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

[transliteration] Regenerating transliteration data files

Browse Source
This commit is contained in:
Al
2016-08-21 19:41:11 -04:00
parent cb4408fea8
commit bde8776bc2
2 changed files with 3113 additions and 2392 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5449
src/transliteration_data.c
View File

File diff suppressed because one or more lines are too long

56
src/transliteration_scripts_data.c
View File

@@ -26,23 +26,31 @@ script_transliteration_rule_t script_transliteration_rules[] = {
{{SCRIPT_TAMIL, ""}, {17, 1}},
{{SCRIPT_MALAYALAM, ""}, {18, 1}},
{{SCRIPT_GUJARATI, ""}, {19, 1}},
{{SCRIPT_LATIN, ""}, {20, 1}},
{{SCRIPT_GEORGIAN, ""}, {21, 2}},
{{SCRIPT_DEVANAGARI, ""}, {23, 1}},
{{SCRIPT_THAI, ""}, {24, 1}},
{{SCRIPT_GREEK, ""}, {25, 3}},
{{SCRIPT_KATAKANA, ""}, {28, 2}},
{{SCRIPT_ORIYA, ""}, {30, 1}},
{{SCRIPT_HIRAGANA, ""}, {31, 1}},
{{SCRIPT_ARABIC, "fa"}, {32, 1}},
{{SCRIPT_ARABIC, ""}, {33, 2}},
{{SCRIPT_ARABIC, "ps"}, {35, 1}},
{{SCRIPT_CANADIAN_ABORIGINAL, ""}, {36, 1}},
{{SCRIPT_BENGALI, ""}, {37, 1}},
{{SCRIPT_GURMUKHI, ""}, {38, 1}},
{{SCRIPT_KANNADA, ""}, {39, 1}},
{{SCRIPT_COMMON, ""}, {40, 1}},
{{SCRIPT_HEBREW, ""}, {41, 2}}
{{SCRIPT_LATIN, "nb"}, {20, 2}},
{{SCRIPT_LATIN, "de"}, {22, 1}},
{{SCRIPT_LATIN, "sv"}, {23, 2}},
{{SCRIPT_LATIN, ""}, {25, 1}},
{{SCRIPT_LATIN, "da"}, {26, 2}},
{{SCRIPT_GEORGIAN, ""}, {28, 2}},
{{SCRIPT_DEVANAGARI, ""}, {30, 1}},
{{SCRIPT_THAI, ""}, {31, 1}},
{{SCRIPT_GREEK, ""}, {32, 3}},
{{SCRIPT_KATAKANA, ""}, {35, 2}},
{{SCRIPT_ORIYA, ""}, {37, 1}},
{{SCRIPT_HIRAGANA, ""}, {38, 1}},
{{SCRIPT_ARABIC, "fa"}, {39, 1}},
{{SCRIPT_ARABIC, ""}, {40, 2}},
{{SCRIPT_ARABIC, "ps"}, {42, 1}},
{{SCRIPT_CANADIAN_ABORIGINAL, ""}, {43, 1}},
{{SCRIPT_BENGALI, ""}, {44, 1}},
{{SCRIPT_GURMUKHI, ""}, {45, 1}},
{{SCRIPT_KANNADA, ""}, {46, 1}},
{{SCRIPT_COMMON, "nb"}, {47, 2}},
{{SCRIPT_COMMON, "de"}, {49, 1}},
{{SCRIPT_COMMON, "sv"}, {50, 2}},
{{SCRIPT_COMMON, ""}, {52, 1}},
{{SCRIPT_COMMON, "da"}, {53, 2}},
{{SCRIPT_HEBREW, ""}, {55, 2}}
};
char *script_transliterators[] = {
@@ -66,6 +74,13 @@ char *script_transliterators[] = {
"tamil-latin",
"malayalam-latin",
"gujarati-latin",
"scandinavian-ascii",
"latin-ascii",
"german-ascii",
"scandinavian-ascii",
"latin-ascii",
"latin-ascii",
"scandinavian-ascii",
"latin-ascii",
"georgian-latin",
"georgian-latin-bgn",
@@ -86,6 +101,13 @@ char *script_transliterators[] = {
"bengali-latin",
"gurmukhi-latin",
"kannada-latin",
"scandinavian-ascii",
"latin-ascii",
"german-ascii",
"scandinavian-ascii",
"latin-ascii",
"latin-ascii",
"scandinavian-ascii",
"latin-ascii",
"hebrew-latin",
"hebrew-latin-bgn"