Files
libpostal/src/transliteration_scripts_data.c

95 lines
2.6 KiB
C

#include "unicode_scripts.h"
typedef struct script_transliteration_rule {
script_language_t script_language;
transliterator_index_t index;
} script_transliteration_rule_t;
script_transliteration_rule_t script_transliteration_rules[] = {
{{SCRIPT_THAANA, ""}, {0, 2}},
{{SCRIPT_GURMUKHI, ""}, {2, 1}},
{{SCRIPT_TELUGU, ""}, {3, 1}},
{{SCRIPT_CYRILLIC, "be"}, {4, 1}},
{{SCRIPT_CYRILLIC, "bg"}, {5, 1}},
{{SCRIPT_CYRILLIC, ""}, {6, 1}},
{{SCRIPT_CYRILLIC, "ru"}, {7, 1}},
{{SCRIPT_CYRILLIC, "uz"}, {8, 1}},
{{SCRIPT_CYRILLIC, "kk"}, {9, 1}},
{{SCRIPT_CYRILLIC, "sr"}, {10, 1}},
{{SCRIPT_CYRILLIC, "mn"}, {11, 1}},
{{SCRIPT_CYRILLIC, "mk"}, {12, 1}},
{{SCRIPT_CYRILLIC, "uk"}, {13, 1}},
{{SCRIPT_CYRILLIC, "ky"}, {14, 1}},
{{SCRIPT_ORIYA, ""}, {15, 1}},
{{SCRIPT_HANGUL, ""}, {16, 2}},
{{SCRIPT_GUJARATI, ""}, {18, 1}},
{{SCRIPT_HAN, ""}, {19, 1}},
{{SCRIPT_ARMENIAN, ""}, {20, 1}},
{{SCRIPT_TAMIL, ""}, {21, 1}},
{{SCRIPT_BENGALI, ""}, {22, 1}},
{{SCRIPT_MALAYALAM, ""}, {23, 1}},
{{SCRIPT_HIRAGANA, ""}, {24, 1}},
{{SCRIPT_KANNADA, ""}, {25, 1}},
{{SCRIPT_LATIN, ""}, {26, 1}},
{{SCRIPT_GEORGIAN, ""}, {27, 2}},
{{SCRIPT_DEVANAGARI, ""}, {29, 1}},
{{SCRIPT_THAI, ""}, {30, 1}},
{{SCRIPT_GREEK, ""}, {31, 3}},
{{SCRIPT_CANADIAN_ABORIGINAL, ""}, {34, 1}},
{{SCRIPT_ARABIC, "fa"}, {35, 1}},
{{SCRIPT_ARABIC, ""}, {36, 2}},
{{SCRIPT_ARABIC, "ps"}, {38, 1}},
{{SCRIPT_HEBREW, ""}, {39, 2}},
{{SCRIPT_KATAKANA, ""}, {41, 2}},
{{SCRIPT_COMMON, ""}, {43, 1}}
};
char *script_transliterators[] = {
"thaana-latin",
"maldivian-latin-bgn",
"gurmukhi-latin",
"telugu-latin",
"belarusian-latin-bgn",
"bulgarian-latin-bgn",
"cyrillic-latin",
"russian-latin-bgn",
"uzbek-latin-bgn",
"kazakh-latin-bgn",
"serbian-latin-bgn",
"mongolian-latin-bgn",
"macedonian-latin-bgn",
"ukrainian-latin-bgn",
"kirghiz-latin-bgn",
"oriya-latin",
"hangul-latin",
"korean-latin-bgn",
"gujarati-latin",
"han-latin",
"armenian-latin-bgn",
"tamil-latin",
"bengali-latin",
"malayam-latin",
"hiragana-latin",
"kannada-latin",
"latin-ascii",
"georgian-latin",
"georgian-latin-bgn",
"devanagari-latin",
"thai-latin",
"greek-latin",
"greek-latin-bgn",
"greek-latin-ungegn",
"canadianaboriginal-latin",
"persian-latin-bgn",
"arabic-latin",
"arabic-latin-bgn",
"pashto-latin-bgn",
"hebrew-latin",
"hebrew-latin-bgn",
"katakana-latin",
"katakana-latin-bgn",
"latin-ascii"
};