tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[transliteration] regenerated scripts data file

Browse Source
This commit is contained in:
Al
2015-06-23 02:10:10 -05:00
parent b21c3a3a2f
commit d4cae97fd3
1 changed files with 41 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

82
src/transliteration_scripts_data.c
View File

@@ -1,49 +1,48 @@
#include "unicode_scripts.h" #include "unicode_scripts.h"
typedef struct script_transliteration_rule { typedef struct script_transliteration_rule {
script_t script; script_language_t script_language;
char *language; transliterator_index_t index;
uint32_t index;
uint32_t len;
} script_transliteration_rule_t; } script_transliteration_rule_t;
script_transliteration_rule_t script_transliteration_rules[] = { script_transliteration_rule_t script_transliteration_rules[] = {
{SCRIPT_THAANA, NULL, 0, 2}, {{SCRIPT_THAANA, NULL}, {0, 2}},
{SCRIPT_GURMUKHI, NULL, 2, 1}, {{SCRIPT_GURMUKHI, NULL}, {2, 1}},
{SCRIPT_TELUGU, NULL, 3, 1}, {{SCRIPT_TELUGU, NULL}, {3, 1}},
{SCRIPT_CYRILLIC, "be", 4, 1}, {{SCRIPT_CYRILLIC, "be"}, {4, 1}},
{SCRIPT_CYRILLIC, "bg", 5, 1}, {{SCRIPT_CYRILLIC, "bg"}, {5, 1}},
{SCRIPT_CYRILLIC, NULL, 6, 1}, {{SCRIPT_CYRILLIC, NULL}, {6, 1}},
{SCRIPT_CYRILLIC, "ru", 7, 1}, {{SCRIPT_CYRILLIC, "ru"}, {7, 1}},
{SCRIPT_CYRILLIC, "uz", 8, 1}, {{SCRIPT_CYRILLIC, "uz"}, {8, 1}},
{SCRIPT_CYRILLIC, "kk", 9, 1}, {{SCRIPT_CYRILLIC, "kk"}, {9, 1}},
{SCRIPT_CYRILLIC, "sr", 10, 1}, {{SCRIPT_CYRILLIC, "sr"}, {10, 1}},
{SCRIPT_CYRILLIC, "mn", 11, 1}, {{SCRIPT_CYRILLIC, "mn"}, {11, 1}},
{SCRIPT_CYRILLIC, "mk", 12, 1}, {{SCRIPT_CYRILLIC, "mk"}, {12, 1}},
{SCRIPT_CYRILLIC, "uk", 13, 1}, {{SCRIPT_CYRILLIC, "uk"}, {13, 1}},
{SCRIPT_CYRILLIC, "ky", 14, 1}, {{SCRIPT_CYRILLIC, "ky"}, {14, 1}},
{SCRIPT_ORIYA, NULL, 15, 1}, {{SCRIPT_ORIYA, NULL}, {15, 1}},
{SCRIPT_HANGUL, NULL, 16, 2}, {{SCRIPT_HANGUL, NULL}, {16, 2}},
{SCRIPT_GUJARATI, NULL, 18, 1}, {{SCRIPT_GUJARATI, NULL}, {18, 1}},
{SCRIPT_HAN, NULL, 19, 1}, {{SCRIPT_HAN, NULL}, {19, 1}},
{SCRIPT_ARMENIAN, NULL, 20, 1}, {{SCRIPT_ARMENIAN, NULL}, {20, 1}},
{SCRIPT_TAMIL, NULL, 21, 1}, {{SCRIPT_TAMIL, NULL}, {21, 1}},
{SCRIPT_BENGALI, NULL, 22, 1}, {{SCRIPT_BENGALI, NULL}, {22, 1}},
{SCRIPT_MALAYALAM, NULL, 23, 1}, {{SCRIPT_MALAYALAM, NULL}, {23, 1}},
{SCRIPT_HIRAGANA, NULL, 24, 1}, {{SCRIPT_HIRAGANA, NULL}, {24, 1}},
{SCRIPT_KANNADA, NULL, 25, 1}, {{SCRIPT_KANNADA, NULL}, {25, 1}},
{SCRIPT_LATIN, NULL, 26, 1}, {{SCRIPT_LATIN, NULL}, {26, 1}},
{SCRIPT_GEORGIAN, NULL, 27, 2}, {{SCRIPT_GEORGIAN, NULL}, {27, 2}},
{SCRIPT_DEVANAGARI, NULL, 29, 1}, {{SCRIPT_DEVANAGARI, NULL}, {29, 1}},
{SCRIPT_THAI, NULL, 30, 1}, {{SCRIPT_THAI, NULL}, {30, 1}},
{SCRIPT_GREEK, NULL, 31, 3}, {{SCRIPT_GREEK, NULL}, {31, 3}},
{SCRIPT_CANADIAN_ABORIGINAL, NULL, 34, 1}, {{SCRIPT_CANADIAN_ABORIGINAL, NULL}, {34, 1}},
{SCRIPT_ARABIC, "fa", 35, 1}, {{SCRIPT_ARABIC, "fa"}, {35, 1}},
{SCRIPT_ARABIC, NULL, 36, 2}, {{SCRIPT_ARABIC, NULL}, {36, 2}},
{SCRIPT_ARABIC, "ps", 38, 1}, {{SCRIPT_ARABIC, "ps"}, {38, 1}},
{SCRIPT_HEBREW, NULL, 39, 2}, {{SCRIPT_HEBREW, NULL}, {39, 2}},
{SCRIPT_KATAKANA, NULL, 41, 2}, {{SCRIPT_KATAKANA, NULL}, {41, 2}},
{SCRIPT_COMMON, NULL, 43, 1} {{SCRIPT_COMMON, NULL}, {43, 1}}
}; };
char *script_transliterators[] = { char *script_transliterators[] = {
@@ -91,4 +90,5 @@ char *script_transliterators[] = {
"katakana-latin", "katakana-latin",
"katakana-latin-bgn", "katakana-latin-bgn",
"latin-ascii" "latin-ascii"
}; }