interscript 0.1.4 → 2.0.5

Sign up to get free protection for your applications and to get access to all the features.
Files changed (183) hide show
  1. checksums.yaml +4 -4
  2. data/.gitignore +11 -0
  3. data/.rspec +3 -0
  4. data/Gemfile +29 -0
  5. data/LICENSE.adoc +31 -0
  6. data/README.md +3 -0
  7. data/Rakefile +53 -0
  8. data/bin/console +14 -0
  9. data/bin/interscript +3 -39
  10. data/bin/maps_analyze_staging +168 -0
  11. data/bin/maps_debug_compilers +58 -0
  12. data/bin/maps_debug_ordering +88 -0
  13. data/bin/maps_debug_ruby_compile +24 -0
  14. data/bin/maps_debug_step_by_step +44 -0
  15. data/bin/maps_optimize_order +112 -0
  16. data/bin/maps_v1_analyze_regexps +45 -0
  17. data/bin/maps_v1_to_v2 +426 -0
  18. data/exe/interscript +6 -0
  19. data/interscript.gemspec +31 -0
  20. data/lib/interscript.rb +76 -128
  21. data/lib/interscript/command.rb +6 -5
  22. data/lib/interscript/compiler.rb +22 -0
  23. data/lib/interscript/compiler/javascript.rb +292 -0
  24. data/lib/interscript/compiler/ruby.rb +262 -0
  25. data/lib/interscript/dsl.rb +67 -0
  26. data/lib/interscript/dsl/aliases.rb +23 -0
  27. data/lib/interscript/dsl/document.rb +46 -0
  28. data/lib/interscript/dsl/group.rb +45 -0
  29. data/lib/interscript/dsl/group/parallel.rb +6 -0
  30. data/lib/interscript/dsl/items.rb +89 -0
  31. data/lib/interscript/dsl/metadata.rb +26 -0
  32. data/lib/interscript/dsl/stage.rb +6 -0
  33. data/lib/interscript/dsl/symbol_mm.rb +11 -0
  34. data/lib/interscript/dsl/tests.rb +12 -0
  35. data/lib/interscript/interpreter.rb +251 -0
  36. data/lib/interscript/node.rb +25 -0
  37. data/lib/interscript/node/alias_def.rb +15 -0
  38. data/lib/interscript/node/dependency.rb +13 -0
  39. data/lib/interscript/node/document.rb +45 -0
  40. data/lib/interscript/node/group.rb +34 -0
  41. data/lib/interscript/node/group/parallel.rb +9 -0
  42. data/lib/interscript/node/group/sequential.rb +2 -0
  43. data/lib/interscript/node/item.rb +52 -0
  44. data/lib/interscript/node/item/alias.rb +42 -0
  45. data/lib/interscript/node/item/any.rb +61 -0
  46. data/lib/interscript/node/item/capture.rb +50 -0
  47. data/lib/interscript/node/item/group.rb +51 -0
  48. data/lib/interscript/node/item/repeat.rb +40 -0
  49. data/lib/interscript/node/item/stage.rb +23 -0
  50. data/lib/interscript/node/item/string.rb +51 -0
  51. data/lib/interscript/node/metadata.rb +18 -0
  52. data/lib/interscript/node/rule.rb +6 -0
  53. data/lib/interscript/node/rule/funcall.rb +18 -0
  54. data/lib/interscript/node/rule/run.rb +15 -0
  55. data/lib/interscript/node/rule/sub.rb +65 -0
  56. data/lib/interscript/node/stage.rb +19 -0
  57. data/lib/interscript/node/tests.rb +15 -0
  58. data/lib/interscript/stdlib.rb +211 -0
  59. data/lib/interscript/utils/regexp_converter.rb +283 -0
  60. data/lib/interscript/version.rb +1 -1
  61. data/requirements.txt +1 -0
  62. metadata +73 -223
  63. data/README.adoc +0 -297
  64. data/bin/rspec +0 -29
  65. data/lib/g2pwrapper.py +0 -34
  66. data/lib/interscript/mapping.rb +0 -125
  67. data/lib/model-7 +0 -0
  68. data/lib/tha-pt-b-7 +0 -0
  69. data/maps/acadsin-zho-Hani-Latn-2002.yaml +0 -38912
  70. data/maps/alalc-aze-Cyrl-Latn-1997.yaml +0 -141
  71. data/maps/alalc-bel-cyrl-latn-1997.yaml +0 -125
  72. data/maps/alalc-ben-Beng-Latn-2017.yaml +0 -130
  73. data/maps/alalc-bul-Cyrl-Latn-1997.yaml +0 -94
  74. data/maps/alalc-ell-Grek-Latn-1997.yaml +0 -625
  75. data/maps/alalc-ell-Grek-Latn-2010.yaml +0 -628
  76. data/maps/alalc-kat-Geok-Latn-1997.yaml +0 -112
  77. data/maps/alalc-kat-Geor-Latn-1997.yaml +0 -146
  78. data/maps/alalc-kor-Hang-Latn-1997.yaml +0 -94
  79. data/maps/alalc-mkd-Cyrl-Latn-2013.yaml +0 -103
  80. data/maps/alalc-mkd-cyrl-latn-1997.yaml +0 -114
  81. data/maps/alalc-rus-Cyrl-Latn-1997.yaml +0 -222
  82. data/maps/alalc-rus-Cyrl-Latn-2012.yaml +0 -162
  83. data/maps/alalc-srp-Cyrl-Latn-1997.yaml +0 -114
  84. data/maps/alalc-srp-cyrl-latn-2013.yaml +0 -135
  85. data/maps/alalc-ukr-Cyrl-Latn-1997.yaml +0 -141
  86. data/maps/alalc-ukr-Cyrl-Latn-2011.yaml +0 -16
  87. data/maps/apcbg-bul-Cyrl-Latn-1995.yaml +0 -283
  88. data/maps/bas-rus-Cyrl-Latn-2017-bss.yaml +0 -175
  89. data/maps/bas-rus-Cyrl-Latn-2017-oss.yaml +0 -169
  90. data/maps/bgn-jpn-Hrkt-Latn-1962.yaml +0 -294
  91. data/maps/bgn-kor-Hang-Latn-1943.yaml +0 -31
  92. data/maps/bgn-kor-Kore-Latn-1943.yaml +0 -31
  93. data/maps/bgna-bul-Cyrl-Latn-2006.yaml +0 -208
  94. data/maps/bgna-bul-Cyrl-Latn-2009.yaml +0 -208
  95. data/maps/bgnpcgn-arm-Armn-Latn-1981.yaml +0 -108
  96. data/maps/bgnpcgn-aze-Cyrl-Latn-1993.yaml +0 -104
  97. data/maps/bgnpcgn-bak-Cyrl-Latn-2007.yaml +0 -184
  98. data/maps/bgnpcgn-bel-cyrl-latn-1979.yaml +0 -285
  99. data/maps/bgnpcgn-bul-Cyrl-Latn-1952.yaml +0 -115
  100. data/maps/bgnpcgn-bul-Cyrl-Latn-2013.yaml +0 -38
  101. data/maps/bgnpcgn-chn-Hans-Latn-1979.yaml +0 -7456
  102. data/maps/bgnpcgn-ell-Grek-Latn-1962.yaml +0 -702
  103. data/maps/bgnpcgn-ell-Grek-Latn-1996.yaml +0 -20
  104. data/maps/bgnpcgn-jpn-Hrkt-Latn-1976.yaml +0 -257
  105. data/maps/bgnpcgn-kat-Geor-Latn-1981.yaml +0 -127
  106. data/maps/bgnpcgn-kat-Geor-Latn-2009.yaml +0 -43
  107. data/maps/bgnpcgn-kor-Hang-Latn-kn-1945.yaml +0 -253
  108. data/maps/bgnpcgn-kor-Hang-Latn-rok-2011.yaml +0 -48
  109. data/maps/bgnpcgn-kor-Kore-Latn-rok-2011.yaml +0 -48
  110. data/maps/bgnpcgn-mkd-Cyrl-Latn-1981.yaml +0 -159
  111. data/maps/bgnpcgn-mkd-Cyrl-Latn-2013.yaml +0 -190
  112. data/maps/bgnpcgn-per-Arab-Latn-1956.yaml +0 -93
  113. data/maps/bgnpcgn-rus-Cyrl-Latn-1947.yaml +0 -314
  114. data/maps/bgnpcgn-srp-Cyrl-Latn-2005.yaml +0 -166
  115. data/maps/bgnpcgn-ukr-Cyrl-Latn-1965.yaml +0 -163
  116. data/maps/bgnpcgn-ukr-Cyrl-Latn-2019.yaml +0 -208
  117. data/maps/by-bel-Cyrl-Latn-1998.yaml +0 -168
  118. data/maps/by-bel-Cyrl-Latn-2007.yaml +0 -115
  119. data/maps/elot-ell-Grek-Latn-743-1982-tl.yaml +0 -685
  120. data/maps/elot-ell-Grek-Latn-743-1982-ts.yaml +0 -681
  121. data/maps/elot-ell-Grek-Latn-743-2001-tl.yaml +0 -20
  122. data/maps/elot-ell-Grek-Latn-743-2001-ts.yaml +0 -32
  123. data/maps/ggg-kat-Geor-Latn-2002.yaml +0 -89
  124. data/maps/gki-bel-cyrl-latn-1992.yaml +0 -33
  125. data/maps/gki-bel-cyrl-latn-2000.yaml +0 -201
  126. data/maps/gost-rus-cyrl-latn-16876-71-1983.yaml +0 -186
  127. data/maps/hk-yue-Hani-Latn-1888.yaml +0 -38497
  128. data/maps/icao-bel-Cyrl-Latn-9303.yaml +0 -141
  129. data/maps/icao-bul-Cyrl-Latn-9303.yaml +0 -122
  130. data/maps/icao-heb-Hebr-Latn-9303.yaml +0 -151
  131. data/maps/icao-mkd-Cyrl-Latn-9303.yaml +0 -117
  132. data/maps/icao-per-Arab-Latn-9303.yaml +0 -104
  133. data/maps/icao-rus-Cyrl-Latn-9303.yaml +0 -118
  134. data/maps/icao-srp-Cyrl-Latn-9303.yaml +0 -117
  135. data/maps/icao-ukr-Cyrl-Latn-9303.yaml +0 -120
  136. data/maps/iso-ell-Grek-Latn-843-1997-t1.yaml +0 -610
  137. data/maps/iso-ell-Grek-Latn-843-1997-t2.yaml +0 -41
  138. data/maps/iso-jpn-Hrkt-Latn-3602-1989.yaml +0 -62
  139. data/maps/iso-rus-Cyrl-Latn-9-1995.yaml +0 -272
  140. data/maps/iso-tha-Thai-Latn-11940-1998.yaml +0 -109
  141. data/maps/kp-kor-Hang-Latn-2002.yaml +0 -901
  142. data/maps/lshk-yue-Hani-Latn-jyutping-1993.yaml +0 -44820
  143. data/maps/mext-jpn-Hrkt-Latn-1954.yaml +0 -411
  144. data/maps/moct-kor-Hang-Latn-2000.yaml +0 -803
  145. data/maps/mofa-jpn-Hrkt-Latn-1989.yaml +0 -541
  146. data/maps/mvd-bel-Cyrl-Latn-2008.yaml +0 -225
  147. data/maps/mvd-bel-Cyrl-Latn-2010.yaml +0 -63
  148. data/maps/mvd-rus-Cyrl-Latn-2008.yaml +0 -110
  149. data/maps/mvd-rus-Cyrl-Latn-2010.yaml +0 -37
  150. data/maps/nil-kor-Hang-Hang-jamo.yaml +0 -11193
  151. data/maps/odni-bel-Cyrl-Latn-2015.yaml +0 -148
  152. data/maps/odni-bul-Cyrl-Latn-2015.yaml +0 -96
  153. data/maps/odni-kat-Geor-Latn-2015.yaml +0 -88
  154. data/maps/odni-rus-Cyrl-Latn-2015.yaml +0 -77
  155. data/maps/odni-srp-Cyrl-Latn-2015.yaml +0 -129
  156. data/maps/odni-ukr-Cyrl-Latn-2015.yaml +0 -157
  157. data/maps/odni-uzb-Cyrl-Latn-2015.yaml +0 -167
  158. data/maps/royin-tha-Thai-Latn-1939-generic.yaml +0 -90
  159. data/maps/royin-tha-Thai-Latn-1968.yaml +0 -179
  160. data/maps/royin-tha-Thai-Latn-1999-chained.yaml +0 -180
  161. data/maps/royin-tha-Thai-Latn-1999.yaml +0 -76
  162. data/maps/sac-zho-Hans-Latn-1979.yaml +0 -24759
  163. data/maps/stategeocadastre-ukr-Cyrl-Latn-1993.yaml +0 -222
  164. data/maps/ua-ukr-Cyrl-Latn-1996.yaml +0 -193
  165. data/maps/un-bel-Cyrl-Latn-2007.yaml +0 -114
  166. data/maps/un-ben-Beng-Latn-2016.yaml +0 -534
  167. data/maps/un-ell-Grek-Latn-1987-tl.yaml +0 -32
  168. data/maps/un-ell-Grek-Latn-1987-ts.yaml +0 -20
  169. data/maps/un-ell-Grek-Latn-phonetic-1987.yaml +0 -780
  170. data/maps/un-mon-Mong-Latn-2013.yaml +0 -93
  171. data/maps/un-rus-Cyrl-Latn-1987.yaml +0 -166
  172. data/maps/un-ukr-cyrl-latn-1998.yaml +0 -30
  173. data/maps/var-jpn-Hrkt-Latn-hepburn-1886.yaml +0 -406
  174. data/maps/var-jpn-Hrkt-Latn-hepburn-1954.yaml +0 -386
  175. data/maps/var-kor-Hang-Latn-mr-1939.yaml +0 -1054
  176. data/maps/var-kor-Kore-Hang-2013.yaml +0 -59754
  177. data/maps/var-kor-Kore-Latn-mr-1939.yaml +0 -37
  178. data/maps/var-tha-Thai-Thai-phonemic.yaml +0 -59
  179. data/maps/var-tha-Thai-Zsym-ipa.yaml +0 -301
  180. data/maps/var-zho-Hani-Latn-1979.yaml +0 -38908
  181. data/spec/interscript/mapping_spec.rb +0 -42
  182. data/spec/interscript_spec.rb +0 -26
  183. data/spec/spec_helper.rb +0 -3
@@ -1,114 +0,0 @@
1
- ---
2
- authority_id: alalc
3
- id: 1997
4
- language: srp
5
- source_script: Cyrl
6
- destination_script: Latn
7
- name: Serbian and Makedonian Romanization, ALA-LC 1997 System
8
- url: http://catdir.loc.gov/catdir/cpso/romanization/serbian.pdf
9
- creation_date: 1997
10
- description: ALA-LC Romanization table for Serbian and Makedonian.
11
-
12
- notes:
13
- - Special characters in romanization
14
- Đ - D with crossbar (upper case). USMARC hexadecimal code A3.
15
- đ - d with crossbar (upper case). USMARC hexadecimal code B3.
16
-
17
- - Character modifiers in romanization
18
- ´ - acute. USMARC hexadecimal code E2.
19
- ˇ - hachek. USMARC hexadecimal code E9.
20
-
21
- tests:
22
- - source: Општина Ердут
23
- expected: Opština Erdut
24
- - source: Општина Двор
25
- expected: Opština Dvor
26
- - source: ЛУЃЕ луѓе
27
- expected: LUǴE luǵe
28
- - source: ЅВЕЗДА ѕвезда Ѕвезда
29
- expected: DZVEZDA dzvezda Dzvezda
30
- - source: ЌАРУВАЊЕ ќарување
31
- expected: ḰARUVANJE ḱaruvanje
32
-
33
- map:
34
- postrules:
35
- # DZ
36
- - pattern: "((?<=[[:upper:]])Dz(?=[[:upper:]])?|(?<=[[:upper:]])?Dz(?=[[:upper:]]))"
37
- result: "DZ"
38
- #LJ
39
- - pattern: "((?<=[[:upper:]])Lj(?=[[:upper:]])?|(?<=[[:upper:]])?Lj(?=[[:upper:]]))"
40
- result: "LJ"
41
- #NJ
42
- - pattern: "((?<=[[:upper:]])Nj(?=[[:upper:]])?|(?<=[[:upper:]])?Nj(?=[[:upper:]]))"
43
- result: "NJ"
44
- #DŽ
45
- - pattern: "((?<=[[:upper:]])Dž(?=[[:upper:]])?|(?<=[[:upper:]])?Dž(?=[[:upper:]]))"
46
- result: "DŽ"
47
-
48
- characters:
49
- "\u0410": "A"
50
- "\u0411": "B"
51
- "\u0412": "V"
52
- "\u0413": "G"
53
- "\u0403": "\u01F4" # Ǵ
54
- "\u0414": "D"
55
- "\u0402": "\u0110" # Đ
56
- "\u0415": "E"
57
- "\u0416": "\u005a\u030c" # Ž
58
- "\u0417": "Z"
59
- "\u0405": "Dz"
60
- "\u0418": "I"
61
- "\u0408": "J"
62
- "\u041A": "K"
63
- "\u040C": "\u1E30" # Ḱ
64
- "\u041B": "L"
65
- "\u0409": "Lj"
66
- "\u041C": "M"
67
- "\u041D": "N"
68
- "\u040A": "Nj"
69
- "\u041E": "O"
70
- "\u041F": "P"
71
- "\u0420": "R"
72
- "\u0421": "S"
73
- "\u0422": "T"
74
- "\u040B": "\u0043\u0301" # Ć
75
- "\u0423": "U"
76
- "\u0424": "F"
77
- "\u0425": "H"
78
- "\u0426": "C"
79
- "\u0427": "\u0043\u030c" # Č
80
- "\u040F": "D\u007a\u030c" # Dž
81
- "\u0428": "\u0053\u030c" # Š
82
- "\u0430": "a"
83
- "\u0431": "b"
84
- "\u0432": "v"
85
- "\u0433": "g"
86
- "\u0453": "\u01F5" # ǵ
87
- "\u0434": "d"
88
- "\u0452": "\u0111" # đ
89
- "\u0435": "e"
90
- "\u0436": "\u007a\u030c" # ž
91
- "\u0437": "z"
92
- "\u0455": "dz"
93
- "\u0438": "i"
94
- "\u0458": "j"
95
- "\u043A": "k"
96
- "\u045C": "\u1E31" # ḱ
97
- "\u043B": "l"
98
- "\u0459": "lj"
99
- "\u043C": "m"
100
- "\u043D": "n"
101
- "\u045A": "nj"
102
- "\u043E": "o"
103
- "\u043F": "p"
104
- "\u0440": "r"
105
- "\u0441": "s"
106
- "\u0442": "t"
107
- "\u045B": "\u0063\u0301" # ć
108
- "\u0443": "u"
109
- "\u0444": "f"
110
- "\u0445": "h"
111
- "\u0446": "c"
112
- "\u0447": "\u0063\u030c" # č
113
- "\u045F": "d\u007a\u030c" # dž
114
- "\u0448": "\u0073\u030c" # š
@@ -1,135 +0,0 @@
1
- ---
2
- authority_id: alalc
3
- id: 2013
4
- language: srp
5
- source_script: Cyrl
6
- destination_script: Latn
7
- name: ROMANIZATION OF SERBIAN, ALA-LC 2013 System
8
- url: https://www.loc.gov/catdir/cpso/romanization/serbian.pdf
9
- creation_date: 2013
10
- description: |
11
- Serbian uses the Cyrillic and Roman alphabets. When the Roman alphabet is used it is the same as Croatian.
12
-
13
- tests:
14
- - source: Шупља Стена
15
- expected: Šuplja Stena
16
- - source: Чукарица
17
- expected: Čukarica
18
- - source: Црна Трава
19
- expected: Crna Trava
20
- - source: Херцег Нови
21
- expected: Herceg Novi
22
- - source: Улцињ
23
- expected: Ulcinj
24
- - source: Ужице
25
- expected: Užice
26
- - source: Тресаначка Река
27
- expected: Tresanačka Reka
28
- - source: Сјеница
29
- expected: Sjenica
30
- - source: Рожаје
31
- expected: Rožaje
32
- - source: Пљевља
33
- expected: Pljevlja
34
- - source: Оџаци
35
- expected: Odžaci
36
- - source: Никшић
37
- expected: Nikšić
38
- - source: Медвеђа
39
- expected: Medveđa
40
- - source: Лозница
41
- expected: Loznica
42
- - source: Књажевац
43
- expected: Knjaževac
44
- - source: Зрењанин
45
- expected: Zrenjanin
46
- - source: Житорађа
47
- expected: Žitorađa
48
- - source: Ервеник
49
- expected: Ervenik
50
- - source: Доње Љупче
51
- expected: Donje Ljupče
52
- - source: Гусиње
53
- expected: Gusinje
54
- - source: ГУСИЊЕ
55
- expected: GUSINJE
56
- - source: Врњачка Бања
57
- expected: Vrnjačka Banja
58
- - source: Бијело Поље
59
- expected: Bijelo Polje
60
- - source: Алибунар
61
- expected: Alibunar
62
-
63
- map:
64
- postrules:
65
- #LJ
66
- - pattern: "((?<=[[:upper:]])Lj(?=[[:upper:]])?|(?<=[[:upper:]])?Lj(?=[[:upper:]]))"
67
- result: "LJ"
68
- #NJ
69
- - pattern: "((?<=[[:upper:]])Nj(?=[[:upper:]])?|(?<=[[:upper:]])?Nj(?=[[:upper:]]))"
70
- result: "NJ"
71
- #DŽ
72
- - pattern: "((?<=[[:upper:]])Dž(?=[[:upper:]])?|(?<=[[:upper:]])?Dž(?=[[:upper:]]))"
73
- result: "DŽ"
74
-
75
- characters:
76
- "\u0410": "A"
77
- "\u0411": "B"
78
- "\u0412": "V"
79
- "\u0413": "G"
80
- "\u0414": "D"
81
- "\u0402": "\u0110" # Đ
82
- "\u0415": "E"
83
- "\u0416": "\u005a\u030c" # Ž
84
- "\u0417": "Z"
85
- "\u0418": "I"
86
- "\u0408": "J"
87
- "\u041A": "K"
88
- "\u041B": "L"
89
- "\u0409": "Lj"
90
- "\u041C": "M"
91
- "\u041D": "N"
92
- "\u040A": "Nj"
93
- "\u041E": "O"
94
- "\u041F": "P"
95
- "\u0420": "R"
96
- "\u0421": "S"
97
- "\u0422": "T"
98
- "\u040B": "\u0043\u0301" # Ć
99
- "\u0423": "U"
100
- "\u0424": "F"
101
- "\u0425": "H"
102
- "\u0426": "C"
103
- "\u0427": "\u0043\u030c" # Č
104
- "\u040F": "D\u007a\u030c" # Dž
105
- "\u0428": "\u0053\u030c" # Š
106
- "\u0430": "a"
107
- "\u0431": "b"
108
- "\u0432": "v"
109
- "\u0433": "g"
110
- "\u0434": "d"
111
- "\u0452": "\u0111" # đ
112
- "\u0435": "e"
113
- "\u0436": "\u007a\u030c" # ž
114
- "\u0437": "z"
115
- "\u0438": "i"
116
- "\u0458": "j"
117
- "\u043A": "k"
118
- "\u043B": "l"
119
- "\u0459": "lj"
120
- "\u043C": "m"
121
- "\u043D": "n"
122
- "\u045A": "nj"
123
- "\u043E": "o"
124
- "\u043F": "p"
125
- "\u0440": "r"
126
- "\u0441": "s"
127
- "\u0442": "t"
128
- "\u045B": "\u0063\u0301" # ć́
129
- "\u0443": "u"
130
- "\u0444": "f"
131
- "\u0445": "h"
132
- "\u0446": "c"
133
- "\u0447": "\u0063\u030c" # č
134
- "\u045F": "d\u007a\u030c" # dž
135
- "\u0448": "\u0073\u030c" # š
@@ -1,141 +0,0 @@
1
- ---
2
- authority_id: alalc
3
- id: 1997
4
- language: ukr
5
- source_script: Cyrl
6
- destination_script: Latn
7
- name: ALA-LC Romanization System 1997
8
- url: http://catdir.loc.gov/catdir/cpso/romanization/ukrainia.pdf
9
- creation_date: 1997
10
- description: |
11
- The ALA-LC Romanization Tables, published by the American Library Association (1885)
12
- and Library of Congress (1905). Used to represent bibliographic information by US and
13
- Canadian libraries, by the British Library since 1975,[3] and in North American publications.
14
- The latest 1997 revision is very similar to the 1905 version.
15
-
16
- notes:
17
- - The z͡h ligature is necessary to distinguish ж from the combination зг.
18
- - The t͡s ligature is necessary to distinguish ц from the combination тс.
19
-
20
- tests:
21
- - source: Автономна Республіка Крим
22
- expected: Avtonomna Respublika Krym
23
- - source: Висунь
24
- expected: Vysunʹ
25
- - source: Долинське
26
- expected: Dolynsʹke
27
- - source: Дубище
28
- expected: Dubyshche
29
- - source: Єнакієве
30
- expected: I͡enakii͡eve
31
- - source: Зупиночний Пункт Мокіївці
32
- expected: Zupynochnyĭ Punkt Mokiïvt͡si
33
- - source: Іванівщина
34
- expected: Ivanivshchyna
35
- - source: Киликиїв
36
- expected: Kylykyïv
37
- - source: Кожанка
38
- expected: Koz͡hanka
39
- - source: Краснянка
40
- expected: Krasni͡anka
41
- - source: Краснівка
42
- expected: Krasnivka
43
- - source: Мале Микільське
44
- expected: Male Mykilʹsʹke
45
- - source: Маломиколаївка
46
- expected: Malomykolaïvka
47
- - source: Нове Село
48
- expected: Nove Selo
49
- - source: Новопавлівка
50
- expected: Novopavlivka
51
- - source: Новошичі
52
- expected: Novoshychi
53
- - source: Новоєфремівка
54
- expected: Novoi͡efremivka
55
- - source: Одеська Область
56
- expected: Odesʹka Oblastʹ
57
- - source: Орлівське
58
- expected: Orlivsʹke
59
- - source: Раневичі
60
- expected: Ranevychi
61
- - source: Рокувата
62
- expected: Rokuvata
63
- - source: Рудаєве
64
- expected: Rudai͡eve
65
- - source: Сахнівці
66
- expected: Sakhnivt͡si
67
- - source: Тернівка
68
- expected: Ternivka
69
- - source: Турбівка
70
- expected: Turbivka
71
- - source: Херсонська Область
72
- expected: Khersonsʹka Oblastʹ
73
-
74
- map:
75
- characters:
76
- "\u0410": "A" # А
77
- "\u0411": "B" # Б
78
- "\u0412": "V" # В
79
- "\u0413": "H" # Г
80
- "\u0490": "G" # Ґ
81
- "\u0414": "D" # Д
82
- "\u0415": "E" # Е
83
- "\u0404": "I\u0361e" # Є -> I͡e
84
- "\u0416": "Z\u0361h" # Ж -> Z͡h
85
- "\u0417": "Z" # З
86
- "\u0418": "Y" # И
87
- "\u0406": "I" # І
88
- "\u0407": "I\u0308" # Ї -> Ï
89
- "\u0419": "I\u0306" # Й -> Ĭ
90
- "\u041a": "K" # К
91
- "\u041b": "L" # Л
92
- "\u041c": "M" # М
93
- "\u041d": "N" # Н
94
- "\u041e": "O" # О
95
- "\u041f": "P" # П
96
- "\u0420": "R" # Р
97
- "\u0421": "S" # С
98
- "\u0422": "T" # Т
99
- "\u0423": "U" # У
100
- "\u0424": "F" # Ф
101
- "\u0425": "Kh" # Х
102
- "\u0426": "T\u0361s" # Ц -> T͡s
103
- "\u0427": "Ch" # Ч
104
- "\u0428": "Sh" # Ш
105
- "\u0429": "Shch" # Щ
106
- "\u042e": "I\u0361u" # Ю -> I͡u
107
- "\u042f": "I\u0361a" # Я -> I͡a
108
- "\u042c": "\u02B9" # Ь -> ʹ
109
- "\u0430": "a" # а
110
- "\u0431": "b" # б
111
- "\u0432": "v" # в
112
- "\u0433": "h" # г
113
- "\u0491": "g" # ґ
114
- "\u0434": "d" # д
115
- "\u0435": "e" # е
116
- "\u0454": "i\u0361e" # є -> i͡e
117
- "\u0436": "z\u0361h" # ж -> z͡h
118
- "\u0437": "z" # з
119
- "\u0438": "y" # и
120
- "\u0456": "i" # і
121
- "\u0457": "i\u0308" # ї -> ï
122
- "\u0439": "i\u0306" # й -> ĭ
123
- "\u043a": "k" # к
124
- "\u043b": "l" # л
125
- "\u043c": "m" # м
126
- "\u043d": "n" # н
127
- "\u043e": "o" # о
128
- "\u043f": "p" # п
129
- "\u0440": "r" # р
130
- "\u0441": "s" # с
131
- "\u0442": "t" # т
132
- "\u0443": "u" # у
133
- "\u0444": "f" # ф
134
- "\u0445": "kh" # х
135
- "\u0446": "t\u0361s" # ц -> t͡s
136
- "\u0447": "ch" # ч
137
- "\u0448": "sh" # ш
138
- "\u0449": "shch" # щ
139
- "\u044e": "i\u0361u" # ю -> i͡u
140
- "\u044f": "i\u0361a" # я -> i͡a
141
- "\u044c": "\u02B9" # Ь -> ʹ
@@ -1,16 +0,0 @@
1
- ---
2
- authority_id: alalc
3
- id: 2011
4
- language: ukr
5
- source_script: Cyrl
6
- destination_script: Latn
7
- name: ALA-LC Romanization System 1997
8
- url: https://www.loc.gov/catdir/cpso/romanization/ukrainia.pdf
9
- creation_date: 2011
10
-
11
- notes:
12
- - The z͡h ligature is necessary to distinguish ж from the combination зг.
13
- - The t͡s ligature is necessary to distinguish ц from the combination тс.
14
-
15
- map:
16
- inherit: alalc-ukr-Cyrl-Latn-1997
@@ -1,283 +0,0 @@
1
- ---
2
- authority_id: apcbg
3
- id: 1995
4
- language: bul
5
- source_script: Cyrl
6
- destination_script: Latn
7
- name: Toponymic Guidelines for Antarctica, Lyubomir Ivanov. Antarctic Place-names Commission of Bulgaria, Sofia, 1995.
8
- url: http://www.apcbg.org/Toponymic-Guidelines.htm
9
- creation_date: 1995
10
- description: |
11
- In Antarctica geographical names are important elements of identification, orientation, localization and navigation,
12
- providing an essential reference system for logistic operations, including search and rescue measures, and for
13
- international scientific research. They facilitate information exchange in the field, in scientific publications and in
14
- administrative measures under the Antarctic Treaty System. Geographical names also reflect the history of Antarctic
15
- exploration.
16
-
17
- tests:
18
- - source: нунатак Абрит
19
- expected: nunatak Abrit
20
- - source: връх Академия
21
- expected: vrah Akademiya
22
- - source: връх Ами Буе
23
- expected: vrah Ami Bue
24
- - source: нос Айтос
25
- expected: nos Aytos
26
- - source: залив Баба Тонка
27
- expected: zaliv Baba Tonka
28
- - source: Балабански камък
29
- expected: Balabanski kamak
30
- - source: Бедечки поток
31
- expected: Bedechki potok
32
- - source: нос Бяга
33
- expected: nos Byaga
34
- - source: остров Качо
35
- expected: ostrov Kacho # wiki ostrov Cacho
36
- - source: Чакъров остров
37
- expected: Chakarov ostrov
38
- - source: връх Дъбник
39
- expected: vrah Dabnik
40
- - source: залив Десислава
41
- expected: zaliv Desislava
42
- - source: ледник Джераси
43
- expected: lednik Dzherasi
44
- - source: Джегова скала
45
- expected: Dzhegova skala
46
- - source: Нунатак Едуард
47
- expected: Nunatak Eduard
48
- - source: Елховска седловина
49
- expected: Elhovska sedlovina
50
- - source: ледник Етър
51
- expected: lednik Etar
52
- - source: нунатак Филип Тотю
53
- expected: nunatak Filip Totyu
54
- - source: ледник Габаре
55
- expected: lednik Gabare
56
- - source: риф Гергини
57
- expected: rif Gergini
58
- - source: Гяуров връх
59
- expected: Gyaurov vrah
60
- - source: Гуцалски рид
61
- expected: Gutsalski rid # not found
62
- - source: Хараламбиев остров
63
- expected: Haralambiev ostrov
64
- - source: връх Ичера
65
- expected: vrah Ichera
66
- - source: полуостров Йоан Павел II
67
- expected: poluostrov Yoan Pavel II
68
- - source: нос Иван Александър
69
- expected: nos Ivan Aleksandar
70
- - source: нос Иречек
71
- expected: nos Irechek
72
- - source: нос Кърджали
73
- expected: nos Kardzhali
74
- - source: седловина Кърнаре
75
- expected: sedlovina Karnare
76
- - source: нунатак Керсеблепт
77
- expected: nunatak Kerseblept
78
- - source: Кондофрейски възвишения
79
- expected: Kondofreyski vazvisheniya
80
- - source: Костинбродски проход
81
- expected: Kostinbrodski prohod
82
- - source: връх Кожух
83
- expected: vrah Kozhuh
84
- - source: Кукерски нунатаци
85
- expected: Kukerski nunatatsi
86
- - source: залив Лазурен бряг
87
- expected: zaliv Lazuren bryag
88
- - source: връх Лудогорие
89
- expected: vrah Ludogorie
90
- - source: Лютибродски скали
91
- expected: Lyutibrodski skali
92
- - source: Масларов нунатак
93
- expected: Maslarov nunatak
94
- - source: Михневски връх
95
- expected: Mihnevski vrah
96
- - source: рид Митино
97
- expected: rid Mitino
98
- - source: езеро Наяда
99
- expected: ezero Nayada
100
- - source: нос Никюп
101
- expected: nos Nikyup
102
- - source: рид Оборище
103
- expected: rid Oborishte
104
- - source: залив Олуша
105
- expected: zaliv Olusha
106
- - source: Оряховски възвишения
107
- expected: Oryahovski vazvisheniya
108
- - source: нунатак Памидово
109
- expected: nunatak Pamidovo
110
- - source: връх Парангалица
111
- expected: vrah Parangalitsa
112
- - source: Първомайски провлак
113
- expected: Parvomayski provlak
114
- - source: ледник Патлейна
115
- expected: lednik Patleyna
116
- - source: полуостров Перник
117
- expected: poluostrov Pernik
118
- - source: връх Петко Войвода
119
- expected: vrah Petko Voyvoda
120
- - source: остров Фанагория
121
- expected: ostrov Fanagoriya
122
- - source: нос Плас
123
- expected: nos Plas
124
- - source: Пресиянов рид
125
- expected: Presiyanov rid
126
- - source: връх Принсипе де Астуриас
127
- expected: vrah Prinsipe de Asturias # wiki: vrah Príncipe de Asturias
128
- - source: нунатак Ръченица
129
- expected: nunatak Rachenitsa
130
- - source: връх Райна Княгиня
131
- expected: vrah Rayna Knyaginya
132
- - source: Рид Ръжана
133
- expected: Rid Razhana
134
- - source: връх Ригс
135
- expected: vrah Rigs
136
- - source: остров Рогулят
137
- expected: ostrov Rogulyat
138
- - source: ледник Сабазий
139
- expected: lednik Sabaziy
140
- - source: ледник Съединение
141
- expected: lednik Saedinenie
142
- - source: нунатак Сенокос
143
- expected: nunatak Senokos
144
- - source: Сейдолски камък
145
- expected: Seydolski kamak
146
- - source: ледник Щерна
147
- expected: lednik Shterna
148
- - source: връх Шишман
149
- expected: vrah Shishman
150
- - source: ледник Сигмен
151
- expected: lednik Sigmen
152
- - source: Седловина Синитово
153
- expected: Sedlovina Sinitovo
154
- - source: Ледник Скаплизо
155
- expected: Lednik Skaplizo
156
- - source: залив Слънчев бряг
157
- expected: zaliv Slanchev bryag
158
- - source: остров Соатрис
159
- expected: ostrov Soatris
160
- - source: планина Софийски Университет
161
- expected: planina Sofiyski Universitet
162
- - source: ледник Сребърна
163
- expected: lednik Srebarna
164
- - source: Средногорски възвишения
165
- expected: Srednogorski vazvisheniya
166
- - source: Св. Евтимиев камък
167
- expected: Sv. Evtimiev kamak
168
- - source: база Св. Климент Охридски
169
- expected: baza Sv. Kliment Ohridski
170
- - source: връх Стъргел
171
- expected: vrah Stargel
172
- - source: нунатак Сторгозия
173
- expected: nunatak Storgoziya # nunatak Storgozia according to wiki
174
- - source: нунатак Сурвакари
175
- expected: nunatak Survakari
176
- - source: ледник Световрачене
177
- expected: lednik Svetovrachene
178
- - source: остров Теменуга
179
- expected: ostrov Temenuga
180
- - source: Тракийски възвишения
181
- expected: Trakiyski vazvisheniya
182
- - source: хълм Цамблак
183
- expected: halm Tsamblak
184
- - source: ледник Урдовиза
185
- expected: lednik Urdoviza
186
- - source: остров Вълчедръм
187
- expected: ostrov Valchedram
188
- - source: острови Вардим
189
- expected: ostrovi Vardim
190
- - source: Владигеров проток
191
- expected: Vladigerov protok
192
- - source: ледник Ябланица
193
- expected: lednik Yablanitsa
194
- - source: залив Ямфорина
195
- expected: zaliv Yamforina
196
- - source: Йовков нос
197
- expected: Yovkov nos
198
- - source: рид Заберново
199
- expected: rid Zabernovo
200
- - source: ледник Збелсурд
201
- expected: lednik Zbelsurd
202
- - source: Жефарович камък
203
- expected: Zhefarovich kamak
204
- - source: връх Зиези
205
- expected: vrah Ziezi
206
- - source: залив Златни пясъци
207
- expected: zaliv Zlatni pyasatsi
208
- - source: ледник Злокучене
209
- expected: lednik Zlokuchene
210
- - source: проток Злогош
211
- expected: protok Zlogosh
212
-
213
- map:
214
- characters:
215
- '\u0410': 'A'
216
- '\u0411': 'B'
217
- '\u0412': 'V'
218
- '\u0413': 'G'
219
- '\u0414': 'D'
220
- '\u0415': 'E'
221
- '\u0416': 'Zh'
222
- '\u0417': 'Z'
223
- '\u0418': 'I'
224
- '\u0419': 'Y'
225
- '\u041a': 'K'
226
- '\u041b': 'L'
227
- '\u041c': 'M'
228
- '\u041d': 'N'
229
- '\u041e': 'O'
230
- '\u041f': 'P'
231
- '\u0420': 'R'
232
- '\u0421': 'S'
233
- '\u0422': 'T'
234
- '\u0423': 'U'
235
- '\u0424': 'F'
236
- '\u0425': 'H'
237
- '\u0426': 'Ts'
238
- '\u0427': 'Ch'
239
- '\u0428': 'Sh'
240
- '\u0429': 'Sht'
241
- '\u042a': 'A'
242
- '\u042c': 'Y'
243
- '\u042e': 'Yu'
244
- '\u042f': 'Ya'
245
-
246
- '\u0430': 'a'
247
- '\u0431': 'b'
248
- '\u0432': 'v'
249
- '\u0433': 'g'
250
- '\u0434': 'd'
251
- '\u0435': 'e'
252
- '\u0436': 'zh'
253
- '\u0437': 'z'
254
- '\u0438': 'i'
255
- '\u0439': 'y'
256
- '\u043a': 'k'
257
- '\u043b': 'l'
258
- '\u043c': 'm'
259
- '\u043d': 'n'
260
- '\u043e': 'o'
261
- '\u043f': 'p'
262
- '\u0440': 'r'
263
- '\u0441': 's'
264
- '\u0442': 't'
265
- '\u0443': 'u'
266
- '\u0444': 'f'
267
- '\u0445': 'h'
268
- '\u0446': 'ts'
269
- '\u0447': 'ch'
270
- '\u0448': 'sh'
271
- '\u0449': 'sht'
272
- '\u044a': 'a'
273
- '\u044c': 'y'
274
- '\u044e': 'yu'
275
- '\u044f': 'ya'
276
-
277
- # note 2
278
- '\u046A': "U\u0306" # Ѫ
279
- '\u046B': "u\u0306" # ѫ
280
-
281
- # note[3]
282
- '\u0462': "YE" # Ѣ
283
- '\u0463': "ye" # ѣ