209 lines
11 KiB
Plaintext
209 lines
11 KiB
Plaintext
//--------------------------------------------------------------------
|
|
// Copyright (c) 1999-2001, International Business Machines
|
|
// Corporation and others. All Rights Reserved.
|
|
//--------------------------------------------------------------------
|
|
// THIS IS A MACHINE-GENERATED FILE
|
|
// Tool: dumpICUrules.bat
|
|
// Source: Transliterator_index.txt
|
|
// Date: Fri Mar 2 12:50:49 2001
|
|
//--------------------------------------------------------------------
|
|
|
|
//--------------------------------------------------------------------
|
|
// N.B.: This file has been generated mechanically from the
|
|
// corresponding ICU4J file, which is the master file that receives
|
|
// primary updates. The colon-delimited fields have been split into
|
|
// separate strings. For 'file' and 'internal' lines, the encoding
|
|
// field has been deleted, since the encoding is processed at build
|
|
// time in ICU4C. Certain large rule sets not intended for general
|
|
// use have been commented out with the notation "Java only".
|
|
//--------------------------------------------------------------------
|
|
|
|
translit_index {
|
|
RuleBasedTransliteratorIDs {
|
|
// Copyright (c) 2001, International Business Machines Corporation and
|
|
// others. All Rights Reserved.
|
|
//
|
|
// TRANSLITERATOR INDEX FILE. This file lists the non-algorithmic
|
|
// system transliterators. It allows arbitrary mappings between
|
|
// transliterator IDs and file names, and also allows the system to
|
|
// define aliases for transliterators, so that "Latin-Hangul", for
|
|
// example, can be implemented transparently as the compound
|
|
// "Latin-Jamo;Jamo-Hangul". Internal IDs may also be defined; these
|
|
// are invisible to the user, but can be composed together by the
|
|
// system to create visible transliterators.
|
|
//
|
|
// Blank lines and lines beginning with '#' are ignored.
|
|
//
|
|
// Lines in this file have one of the following forms (text not
|
|
// enclosed by <> is literal):
|
|
//
|
|
// <id>:file:<resource>:<encoding>:<direction>
|
|
// <id>:internal:<resource>:<encoding>:<direction>
|
|
// <id>:alias:<getInstanceArg>
|
|
//
|
|
// <id> is the ID of the system transliterator being defined. These
|
|
// are public IDs enumerated by Transliterator.getAvailableIDs(),
|
|
// unless the second field is "internal".
|
|
//
|
|
// <resource> is a ResourceReader resource name. Currently these refer
|
|
// to file names under com/ibm/text/resources. This string is passed
|
|
// directly to ResourceReader, together with <encoding>.
|
|
//
|
|
// <encoding> is the character encoding to use when reading <resource>;
|
|
// passed directly to ResourceReader. E.g., "UTF8".
|
|
//
|
|
// <direction> is either "FORWARD" or "REVERSE".
|
|
//
|
|
// <getInstanceArg> is a string to be passed directly to
|
|
// Transliterator.getInstance(). The returned Transliterator object
|
|
// then has its ID changed to <id> and is returned.
|
|
|
|
|
|
// Bidirectional rule files
|
|
|
|
{ "Fullwidth-Halfwidth", "file", "fullhalf", "FORWARD" },
|
|
{ "Halfwidth-Fullwidth", "file", "fullhalf", "REVERSE" },
|
|
|
|
{ "Latin-Arabic", "file", "larabic", "FORWARD" },
|
|
{ "Arabic-Latin", "file", "larabic", "REVERSE" },
|
|
|
|
{ "Latin-Cyrillic", "file", "lcyril", "FORWARD" },
|
|
{ "Cyrillic-Latin", "file", "lcyril", "REVERSE" },
|
|
|
|
{ "Latin-Devanagari", "file", "ldevan", "FORWARD" },
|
|
{ "Devanagari-Latin", "file", "ldevan", "REVERSE" },
|
|
|
|
{ "Latin-Greek", "file", "lgreek", "FORWARD" },
|
|
{ "Greek-Latin", "file", "lgreek", "REVERSE" },
|
|
|
|
{ "Latin-Hebrew", "file", "lhebrew", "FORWARD" },
|
|
{ "Hebrew-Latin", "file", "lhebrew", "REVERSE" },
|
|
|
|
{ "Latin-Jamo", "file", "ljamo", "FORWARD" },
|
|
{ "Jamo-Latin", "file", "ljamo", "REVERSE" },
|
|
|
|
{ "Latin-Kana", "file", "lkana", "FORWARD" },
|
|
{ "Kana-Latin", "file", "lkana", "REVERSE" },
|
|
|
|
{ "Hiragana-Katakana", "file", "kana", "FORWARD" },
|
|
{ "Katakana-Hiragana", "file", "kana", "REVERSE" },
|
|
|
|
{ "StraightQuotes-CurlyQuotes", "file", "quotes", "FORWARD" },
|
|
{ "CurlyQuotes-StraightQuotes", "file", "quotes", "REVERSE" },
|
|
|
|
// One way rules (forward only)
|
|
|
|
// Java only: { "Han-Pinyin", "file", "-", "FORWARD" },
|
|
// Java only: { "Kanji-English", "file", "-", "FORWARD" },
|
|
// Java only: { "Kanji-OnRomaji", "file", "-", "FORWARD" },
|
|
{ "KeyboardEscape-Latin1", "file", "kbdescl1", "FORWARD" },
|
|
|
|
// Replaced by algorithmic transliterator:
|
|
// { "UnicodeName-UnicodeChar", "file", "ucname", "FORWARD" },
|
|
|
|
// Compound rules
|
|
|
|
/// TODO
|
|
/// Add the appropriate compound filters here
|
|
{ "Latin-Hangul", "alias", "Latin-Jamo;[\u1100-\u11FF]NFC", "" },
|
|
{ "Hangul-Latin", "alias", "[\uAC00-\uD7AF]NFD;Jamo-Latin", "" },
|
|
|
|
// Inter-Indic composed rules
|
|
|
|
{ "Devanagari-InterIndic", "internal", "Devanagari_InterIndic", "FORWARD" },
|
|
{ "Bengali-InterIndic", "internal", "Bengali_InterIndic", "FORWARD" },
|
|
{ "Gurmukhi-InterIndic", "internal", "Gurmukhi_InterIndic", "FORWARD" },
|
|
{ "Gujarati-InterIndic", "internal", "Gujarati_InterIndic", "FORWARD" },
|
|
{ "Oriya-InterIndic", "internal", "Oriya_InterIndic", "FORWARD" },
|
|
{ "Tamil-InterIndic", "internal", "Tamil_InterIndic", "FORWARD" },
|
|
{ "Telugu-InterIndic", "internal", "Telugu_InterIndic", "FORWARD" },
|
|
{ "Kannada-InterIndic", "internal", "Kannada_InterIndic", "FORWARD" },
|
|
{ "Malayalam-InterIndic", "internal", "Malayalam_InterIndic", "FORWARD" },
|
|
|
|
{ "InterIndic-Devanagari", "internal", "InterIndic_Devanagari", "FORWARD" },
|
|
{ "InterIndic-Bengali", "internal", "InterIndic_Bengali", "FORWARD" },
|
|
{ "InterIndic-Gurmukhi", "internal", "InterIndic_Gurmukhi", "FORWARD" },
|
|
{ "InterIndic-Gujarati", "internal", "InterIndic_Gujarati", "FORWARD" },
|
|
{ "InterIndic-Oriya", "internal", "InterIndic_Oriya", "FORWARD" },
|
|
{ "InterIndic-Tamil", "internal", "InterIndic_Tamil", "FORWARD" },
|
|
{ "InterIndic-Telugu", "internal", "InterIndic_Telugu", "FORWARD" },
|
|
{ "InterIndic-Kannada", "internal", "InterIndic_Kannada", "FORWARD" },
|
|
{ "InterIndic-Malayalam", "internal", "InterIndic_Malayalam", "FORWARD" },
|
|
|
|
{ "Devanagari-Bengali", "alias", "Devanagari-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Devanagari-Gurmukhi", "alias", "Devanagari-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Devanagari-Gujarati", "alias", "Devanagari-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Devanagari-Oriya", "alias", "Devanagari-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Devanagari-Tamil", "alias", "Devanagari-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Devanagari-Telugu", "alias", "Devanagari-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Devanagari-Kannada", "alias", "Devanagari-InterIndic;InterIndic-Kannada", "" },
|
|
{ "Devanagari-Malayalam", "alias", "Devanagari-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Bengali-Devanagari", "alias", "Bengali-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Bengali-Gurmukhi", "alias", "Bengali-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Bengali-Gujarati", "alias", "Bengali-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Bengali-Oriya", "alias", "Bengali-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Bengali-Tamil", "alias", "Bengali-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Bengali-Telugu", "alias", "Bengali-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Bengali-Kannada", "alias", "Bengali-InterIndic;InterIndic-Kannada", "" },
|
|
{ "Bengali-Malayalam", "alias", "Bengali-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Gurmukhi-Devanagari", "alias", "Gurmukhi-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Gurmukhi-Bengali", "alias", "Gurmukhi-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Gurmukhi-Gujarati", "alias", "Gurmukhi-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Gurmukhi-Oriya", "alias", "Gurmukhi-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Gurmukhi-Tamil", "alias", "Gurmukhi-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Gurmukhi-Telugu", "alias", "Gurmukhi-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Gurmukhi-Kannada", "alias", "Gurmukhi-InterIndic;InterIndic-Kannada", "" },
|
|
{ "Gurmukhi-Malayalam", "alias", "Gurmukhi-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Gujarati-Devanagari", "alias", "Gujarati-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Gujarati-Bengali", "alias", "Gujarati-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Gujarati-Gurmukhi", "alias", "Gujarati-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Gujarati-Oriya", "alias", "Gujarati-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Gujarati-Tamil", "alias", "Gujarati-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Gujarati-Telugu", "alias", "Gujarati-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Gujarati-Kannada", "alias", "Gujarati-InterIndic;InterIndic-Kannada", "" },
|
|
{ "Gujarati-Malayalam", "alias", "Gujarati-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Oriya-Devanagari", "alias", "Oriya-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Oriya-Bengali", "alias", "Oriya-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Oriya-Gurmukhi", "alias", "Oriya-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Oriya-Gujarati", "alias", "Oriya-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Oriya-Tamil", "alias", "Oriya-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Oriya-Telugu", "alias", "Oriya-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Oriya-Kannada", "alias", "Oriya-InterIndic;InterIndic-Kannada", "" },
|
|
{ "Oriya-Malayalam", "alias", "Oriya-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Tamil-Devanagari", "alias", "Tamil-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Tamil-Bengali", "alias", "Tamil-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Tamil-Gurmukhi", "alias", "Tamil-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Tamil-Gujarati", "alias", "Tamil-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Tamil-Oriya", "alias", "Tamil-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Tamil-Telugu", "alias", "Tamil-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Tamil-Kannada", "alias", "Tamil-InterIndic;InterIndic-Kannada", "" },
|
|
{ "Tamil-Malayalam", "alias", "Tamil-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Telugu-Devanagari", "alias", "Telugu-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Telugu-Bengali", "alias", "Telugu-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Telugu-Gurmukhi", "alias", "Telugu-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Telugu-Gujarati", "alias", "Telugu-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Telugu-Oriya", "alias", "Telugu-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Telugu-Tamil", "alias", "Telugu-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Telugu-Kannada", "alias", "Telugu-InterIndic;InterIndic-Kannada", "" },
|
|
{ "Telugu-Malayalam", "alias", "Telugu-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Kannada-Devanagari", "alias", "Kannada-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Kannada-Bengali", "alias", "Kannada-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Kannada-Gurmukhi", "alias", "Kannada-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Kannada-Gujarati", "alias", "Kannada-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Kannada-Oriya", "alias", "Kannada-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Kannada-Tamil", "alias", "Kannada-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Kannada-Telugu", "alias", "Kannada-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Kannada-Malayalam", "alias", "Kannada-InterIndic;InterIndic-Malayalam", "" },
|
|
{ "Malayalam-Devanagari", "alias", "Malayalam-InterIndic;InterIndic-Devanagari", "" },
|
|
{ "Malayalam-Bengali", "alias", "Malayalam-InterIndic;InterIndic-Bengali", "" },
|
|
{ "Malayalam-Gurmukhi", "alias", "Malayalam-InterIndic;InterIndic-Gurmukhi", "" },
|
|
{ "Malayalam-Gujarati", "alias", "Malayalam-InterIndic;InterIndic-Gujarati", "" },
|
|
{ "Malayalam-Oriya", "alias", "Malayalam-InterIndic;InterIndic-Oriya", "" },
|
|
{ "Malayalam-Tamil", "alias", "Malayalam-InterIndic;InterIndic-Tamil", "" },
|
|
{ "Malayalam-Telugu", "alias", "Malayalam-InterIndic;InterIndic-Telugu", "" },
|
|
{ "Malayalam-Kannada", "alias", "Malayalam-InterIndic;InterIndic-Kannada", "" },
|
|
|
|
// eof
|
|
}
|
|
}
|