|
105
|
-
|
106
|
-
- source: حَقدَرَه
|
107
|
-
expected: Haqdarah
|
108
|
-
|
109
|
-
- source: کَچکِینَہ
|
110
|
-
expected: Kachkīnaḥ
|
111
|
-
|
112
|
-
- source: بَاگَن
|
113
|
-
expected: Bāgan
|
114
|
-
|
115
|
-
- source: بُلبَلَک
|
116
|
-
expected: Bulbalak
|
117
|
-
|
118
|
-
- source: بِلیَامِین
|
119
|
-
expected: Bilyāmīn
|
120
|
-
|
121
|
-
- source: نَہر
|
122
|
-
expected: Nahr
|
123
|
-
|
124
|
-
- source: اَرَوْالِی
|
125
|
-
expected: Arawālī
|
126
|
-
|
127
|
-
- source: مَہردِی
|
128
|
-
expected: Mahrdī
|
129
|
-
|
130
|
-
- source: بَڑھ
|
131
|
-
expected: Baṙh
|
132
|
-
|
133
|
-
- source: یَاردَا کَلے
|
134
|
-
expected: Yārdā Kale
|
135
|
-
|
136
|
-
- source: بهَائِي خَان
|
137
|
-
expected: Bhā-ī Ḳhān
|
138
|
-
|
139
|
-
- source: پھاشک
|
140
|
-
expected: Phāshk
|
141
|
-
|
142
|
-
- source: تھَلّ
|
143
|
-
expected: Thall
|
144
|
-
|
145
|
-
- source: پَٹھان ريَا
|
146
|
-
expected: Paṭhān Ryā
|
147
|
-
|
148
|
-
- source: جھِیل
|
149
|
-
expected: Jhīl
|
150
|
-
|
151
|
-
- source: غَزْنِي سْپِين
|
152
|
-
expected: Ghaznī Spīn
|
153
|
-
|
154
|
-
- source: بَادشَاه چھُم
|
155
|
-
expected: Bādshāh Chhum
|
156
|
-
|
157
|
-
- source: سِندھ
|
158
|
-
expected: Sindh
|
159
|
-
|
160
|
-
- source: ڈھَنڈ
|
161
|
-
expected: Ḍhanḍ
|
162
|
-
|
163
|
-
- source: خَان گھَڑِی
|
164
|
-
expected: Ḳhān Ghaṙī
|
165
|
-
|
166
|
-
- source: غُلَامَک كَلے
|
167
|
-
expected: Ghulāmak Kale
|
168
|
-
|
169
|
-
- source: خَپیَنگا
|
170
|
-
expected: Ḳhapyangā
|
171
|
-
|
172
|
-
- source: گَندَه كَلے
|
173
|
-
expected: Gandah Kale
|
174
|
-
|
175
|
-
- source: مَورپِتھِی
|
176
|
-
expected: Maurpithī
|
177
|
-
|
178
|
-
- source: درے پلارِی
|
179
|
-
expected: Dre Plārī
|
180
|
-
|
181
|
-
- source: آگرَہ
|
182
|
-
expected: Āgraḥ
|
183
|
-
|
184
|
-
- source: ڈَنڈَر
|
185
|
-
expected: Ḍanḍar
|
186
|
-
|
187
|
-
- source: گُبازانَہ
|
188
|
-
expected: Gubāzānaḥ
|
189
|
-
|
190
|
-
- source: حَےدَر عَلِی كَلے
|
191
|
-
expected: Haidar ‘Alī Kale
|
192
|
-
|
193
|
-
- source: تَودَہ چِینَہ
|
194
|
-
expected: Taudaḥ Chīnaḥ
|
195
|
-
|
196
|
-
- source: مُوسى خَان كَلے
|
197
|
-
expected: Mūsá Ḳhān Kale
|
198
|
-
|
199
|
-
- source: مُلَّا بَاغ
|
200
|
-
expected: Mullā Bāgh
|
201
|
-
|
202
|
-
map:
|
203
|
-
postrules:
|
204
|
-
- pattern: (?<=\b)(?<!\b[‘|’|'|-])[\u0061-\uFFFF]
|
205
|
-
result: "upcase"
|
206
|
-
# don't capitalize defined article in the middle of a sentence
|
207
|
-
- pattern : ' At T' # الت
|
208
|
-
result: ' at T'
|
209
|
-
- pattern : ' As̄ S̄' # الث
|
210
|
-
result: ' as̄ S̄'
|
211
|
-
- pattern : ' Ad D' # الد
|
212
|
-
result: ' ad D'
|
213
|
-
- pattern : ' Az Z' # الذ
|
214
|
-
result: ' az Z'
|
215
|
-
- pattern : ' Ar R' # الر
|
216
|
-
result: ' ar R'
|
217
|
-
- pattern : ' Az Z' # الز
|
218
|
-
result: ' az Z'
|
219
|
-
- pattern : ' As S' # الس
|
220
|
-
result: ' as S'
|
221
|
-
- pattern : ' Ash Sh' # الش
|
222
|
-
result: ' ash Sh'
|
223
|
-
- pattern : ' As S' # الص
|
224
|
-
result: ' as S'
|
225
|
-
- pattern : ' Az Z' # الض
|
226
|
-
result: ' az Z'
|
227
|
-
- pattern : ' At T' # الط
|
228
|
-
result: ' at T'
|
229
|
-
- pattern : ' Az Z' # الظ
|
230
|
-
result: ' az Z'
|
231
|
-
- pattern : ' Al L' # الل
|
232
|
-
result: ' al L'
|
233
|
-
- pattern : ' An N' # الن
|
234
|
-
result: ' an N'
|
235
|
-
- pattern: " Al " # ال
|
236
|
-
result: " al "
|
237
|
-
characters:
|
238
|
-
# special rules
|
239
|
-
|
240
|
-
'\s(?=\u0622\u0628\u064E\u0627\u062F)': '' # space followed by abad is removed
|
241
|
-
'\ufdf2': 'Allāh' # See note 5
|
242
|
-
|
243
|
-
# Vowels, Diphthongs, and Diacritical Marks
|
244
|
-
'\u064e' : 'a' # َ fatha
|
245
|
-
'\u064e\u0627' : 'ā' # ـَا fatha followed by ا
|
246
|
-
'\u0627' : 'ā' # ا
|
247
|
-
'\u0649\u0670' : 'ā' # ىٰ
|
248
|
-
'\u06D2\u0670' : 'ā' # ےٰ
|
249
|
-
'\u0622' : 'ā' # آ
|
250
|
-
'\b\u0627' : '' # ا
|
251
|
-
'\u064e(?=\u0629)' : '' # َ fatha followed by ta' marboota
|
252
|
-
'\u064e(?=a[h|t])' : '' # َ fatha followed by ta' marboota, handling different order of conversion
|
253
|
-
|
254
|
-
'\u0652' : '' # ْ sokoon
|
255
|
-
'\u0659': 'ê'
|
256
|
-
|
257
|
-
'\u0650' : 'i' # karsra
|
258
|
-
'\u0650[\u064a|\u06cc]' : 'ī' # ـِي kasra followed by ي
|
259
|
-
'\u0650\u06d2\u0652' : 'e' # ـے
|
260
|
-
'\u0650\u06d2' : 'e' # ـے
|
261
|
-
'\u06d2' : 'e' # ـے
|
262
|
-
|
263
|
-
'\u064f' : 'u' # ُ damma
|
264
|
-
'\u064f\u0648' : 'ū' # ـُو damma followed by و
|
265
|
-
'\u064f\u0648\u0652' : 'o' # ـَوْ
|
266
|
-
|
267
|
-
|
268
|
-
'\u064e\u06d2' : 'ai' # ـے
|
269
|
-
'\u064e\u0648' : 'au' # ـَو
|
270
|
-
'\u0670': 'á' # ىٰ
|
271
|
-
'\u0649': 'á' # ىٰ
|
272
|
-
|
273
|
-
# shadda
|
274
|
-
'\u0628\u0651' : 'bb' # ب
|
275
|
-
'\u062a\u0651' : 'tt' # ت
|
276
|
-
'\u062b\u0651' : 'ss' # ث
|
277
|
-
'\u062c\u0651' : 'jj' # ج
|
278
|
-
'\u062d\u0651' : 'hh' # ح
|
279
|
-
'\u062e\u0651' : 'ḳhḳh' # خ
|
280
|
-
'\u062f\u0651' : 'dd' # د
|
281
|
-
'\u0630\u0651' : 'zz' # ذ
|
282
|
-
'\u0631\u0651' : 'rr' # ر
|
283
|
-
'\u0632\u0651' : 'zz' # ز
|
284
|
-
'\u0633\u0651' : 'ss' # س
|
285
|
-
'\u0634\u0651' : 'sh' # ش
|
286
|
-
'\u0635\u0651' : 'ss' # ص
|
287
|
-
'\u0636\u0651' : 'ḏḏ' # ض
|
288
|
-
'\u0637\u0651' : 'tt' # ط
|
289
|
-
'\u0638\u0651' : 'zz' # ظ
|
290
|
-
'\u063a\u0651' : 'ghgh' # غ
|
291
|
-
'\u0641\u0651' : 'ff' # ف
|
292
|
-
'\u0642\u0651' : 'qq' # ق
|
293
|
-
'\u0643\u0651' : 'kk' # ك
|
294
|
-
'\u0644\u0651' : 'll' # ل
|
295
|
-
'\u0645\u0651' : 'mm' # م
|
296
|
-
'\u0646\u0651' : 'nn' # ن
|
297
|
-
'\u0647\u0651' : 'hh' # ه
|
298
|
-
'\u0648\u0651' : 'vv' # و
|
299
|
-
'[\u064a|\u06cc]\u0651' : 'yy' # ي
|
300
|
-
|
301
|
-
# NOTE 1
|
302
|
-
'\u0650\b' : '-e' # ِ kasra
|
303
|
-
'\u0674' : '-e' # ٴ
|
304
|
-
'\u0654' : '-e' # ٔ
|
305
|
-
|
306
|
-
'\u0650\u064a\u0651\u064e' : 'īy' # ـِيَّ
|
307
|
-
'\u0650\u064a(?=\u064e|u064f)' : 'iy' # ـِي kasra followed by ي
|
308
|
-
'\u064e\u0649' : 'ay' # ـَى fatha followed by ى which is ا not ي
|
309
|
-
'\u064e\u0648\u0652' : 'aw' # ـَوْ
|
310
|
-
'\u064e\u064a\u0652' : 'ay' # ـَيْ
|
311
|
-
'\u0650\u06cc\u0651\u064e' : 'īy' # ـِيَّ
|
312
|
-
'\u064e\u064a' : 'aī' # ـَي
|
313
|
-
'\u064e\u06cc' : 'aī' # ـَي
|
314
|
-
# - '-ye'
|
315
|
-
|
316
|
-
|
317
|
-
# ta' marboota
|
318
|
-
'\u0629' : 'at' # ة in the middle of the sentence
|
319
|
-
'\u0629$' : 'ah'
|
320
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{2})\u0629' : 'ah'
|
321
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{3})\u0629' : 'ah'
|
322
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{4})\u0629' : 'ah'
|
323
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{5})\u0629' : 'ah'
|
324
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{6})\u0629' : 'ah'
|
325
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{7})\u0629' : 'ah'
|
326
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{8})\u0629' : 'ah'
|
327
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{9})\u0629' : 'ah'
|
328
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{10})\u0629' : 'ah'
|
329
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{11})\u0629' : 'ah'
|
330
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{12})\u0629' : 'ah'
|
331
|
-
'(?<=\b\u0627\u0644[\u0600-\u06ff]{13})\u0629' : 'ah'
|
332
|
-
|
333
|
-
|
334
|
-
|
335
|
-
'\u0621' : '-' # ء
|
336
|
-
'\u0624' : '-' # ؤ
|
337
|
-
'\u0626' : '-' # ئ
|
338
|
-
|
339
|
-
'\u0623' : '' # أ
|
340
|
-
'\u0625' : '' # إ
|
341
|
-
# See note B
|
342
|
-
'\b\u0627\u0644' : 'al ' # ال
|
343
|
-
# '\uFE8E' : '' # ﺎ
|
344
|
-
|
345
|
-
# Sun letters
|
346
|
-
'\b\u0627\u0644\u062a' : 'at t' # الت
|
347
|
-
'\b\u0627\u0644\u062b' : 'as s' # الث
|
348
|
-
'\b\u0627\u0644\u062f' : 'ad d' # الد
|
349
|
-
'\b\u0627\u0644\u0630' : 'az z' # الذ
|
350
|
-
'\b\u0627\u0644\u0631' : 'ar r' # الر
|
351
|
-
'\b\u0627\u0644\u0632' : 'az z' # الز
|
352
|
-
'\b\u0627\u0644\u0633' : 'as s' # الس
|
353
|
-
'\b\u0627\u0644\u0634' : 'ash sh' # الش
|
354
|
-
'\b\u0627\u0644\u0635' : 'as s' # الص
|
355
|
-
'\b\u0627\u0644\u0636' : 'az z' # الض
|
356
|
-
'\b\u0627\u0644\u0637' : 'at t' # الط
|
357
|
-
'\b\u0627\u0644\u0638' : 'az z' # الظ
|
358
|
-
'\b\u0627\u0644\u0644' : 'al l' # الل
|
359
|
-
'\b\u0627\u0644\u0646' : 'an n' # الن
|
360
|
-
|
361
|
-
|
362
|
-
# consonant characters
|
363
|
-
|
364
|
-
'\u0628' : 'b' # ب
|
365
|
-
'\u067E' : 'p' # پ
|
366
|
-
'\u062a' : 't' # ت
|
367
|
-
'\u0679' : 'ṭ' # ٹ
|
368
|
-
'\u062B' : 's' # ث
|
369
|
-
'\u062c' : 'j' # ج
|
370
|
-
'\u0686' : 'ch' # چ
|
371
|
-
'\u062d' : 'h' # ح
|
372
|
-
'\u062e' : 'ḳh' # خ
|
373
|
-
'\u062f' : 'd' # د
|
374
|
-
'\u0688' : 'ḍ' # ڈ
|
375
|
-
'\u0630' : 'z' # ذ
|
376
|
-
'\u0631' : 'r' # ر
|
377
|
-
'\u0691' : 'ṙ' # ڑ
|
378
|
-
'\u0632' : 'z' # ز
|
379
|
-
'\u0698' : 'ỵ' # ژ
|
380
|
-
'\u0633' : 's' # س
|
381
|
-
'\u0634' : 'sh' # ش
|
382
|
-
'\u0635' : 's' # ص
|
383
|
-
'\u0636' : 'z' # ض
|
384
|
-
'\u0637' : 't' # ط
|
385
|
-
'\u0638' : 'z' # ظ
|
386
|
-
'\u0639' : '‘' # ع
|
387
|
-
'\u064e\u0639' : '‘ā' # ع NOTE A
|
388
|
-
'\u063a' : 'gh' # غ
|
389
|
-
'\u0641' : 'f' # ف
|
390
|
-
'\u0642' : 'q' # ق
|
391
|
-
'\u0643' : 'k' # ك
|
392
|
-
'\u06A9' : 'k' # ک
|
393
|
-
'\u06AF' : 'g' # گ
|
394
|
-
'\u0644' : 'l' # ل
|
395
|
-
'\u0645' : 'm' # م
|
396
|
-
'[\u06BA|\u0646]' : 'n' # ن, ں
|
397
|
-
'[\ufba9|\u06c1]' : 'h' # ہ , ﮩ
|
398
|
-
'(?<=[\u064e|\u0650|\u064f])[\ufba9|\u06c1]\b' : 'ḥ' # ہ , ﮩ NOTE C
|
399
|
-
'[\u0647|\u06be]' : 'h' # ه, ھ
|
400
|
-
'\u0648' : 'v' # و
|
401
|
-
'(?<=\u062e)\u0648' : 'ẉ' # و NOTE B
|
402
|
-
'[\u064a|\u06cc]' : 'y' # ي
|
403
|
-
# '\u0649' : 'y' # ي
|
404
|
-
'\u06D0' : 'ē' # ې
|
405
|
-
'\u06CD' : 'êy' # ۍ
|