diff options
author | Thomas Lange <tl@openoffice.org> | 2007-01-18 13:07:31 +0000 |
---|---|---|
committer | Thomas Lange <tl@openoffice.org> | 2007-01-18 13:07:31 +0000 |
commit | a80761fc9ef1cd9bb6e0b7a8853b8a2adadb69b2 (patch) | |
tree | 84bae0f8de0b35f8354eee91adcb6968bef2cfbf /libtextcat | |
parent | 60133b93d2ef9cc2fd69fe5b717bd8ebb3475c28 (diff) |
#i73173# integrate Google SoC language-guessing
Diffstat (limited to 'libtextcat')
-rw-r--r-- | libtextcat/data/new_fingerprints/fpdb.conf | 130 |
1 files changed, 65 insertions, 65 deletions
diff --git a/libtextcat/data/new_fingerprints/fpdb.conf b/libtextcat/data/new_fingerprints/fpdb.conf index b72e103ddffb..7520dfd315ac 100644 --- a/libtextcat/data/new_fingerprints/fpdb.conf +++ b/libtextcat/data/new_fingerprints/fpdb.conf @@ -15,68 +15,68 @@ # this file have been modified (to OOo by Jocelyn MERAND joc.mer@gmail.com) to include country and encoding # guess strings are made as following : language-country-encoding -afrikaans.lm af---utf8 -albanian.lm sq---utf8 -amharic_utf.lm am---utf8 -arabic.lm ar---utf8 -basque.lm eu---utf8 -belarus.lm be---utf8 -bosnian.lm bs---utf8 -breton.lm br---utf8 -catalan.lm ca---utf8 -chinese_simplified.lm zh-CN--utf8 -chinese_traditional.lm zh-TW--utf8 -croatian.lm hr---utf8 -czech.lm cs---utf8 -danish.lm da---utf8 -dutch.lm nl---utf8 -english.lm en---utf8 -esperanto.lm eo---utf8 -estonian.lm et---utf8 -finnish.lm fi---utf8 -french.lm fr---utf8 -frisian.lm fy---utf8 -georgian.lm ka---utf8 -german.lm de---utf8 -greek.lm el---utf8 -hebrew.lm he---utf8 -hindi.lm hi---utf8 -hungarian.lm hu---utf8 -icelandic.lm is---utf8 -indonesian.lm id---utf8 -irish_gaelic.lm ga---utf8 -italian.lm it---utf8 -japanese.lm ja---utf8 -korean.lm ko---utf8 -latin.lm la---utf8 -latvian.lm lv---utf8 -lithuanian.lm lt---utf8 -malay.lm ms---utf8 -manx_gaelic.lm gv---utf8 -marathi.lm mr---utf8 -nepali.lm ne---utf8 -norwegian.lm nb---utf8 # Norwegian (Bokmal) -persian.lm fa---utf8 # Farsi -polish.lm pl---utf8 -portuguese.lm pt-PT--utf8 -quechua.lm qu---utf8 -romanian.lm ro---utf8 -romansh.lm rm---utf8 -russian.lm ru---utf8 -sanskrit.lm sa---utf8 -scots.lm sco---utf8 -scots_gaelic.lm gd---utf8 -serbian_ascii.lm sh-YU--utf8 -slovak_ascii.lm sk-SK--utf8 -slovenian.lm sl---utf8 -spanish.lm es---utf8 -swahili.lm sw---utf8 -swedish.lm sv---utf8 -tagalog.lm tl---utf8 -tamil.lm ta---utf8 -thai.lm th---utf8 -turkish.lm tr---utf8 -ukrainian.lm uk---utf8 -vietnamese.lm vi---utf8 -welsh.lm cy---utf8 -yiddish_utf.lm yi---utf8 +afrikaans.lm af--utf8 +albanian.lm sq--utf8 +amharic_utf.lm am--utf8 +arabic.lm ar--utf8 +basque.lm eu--utf8 +belarus.lm be--utf8 +bosnian.lm bs--utf8 +breton.lm br--utf8 +catalan.lm ca--utf8 +chinese_simplified.lm zh-CN-utf8 +chinese_traditional.lm zh-TW-utf8 +croatian.lm hr--utf8 +czech.lm cs--utf8 +danish.lm da--utf8 +dutch.lm nl--utf8 +english.lm en--utf8 +esperanto.lm eo--utf8 +estonian.lm et--utf8 +finnish.lm fi--utf8 +french.lm fr--utf8 +frisian.lm fy--utf8 +georgian.lm ka--utf8 +german.lm de--utf8 +greek.lm el--utf8 +hebrew.lm he--utf8 +hindi.lm hi--utf8 +hungarian.lm hu--utf8 +icelandic.lm is--utf8 +indonesian.lm id--utf8 +irish_gaelic.lm ga--utf8 +italian.lm it--utf8 +japanese.lm ja--utf8 +korean.lm ko--utf8 +latin.lm la--utf8 +latvian.lm lv--utf8 +lithuanian.lm lt--utf8 +malay.lm ms--utf8 +manx_gaelic.lm gv--utf8 +marathi.lm mr--utf8 +nepali.lm ne--utf8 +norwegian.lm nb--utf8 # Norwegian (Bokmal) +persian.lm fa--utf8 # Farsi +polish.lm pl--utf8 +portuguese.lm pt-PT-utf8 +quechua.lm qu--utf8 +romanian.lm ro--utf8 +romansh.lm rm--utf8 +russian.lm ru--utf8 +sanskrit.lm sa--utf8 +scots.lm sco--utf8 +scots_gaelic.lm gd--utf8 +serbian_ascii.lm sh-YU-utf8 +slovak_ascii.lm sk-SK-utf8 +slovenian.lm sl--utf8 +spanish.lm es--utf8 +swahili.lm sw--utf8 +swedish.lm sv--utf8 +tagalog.lm tl--utf8 +tamil.lm ta--utf8 +thai.lm th--utf8 +turkish.lm tr--utf8 +ukrainian.lm uk--utf8 +vietnamese.lm vi--utf8 +welsh.lm cy--utf8 +yiddish_utf.lm yi--utf8 |