Skip to content

W
wine-winehq
Commit 8cdb593f authored by Alexandre Julliard's avatar Alexandre Julliard
Browse files
Options

make_unicode: Generate locale.nls using the Unicode CLDR data.

Signed-off-by: 's avatarAlexandre Julliard <julliard@winehq.org>
parent c44ba873
Hide whitespace changes
Inline Side-by-side
Showing with 1875 additions and 1 deletion
loader/wine.inf.in
View file @ 8cdb593f
......@@ -5617,6 +5617,7 @@ c_936.nls
c_949.nls
c_950.nls
l_intl.nls
locale.nls
normidna.nls
normnfc.nls
normnfd.nls
......
nls/Makefile.in
View file @ 8cdb593f
......@@ -68,6 +68,7 @@ SOURCES = \
c_949.nls \
c_950.nls \
l_intl.nls \
locale.nls \
normidna.nls \
normnfc.nls \
normnfd.nls \
......
nls/locale.nls 0 → 100644
View file @ 8cdb593f
File added
tools/make_unicode
View file @ 8cdb593f
......@@ -20,6 +20,8 @@
#
use strict;
use XML::LibXML;
use Encode;
# base URLs for www.unicode.org files
my $UNIVERSION = "14.0.0";
......@@ -31,6 +33,14 @@ my $REPORTS = "http://www.unicode.org/reports";
my $MSDATA = "https://download.microsoft.com/download/C/F/7/CF713A5E-9FBC-4FD6-9246-275F65C0E498";
my $MSCODEPAGES = "$MSDATA/Windows Supported Code Page Data Files.zip";
#my $CLDRDATA = "https://www.unicode.org/Public/cldr/40/cldr-common-40.0.zip";
my $CLDRVERSION = "40";
my $CLDRDATA = "https://github.com/unicode-org/cldr/archive/refs/tags/release-$CLDRVERSION.zip";
my $CLDR33DATA = "https://www.unicode.org/Public/cldr/33/cldr-common-33.0.zip";
my $ISO639VERSION = "20220120";
my $ISO639 = "https://iso639-3.sil.org/sites/iso639-3/files/downloads/iso-639-3_Code_Tables_$ISO639VERSION.zip";
# Sort keys file
my $SORTKEYS = "tr10/allkeys.txt";
......@@ -429,6 +439,1054 @@ my %joining_types =
"T" => 6, # Transparent
);
my @locales =
(
{ name => "", lcid => 0x0000007f, file => "root", territory => "IV", sabbrevlangname => "IVL", sopentypelang =>"dflt" },
{ name => "aa", dir => "seed", sopentypelang => "AFR" },
{ name => "aa-DJ", dir => "seed" },
{ name => "aa-ER", dir => "seed" },
{ name => "aa-ET", dir => "seed" },
{ name => "af", lcid => 0x00000036, oemcp => 850, sabbrevlangname => "AFK", sopentypelang => "AFK" },
{ name => "af-NA" },
{ name => "af-ZA", lcid => 0x00000436 },
{ name => "agq" },
{ name => "agq-CM" },
{ name => "ak", sopentypelang => "TWI" },
{ name => "ak-GH" },
{ name => "am", lcid => 0x0000005e, sabbrevlangname => "AMH" },
{ name => "am-ET", lcid => 0x0000045e },
{ name => "ar", lcid => 0x00000001, territory => "SA", oemcp => 720, group => 13 },
{ name => "ar-001" },
{ name => "ar-AE", lcid => 0x00003801, sabbrevlangname => "ARU" },
{ name => "ar-BH", lcid => 0x00003c01, sabbrevlangname => "ARH" },
{ name => "ar-DJ" },
{ name => "ar-DZ", lcid => 0x00001401, sabbrevlangname => "ARG" },
{ name => "ar-EG", lcid => 0x00000c01, sabbrevlangname => "ARE" },
{ name => "ar-EH" },
{ name => "ar-ER" },
{ name => "ar-IL" },
{ name => "ar-IQ", lcid => 0x00000801, sabbrevlangname => "ARI" },
{ name => "ar-JO", lcid => 0x00002c01, sabbrevlangname => "ARJ" },
{ name => "ar-KM" },
{ name => "ar-KW", lcid => 0x00003401, sabbrevlangname => "ARK" },
{ name => "ar-LB", lcid => 0x00003001, sabbrevlangname => "ARB" },
{ name => "ar-LY", lcid => 0x00001001, sabbrevlangname => "ARL" },
{ name => "ar-MA", lcid => 0x00001801, sabbrevlangname => "ARM" },
{ name => "ar-MR" },
{ name => "ar-OM", lcid => 0x00002001, sabbrevlangname => "ARO" },
{ name => "ar-PS" },
{ name => "ar-QA", lcid => 0x00004001, sabbrevlangname => "ARQ" },
{ name => "ar-SA", lcid => 0x00000401, sabbrevlangname => "ARA" },
{ name => "ar-SD" },
{ name => "ar-SO" },
{ name => "ar-SS" },
{ name => "ar-SY", lcid => 0x00002801, sabbrevlangname => "ARS" },
{ name => "ar-TD" },
{ name => "ar-TN", lcid => 0x00001c01, sabbrevlangname => "ART" },
{ name => "ar-YE", lcid => 0x00002401, sabbrevlangname => "ARY" },
{ name => "arn", lcid => 0x0000007a, oemcp => 850, ebcdiccp => 20284, slist => ",", dir => "seed", sabbrevlangname => "MPD", sopentypelang => "MAP" },
{ name => "arn-CL", lcid => 0x0000047a, dir => "seed" },
{ name => "arn-Latn", alias => "arn" },
{ name => "arn-Latn-CL", alias => "arn-CL" },
{ name => "as", lcid => 0x0000004d, slist => ",", group => 15 },
{ name => "as-IN", lcid => 0x0000044d },
{ name => "asa" },
{ name => "asa-TZ" },
{ name => "ast" },
{ name => "ast-ES" },
{ name => "az", lcid => 0x0000002c, oemcp => 857, ebcdiccp => 20905, group => 2 },
{ name => "az-Cyrl", lcid => 0x0000742c, oemcp => 866, ebcdiccp => 20880, group => 5, sabbrevlangname => "AZC" },
{ name => "az-Cyrl-AZ", lcid => 0x0000082c },
{ name => "az-Latn", lcid => 0x0000782c },
{ name => "az-Latn-AZ", lcid => 0x0000042c },
{ name => "ba", lcid => 0x0000006d, oemcp => 866, group => 5, dir => "seed", sabbrevlangname => "BAS", sopentypelang => "BSH" },
{ name => "ba-Cyrl", alias => "ba" },
{ name => "ba-Cyrl-RU", alias => "ba-RU" },
{ name => "ba-RU", lcid => 0x0000046d, dir => "seed" },
{ name => "bas" },
{ name => "bas-CM" },
{ name => "be", lcid => 0x00000023, oemcp => 866, ebcdiccp => 500, group => 5 },
{ name => "be-BY", lcid => 0x00000423 },
{ name => "bem" },
{ name => "bem-ZM" },
{ name => "bez" },
{ name => "bez-TZ" },
{ name => "bg", lcid => 0x00000002, oemcp => 866, ebcdiccp => 21025, group => 5, sabbrevlangname => "BGR", sopentypelang => "BGR" },
{ name => "bg-BG", lcid => 0x00000402 },
{ name => "bin", lcid => 0x00000066, oemcp => 850, dir => "exemplars", sabbrevlangname => "ZZZ", sopentypelang => "EDO" },
{ name => "bin-NG", lcid => 0x00000466, file => "bin", dir => "exemplars" },
{ name => "bm", sopentypelang => "BMB" },
{ name => "bm-Latn", file => "bm" },
{ name => "bm-Latn-ML", file => "bm_ML" },
{ name => "bm-ML", alias => "bm-Latn-ML" },
{ name => "bn", lcid => 0x00000045, slist => ",", group => 15, sabbrevlangname => "BNB" },
{ name => "bn-BD", lcid => 0x00000845 },
{ name => "bn-IN", lcid => 0x00000445, sabbrevlangname => "BNG" },
{ name => "bo", lcid => 0x00000051, slist => ",", group => 15, sabbrevlangname => "BOB", sopentypelang => "TIB" },
{ name => "bo-CN", lcid => 0x00000451 },
{ name => "bo-IN", slist => "," },
{ name => "bo-Tibt", alias => "bo" },
{ name => "bo-Tibt-CN", alias => "bo-CN" },
{ name => "bo-Tibt-IN", alias => "bo-IN" },
{ name => "br", lcid => 0x0000007e, oemcp => 850, ebcdiccp => 20297 },
{ name => "br-FR", lcid => 0x0000047e },
{ name => "br-Latn", alias => "br" },
{ name => "br-Latn-FR", alias => "br-FR" },
{ name => "brx" },
{ name => "brx-IN" },
{ name => "bs", lcid => 0x0000781a, oemcp => 852, maccp => 10082, ebcdiccp => 870, group => 2, sabbrevlangname => "BSB" },
{ name => "bs-Cyrl", lcid => 0x0000641a, oemcp => 855, group => 5, sabbrevlangname => "BSC" },
{ name => "bs-Cyrl-BA", lcid => 0x0000201a },
{ name => "bs-Latn", lcid => 0x0000681a },
{ name => "bs-Latn-BA", lcid => 0x0000141a },
{ name => "byn", dir => "seed", sopentypelang => "BIL" },
{ name => "byn-ER", dir => "seed" },
{ name => "ca", lcid => 0x00000003, oemcp => 850 },
{ name => "ca-AD", maccp => 65001 },
{ name => "ca-ES", lcid => 0x00000403 },
{ name => "ca-ES-valencia", lcid => 0x00000803, file => "ca_ES_VALENCIA", sabbrevlangname => "VAL" },
{ name => "ca-FR", maccp => 65001 },
{ name => "ca-IT", maccp => 65001 },
{ name => "ccp" },
{ name => "ccp-BD", alias => "ccp-Cakm-BD" },
{ name => "ccp-Cakm", file => "ccp" },
{ name => "ccp-Cakm-BD", file => "ccp_BD" },
{ name => "ccp-Cakm-IN", file => "ccp_IN" },
{ name => "ccp-IN", alias => "ccp-Cakm-IN" },
{ name => "ce" },
{ name => "ce-RU" },
{ name => "ceb" },
{ name => "ceb-Latn", file => "ceb" },
{ name => "ceb-Latn-PH", file => "ceb_PH" },
{ name => "ceb-PH", alias => "ceb-Latn-PH" },
{ name => "cgg" },
{ name => "cgg-UG" },
{ name => "chr", lcid => 0x0000005c, slist => ",", sabbrevlangname => "CRE" },
{ name => "chr-Cher", lcid => 0x00007c5c, file => "chr" },
{ name => "chr-Cher-US", lcid => 0x0000045c, file => "chr_US" },
{ name => "chr-US", alias => "chr-Cher-US" },
{ name => "ckb", alias => "ku" },
{ name => "ckb-IQ", alias => "ku-Arab-IQ" },
{ name => "ckb-IR", alias => "ku-Arab-IR" },
{ name => "co", lcid => 0x00000083, oemcp => 850, ebcdiccp => 20297, dir => "seed" },
{ name => "co-FR", lcid => 0x00000483, dir => "seed" },
{ name => "co-Latn", alias => "co" },
{ name => "co-Latn-FR", alias => "co-FR" },
{ name => "cs", lcid => 0x00000005, oemcp => 852, group => 2, sabbrevlangname => "CSY", sopentypelang => "CSY" },
{ name => "cs-CZ", lcid => 0x00000405 },
{ name => "cu", dir => "seed", sopentypelang => "CSL" },
{ name => "cu-RU", dir => "seed" },
{ name => "cy", lcid => 0x00000052, oemcp => 850, ebcdiccp => 20285, sabbrevlangname => "CYM", sopentypelang => "WEL" },
{ name => "cy-GB", lcid => 0x00000452 },
{ name => "da", lcid => 0x00000006, oemcp => 850, ebcdiccp => 20277 },
{ name => "da-DK", lcid => 0x00000406 },
{ name => "da-GL", maccp => 65001 },
{ name => "dav" },
{ name => "dav-KE" },
{ name => "de", lcid => 0x00000007, oemcp => 850, ebcdiccp => 20273 },
{ name => "de-AT", lcid => 0x00000c07, sabbrevlangname => "DEA" },
{ name => "de-BE" },
{ name => "de-CH", lcid => 0x00000807, sabbrevlangname => "DES" },
{ name => "de-DE", lcid => 0x00000407 },
{ name => "de-DE_phoneb", lcid => 0x00010407, alias => "de-DE" },
{ name => "de-DE-u-co-phonebk", alias => "de-DE_phoneb" },
{ name => "de-IT", oemcp => 65001 },
{ name => "de-LI", lcid => 0x00001407, sabbrevlangname => "DEC" },
{ name => "de-LU", lcid => 0x00001007, sabbrevlangname => "DEL" },
{ name => "dje", sopentypelang => "DJR" },
{ name => "dje-NE" },
{ name => "doi" },
{ name => "doi-IN" },
{ name => "dsb", lcid => 0x00007c2e, sparent => "hsb", oemcp => 850, ebcdiccp => 870, sabbrevlangname => "DSB", sopentypelang => "LSB" },
{ name => "dsb-DE", lcid => 0x0000082e },
{ name => "dua" },
{ name => "dua-CM" },
{ name => "dv", lcid => 0x00000065, slist => "\x{060c}", group => 13, dir => "seed" },
{ name => "dv-MV", lcid => 0x00000465, dir => "seed" },
{ name => "dyo" },
{ name => "dyo-SN" },
{ name => "dz", sopentypelang => "DZN" },
{ name => "dz-BT", lcid => 0x00000c51, sabbrevlangname => "ZZZ" },
{ name => "ebu" },
{ name => "ebu-KE" },
{ name => "ee" },
{ name => "ee-GH" },
{ name => "ee-TG" },
{ name => "el", lcid => 0x00000008, oemcp => 737, group => 4 },
{ name => "el-CY" },
{ name => "el-GR", lcid => 0x00000408 },
{ name => "en", lcid => 0x00000009, oemcp => 437, slist => ",", sabbrevlangname => "ENU" },
{ name => "en-001", oemcp => 850 },
{ name => "en-029", lcid => 0x00002409, file => "en", oemcp => 850, sintlsymbol => "XCD", sabbrevlangname => "ENB" },
{ name => "en-150", oemcp => 65001 },
{ name => "en-AE", lcid => 0x00004c09, oemcp => 65001, sabbrevlangname => "ZZZ" },
{ name => "en-AG", oemcp => 850 },
{ name => "en-AI", oemcp => 850 },
{ name => "en-AS", oemcp => 850 },
{ name => "en-AT", oemcp => 65001 },
{ name => "en-AU", lcid => 0x00000c09, oemcp => 850, sabbrevlangname => "ENA" },
{ name => "en-BB", oemcp => 850 },
{ name => "en-BE", oemcp => 850 },
{ name => "en-BI", oemcp => 65001 },
{ name => "en-BM", oemcp => 850 },
{ name => "en-BS", oemcp => 850 },
{ name => "en-BW", oemcp => 850 },
{ name => "en-BZ", lcid => 0x00002809, oemcp => 850, sabbrevlangname => "ENL" },
{ name => "en-CA", lcid => 0x00001009, oemcp => 850, ebcdiccp => 37, sabbrevlangname => "ENC" },
{ name => "en-CC", oemcp => 850 },
{ name => "en-CH", oemcp => 65001 },
{ name => "en-CK", oemcp => 850 },
{ name => "en-CM", oemcp => 850 },
{ name => "en-CX", oemcp => 850 },
{ name => "en-CY", oemcp => 65001 },
{ name => "en-DE", oemcp => 65001 },
{ name => "en-DG", oemcp => 850 },
{ name => "en-DK", oemcp => 65001 },
{ name => "en-DM", oemcp => 850 },
{ name => "en-ER", oemcp => 850 },
{ name => "en-FI", oemcp => 65001 },
{ name => "en-FJ", oemcp => 850 },
{ name => "en-FK", oemcp => 850 },
{ name => "en-FM", oemcp => 850 },
{ name => "en-GB", lcid => 0x00000809, oemcp => 850, ebcdiccp => 20285, sabbrevlangname => "ENG" },
{ name => "en-GD", oemcp => 850 },
{ name => "en-GG", oemcp => 850 },
{ name => "en-GH", oemcp => 850 },
{ name => "en-GI", oemcp => 850 },
{ name => "en-GM", oemcp => 850 },
{ name => "en-GU", oemcp => 850 },
{ name => "en-GY", oemcp => 850 },
{ name => "en-HK", lcid => 0x00003c09, oemcp => 850, sabbrevlangname => "ENH" },
{ name => "en-ID", lcid => 0x00003809, file => "en", oemcp => 850, sabbrevlangname => "ZZZ" },
{ name => "en-IE", lcid => 0x00001809, oemcp => 850, sabbrevlangname => "ENI" },
{ name => "en-IL", oemcp => 65001 },
{ name => "en-IM", oemcp => 850 },
{ name => "en-IN", lcid => 0x00004009, sabbrevlangname => "ENN" },
{ name => "en-IO", oemcp => 850 },
{ name => "en-JE", oemcp => 850 },
{ name => "en-JM", lcid => 0x00002009, oemcp => 850, sabbrevlangname => "ENJ" },
{ name => "en-KE", oemcp => 850 },
{ name => "en-KI", oemcp => 850 },
{ name => "en-KN", oemcp => 850 },
{ name => "en-KY", oemcp => 850 },
{ name => "en-LC", oemcp => 850 },
{ name => "en-LR", oemcp => 850 },
{ name => "en-LS", oemcp => 850 },
{ name => "en-MG", oemcp => 850 },
{ name => "en-MH", oemcp => 850 },
{ name => "en-MO", oemcp => 850 },
{ name => "en-MP", oemcp => 850 },
{ name => "en-MS", oemcp => 850 },
{ name => "en-MT", oemcp => 850 },
{ name => "en-MU", oemcp => 850 },
{ name => "en-MW", oemcp => 850 },
{ name => "en-MY", lcid => 0x00004409, sabbrevlangname => "ENM" },
{ name => "en-NA", oemcp => 850 },
{ name => "en-NF", oemcp => 850 },
{ name => "en-NG", oemcp => 850 },
{ name => "en-NL", oemcp => 65001 },
{ name => "en-NR", oemcp => 850 },
{ name => "en-NU", oemcp => 850 },
{ name => "en-NZ", lcid => 0x00001409, oemcp => 850, sabbrevlangname => "ENZ" },
{ name => "en-PG", oemcp => 850 },
{ name => "en-PH", lcid => 0x00003409, ebcdiccp => 500, sabbrevlangname => "ENP" },
{ name => "en-PK", oemcp => 850 },
{ name => "en-PN", oemcp => 850 },
{ name => "en-PR", oemcp => 850 },
{ name => "en-PW", oemcp => 850 },
{ name => "en-RW", oemcp => 850 },
{ name => "en-SB", oemcp => 850 },
{ name => "en-SC", oemcp => 850 },
{ name => "en-SD", oemcp => 850 },
{ name => "en-SE", oemcp => 65001 },
{ name => "en-SG", lcid => 0x00004809, sabbrevlangname => "ENE" },
{ name => "en-SH", oemcp => 850 },
{ name => "en-SI", oemcp => 65001 },
{ name => "en-SL", oemcp => 850 },
{ name => "en-SS", oemcp => 850 },
{ name => "en-SX", oemcp => 850 },
{ name => "en-SZ", oemcp => 850 },
{ name => "en-TC", oemcp => 850 },
{ name => "en-TK", oemcp => 850 },
{ name => "en-TO", oemcp => 850 },
{ name => "en-TT", lcid => 0x00002c09, oemcp => 850, sabbrevlangname => "ENT" },
{ name => "en-TV", oemcp => 850 },
{ name => "en-TZ", oemcp => 850 },
{ name => "en-UG", oemcp => 850 },
{ name => "en-UM", oemcp => 850 },
{ name => "en-US", lcid => 0x00000409 },
{ name => "en-VC", oemcp => 850 },
{ name => "en-VG", oemcp => 850 },
{ name => "en-VI", oemcp => 850 },
{ name => "en-VU", oemcp => 850 },
{ name => "en-WS", oemcp => 850 },
{ name => "en-ZA", lcid => 0x00001c09, ebcdiccp => 500, sabbrevlangname => "ENS" },
{ name => "en-ZM", oemcp => 850 },
{ name => "en-ZW", lcid => 0x00003009, ebcdiccp => 500, sabbrevlangname => "ENW" },
{ name => "eo", sopentypelang => "NTO" },
{ name => "eo-001" },
{ name => "es", lcid => 0x0000000a, oemcp => 850, ebcdiccp => 20284, sabbrevlangname => "ESP", sopentypelang => "ESP" },
{ name => "es-419", lcid => 0x0000580a, sabbrevlangname => "ESJ" },
{ name => "es-AR", lcid => 0x00002c0a, sabbrevlangname => "ESS" },
{ name => "es-BO", lcid => 0x0000400a, sabbrevlangname => "ESB" },
{ name => "es-BR", oemcp => 65001 },
{ name => "es-BZ", oemcp => 65001 },
{ name => "es-CL", lcid => 0x0000340a, sabbrevlangname => "ESL" },
{ name => "es-CO", lcid => 0x0000240a, sabbrevlangname => "ESO" },
{ name => "es-CR", lcid => 0x0000140a, sabbrevlangname => "ESC" },
{ name => "es-CU", lcid => 0x00005c0a, sabbrevlangname => "ESK" },
{ name => "es-DO", lcid => 0x00001c0a, sabbrevlangname => "ESD" },
{ name => "es-EA" },
{ name => "es-EC", lcid => 0x0000300a, sabbrevlangname => "ESF" },
{ name => "es-ES", lcid => 0x00000c0a, sabbrevlangname => "ESN" },
{ name => "es-ES_tradnl", lcid => 0x0000040a, file => "es_ES" },
{ name => "es-ES-u-co-trad", alias => "es-ES_tradnl" },
{ name => "es-GQ" },
{ name => "es-GT", lcid => 0x0000100a, sabbrevlangname => "ESG" },
{ name => "es-HN", lcid => 0x0000480a, sabbrevlangname => "ESH" },
{ name => "es-IC" },
{ name => "es-MX", lcid => 0x0000080a, sabbrevlangname => "ESM" },
{ name => "es-NI", lcid => 0x00004c0a, sabbrevlangname => "ESI" },
{ name => "es-PA", lcid => 0x0000180a, sabbrevlangname => "ESA" },
{ name => "es-PE", lcid => 0x0000280a, sabbrevlangname => "ESR" },
{ name => "es-PH" },
{ name => "es-PR", lcid => 0x0000500a, sabbrevlangname => "ESU" },
{ name => "es-PY", lcid => 0x00003c0a, sabbrevlangname => "ESZ" },
{ name => "es-SV", lcid => 0x0000440a, sabbrevlangname => "ESE" },
{ name => "es-US", lcid => 0x0000540a, sabbrevlangname => "EST" },
{ name => "es-UY", lcid => 0x0000380a, sabbrevlangname => "ESY" },
{ name => "es-VE", lcid => 0x0000200a, sabbrevlangname => "ESV" },
{ name => "et", lcid => 0x00000025, oemcp => 775, group => 3, sabbrevlangname => "ETI", sopentypelang => "ETI" },
{ name => "et-EE", lcid => 0x00000425 },
{ name => "eu", lcid => 0x0000002d, oemcp => 850, maccp => 65001, sabbrevlangname => "EUQ", sopentypelang => "EUQ" },
{ name => "eu-ES", lcid => 0x0000042d },
{ name => "ewo" },
{ name => "ewo-CM" },
{ name => "fa", lcid => 0x00000029, inegnumber => 3, oemcp => 720, slist => "\x{061b}", group => 13, sabbrevlangname => "FAR", sopentypelang => "FAR" },
{ name => "fa-AF", alias => "prs-AF" },
{ name => "fa-IR", lcid => 0x00000429 },
{ name => "ff", lcid => 0x00000067, oemcp => 850, ebcdiccp => 20297 },
{ name => "ff-CM", alias => "ff-Latn-CM" },
{ name => "ff-GN", alias => "ff-Latn-GN" },
{ name => "ff-MR", alias => "ff-Latn-MR" },
{ name => "ff-NG", alias => "ff-Latn-NG" },
{ name => "ff-SN", alias => "ff-Latn-SN" },
{ name => "ff-Adlm" },
{ name => "ff-Adlm-BF" },
{ name => "ff-Adlm-CM" },
{ name => "ff-Adlm-GH" },
{ name => "ff-Adlm-GM" },
{ name => "ff-Adlm-GN" },
{ name => "ff-Adlm-GW" },
{ name => "ff-Adlm-LR" },
{ name => "ff-Adlm-MR" },
{ name => "ff-Adlm-NE" },
{ name => "ff-Adlm-NG" },
{ name => "ff-Adlm-SL" },
{ name => "ff-Adlm-SN" },
{ name => "ff-Latn", lcid => 0x00007c67 },
{ name => "ff-Latn-BF", oemcp => 65001 },
{ name => "ff-Latn-CM" },
{ name => "ff-Latn-GH", oemcp => 65001 },
{ name => "ff-Latn-GM", oemcp => 65001 },
{ name => "ff-Latn-GN" },
{ name => "ff-Latn-GW", oemcp => 65001 },
{ name => "ff-Latn-LR", oemcp => 65001 },
{ name => "ff-Latn-MR" },
{ name => "ff-Latn-NE", oemcp => 65001 },
{ name => "ff-Latn-NG", lcid => 0x00000467, sabbrevlangname => "ZZZ" },
{ name => "ff-Latn-SL", oemcp => 65001 },
{ name => "ff-Latn-SN", lcid => 0x00000867 },
{ name => "fi", lcid => 0x0000000b, oemcp => 850, ebcdiccp => 20278 },
{ name => "fi-FI", lcid => 0x0000040b },
{ name => "fil", lcid => 0x00000064, oemcp => 437, ebcdiccp => 500, sabbrevlangname => "FPO", sopentypelang => "PIL" },
{ name => "fil-PH", lcid => 0x00000464 },
{ name => "fil-Latn", alias => "fil" },
{ name => "fil-Latn-PH", alias => "fil-PH" },
{ name => "fo", lcid => 0x00000038, oemcp => 850, maccp => 10079, ebcdiccp => 20277, sabbrevlangname => "FOS", sopentypelang => "FOS" },
{ name => "fo-DK", oemcp => 65001, maccp => 65001 },
{ name => "fo-FO", lcid => 0x00000438 },
{ name => "fr", lcid => 0x0000000c, oemcp => 850, ebcdiccp => 20297 },
{ name => "fr-029", lcid => 0x00001c0c, file => "fr", sintlsymbol => "XCD", sabbrevlangname => "ZZZ" },
{ name => "fr-BE", lcid => 0x0000080c, sabbrevlangname => "FRB" },
{ name => "fr-BF" },
{ name => "fr-BI" },
{ name => "fr-BJ" },
{ name => "fr-BL" },
{ name => "fr-CA", lcid => 0x00000c0c, sabbrevlangname => "FRC" },
{ name => "fr-CD", lcid => 0x0000240c, sabbrevlangname => "FRD" },
{ name => "fr-CF" },
{ name => "fr-CG" },
{ name => "fr-CH", lcid => 0x0000100c, sabbrevlangname => "FRS" },
{ name => "fr-CI", lcid => 0x0000300c, sabbrevlangname => "FRI" },
{ name => "fr-CM", lcid => 0x00002c0c, sabbrevlangname => "FRE" },
{ name => "fr-DJ" },
{ name => "fr-DZ" },
{ name => "fr-FR", lcid => 0x0000040c },
{ name => "fr-GA" },
{ name => "fr-GF" },
{ name => "fr-GN" },
{ name => "fr-GP" },
{ name => "fr-GQ" },
{ name => "fr-HT", lcid => 0x00003c0c, sabbrevlangname => "FRH" },
{ name => "fr-KM" },
{ name => "fr-LU", lcid => 0x0000140c, sabbrevlangname => "FRL" },
{ name => "fr-MA", lcid => 0x0000380c, sabbrevlangname => "FRO" },
{ name => "fr-MC", lcid => 0x0000180c, sabbrevlangname => "FRM" },
{ name => "fr-MF" },
{ name => "fr-MG" },
{ name => "fr-ML", lcid => 0x0000340c, sabbrevlangname => "FRF" },
{ name => "fr-MQ" },
{ name => "fr-MR" },
{ name => "fr-MU" },
{ name => "fr-NC" },
{ name => "fr-NE" },
{ name => "fr-PF" },
{ name => "fr-PM" },
{ name => "fr-RE", lcid => 0x0000200c, sabbrevlangname => "FRR" },
{ name => "fr-RW" },
{ name => "fr-SC" },
{ name => "fr-SN", lcid => 0x0000280c, sabbrevlangname => "FRN" },
{ name => "fr-SY" },
{ name => "fr-TD" },
{ name => "fr-TG" },
{ name => "fr-TN" },
{ name => "fr-VU" },
{ name => "fr-WF" },
{ name => "fr-YT" },
{ name => "fur", sopentypelang => "FRL" },
{ name => "fur-IT" },
{ name => "fy", lcid => 0x00000062, oemcp => 850, sabbrevlangname => "FYN", sopentypelang => "FRI" },
{ name => "fy-NL", lcid => 0x00000462 },
{ name => "ga", lcid => 0x0000003c, oemcp => 850, sabbrevlangname => "IRE", sopentypelang => "IRI" },
{ name => "ga-GB" },
{ name => "ga-IE", lcid => 0x0000083c },
{ name => "gd", lcid => 0x00000091, oemcp => 850, ebcdiccp => 20285, sopentypelang => "GAE" },
{ name => "gd-GB", lcid => 0x00000491 },
{ name => "gd-Latn", alias => "gd" },
{ name => "gl", lcid => 0x00000056, oemcp => 850, sabbrevlangname => "GLC", sopentypelang => "GAL" },
{ name => "gl-ES", lcid => 0x00000456 },
{ name => "gn", lcid => 0x00000074, oemcp => 850, ebcdiccp => 20284, slist => ",", dir => "seed", sopentypelang => "GUA" },
{ name => "gn-PY", lcid => 0x00000474, dir => "seed" },
{ name => "gsw", lcid => 0x00000084, oemcp => 850, ebcdiccp => 20297, sabbrevlangname => "ZZZ", sopentypelang => "ALS" },
{ name => "gsw-CH" },
{ name => "gsw-FR", lcid => 0x00000484, sabbrevlangname => "GSW" },
{ name => "gsw-LI" },
{ name => "gu", lcid => 0x00000047, slist => ",", group => 15 },
{ name => "gu-IN", lcid => 0x00000447 },
{ name => "guz" },
{ name => "guz-KE" },
{ name => "gv", sopentypelang => "MNX" },
{ name => "gv-IM" },
{ name => "ha", lcid => 0x00000068, oemcp => 437 },
{ name => "ha-GH", alias => "ha-Latn-GH" },
{ name => "ha-Latn", lcid => 0x00007c68, file => "ha" },
{ name => "ha-Latn-GH", file => "ha_GH", ebcdiccp => 500 },
{ name => "ha-Latn-NE", file => "ha_NE", ebcdiccp => 500 },
{ name => "ha-Latn-NG", lcid => 0x00000468, file => "ha_NG" },
{ name => "ha-NE", alias => "ha-Latn-NE" },
{ name => "ha-NG", alias => "ha-Latn-NG" },
{ name => "haw", lcid => 0x00000075, oemcp => 437 },
{ name => "haw-Latn", alias => "haw" },
{ name => "haw-Latn-US", alias => "haw-US" },
{ name => "haw-US", lcid => 0x00000475 },
{ name => "he", lcid => 0x0000000d, oemcp => 862, slist => ",", group => 12, sopentypelang => "IWR" },
{ name => "he-IL", lcid => 0x0000040d },
{ name => "hi", lcid => 0x00000039, slist => ",", group => 15 },
{ name => "hi-IN", lcid => 0x00000439 },
{ name => "hr", lcid => 0x0000001a, inegnumber => 2, oemcp => 852, maccp => 10082, group => 2 },
{ name => "hr-BA", lcid => 0x0000101a, ebcdiccp => 870, inegnumber => 1, sabbrevlangname => "HRB" },
{ name => "hr-HR", lcid => 0x0000041a },
{ name => "hsb", lcid => 0x0000002e, oemcp => 850, ebcdiccp => 870, sopentypelang => "USB" },
{ name => "hsb-DE", lcid => 0x0000042e },
{ name => "hu", lcid => 0x0000000e, oemcp => 852, group => 2 },
{ name => "hu-HU", lcid => 0x0000040e },
{ name => "hu-HU_technl", lcid => 0x0001040e, alias => "hu-HU" },
{ name => "hy", lcid => 0x0000002b, slist => ",", group => 17 },
{ name => "hy-AM", lcid => 0x0000042b },
{ name => "ia" },
{ name => "ia-001" },
## name => "ibb", lcid => 0x00000069 },
## name => "ibb-NG", lcid => 0x00000469 },
{ name => "id", lcid => 0x00000021, oemcp => 850 },
{ name => "id-ID", lcid => 0x00000421 },
{ name => "ig", lcid => 0x00000070, oemcp => 437 },
{ name => "ig-Latn", alias => "ig" },
{ name => "ig-Latn-NG", alias => "ig-NG" },
{ name => "ig-NG", lcid => 0x00000470 },
{ name => "ii", lcid => 0x00000078, group => 9, sopentypelang => "YIM" },
{ name => "ii-CN", lcid => 0x00000478 },
{ name => "ii-Yiii", alias => "ii" },
{ name => "ii-Yiii-CN", alias => "ii-CN" },
{ name => "is", lcid => 0x0000000f, oemcp => 850, maccp => 10079, ebcdiccp => 20871 },
{ name => "is-IS", lcid => 0x0000040f },
{ name => "it", lcid => 0x00000010, oemcp => 850, ebcdiccp => 20280 },
{ name => "it-CH", lcid => 0x00000810, ebcdiccp => 500, sabbrevlangname => "ITS" },
{ name => "it-IT", lcid => 0x00000410 },
{ name => "it-SM" },
{ name => "it-VA", oemcp => 65001 },
{ name => "iu", lcid => 0x0000005d, oemcp => 437, slist => ",", sortlocale => "iu-Latn-CA", dir => "seed", sabbrevlangname => "IUK", sopentypelang => "INU" },
{ name => "iu-Cans", lcid => 0x0000785d, file => "iu", oemcp => 65001, dir => "seed", sabbrevlangname => "IUS" },
{ name => "iu-Cans-CA", lcid => 0x0000045d, file => "iu_CA", dir => "seed" },
{ name => "iu-Latn", lcid => 0x00007c5d, dir => "seed" },
{ name => "iu-Latn-CA", lcid => 0x0000085d, dir => "seed" },
{ name => "ja", lcid => 0x00000011, ireadinglayout => 2, oemcp => 932, slist => ",", sscripts => "Hani Hira Jpan Kana", group => 7, sopentypelang => "JAN" },
{ name => "ja-JP", lcid => 0x00000411 },
{ name => "ja-JP_radstr", lcid => 0x00040411, alias => "ja-JP" },
{ name => "ja-JP-u-co-unihan", alias => "ja-JP_radstr" },
{ name => "jgo" },
{ name => "jgo-CM" },
{ name => "jmc" },
{ name => "jmc-TZ" },
{ name => "jv", oemcp => 850 },
{ name => "jv-ID", alias => "jv-Latn-ID" },
## name => "jv-Java" },
## name => "jv-Java-ID" },
{ name => "jv-Latn", file => "jv" },
{ name => "jv-Latn-ID", file => "jv_ID" },
{ name => "ka", lcid => 0x00000037, group => 16 },
{ name => "ka-GE", lcid => 0x00000437 },
{ name => "ka-GE_modern", lcid => 0x00010437, alias => "ka-GE" },
{ name => "kab", sopentypelang => "KAB0" },
{ name => "kab-DZ" },
{ name => "kam", sopentypelang => "KMB" },
{ name => "kam-KE" },
{ name => "kde" },
{ name => "kde-TZ" },
{ name => "kea" },
{ name => "kea-CV" },
{ name => "kgp" },
{ name => "kgp-BR" },
{ name => "khq" },
{ name => "khq-ML" },
{ name => "ki" },
{ name => "ki-KE" },
{ name => "kk", lcid => 0x0000003f, group => 5, sabbrevlangname => "KKZ" },
{ name => "kk-Cyrl", alias => "kk" },
{ name => "kk-Cyrl-KZ", alias => "kk-KZ" },
{ name => "kk-KZ", lcid => 0x0000043f },
{ name => "kkj" },
{ name => "kkj-CM" },
{ name => "kl", lcid => 0x0000006f, oemcp => 850, ebcdiccp => 20277, sopentypelang => "GRN" },
{ name => "kl-GL", lcid => 0x0000046f },
{ name => "kln", sopentypelang => "KAL" },
{ name => "kln-KE" },
{ name => "km", lcid => 0x00000053, inegnumber => 2, slist => ",", group => 15 },
{ name => "km-KH", lcid => 0x00000453 },
{ name => "kn", lcid => 0x0000004b, slist => ",", group => 15, sabbrevlangname => "KDI" },
{ name => "kn-IN", lcid => 0x0000044b },
{ name => "ko", lcid => 0x00000012, ireadinglayout => 2, slist => ",", oemcp => 949, ebcdiccp => 20833, sscripts => "Hang Hani Kore", group => 8 },
{ name => "ko-KP", oemcp => 65001 },
{ name => "ko-KR", lcid => 0x00000412 },
{ name => "kok", lcid => 0x00000057, slist => ",", group => 15, sabbrevlangname => "KNK" },
{ name => "kok-IN", lcid => 0x00000457 },
{ name => "kr", lcid => 0x00000071, sortlocale => "kr-Latn-NG", oemcp => 850, dir => "exemplars", sabbrevlangname => "ZZZ", sopentypelang => "KNR" },
{ name => "kr-Latn", file => "kr", dir => "exemplars" },
{ name => "kr-Latn-NG", lcid => 0x00000471, file => "kr", dir => "exemplars" },
{ name => "kr-NG", alias => "kr-Latn-NG" },
{ name => "ks", lcid => 0x00000060, group => 15, sabbrevlangname => "ZZZ", sopentypelang => "KSH" },
{ name => "ks-Arab", lcid => 0x00000460 },
{ name => "ks-Arab-IN" },
{ name => "ks-Deva", slist => ",", dir => "seed" },
{ name => "ks-Deva-IN", lcid => 0x00000860, dir => "seed" },
{ name => "ks-IN", alias => "ks-Arab-IN" },
{ name => "ksb" },
{ name => "ksb-TZ" },
{ name => "ksf" },
{ name => "ksf-CM" },
{ name => "ksh", sopentypelang => "KSH0" },
{ name => "ksh-DE" },
{ name => "ku", lcid => 0x00000092, file => "ckb", slist => "\x{061b}", sortlocale => "ku-Arab-IQ", oemcp => 720 },
{ name => "ku-Arab", lcid => 0x00007c92, file => "ckb", group => 13 },
{ name => "ku-Arab-IQ", lcid => 0x00000492, file => "ckb_IQ" },
{ name => "ku-Arab-IR", file => "ckb_IR", oemcp => 65001 },
{ name => "kw" },
{ name => "kw-GB" },
{ name => "ky", lcid => 0x00000040, oemcp => 866, group => 5, sabbrevlangname => "KYR" },
{ name => "ky-Cyrl", alias => "ky" },
{ name => "ky-Cyrl-KG", alias => "ky-KG" },
{ name => "ky-KG", lcid => 0x00000440 },
{ name => "la", lcid => 0x00000076, oemcp => 437, slist => ",", dir => "seed", sabbrevlangname => "ZZZ" },
{ name => "la-001", lcid => 0x00000476, file => "la", dir => "seed" },
{ name => "lag" },
{ name => "lag-TZ" },
{ name => "lb", lcid => 0x0000006e, oemcp => 850, ebcdiccp => 20297, sabbrevlangname => "LBX" },
{ name => "lb-LU", lcid => 0x0000046e },
{ name => "lg" },
{ name => "lg-UG" },
{ name => "lkt" },
{ name => "lkt-US" },
{ name => "ln" },
{ name => "ln-AO" },
{ name => "ln-CD" },
{ name => "ln-CF" },
{ name => "ln-CG" },
{ name => "lo", lcid => 0x00000054, group => 15 },
{ name => "lo-LA", lcid => 0x00000454 },
{ name => "lrc" },
{ name => "lrc-IQ" },
{ name => "lrc-IR" },
{ name => "lt", lcid => 0x00000027, oemcp => 775, group => 3, sabbrevlangname => "LTH", sopentypelang => "LTH" },
{ name => "lt-LT", lcid => 0x00000427 },
{ name => "lu" },
{ name => "lu-CD" },
{ name => "luo" },
{ name => "luo-KE" },
{ name => "luy", sopentypelang => "LUH" },
{ name => "luy-KE" },
{ name => "lv", lcid => 0x00000026, oemcp => 775, group => 3, sabbrevlangname => "LVI", sopentypelang => "LVI" },
{ name => "lv-LV", lcid => 0x00000426 },
{ name => "mai" },
{ name => "mai-IN" },
{ name => "mas" },
{ name => "mas-KE" },
{ name => "mas-TZ" },
{ name => "mer" },
{ name => "mer-KE" },
{ name => "mfe" },
{ name => "mfe-MU" },
{ name => "mg" },
{ name => "mg-MG" },
{ name => "mgh" },
{ name => "mgh-MZ" },
{ name => "mgo" },
{ name => "mgo-CM" },
{ name => "mi", lcid => 0x00000081, slist => "," },
{ name => "mi-Latn", alias => "mi" },
{ name => "mi-Latn-NZ", alias => "mi-NZ" },
{ name => "mi-NZ", lcid => 0x00000481 },
{ name => "mk", lcid => 0x0000002f, oemcp => 866, ebcdiccp => 500, group => 5, sabbrevlangname => "MKI" },
{ name => "mk-MK", lcid => 0x0000042f },
{ name => "ml", lcid => 0x0000004c, group => 15, sabbrevlangname => "MYM", sopentypelang => "MLR" },
{ name => "ml-IN", lcid => 0x0000044c },
{ name => "mn", lcid => 0x00000050, oemcp => 866, sopentypelang => "MNG" },
{ name => "mn-Cyrl", lcid => 0x00007850, file => "mn", sabbrevlangname => "MNN" },
{ name => "mn-Cyrl-MN", alias => "mn-MN" },
{ name => "mn-MN", lcid => 0x00000450, sparent => "mn-Cyrl", group => 5 },
{ name => "mn-Mong", lcid => 0x00007c50, oemcp => 65001, slist => ",", group => 15, dir => "seed", sabbrevlangname => "MNG" },
{ name => "mn-Mong-CN", lcid => 0x00000850, dir => "seed" },
{ name => "mn-Mong-MN", lcid => 0x00000c50, dir => "seed", sabbrevlangname => "MNM" },
{ name => "mni", lcid => 0x00000058, slist => ",", sabbrevlangname => "ZZZ" },
{ name => "mni-IN", lcid => 0x00000458, file => "mni_Beng_IN" },
{ name => "moh", lcid => 0x0000007c, oemcp => 850, ebcdiccp => 37, slist => ",", dir => "seed", sabbrevlangname => "MWK" },
{ name => "moh-CA", lcid => 0x0000047c, dir => "seed" },
{ name => "moh-Latn", alias => "moh" },
{ name => "moh-Latn-CA", alias => "moh-CA" },
{ name => "mr", lcid => 0x0000004e, slist => ",", group => 15 },
{ name => "mr-IN", lcid => 0x0000044e },
{ name => "ms", lcid => 0x0000003e, oemcp => 850, sabbrevlangname => "MSL", sopentypelang => "MLY" },
{ name => "ms-BN", lcid => 0x0000083e, sabbrevlangname => "MSB" },
{ name => "ms-ID" },
{ name => "ms-Latn", alias => "ms" },
{ name => "ms-Latn-BN", alias => "ms-BN" },
{ name => "ms-Latn-MY", alias => "ms-MY" },
{ name => "ms-Latn-SG", alias => "ms-SG" },
{ name => "ms-MY", lcid => 0x0000043e },
{ name => "ms-SG" },
{ name => "mt", lcid => 0x0000003a, sopentypelang => "MTS" },
{ name => "mt-MT", lcid => 0x0000043a },
{ name => "mua" },
{ name => "mua-CM" },
{ name => "my", lcid => 0x00000055, sopentypelang => "BRM" },
{ name => "my-MM", lcid => 0x00000455 },
{ name => "mzn" },
{ name => "mzn-IR" },
{ name => "naq" },
{ name => "naq-NA" },
{ name => "nb", lcid => 0x00007c14, oemcp => 850, ebcdiccp => 20277, sabbrevlangname => "NOR", sopentypelang => "NOR" },
{ name => "nb-NO", lcid => 0x00000414 },
{ name => "nb-SJ" },
{ name => "nd", sopentypelang => "NDB" },
{ name => "nd-ZW" },
{ name => "nds" },
{ name => "nds-DE" },
{ name => "nds-NL" },
{ name => "ne", lcid => 0x00000061, slist => "," },
{ name => "ne-IN", lcid => 0x00000861, sabbrevlangname => "NEI" },
{ name => "ne-NP", lcid => 0x00000461, group => 15 },
{ name => "nl", lcid => 0x00000013, oemcp => 850 },
{ name => "nl-AW" },
{ name => "nl-BE", lcid => 0x00000813, sabbrevlangname => "NLB" },
{ name => "nl-BQ" },
{ name => "nl-CW" },
{ name => "nl-NL", lcid => 0x00000413 },
{ name => "nl-SR" },
{ name => "nl-SX" },
{ name => "nmg" },
{ name => "nmg-CM" },
{ name => "nn", lcid => 0x00007814, oemcp => 850, ebcdiccp => 20277, sabbrevlangname => "NON", sopentypelang => "NYN" },
{ name => "nn-NO", lcid => 0x00000814 },
{ name => "nnh" },
{ name => "nnh-CM" },
{ name => "no", lcid => 0x00000014, oemcp => 850, ebcdiccp => 20277, sortlocale => "nb-NO" },
{ name => "nqo", idigits => 3, inegnumber => 3, slist => "\x{060c}", dir => "seed", sopentypelang => "NKO" },
{ name => "nqo-GN", dir => "seed" },
{ name => "nr", dir => "seed", sopentypelang => "NDB" },
{ name => "nr-ZA", dir => "seed" },
{ name => "nso", lcid => 0x0000006c, oemcp => 850, dir => "seed", sopentypelang => "SOT" },
{ name => "nso-ZA", lcid => 0x0000046c, dir => "seed" },
{ name => "nus" },
{ name => "nus-SD", alias => "nus-SS" },
{ name => "nus-SS" },
{ name => "nyn", sopentypelang => "NKL" },
{ name => "nyn-UG" },
{ name => "oc", lcid => 0x00000082, oemcp => 850, ebcdiccp => 20297, dir => "seed" },
{ name => "oc-FR", lcid => 0x00000482, dir => "seed" },
{ name => "oc-Latn", alias => "oc" },
{ name => "oc-Latn-FR", alias => "oc-FR" },
{ name => "om", lcid => 0x00000072, sopentypelang => "ORO" },
{ name => "om-ET", lcid => 0x00000472 },
{ name => "om-KE" },
{ name => "or", lcid => 0x00000048, slist => ",", group => 15 },
{ name => "or-IN", lcid => 0x00000448 },
{ name => "os" },
{ name => "os-GE" },
{ name => "os-RU" },
{ name => "pa", lcid => 0x00000046, slist => "," },
{ name => "pa-Arab", lcid => 0x00007c46, slist => ";", inegnumber => 2, oemcp => 720, group => 13, sabbrevlangname => "PAP" },
{ name => "pa-Arab-PK", lcid => 0x00000846 },
{ name => "pa-Guru" },
{ name => "pa-Guru-IN", alias => "pa-IN" },
{ name => "pa-IN", lcid => 0x00000446, sparent => "pa-Guru", file => "pa_Guru_IN", group => 15 },
## name => "pap", lcid => 0x00000079 },
## name => "pap-029", lcid => 0x00000479 },
{ name => "pcm" },
{ name => "pcm-NG" },
{ name => "pl", lcid => 0x00000015, oemcp => 852, ebcdiccp => 20880, group => 2, sabbrevlangname => "PLK", sopentypelang => "PLK" },
{ name => "pl-PL", lcid => 0x00000415 },
## name => "prg" },
## name => "prg-001" },
{ name => "prs", lcid => 0x0000008c, file => "fa", inegnumber => 3, oemcp => 720, group => 13, sopentypelang => "DRI" },
{ name => "prs-AF", lcid => 0x0000048c, file => "fa_AF" },
{ name => "prs-Arab", alias => "prs" },
{ name => "prs-Arab-AF", alias => "prs-AF" },
{ name => "ps", lcid => 0x00000063, group => 13, sabbrevlangname => "PAS", sopentypelang => "PAS" },
{ name => "ps-AF", lcid => 0x00000463 },
{ name => "ps-PK" },
{ name => "pt", lcid => 0x00000016, oemcp => 850, sabbrevlangname => "PTB", sopentypelang => "PTG" },
{ name => "pt-AO" },
{ name => "pt-BR", lcid => 0x00000416 },
{ name => "pt-CH", oemcp => 65001 },
{ name => "pt-CV" },
{ name => "pt-GQ", oemcp => 65001 },
{ name => "pt-GW" },
{ name => "pt-LU", oemcp => 65001 },
{ name => "pt-MO" },
{ name => "pt-MZ" },
{ name => "pt-PT", lcid => 0x00000816, sabbrevlangname => "PTG" },
{ name => "pt-ST" },
{ name => "pt-TL" },
## name => qps-Latn-x-sh", lcid => 0x80000901 },
## name => qps-ploc", lcid => 0x80000501 },
## name => qps-ploca", lcid => 0x800005fe },
## name => qps-plocm", lcid => 0x800009ff },
{ name => "qu", alias => "quz" },
{ name => "qu-BO", alias => "quz-BO" },
{ name => "qu-EC", alias => "quz-EC" },
{ name => "qu-PE", alias => "quz-PE" },
{ name => "quc", lcid => 0x00000086, oemcp => 850, ebcdiccp => 20284, slist => ",", dir => "seed" },
{ name => "quc-Latn", lcid => 0x00007c86, file => "quc", dir => "seed" },
{ name => "quc-Latn-GT", lcid => 0x00000486, file => "quc_GT", dir => "seed" },
{ name => "qut", alias => "quc" },
{ name => "qut-GT", alias => "quc-Latn-GT" },
{ name => "quz", lcid => 0x0000006b, file => "qu", territory => "BO", oemcp => 850, ebcdiccp => 20284, slist => "," },
{ name => "quz-BO", lcid => 0x0000046b, file => "qu_BO" },
{ name => "quz-EC", lcid => 0x0000086b, file => "qu_EC" },
{ name => "quz-Latn", alias => "quz" },
{ name => "quz-Latn-BO", alias => "quz-BO" },
{ name => "quz-Latn-EC", alias => "quz-EC" },
{ name => "quz-Latn-PE", alias => "quz-PE" },
{ name => "quz-PE", lcid => 0x00000c6b, file => "qu_PE" },
{ name => "rm", lcid => 0x00000017, oemcp => 850, ebcdiccp => 20273, sabbrevlangname => "RMC", sopentypelang => "RMS" },
{ name => "rm-CH", lcid => 0x00000417 },
{ name => "rn" },
{ name => "rn-BI" },
{ name => "ro", lcid => 0x00000018, oemcp => 852, ebcdiccp => 20880, sabbrevlangname => "ROM", sopentypelang => "ROM" },
{ name => "ro-MD", lcid => 0x00000818, maccp => 65001, sabbrevlangname => "ROD" },
{ name => "ro-RO", lcid => 0x00000418, group => 2 },
{ name => "rof" },
{ name => "rof-TZ" },
{ name => "ru", lcid => 0x00000019, oemcp => 866 },
{ name => "ru-BY", maccp => 65001 },
{ name => "ru-KG", maccp => 65001 },
{ name => "ru-KZ", maccp => 65001 },
{ name => "ru-MD", lcid => 0x00000819, maccp => 65001, sabbrevlangname => "RUM" },
{ name => "ru-RU", lcid => 0x00000419, group => 5 },
{ name => "ru-UA", maccp => 65001 },
{ name => "rw", lcid => 0x00000087, oemcp => 437, sopentypelang => "RUA" },
{ name => "rw-RW", lcid => 0x00000487 },
{ name => "rwk" },
{ name => "rwk-TZ" },
{ name => "sa", lcid => 0x0000004f, slist => ",", group => 15 },
{ name => "sa-Deva", alias => "sa" },
{ name => "sa-Deva-IN", alias => "sa-IN" },
{ name => "sa-IN", lcid => 0x0000044f },
{ name => "sah", lcid => 0x00000085, oemcp => 866, group => 5, sopentypelang => "YAK" },
{ name => "sah-Cyrl", alias => "sah" },
{ name => "sah-Cyrl-RU", alias => "sah-RU" },
{ name => "sah-RU", lcid => 0x00000485 },
{ name => "saq" },
{ name => "saq-KE" },
{ name => "sat" },
{ name => "sat-Olck" },
{ name => "sat-Olck-IN" },
{ name => "sbp" },
{ name => "sbp-TZ" },
{ name => "sc" },
{ name => "sc-IT" },
{ name => "sd", lcid => 0x00000059, inegnumber => 3, oemcp => 720, sabbrevlangname => "SIP" },
{ name => "sd-Arab", lcid => 0x00007c59, group => 13 },
{ name => "sd-Arab-PK", lcid => 0x00000859 },
{ name => "sd-Deva", inegnumber => 1, slist => ",", oemcp => 65001, group => 15 },
{ name => "sd-Deva-IN", lcid => 0x00000459, sabbrevlangname => "ZZZ" },
{ name => "sd-PK", alias => "sd-Arab-PK" },
{ name => "se", lcid => 0x0000003b, oemcp => 850, ebcdiccp => 20277, sopentypelang => "NSM" },
{ name => "se-FI", lcid => 0x00000c3b, ebcdiccp => 20278, sabbrevlangname => "SMG" },
{ name => "se-NO", lcid => 0x0000043b },
{ name => "se-SE", lcid => 0x0000083b, ebcdiccp => 20278, sabbrevlangname => "SMF" },
{ name => "se-Latn", alias => "se" },
{ name => "se-Latn-FI", alias => "se-FI" },
{ name => "se-Latn-NO", alias => "se-NO" },
{ name => "se-Latn-SE", alias => "se-SE" },
{ name => "seh" },
{ name => "seh-MZ" },
{ name => "ses" },
{ name => "ses-ML" },
{ name => "sg", sopentypelang => "SGO" },
{ name => "sg-CF" },
{ name => "shi" },
{ name => "shi-Latn" },
{ name => "shi-Latn-MA" },
{ name => "shi-Tfng" },
{ name => "shi-Tfng-MA" },
{ name => "si", lcid => 0x0000005b, group => 15, sopentypelang => "SNH" },
{ name => "si-LK", lcid => 0x0000045b },
{ name => "sk", lcid => 0x0000001b, oemcp => 852, ebcdiccp => 20880, group => 2, sabbrevlangname => "SKY", sopentypelang => "SKY" },
{ name => "sk-SK", lcid => 0x0000041b },
{ name => "sl", lcid => 0x00000024, oemcp => 852, ebcdiccp => 20880, group => 2 },
{ name => "sl-SI", lcid => 0x00000424 },
{ name => "sma", lcid => 0x0000783b, sparent => "se", ebcdiccp => 20278, dir => "seed", sabbrevlangname => "SMB", sopentypelang => "SSM" },
{ name => "sma-Latn", alias => "sma" },
{ name => "sma-Latn-NO", alias => "sma-NO" },
{ name => "sma-Latn-SE", alias => "sma-SE" },
{ name => "sma-NO", lcid => 0x0000183b, ebcdiccp => 20277, dir => "seed", sabbrevlangname => "SMA" },
{ name => "sma-SE", lcid => 0x00001c3b, dir => "seed" },
{ name => "smj", lcid => 0x00007c3b, sparent => "se", ebcdiccp => 20278, dir => "seed", sabbrevlangname => "SMK", sopentypelang => "LSM" },
{ name => "smj-Latn", alias => "smj" },
{ name => "smj-Latn-NO", alias => "smj-NO" },
{ name => "smj-Latn-SE", alias => "smj-SE" },
{ name => "smj-NO", lcid => 0x0000103b, ebcdiccp => 20277, dir => "seed", sabbrevlangname => "SMJ" },
{ name => "smj-SE", lcid => 0x0000143b, dir => "seed" },
{ name => "smn", lcid => 0x0000703b, sparent => "se", ebcdiccp => 20278, sopentypelang => "ISM" },
{ name => "smn-FI", lcid => 0x0000243b },
{ name => "smn-Latn", alias => "smn" },
{ name => "smn-Latn-FI", alias => "smn-FI" },
{ name => "sms", lcid => 0x0000743b, sparent => "se", ebcdiccp => 20278, dir => "seed", sopentypelang => "SKS" },
{ name => "sms-FI", lcid => 0x0000203b, dir => "seed" },
{ name => "sms-Latn", alias => "sms" },
{ name => "sms-Latn-FI", alias => "sms-FI" },
{ name => "sn", sopentypelang => "SNA0" },
{ name => "sn-Latn", file => "sn" },
{ name => "sn-Latn-ZW", file => "sn_ZW" },
{ name => "sn-ZW", alias => "sn-Latn-ZW" },
{ name => "so", lcid => 0x00000077, sopentypelang => "SML" },
{ name => "so-DJ" },
{ name => "so-ET" },
{ name => "so-KE" },
{ name => "so-SO", lcid => 0x00000477 },
{ name => "sq", lcid => 0x0000001c, oemcp => 852, ebcdiccp => 20880, group => 2 },
{ name => "sq-AL", lcid => 0x0000041c },
{ name => "sq-MK" },
{ name => "sq-XK" },
{ name => "sr", lcid => 0x00007c1a, sortlocale => "sr-Latn-RS", oemcp => 852, group => 2, sabbrevlangname => "SRB", sopentypelang => "SRB" },
{ name => "sr-Cyrl", lcid => 0x00006c1a, oemcp => 855, ebcdiccp => 21025, group => 5, sabbrevlangname => "SRO" },
{ name => "sr-Cyrl-BA", lcid => 0x00001c1a, sabbrevlangname => "SRN" },
{ name => "sr-Cyrl-ME", lcid => 0x0000301a, sabbrevlangname => "SRQ" },
{ name => "sr-Cyrl-RS", lcid => 0x0000281a },
{ name => "sr-Cyrl-XK" },
{ name => "sr-Latn", lcid => 0x0000701a, sabbrevlangname => "SRM" },
{ name => "sr-Latn-BA", lcid => 0x0000181a, maccp => 10082, ebcdiccp => 870, sabbrevlangname => "SRS" },
{ name => "sr-Latn-ME", lcid => 0x00002c1a, sabbrevlangname => "SRP" },
{ name => "sr-Latn-RS", lcid => 0x0000241a, sabbrevlangname => "SRM" },
{ name => "sr-Latn-XK" },
## name => "sr-Cyrl-CS", lcid => 0x00000c1a },
## name => "sr-Latn-CS", lcid => 0x0000081a },
{ name => "ss", dir => "seed", sopentypelang => "SWZ" },
{ name => "ss-SZ", dir => "seed" },
{ name => "ss-ZA", dir => "seed" },
{ name => "ssy", dir => "seed" },
{ name => "ssy-ER", dir => "seed" },
{ name => "st", lcid => 0x00000030, dir => "seed" },
{ name => "st-LS", dir => "seed" },
{ name => "st-ZA", lcid => 0x00000430, dir => "seed" },
{ name => "su" },
{ name => "su-Latn" },
{ name => "su-Latn-ID" },
{ name => "sv", lcid => 0x0000001d, oemcp => 850, ebcdiccp => 20278, sabbrevlangname => "SVE", sopentypelang => "SVE" },
{ name => "sv-AX" },
{ name => "sv-FI", lcid => 0x0000081d, sabbrevlangname => "SVF" },
{ name => "sv-SE", lcid => 0x0000041d, sabbrevlangname => "SVE" },
{ name => "sw", lcid => 0x00000041, territory => "KE", oemcp => 437, ebcdiccp => 500, sabbrevlangname => "SWK", sopentypelang => "SWK" },
{ name => "sw-CD" },
{ name => "sw-KE", lcid => 0x00000441 },
{ name => "sw-TZ" },
{ name => "sw-UG" },
{ name => "swc-CD", alias => "sw-CD" },
{ name => "syr", lcid => 0x0000005a, slist => ",", group => 13, dir => "seed" },
{ name => "syr-SY", lcid => 0x0000045a, dir => "seed" },
{ name => "syr-Syrc", alias => "syr" },
{ name => "syr-Syrc-SY", alias => "syr-SY" },
{ name => "ta", lcid => 0x00000049, slist => ",", group => 15, sabbrevlangname => "TAI" },
{ name => "ta-IN", lcid => 0x00000449 },
{ name => "ta-LK", lcid => 0x00000849, sabbrevlangname => "TAM" },
{ name => "ta-MY" },
{ name => "ta-SG" },
{ name => "te", lcid => 0x0000004a, group => 15 },
{ name => "te-IN", lcid => 0x0000044a },
{ name => "teo" },
{ name => "teo-KE" },
{ name => "teo-UG" },
{ name => "tg", lcid => 0x00000028, oemcp => 866, group => 5, sabbrevlangname => "TAJ", sopentypelang => "TAJ" },
{ name => "tg-Cyrl", lcid => 0x00007c28, file => "tg" },
{ name => "tg-Cyrl-TJ", lcid => 0x00000428, file => "tg_TJ" },
{ name => "tg-TJ", alias => "tg-Cyrl-TJ" },
{ name => "th", lcid => 0x0000001e, oemcp => 874, ebcdiccp => 20838, slist => ",", group => 11 },
{ name => "th-TH", lcid => 0x0000041e },
{ name => "ti", lcid => 0x00000073, territory => "ER", sopentypelang => "TGY" },
{ name => "ti-ER", lcid => 0x00000873 },
{ name => "ti-ET", lcid => 0x00000473, sabbrevlangname => "TIE" },
{ name => "tig", dir => "seed", sopentypelang => "TGR" },
{ name => "tig-ER", dir => "seed" },
{ name => "tk", lcid => 0x00000042, oemcp => 852, ebcdiccp => 20880, group => 2, sopentypelang => "TKM" },
{ name => "tk-Latn", alias => "tk" },
{ name => "tk-Latn-TM", alias => "tk-TM" },
{ name => "tk-TM", lcid => 0x00000442 },
{ name => "tn", lcid => 0x00000032, oemcp => 850, dir => "seed", sopentypelang => "TNA" },
{ name => "tn-BW", lcid => 0x00000832, dir => "seed", sabbrevlangname => "TSB" },
{ name => "tn-ZA", lcid => 0x00000432, dir => "seed" },
{ name => "to", sopentypelang => "TGN" },
{ name => "to-TO" },
{ name => "tr", lcid => 0x0000001f, oemcp => 857, ebcdiccp => 20905, group => 6, sabbrevlangname => "TRK", sopentypelang => "TRK" },
{ name => "tr-CY" },
{ name => "tr-TR", lcid => 0x0000041f },
{ name => "ts", lcid => 0x00000031, dir => "seed", sopentypelang => "TSG" },
{ name => "ts-ZA", lcid => 0x00000431, dir => "seed" },
{ name => "tt", lcid => 0x00000044, oemcp => 866, group => 5, sabbrevlangname => "TTT" },
{ name => "tt-Cyrl", alias => "tt" },
{ name => "tt-Cyrl-RU", alias => "tt-RU" },
{ name => "tt-RU", lcid => 0x00000444 },
{ name => "twq" },
{ name => "twq-NE" },
{ name => "tzm", lcid => 0x0000005f, sortlocale => "tzm-Latn-DZ", oemcp => 850, ebcdiccp => 20297, sabbrevlangname => "TZA" },
{ name => "tzm-Latn", lcid => 0x00007c5f, territory => "DZ", file => "tzm" },
{ name => "tzm-Latn-MA", file => "tzm_MA", oemcp => 65001 },
{ name => "tzm-Latn-DZ", lcid => 0x0000085f, file => "tzm" },
{ name => "tzm-MA", alias => "tzm-Latn-MA" },
## name => "tzm-Arab", group => 13 },
## name => "tzm-Arab-MA", lcid => 0x0000045f },
## name => "tzm-Tfng", lcid => 0x0000785f },
## name => "tzm-Tfng-MA", lcid => 0x0000105f },
{ name => "ug", lcid => 0x00000080, oemcp => 720, slist => ",", group => 13, sopentypelang => "UYG" },
{ name => "ug-Arab", alias => "ug" },
{ name => "ug-Arab-CN", alias => "ug-CN" },
{ name => "ug-CN", lcid => 0x00000480 },
{ name => "uk", lcid => 0x00000022, oemcp => 866, maccp => 10017, ebcdiccp => 500, group => 5 },
{ name => "uk-UA", lcid => 0x00000422 },
{ name => "ur", lcid => 0x00000020, oemcp => 720 },
{ name => "ur-IN", lcid => 0x00000820, maccp => 65001, sabbrevlangname => "URI" },
{ name => "ur-PK", lcid => 0x00000420, group => 13 },
{ name => "uz", lcid => 0x00000043, oemcp => 857, maccp => 10029, group => 2 },
{ name => "uz-Arab", oemcp => 65001, maccp => 65001 },
{ name => "uz-Arab-AF" },
{ name => "uz-Cyrl", lcid => 0x00007843, oemcp => 866, maccp => 10007, group => 5, sabbrevlangname => "UZC" },
{ name => "uz-Cyrl-UZ", lcid => 0x00000843 },
{ name => "uz-Latn", lcid => 0x00007c43 },
{ name => "uz-Latn-UZ", lcid => 0x00000443 },
{ name => "vai" },
{ name => "vai-Latn" },
{ name => "vai-Latn-LR" },
{ name => "vai-Vaii" },
{ name => "vai-Vaii-LR" },
{ name => "ve", lcid => 0x00000033, dir => "seed", sabbrevlangname => "ZZZ" },
{ name => "ve-ZA", lcid => 0x00000433, dir => "seed" },
{ name => "vi", lcid => 0x0000002a, oemcp => 1258, slist => ",", group => 14, sabbrevlangname => "VIT", sopentypelang => "VIT" },
{ name => "vi-VN", lcid => 0x0000042a },
{ name => "vo", dir => "seed" },
{ name => "vo-001", dir => "seed" },
{ name => "vun" },
{ name => "vun-TZ" },
{ name => "wae" },
{ name => "wae-CH" },
{ name => "wal", dir => "seed" },
{ name => "wal-ET", dir => "seed" },
{ name => "wo", lcid => 0x00000088, oemcp => 850, ebcdiccp => 20297, sopentypelang => "WLF" },
{ name => "wo-Latn", alias => "wo" },
{ name => "wo-Latn-SN", alias => "wo-SN" },
{ name => "wo-SN", lcid => 0x00000488 },
{ name => "x-IV_mathan", lcid => 0x0001007f, alias => "" },
{ name => "xh", lcid => 0x00000034, oemcp => 850, sopentypelang => "XHS" },
{ name => "xh-ZA", lcid => 0x00000434 },
{ name => "xog" },
{ name => "xog-UG" },
{ name => "yav" },
{ name => "yav-CM" },
{ name => "yi", lcid => 0x0000003d, sabbrevlangname => "ZZZ", sopentypelang => "JII" },
{ name => "yi-001", lcid => 0x0000043d },
{ name => "yo", lcid => 0x0000006a, oemcp => 437, sopentypelang => "YBA" },
{ name => "yo-BJ", ebcdiccp => 500 },
{ name => "yo-Latn", alias => "yo" },
{ name => "yo-Latn-NG", alias => "yo-NG" },
{ name => "yo-NG", lcid => 0x0000046a },
{ name => "yrl" },
{ name => "yrl-BR" },
{ name => "yrl-CO" },
{ name => "yrl-VE" },
{ name => "yue" },
{ name => "yue-Hans" },
{ name => "yue-Hans-CN" },
{ name => "yue-Hant" },
{ name => "yue-Hant-HK" },
{ name => "zgh" },
{ name => "zgh-MA", alias => "zgh-Tfng-MA" },
{ name => "zgh-Tfng", file => "zgh" },
{ name => "zgh-Tfng-MA", file => "zgh_MA" },
{ name => "zh", lcid => 0x00007804, ireadinglayout => 2, oemcp => 936, slist => ",", sscripts => "Hani Hans", sabbrevlangname => "CHS", sopentypelang => "ZHS" },
{ name => "zh-CN", lcid => 0x00000804, file => "zh_Hans_CN", sparent => "zh-Hans" },
{ name => "zh-CN_phoneb", lcid => 0x00050804, alias => "zh-CN" },
{ name => "zh-CN_stroke", lcid => 0x00020804, alias => "zh-CN" },
{ name => "zh-Hans", lcid => 0x00000004, group => 10 },
{ name => "zh-Hans-CN", alias => "zh-CN" },
{ name => "zh-Hans-CN-u-co-phonebk", alias => "zh-CN_phoneb" },
{ name => "zh-Hans-CN-u-co-stroke", alias => "zh-CN_stroke" },
{ name => "zh-Hans-HK", slist => ";" },
{ name => "zh-Hans-MO", slist => ";" },
{ name => "zh-Hans-SG", alias => "zh-SG" },
{ name => "zh-Hans-SG-u-co-phonebk", alias => "zh-SG_phoneb" },
{ name => "zh-Hans-SG-u-co-stroke", alias => "zh-SG_stroke" },
{ name => "zh-Hant", lcid => 0x00007c04, sortlocale => "zh-HK", ireadinglayout => 2, oemcp => 950, slist => ",", sscripts => "Hani Hant", group => 9, sabbrevlangname => "CHT", sopentypelang => "ZHH" },
{ name => "zh-Hant-HK", alias => "zh-HK" },
{ name => "zh-Hant-HK-u-co-unihan", alias => "zh-HK_radstr" },
{ name => "zh-Hant-MO", alias => "zh-MO" },
{ name => "zh-Hant-MO-u-co-stroke", alias => "zh-MO_stroke" },
{ name => "zh-Hant-MO-u-co-unihan", alias => "zh-MO_radstr" },
{ name => "zh-Hant-TW", alias => "zh-TW" },
{ name => "zh-Hant-TW-u-co-phonetic", alias => "zh-TW_pronun" },
{ name => "zh-Hant-TW-u-co-unihan", alias => "zh-TW_radstr" },
{ name => "zh-HK", lcid => 0x00000c04, file => "zh_Hant_HK", sparent => "zh-Hant", sabbrevlangname => "ZHH" },
{ name => "zh-HK_radstr", lcid => 0x00040c04, alias => "zh-HK" },
{ name => "zh-MO", lcid => 0x00001404, file => "zh_Hant_MO", sparent => "zh-Hant", sabbrevlangname => "ZHM", sopentypelang => "ZHT" },
{ name => "zh-MO_radstr", lcid => 0x00041404, alias => "zh-MO" },
{ name => "zh-MO_stroke", lcid => 0x00021404, alias => "zh-MO" },
{ name => "zh-SG", lcid => 0x00001004, file => "zh_Hans_SG", sparent => "zh-Hans", sabbrevlangname => "ZHI" },
{ name => "zh-SG_phoneb", lcid => 0x00051004, alias => "zh-SG" },
{ name => "zh-SG_stroke", lcid => 0x00021004, alias => "zh-SG" },
{ name => "zh-TW", lcid => 0x00000404, file => "zh_Hant_TW", sparent => "zh-Hant", sopentypelang => "ZHT" },
{ name => "zh-TW_pronun", lcid => 0x00030404, alias => "zh-TW" },
{ name => "zh-TW_radstr", lcid => 0x00040404, alias => "zh-TW" },
{ name => "zu", lcid => 0x00000035, oemcp => 850 },
{ name => "zu-ZA", lcid => 0x00000435 },
);
my @cp2uni = ();
my @glyph2uni = ();
my @lead_bytes = ();
......@@ -511,6 +1569,17 @@ sub open_data_file($$)
}
################################################################
# load a unicode.org file as XML data
sub load_xml_data_file($$)
{
my ($base, $name) = @_;
my $FILE = open_data_file( $base, $name );
my $xml = XML::LibXML->load_xml( IO => $FILE );
close FILE;
return $xml;
}
################################################################
# recursively get the decomposition for a character
sub get_decomposition($$);
sub get_decomposition($$)
......@@ -2761,6 +3830,808 @@ sub dump_sortkey_table($$)
print OUTPUT $table, $casemaps, $chartypes, $sorttables;
close OUTPUT;
save_file($filename);
return $chartypes;
}
my %lcnames;
sub locale_parent($)
{
my $loc = shift;
return undef unless $loc;
return $lcnames{$loc}->{sparent} if defined $lcnames{$loc} && defined $lcnames{$loc}->{sparent};
return $lcnames{$loc}->{parent} if defined $lcnames{$loc} && defined $lcnames{$loc}->{parent};
if ($loc =~ /(.*)-[0-9A-Za-z]+/) { return $1; }
return "";
}
sub compare_locales
{
(my $n1 = $a) =~ tr/A-Z_/a-z-/;
(my $n2 = $b) =~ tr/A-Z_/a-z-/;
return $n1 cmp $n2;
}
# query an xml key
sub xml_query($$)
{
my ($xml, $query) = @_;
my $ret = $xml->find( $query );
return undef unless $ret;
printf STDERR "multiple entries for %s\n", $query if (@{$ret} > 1);
return @{$ret}[0]->textContent;
}
# query an xml key for a locale, with fallback to the parents
sub loc_query($$)
{
my ($loc, $query) = @_;
$loc = $lcnames{"en-US"} unless $loc->{name}; # fallback to "en-US" for root locale
for (my $cur = $loc->{name}; defined $cur; $cur = locale_parent( $cur ))
{
next unless defined $lcnames{$cur};
my $xml = $lcnames{$cur}->{xml};
my $ret = $xml->find( $query );
next unless $ret;
printf STDERR "%s: multiple entries for %s\n", $cur, $query if (@{$ret} > 1);
next if @{$ret}[0]->textContent eq "\x{2191}\x{2191}\x{2191}"; # "↑↑↑"
return @{$ret}[0]->textContent;
}
return undef;
}
# retrieve a locale field entry by going up the parents tree
sub locale_entry($$$)
{
my ($loc, $field, $def) = @_;
return $loc->{$field} if defined $loc->{$field};
unless ($loc->{name}) # fallback to "en-US" for root locale
{
$loc = $lcnames{"en-US"};
return $loc->{$field} if defined $loc->{$field};
}
while (defined $loc->{alias}) # resolve aliases
{
$loc = $lcnames{$loc->{alias}};
return $loc->{$field} if defined $loc->{$field};
}
my $cur = $loc->{name};
while ($cur)
{
if (defined $lcnames{$cur} && defined $lcnames{$cur}->{sparent})
{
$cur = $lcnames{$cur}->{sparent};
}
elsif ($cur =~ /(.*)-[0-9A-Za-z]+/)
{
$cur = $1;
}
else
{
return $def;
}
return $lcnames{$cur}->{$field} if defined $lcnames{$cur} && defined $lcnames{$cur}->{$field};
}
return $def;
}
my $string_data;
sub add_str_data($)
{
my $txt = shift;
my $ret = index( $string_data, $txt );
if ($ret == -1)
{
$ret = length($string_data);
$string_data .= $txt
}
return $ret / 2;
}
sub add_string($)
{
my $str = shift;
return 0 unless defined($str) && $str ne "";
my $utf = encode( "UTF16LE", $str );
return add_str_data( (pack "S<", length($utf) / 2) . $utf . (pack "S", 0) );
}
sub add_fontsig(@)
{
return add_str_data( pack "S<L<*", scalar(@_) * 2, @_ );
}
sub add_strarray(@)
{
return 0 unless @_;
return add_str_data( pack "S<L<*", scalar @_, map { add_string($_) } @_);
}
sub format_to_grouping($)
{
my $format = shift;
if ($format =~ /#,(#+),(#+0)/) { return chr(length($2)) . chr(length($1)); }
if ($format =~ /#,(#+0)/) { return chr(length($1)); }
# printf STDERR "unknown format %s\n", $format;
return chr(3);
}
sub parse_currency_format($$)
{
my $name = shift;
my ($posfmt, $negfmt) = split /;/, shift;
my @pospatterns = ( "\xa4[^\xa0]*#", # $1.1
"00[^\xa0]*\xa4", # 1.1$
"\xa4.*\xa0.*#", # $ 1.1
"00.*\xa0.*\xa4" ); # 1.1 $
my @negpatterns = ( "\\(\xa4[^\xa0]*#", # ($1.1)
"-\xa4[^\xa0]*#", # -$1.1
"\xa4[^\xa0]*-#", # $-1.1
"\xa4[^\xa0]*#.*00-", # $1.1-
"00[^\xa0]*\xa4\\)", # (1.1$)
"-#.*00[^\xa0]*\xa4", # -1.1$
"00-[^\xa0]*\xa4", # 1.1-$
"00[^\xa0]*\xa4-", # 1.1$-
"-#.*00.*\xa0.*\xa4", # -1.1 $
"-\xa4.*\xa0.*#", # -$ 1.1
"00.*\xa0.*\xa4-", # 1.1 $-
"\xa4.*\xa0.*#.*00-", # $ 1.1-
"\xa4.*\xa0.*-#", # $ -1.1
"00-.*\xa0.*\xa4", # 1.1- $
"\\(\xa4.*\xa0.*#", # ($ 1.1)
"00.*\xa0.*\xa4\\)"); # (1.1 $)
my ($pos, $neg);
for ($pos = 0; $pos < @pospatterns; $pos++)
{
last if ($posfmt =~ /$pospatterns[$pos]/);
}
#printf STDERR "$name: unknown format '%s'\n", $posfmt if ($pos == @pospatterns);
$pos = 0 if ($pos == @pospatterns);
if (defined $negfmt)
{
for ($neg = 0; $neg < @negpatterns; $neg++)
{
last if ($negfmt =~ /$negpatterns[$neg]/);
}
#printf STDERR "$name: unknown format '%s'\n", $negfmt if ($neg == @negpatterns);
$neg = 0 if ($neg == @negpatterns);
}
elsif ($pos == 0) { $neg = 1; }
elsif ($pos == 1) { $neg = 5; }
elsif ($pos == 2) { $neg = 9; }
elsif ($pos == 3) { $neg = 8; }
return ($pos, $neg);
}
sub parse_percent_format($)
{
my $fmt = shift;
my @patterns = ( "0.+%", # 1 %
"0%", # 1%
"%#", # %1
"%.+#" ); # % 1
my $pos;
for ($pos = 0; $pos < @patterns; $pos++)
{
last if ($fmt =~ /$patterns[$pos]/);
}
printf STDERR "unknown format '%s'\n", $fmt if ($pos == @patterns);
return ($pos, ($pos == 3) ? 7 : $pos);
}
sub convert_date_format($)
{
my $fmt = shift;
$fmt =~ s/G/g/;
$fmt =~ s/LLLL/MMMM/;
$fmt =~ s/LLL/MMM/;
$fmt =~ s/E+/dddd/;
$fmt =~ s/ccc+/dddd/;
$fmt =~ s/([^y])y([^y])/$1yyyy$2/;
$fmt =~ s/^y([^y])/yyyy$1/;
$fmt =~ s/([^y])y$/$1yyyy/;
return $fmt;
}
sub convert_time_format($)
{
my $fmt = shift;
$fmt =~ s/a+/tt/;
$fmt =~ s/B+/tt/;
return $fmt;
}
sub load_iso639()
{
my %iso639;
my $DATA = open_data_file( $ISO639, "iso-639-3_Code_Tables_$ISO639VERSION/iso-639-3.tab" );
while (<$DATA>)
{
if (/^\s*[a-z]{3}\s+[a-z]{3}\s+([a-z]{3})\s+([a-z]{2})\s/) { $iso639{$2} = $1; }
}
close $DATA;
return %iso639;
}
################################################################
# build the locale table for locale.nls
sub build_locale_data()
{
my $base = "cldr-release-$CLDRVERSION";
my $suppl = load_xml_data_file( $CLDRDATA, "$base/common/supplemental/supplementalData.xml" );
my $subtags = load_xml_data_file( $CLDRDATA, "$base/common/supplemental/likelySubtags.xml" );
my $numbers = load_xml_data_file( $CLDRDATA, "$base/common/supplemental/numberingSystems.xml" );
# obsolete phone data from CLDR version 33
my $phone = load_xml_data_file( $CLDR33DATA, "common/supplemental/telephoneCodeData.xml" );
my %iso639 = load_iso639();
$string_data = pack "S2", 0, 0; # offset 0 == empty string
%lcnames = map { $_->{name} => $_ } @locales;
my %lcids;
foreach my $loc (@locales) { $lcids{$loc->{lcid}} = $loc if defined $loc->{lcid}; }
my %days = ( "sun" => 0, "mon" => 1, "tue" => 2, "wed" => 3, "thu" => 4, "fri" => 5, "sat" => 6 );
# assign locale parents
foreach my $loc (@locales)
{
next if $loc->{name} eq "";
next if defined $loc->{parent};
(my $unix_name = $loc->{name}) =~ s/-/_/g;
my $parent = xml_query( $suppl, "/supplementalData/parentLocales/parentLocale[contains(concat(' ',\@locales,' '),' $unix_name ')]/\@parent" );
if ($parent)
{
$parent =~ s/_/-/g;
$parent = "" if $parent eq "root";
}
elsif ($loc->{name} =~ /(.*)-[0-9A-Za-z]+/) { $parent = $1; }
$loc->{parent} = $parent || "";
}
# load per-locale XML files
foreach my $loc (@locales)
{
next if defined $loc->{alias};
(my $file = $loc->{file} || $loc->{name}) =~ s/-/_/g;
$file = "$base/" . ($loc->{dir} || "common") . "/main/$file.xml";
my $xml = load_xml_data_file( $CLDRDATA, $file );
$loc->{xml} = $xml;
$loc->{language} ||= xml_query( $xml, "/ldml/identity/language/\@type" );
$loc->{territory} ||= xml_query( $xml, "/ldml/identity/territory/\@type" );
$loc->{script} = xml_query( $xml, "/ldml/identity/script/\@type" );
if (!defined($loc->{territory}) && $loc->{name} =~ /-([A-Z]{2}|[0-9]{3})$/) { $loc->{territory} = $1; }
if (!defined($loc->{script}) && $loc->{name} =~ /-([A-Z][a-z]{3})(-[A-Z]{2})?$/) { $loc->{script} = $1; }
}
# assign a default territory and sort locale
foreach my $loc (@locales)
{
next if defined $loc->{alias};
next if defined $loc->{territory};
my $id = $loc->{sortlocale};
if (defined $id && ($id =~ /[-_]([A-Z0-9]+)$/))
{
$loc->{territory} = $1;
next;
}
my @children = grep /^$loc->{name}-[A-Z0-9]+$/ && !defined $lcnames{$_}->{alias}, keys %lcnames;
if (@children == 1)
{
$id = $children[0];
}
else
{
my $name = $loc->{file} || $loc->{name};
$name =~ s/-(Arab|Cyrl|Deva|Guru|Hans|Hant|Latn|Tfng|Vaii)$//;
$name =~ s/-/_/g;
$id = xml_query( $subtags, "/supplementalData/likelySubtags/likelySubtag[\@from='$name']/\@to" );
$id =~ s/_/-/g if $id;
}
if ($id =~ /[-_]([A-Z0-9]+)$/)
{
$loc->{territory} = $1;
next if defined $loc->{sortlocale};
next unless $id =~ /^$loc->{name}/;
while (defined $lcnames{$id} && defined $lcnames{$id}->{alias}) { $id = $lcnames{$id}->{alias}; }
$loc->{sortlocale} = $id if defined $lcnames{$id};
next;
}
print STDERR "no territory found for $loc->{name}\n";
}
# assign default lcid to aliases
foreach my $loc (@locales)
{
next unless defined $loc->{alias};
next if defined $loc->{lcid};
my $alias = $loc->{alias};
my $lcid = $lcnames{$alias}->{lcid} || 0x1000;
$loc->{lcid} = $lcid | 0x80000000;
}
# assign sort aliases to parent locale
foreach my $loc (@locales)
{
next unless $loc->{name} =~ /_/;
next unless defined $loc->{alias};
my $alias = $loc->{alias};
my $parent = $lcnames{$alias};
my $basename = $parent->{name};
while (1)
{
@{$parent->{sortnames}}[($loc->{lcid} >> 16) - 1] = $loc->{name};
$alias = locale_parent( $alias );
last unless $alias && defined $lcnames{$alias};
$parent = $lcnames{$alias};
last if defined $parent->{sortbase} && $parent->{sortbase} ne $basename;
$parent->{sortbase} = $basename;
}
}
# assign an array index to all locales
my $idx = 0;
foreach my $loc (@locales)
{
next if defined $loc->{alias};
$loc->{idx} = $idx++;
}
foreach my $loc (@locales)
{
my $alias = $loc->{alias};
next unless defined $alias;
while (defined $lcnames{$alias}->{alias}) { $alias = $lcnames{$alias}->{alias}; }
$loc->{idx} = $lcnames{$alias}->{idx};
}
# output lcids table
my $lcid_data = "";
foreach my $id (sort { $a <=> $b } keys %lcids)
{
my $loc = $lcids{$id};
$lcid_data .= pack "L<S<2", $id, $loc->{idx}, add_string($loc->{name});
}
# output lcnames table
my $lcname_data = "";
foreach my $name (sort compare_locales keys %lcnames)
{
my $loc = $lcnames{$name};
$lcname_data .= pack "S<2L<", add_string($name), $loc->{idx}, $loc->{lcid} || 0x1000;
}
# output locales array
my $locale_data = "";
my $default_lcid = 0x8001;
foreach my $loc (@locales)
{
next if defined $loc->{alias};
my $sname = $loc->{name};
my $language = $loc->{language};
my $territory = $loc->{territory};
my $script = $loc->{script};
my $neutral = ($sname && $sname !~ /-$territory/);
my $sparent = $loc->{sparent} || (($sname =~ /(.*)-[0-9A-Za-z]+/) ? $1 : $loc->{parent});
my $unique_lcid = $loc->{lcid};
unless (defined $unique_lcid) { $unique_lcid = $default_lcid++; }
my $territory_match = "contains(concat(' ',normalize-space(\@territories),' '),' $territory ')";
# languages and scripts
my $ssortlocale = $loc->{sortlocale} || ($neutral ? "$sname-$territory" : $sname);
my $idefaultlanguage = defined $lcnames{$ssortlocale} ? $lcnames{$ssortlocale}->{lcid} : undef;
$idefaultlanguage = $lcnames{"en-US"}->{lcid} unless $ssortlocale;
(my $siso639langname = $sname) =~ s/-.*$//;
my $siso639langname2 = $iso639{$siso639langname} || $siso639langname;
my $sopentypelang = sprintf "%-4s", locale_entry( $loc, "sopentypelang", uc $siso639langname2 );
my $sabbrevlangname = defined $loc->{lcid} ? locale_entry( $loc, "sabbrevlangname", uc $siso639langname2 ) : "ZZZ";
my $siso3166ctryname2 = $territory =~ /^\d+$/ ? $territory : xml_query( $suppl, "/supplementalData/codeMappings/territoryCodes[\@type='$territory']/\@alpha3");
my $senglanguage = loc_query( $lcnames{en}, "/ldml/localeDisplayNames/languages/language[\@type='$language' and not(\@alt)]" ) || "";
my $sengcountry = loc_query( $lcnames{en}, "/ldml/localeDisplayNames/territories/territory[\@type='$territory' and not(\@alt)]" ) || "";
my $snativelangname = loc_query( $loc, "/ldml/localeDisplayNames/languages/language[\@type='$language' and not(\@alt)]" );
my $snativectryname = loc_query( $loc, "/ldml/localeDisplayNames/territories/territory[\@type='$territory' and not(\@alt)]" );
$sengcountry =~ s/South Korea/Korea/;
$snativelangname ||= $senglanguage;
$snativectryname ||= $sengcountry;
if ($script)
{
my $engscript = loc_query( $lcnames{en}, "/ldml/localeDisplayNames/scripts/script[\@type='$script' and not(\@alt)]" );
my $nativescript = loc_query( $loc, "/ldml/localeDisplayNames/scripts/script[\@type='$script' and not(\@alt)]" );
$senglanguage .= " ($engscript)" if $engscript;
$snativelangname .= " ($nativescript)" if $nativescript;
}
my $sengdisplayname = $neutral ? $senglanguage : "$senglanguage ($sengcountry)";
my $snativedisplayname = $neutral ? $snativelangname : "$snativelangname ($snativectryname)";
$sengdisplayname =~ s/\) \(/, /;
$snativedisplayname =~ s/\) \(/, /;
my $sscripts = locale_entry( $loc, "sscripts", $script ) || xml_query( $suppl, "/supplementalData/languageData/language[\@type='$language' and not(\@alt)]/\@scripts" );
$sscripts = (join ";", (sort split / /, ($sscripts || "Latn"))) . ";";
my $ireadinglayout = locale_entry( $loc, "ireadinglayout", 0 );
my $charlayout = loc_query( $loc, "/ldml/layout/orientation/characterOrder" );
if ($charlayout eq "right-to-left")
{
$ireadinglayout = 1;
}
elsif ($charlayout eq "top-to-bottom")
{
my $linelayout = loc_query( $loc, "/ldml/layout/orientation/lineOrder" );
$ireadinglayout = $linelayout eq "right-to-left" ? 2 : 3;
}
my $icountry = xml_query( $phone, "(/supplementalData/telephoneCodeData/codesByTerritory[\@territory='$territory']/telephoneCountryCode)[1]/\@code" ) || 1;
# numbers
my $sdecimal = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/decimal" );
my $slist = locale_entry( $loc, "slist", ";" );
my $smondecimalsep = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/currencyDecimal" ) || $sdecimal;
my $sthousand = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/group" );
$sthousand =~ s/\x{202f}/\x{00a0}/;
my $smonthousandsep = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/currencyGroup" ) || $sthousand;
my $spositivesign = "";
my $snegativesign = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/minusSign" );
my $spercent = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/percentSign" );
my $snan = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/nan" );
my $sposinfinity = loc_query( $loc, "/ldml/numbers/symbols[\@numberSystem='latn']/infinity" );
my $sneginfinity = $sposinfinity ? "-$sposinfinity" : "";
my $sgrouping = format_to_grouping( loc_query( $loc, "/ldml/numbers/decimalFormats[\@numberSystem='latn']/decimalFormatLength[not(\@type)]/decimalFormat/pattern" ));
my $percentformat = loc_query( $loc, "/ldml/numbers/percentFormats[\@numberSystem='latn']/percentFormatLength[not(\@type)]/percentFormat/pattern" );
my $currencyformat = loc_query( $loc, "/ldml/numbers/currencyFormats[\@numberSystem='latn']/currencyFormatLength[not(\@type)]/currencyFormat[\@type='accounting']/pattern" ) ||
loc_query( $loc, "/ldml/numbers/currencyFormats[\@numberSystem='latn']/currencyFormatLength[not(\@type)]/currencyFormat[\@type='standard']/pattern" );
my $smongrouping = format_to_grouping( $currencyformat );
my ($icurrency, $inegcurr) = parse_currency_format( $sname, $currencyformat );
my ($ipospercent, $inegpercent) = parse_percent_format( $percentformat );
my $native_numbering = loc_query( $loc, "/ldml/numbers/otherNumberingSystems/native" );
my @snativedigits = split //, xml_query( $numbers, "/supplementalData/numberingSystems/numberingSystem[\@id='$native_numbering']/\@digits" );
my $digitsubstitution = !(ord($snativedigits[0]) >= 0x600 && ord($snativedigits[0]) <= 0x6ff);
my $measure = defined xml_query( $suppl, "/supplementalData/measurementData/measurementSystem[\@type='US' and $territory_match]" );
my $papersize = defined xml_query( $suppl, "/supplementalData/measurementData/paperSize[\@type='US-Letter' and $territory_match]" );
# currencies
my $sintlsymbol = $loc->{sintlsymbol} || xml_query( $suppl, "(/supplementalData/currencyData/region[\@iso3166='$territory']/currency[not(\@to)])[1]/\@iso4217") || "XDR";
my $scurrency = $loc->{scurrency} || loc_query( $loc, "/ldml/numbers/currencies/currency[\@type='$sintlsymbol']/symbol[\@alt='narrow']" );
$scurrency ||= loc_query( $loc, "/ldml/numbers/currencies/currency[\@type='$sintlsymbol']/symbol[not(\@alt)]" );
$scurrency ||= $sintlsymbol;
my $sengcurrname = $loc->{sengcurrname} || loc_query( $lcnames{en}, "/ldml/numbers/currencies/currency[\@type='$sintlsymbol']/displayName[not(\@count)]" );
my $snativecurrname = $loc->{sengcurrname} || loc_query( $loc, "/ldml/numbers/currencies/currency[\@type='$sintlsymbol']/displayName[not(\@count)]" ) || $sengcurrname;
my $icurrdigits = xml_query( $suppl, "/supplementalData/currencyData/fractions/info[\@iso4217='$sintlsymbol']/\@digits" );
$icurrdigits = 2 unless defined $icurrdigits;
# calendars
my $firstday = xml_query( $suppl, "/supplementalData/weekData/firstDay[not(\@alt) and $territory_match]/\@day" );
my $ifirstdayofweek = $firstday ? $days{$firstday} : 1;
my $firstweekofyear = (xml_query( $suppl, "/supplementalData/weekData/minDays[$territory_match]/\@count" ) || 0) == 4 ? 2 : 0;
my $serastring = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/eras/eraAbbr/era[\@type='1' and not(\@alt)]" );
my (@sdayname, @sabbrevdayname, @sshortestdayname);
foreach my $d (sort { $days{$a} <=> $days{$b} } keys %days)
{
my $n = $days{$d};
my %name;
foreach my $type (qw(wide abbreviated short))
{
$name{$type} = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/days/dayContext[\@type='format']/dayWidth[\@type='$type']/day[\@type='$d' and not(\@alt)]" );
}
push @sdayname, $name{wide};
push @sabbrevdayname, $name{abbreviated} || $name{wide};
push @sshortestdayname, $name{short} || $name{abbreviated} || $name{wide};
}
my (@smonthname, @sabbrevmonthname, @sgenitivemonth, @sabbrevgenitivemonth);
foreach my $n (1..13)
{
my $name = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/months/monthContext[\@type='stand-alone']/monthWidth[\@type='wide']/month[\@type='$n']" );
my $abbrev = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/months/monthContext[\@type='stand-alone']/monthWidth[\@type='abbreviated']/month[\@type='$n']" );
my $genitive = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/months/monthContext[\@type='format']/monthWidth[\@type='wide']/month[\@type='$n']" );
my $abbrevgen = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/months/monthContext[\@type='format']/monthWidth[\@type='abbreviated']/month[\@type='$n']" );
push @smonthname, $name || $genitive || "";
push @sabbrevmonthname, $abbrev || $abbrevgen || $name || $genitive || "";
push @sgenitivemonth, $genitive || "";
push @sabbrevgenitivemonth, $abbrevgen || $genitive || "";
}
@sgenitivemonth = () if join("|",@smonthname) eq join("|",@sgenitivemonth);
@sabbrevgenitivemonth = () if join("|",@sabbrevmonthname) eq join("|",@sabbrevgenitivemonth);
my %caltypes = ( "gregorian" => 1, "japanese" => 3, "chinese" => 4, "dangi" => 5, "islamic" => 6, "buddhist" => 7, "hebrew" => 8,
"persian" => 22, "islamic-civil" => 23, "islamic-umalqura" => 23 );
my $calpref = xml_query( $suppl, "/supplementalData/calendarPreferenceData/calendarPreference[$territory_match]/\@ordering" ) || "gregorian";
my $icalendartype;
my @scalnames;
foreach my $c (split /\s+/, $calpref)
{
next unless defined $caltypes{$c};
$icalendartype .= chr($caltypes{$c});
$scalnames[$caltypes{$c} - 1] = loc_query( $loc, "/ldml/localeDisplayNames/types/type[\@key='calendar' and \@type='$c']" );
}
# date/time formats
my $s1159 = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dayPeriods/dayPeriodContext[\@type='format']/dayPeriodWidth[\@type='abbreviated']/dayPeriod[\@type='am' and not(\@alt)]" );
my $s2359 = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dayPeriods/dayPeriodContext[\@type='format']/dayPeriodWidth[\@type='abbreviated']/dayPeriod[\@type='pm' and not (\@alt)]" );
my $sshortestam = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dayPeriods/dayPeriodContext[\@type='format']/dayPeriodWidth[\@type='narrow']/dayPeriod[\@type='am' and not(\@alt)]" );
my $sshortestpm = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dayPeriods/dayPeriodContext[\@type='format']/dayPeriodWidth[\@type='narrow']/dayPeriod[\@type='pm' and not (\@alt)]" );
my @stimeformat = (loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/timeFormats/timeFormatLength[\@type='medium']/timeFormat/pattern[not(\@alt)]" ));
push @stimeformat, loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='Hms' and not(\@alt)]" );
pop @stimeformat if $stimeformat[0] eq $stimeformat[1];
@stimeformat = map convert_time_format($_), @stimeformat;
my @sshorttime = (loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/timeFormats/timeFormatLength[\@type='short']/timeFormat/pattern[not(\@alt)]" ));
push @sshorttime, loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='Hm' and not(\@alt)]" );
pop @sshorttime if $sshorttime[0] eq $sshorttime[1];
@sshorttime = map convert_time_format($_), @sshorttime;
my @sshortdate = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='yMd' and not(\@alt)]" );
push @sshortdate, loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='yMMMd' and not(\@alt)]" );
@sshortdate = map convert_date_format($_), @sshortdate;
my @slongdate = (loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateFormats/dateFormatLength[\@type='full']/dateFormat/pattern[not(\@alt)]" ));
push @slongdate, loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateFormats/dateFormatLength[\@type='long']/dateFormat/pattern[not(\@alt)]" );
@slongdate = map convert_date_format($_), @slongdate;
my @smonthday = (loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='MMMMd' and not(\@alt)]" ));
push @smonthday, loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='Md' and not(\@alt)]" );
push @smonthday, loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='MMMd' and not(\@alt)]" );
@smonthday = map convert_date_format($_), @smonthday;
my @syearmonth = map convert_date_format($_), loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='yMMMM' and not(\@alt)]" );
my @sduration = map convert_time_format( lc $_ ), loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='Hms' and not(\@alt)]" );
my $srelativelongdate = loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='MMMMEd' and not(\@alt)]" ) ||
loc_query( $loc, "/ldml/dates/calendars/calendar[\@type='gregorian']/dateTimeFormats/availableFormats/dateFormatItem[\@id='MMMEd' and not(\@alt)]" );
$srelativelongdate = convert_date_format( $srelativelongdate );
# codepages
my %ansicpmap = ( 437 => 1252, 720 => 1256, 737 => 1253, 775 => 1257, 850 => 1252,
852 => 1250, 855 => 1251, 866 => 1251, 857 => 1254, 862 => 1255 );
my %maccpmap = ( 437 => 10000, 720 => 10004, 737 => 10006, 775 => 10029, 850 => 10000,
852 => 10029, 855 => 10007, 857 => 10081, 862 => 10005, 866 => 10007,
874 => 10021, 932 => 10001, 936 => 10008, 949 => 10003, 950 => 10002,
1258 => 10000 );
my %ebcdiccpmap = ( 437 => 37, 720 => 20420, 737 => 20273, 866 => 20880, 932 => 20290 );
my %codepagemasks = ( 874 => [ 0x01000000, 0x00000000, 0x00000000, 0, 0x00010000, 0x00000000, 0x00010000, 0x00000000 ],
932 => [ 0x00000000, 0x28c70000, 0x00000010, 0, 0x00020000, 0x00000000, 0x00020000, 0x00000000 ],
936 => [ 0x00000000, 0x28010000, 0x00000002, 0, 0x00040000, 0x00000000, 0x00040000, 0x00000000 ],
949 => [ 0x00000000, 0x00000000, 0x00000000, 0, 0x00080000, 0x00000000, 0x00080000, 0x00000000 ],
950 => [ 0x00000000, 0x28c10000, 0x00000012, 0, 0x00100000, 0x00000000, 0x00100000, 0x00000000 ],
1258 => [ 0x2000000f, 0x00000000, 0x00000000, 0, 0x00000100, 0x00008000, 0x00000100, 0x00008000 ],
866 => [ 0x00000200, 0x00000000, 0x00000000, 0, 0x00000004, 0x00020000, 0x00000004, 0x02020000 ],
862 => [ 0x00000800, 0x40000000, 0x00000000, 0, 0x00000020, 0x00200000, 0x00000020, 0x00200000 ],
857 => [ 0x0000001f, 0x00000000, 0x00000000, 0, 0x00000010, 0x01000000, 0x00000010, 0x01000000 ],
855 => [ 0x00000200, 0x00000000, 0x00000000, 0, 0x00000004, 0x02000000, 0x00000004, 0x02000000 ],
852 => [ 0x00000027, 0x00000000, 0x00000000, 0, 0x00000002, 0x04000000, 0x00000002, 0x04000000 ],
775 => [ 0x00000007, 0x00000000, 0x00000000, 0, 0x00000080, 0x08000000, 0x00000080, 0x08000000 ],
737 => [ 0x00000080, 0x00000000, 0x00000000, 0, 0x00000008, 0x10000000, 0x00000008, 0x10010000 ],
720 => [ 0x00002000, 0x00000000, 0x00000000, 0, 0x00000040, 0x20000000, 0x00000040, 0x20080000 ],
850 => [ 0x00000003, 0x00000000, 0x00000000, 0, 0x00000001, 0x40000000, 0x0000019f, 0xdfd70000 ],
437 => [ 0x00000003, 0x00000000, 0x00000000, 0, 0x00000001, 0x80000000, 0x0000019f, 0xdfd70000 ],
65001 => [ 0x00000000, 0x00000000, 0x00000000, 0, 0x00000000, 0x00000000, 0x0000019f, 0xdfd70000 ] );
my $oemcp = locale_entry( $loc, "oemcp", 65001 );
my $maccp = locale_entry( $loc, "maccp", undef ) || $maccpmap{$oemcp} || 65001;
my $ebcdiccp = locale_entry( $loc, "ebcdiccp", undef ) || $ebcdiccpmap{$oemcp} || 500;
$ebcdiccp = 500 if (defined $loc->{oemcp} && $loc->{oemcp} == 65001) || (defined $loc->{maccp} && $loc->{maccp} == 65001);
my $ansicp = $ansicpmap{$oemcp} || $oemcp;
my @fontsig = (0) x 8;
my $sig = locale_entry( $loc, "fontsig", [] );
foreach my $i (0..7) { $fontsig[$i] |= $codepagemasks{$oemcp}->[$i]; }
foreach my $i (0..$#{$sig}) { $fontsig[$i] |= $sig->[$i]; }
$fontsig[3] |= 1 << 31;
$fontsig[3] |= 1 << 27 if $ireadinglayout == 1;
$fontsig[3] |= 1 << 28 if $ireadinglayout == 3;
# special cases for invariant locale
unless ($loc->{name})
{
$siso639langname = "iv";
$siso639langname2 = "ivl";
$senglanguage = $snativelangname = "Invariant Language";
$sengcountry = $snativectryname = "Invariant Country";
$sengdisplayname = "Invariant Language (Invariant Country)";
$snativedisplayname = "Invariant Language (Invariant Region)";
$sengcurrname = $snativecurrname = "International Monetary Fund";
$scurrency = "\x{00a4}";
$ifirstdayofweek = 0;
@stimeformat = ("HH:mm:ss");
@sshortdate = ("MM/dd/yyyy", "yyyy-MM-dd");
@slongdate = ("dddd, dd MMMM yyyy");
@syearmonth = ("yyyy MMMM");
@smonthday = ("MMMM dd", "MMMM d", "M/d", "MMM d");
@sshorttime = ("HH:mm", "hh:mm tt", "H:mm", "h:mm tt");
$srelativelongdate = "dddd, MMMM dd";
$sposinfinity = "Infinity";
$sneginfinity = "-Infinity";
$spositivesign = "+";
$ipospercent = $inegpercent = 0;
}
# output data
$locale_data .= pack "L<2",
add_string( $sname ), # name
add_string( $sopentypelang ); # LOCALE_SOPENTYPELANGUAGETAG
$locale_data .= pack "S<14",
$loc->{lcid} || 0x1000, # LOCALE_ILANGUAGE
$unique_lcid, # unique_lcid
locale_entry( $loc, "idigits", 2 ), # LOCALE_IDIGITS
locale_entry( $loc, "inegnumber", 1 ), # LOCALE_INEGNUMBER
$icurrdigits, # LOCALE_ICURRDIGITS
$icurrency, # LOCALE_ICURRENCY
$inegcurr, # LOCALE_INEGCURR
locale_entry( $loc, "ilzero", 1 ), # LOCALE_ILZERO
!$neutral, # LOCALE_INEUTRAL
$ifirstdayofweek, # LOCALE_IFIRSTDAYOFWEEK
$firstweekofyear, # LOCALE_IFIRSTWEEKOFYEAR
$icountry, # LOCALE_ICOUNTRY,
$measure, # LOCALE_IMEASURE
$digitsubstitution; # LOCALE_IDIGITSUBSTITUTION
$locale_data .= pack "L<18",
add_string( $sgrouping ), # LOCALE_SGROUPING
add_string( $smongrouping ), # LOCALE_SMONGROUPING
add_string( $slist ), # LOCALE_SLIST
add_string( $sdecimal ), # LOCALE_SDECIMAL
add_string( $sthousand ), # LOCALE_STHOUSAND
add_string( $scurrency ), # LOCALE_SCURRENCY
add_string( $smondecimalsep ), # LOCALE_SMONDECIMALSEP
add_string( $smonthousandsep ), # LOCALE_SMONTHOUSANDSEP
add_string( $spositivesign ), # LOCALE_SPOSITIVESIGN
add_string( $snegativesign ), # LOCALE_SNEGATIVESIGN
add_string( $s1159 ), # LOCALE_S1159
add_string( $s2359 ), # LOCALE_S2359
add_strarray( @snativedigits ), # LOCALE_SNATIVEDIGITS
add_strarray( @stimeformat ), # LOCALE_STIMEFORMAT
add_strarray( @sshortdate ), # LOCALE_SSHORTDATE
add_strarray( @slongdate ), # LOCALE_SLONGDATE
add_strarray( @syearmonth ), # LOCALE_SYEARMONTH
add_strarray( @sduration ); # LOCALE_SDURATION
$locale_data .= pack "S<8",
$idefaultlanguage || 0x1000, # LOCALE_IDEFAULTLANGUAGE
$ansicp, # LOCALE_IDEFAULTANSICODEPAGE
$oemcp, # LOCALE_IDEFAULTCODEPAGE
$maccp, # LOCALE_IDEFAULTMACCODEPAGE
$ebcdiccp, # LOCALE_IDEFAULTEBCDICCODEPAGE
0, # FIXME # LOCALE_IGEOID
$papersize ? 1 : 9, # LOCALE_IPAPERSIZE
0; # FIXME # islamic_cal
$locale_data .= pack "L<24",
add_string( $icalendartype ), # LOCALE_ICALENDARTYPE
add_string( $sabbrevlangname ), # LOCALE_SABBREVLANGNAME
add_string( $siso639langname ), # LOCALE_SISO639LANGNAME
add_string( $senglanguage ), # LOCALE_SENGLANGUAGE
add_string( $snativelangname ), # LOCALE_SNATIVELANGNAME
add_string( $sengcountry ), # LOCALE_SENGCOUNTRY
add_string( $snativectryname ), # LOCALE_SNATIVECTRYNAME
add_string( $siso3166ctryname2 ), # LOCALE_SABBREVCTRYNAME
add_string( $territory ), # LOCALE_SISO3166CTRYNAME
add_string( $sintlsymbol ), # LOCALE_SINTLSYMBOL
add_string( $sengcurrname ), # LOCALE_SENGCURRNAME
add_string( $snativecurrname ), # LOCALE_SNATIVECURRNAME
add_fontsig( @fontsig ), # LOCALE_FONTSIGNATURE
add_string( $siso639langname2 ), # LOCALE_SISO639LANGNAME2
add_string( $siso3166ctryname2 ), # LOCALE_SISO3166CTRYNAME2
add_string( $sparent ), # LOCALE_SPARENT
add_strarray( @sdayname ), # LOCALE_SDAYNAME
add_strarray( @sabbrevdayname ), # LOCALE_SABBREVDAYNAME
add_strarray( @smonthname ), # LOCALE_SMONTHNAME
add_strarray( @sabbrevmonthname ), # LOCALE_SABBREVMONTHNAME
add_strarray( @sgenitivemonth ), # LOCALE_SGENITIVEMONTH
add_strarray( @sabbrevgenitivemonth ), # LOCALE_SABBREVGENITIVEMONTH
add_strarray( @scalnames ), # LOCALE_SCALNAMES
add_strarray( @{$loc->{sortnames}} ); # LOCALE_SSORTNAMES
$locale_data .= pack "S<6",
$inegpercent, # LOCALE_INEGATIVEPERCENT
$ipospercent, # LOCALE_IPOSITIVEPERCENT
0, # unknown
$ireadinglayout, # LOCALE_IREADINGLAYOUT
0x2a, # unknown
0x2a; # unknown
$locale_data .= pack "L<24",
0, # unknown
add_string( $sengdisplayname ), # LOCALE_SENGLISHDISPLAYNAME
add_string( $snativedisplayname ), # LOCALE_SNATIVEDISPLAYNAME
add_string( $spercent ), # LOCALE_SPERCENT
add_string( $snan ), # LOCALE_SNAN
add_string( $sposinfinity ), # LOCALE_SPOSINFINITY
add_string( $sneginfinity ), # LOCALE_SNEGINFINITY
0, # unknown
add_string( $serastring ), # CAL_SERASTRING
add_string( $serastring ), # CAL_SABBREVERASTRING
0, # unknown
add_string( $ssortlocale ), # LOCALE_SCONSOLEFALLBACKNAME
add_strarray( @sshorttime ), # LOCALE_SSHORTTIME
add_strarray( @sshortestdayname ), # CAL_SSHORTESTDAYNAME
0, # unknown
add_string( $ssortlocale ), # LOCALE_SSORTLOCALE
add_string( "0409:00000409" ), # FIXME # LOCALE_SKEYBOARDSTOINSTALL
add_string( $sscripts ), # LOCALE_SSCRIPTS
add_string( $srelativelongdate ), # LOCALE_SRELATIVELONGDATE
0, # FIXME # LOCALE_IGEOID2
add_string( $sshortestam || "a" ), # LOCALE_SSHORTESTAM
add_string( $sshortestpm || "p" ), # LOCALE_SSHORTESTPM
add_strarray( @smonthday ), # LOCALE_SMONTHDAY
add_string( "k0-windows-us" ) # FIXME # keyboard_layout
}
my $nb_lcids = scalar keys %lcids;
my $nb_locales = scalar grep { !defined $_->{alias} } @locales;
my $nb_lcnames = scalar keys %lcnames;
my $locale_size = length($locale_data) / $nb_locales;
my $nb_calendars = 0;
my $lcids_offset = 19 * 4; # size of header
my $lcnames_offset = $lcids_offset + length $lcid_data;
my $locales_offset = $lcnames_offset + length $lcname_data;
my $calendar_offset = $locales_offset + length $locale_data;
my $strings_offset = $calendar_offset;
my $locale_header = pack "L<7S<4L<S<2L<3S<2L<4",
8, # offset
0,
7, # version
0x5344534e, # magic
0, 0, 0,
0,
$nb_lcids,
$nb_locales,
$locale_size,
$locales_offset,
$nb_lcnames,
0,
$lcids_offset,
$lcnames_offset,
0,
$nb_calendars,
0,
$calendar_offset,
$strings_offset,
0, 0;
return align_string( 4, $locale_header . $lcid_data . $lcname_data . $locale_data . $string_data );
}
################################################################
# build a binary locale table
sub dump_locales($$)
{
my ($filename, $chartypes) = @_;
printf "Building $filename\n";
my $locale_data = build_locale_data();
my $charmaps_data = ""; # FIXME
my $geoids_data = ""; # FIXME
my $scripts_data = ""; # FIXME
my @header = ( 0 ) x 8;
$header[0] = 4 * scalar @header; # chartypes offset
$header[4] = $header[0] + length $chartypes; # locales offset
$header[5] = $header[4] + length $locale_data; # charmaps offset
$header[6] = $header[5] + length $charmaps_data; # geoids offset
$header[7] = $header[6] + length $geoids_data; # scripts offset
open OUTPUT, ">$filename.new" or die "Cannot create $filename";
print OUTPUT pack "L<*", @header;
print OUTPUT $chartypes, $locale_data, $charmaps_data, $geoids_data, $scripts_data;
close OUTPUT;
save_file($filename);
}
......@@ -2841,7 +4712,8 @@ dump_norm_table( "nls/normnfd.nls" );
dump_norm_table( "nls/normnfkc.nls" );
dump_norm_table( "nls/normnfkd.nls" );
dump_norm_table( "nls/normidna.nls" );
dump_sortkey_table( "nls/sortdefault.nls", "Windows 10 Sorting Weight Table.txt" );
my $chartypes = dump_sortkey_table( "nls/sortdefault.nls", "Windows 10 Sorting Weight Table.txt" );
dump_locales( "nls/locale.nls", $chartypes );
foreach my $file (@allfiles) { dump_msdata_codepage( $file ); }
dump_eucjp_codepage();
dump_registry_script( "dlls/kernelbase/kernelbase.rgs", %registry_keys );
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment