From 9724dcdf4b113824dd42f77b4448248154af3ef9 Mon Sep 17 00:00:00 2001 From: Frans de Jonge Date: Tue, 11 Jun 2024 22:17:49 +0200 Subject: [PATCH] IsoLanguage: add Tesseract language names --- frontend/ui/data/isolanguage.lua | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/frontend/ui/data/isolanguage.lua b/frontend/ui/data/isolanguage.lua index 6482c892a..1b13c7353 100644 --- a/frontend/ui/data/isolanguage.lua +++ b/frontend/ui/data/isolanguage.lua @@ -6,6 +6,7 @@ local locale_lang_map = { -- @translators Most of these language name have already been translated at . Click "Automatic suggestions" to see them below the textfield. aar = _("Afar"), afr = _("Afrikaans"), + amh = _("Amharic"), ara = _("Arabic"), -- macrolanguage asm = _("Assamese"), aze = _("Azerbaijani"), -- macrolanguage @@ -13,15 +14,19 @@ local locale_lang_map = { bel = _("Belarusian"), ben = _("Bengali"), bod = _("Tibetan"), + bos = _("Bosnian"), bre = _("Breton"), bul = _("Bulgarian"), + cat = _("Catalan"), ces = _("Czech"), chu = _("Church Slavic"), cor = _("Cornish"), + cos = _("Corsican"), cym = _("Welsh"), dan = _("Danish"), deu = _("German"), div = _("Dhivehi"), + dzo = _("Dzongkha"), ell = _("Modern Greek"), -- (1453-) eng = _("English"), epo = _("Esperanto"), -- constructed language @@ -29,6 +34,7 @@ local locale_lang_map = { eus = _("Basque"), fao = _("Faroese"), fas = _("Persian"), -- macrolanguage + fil = _("Filipino"), fin = _("Finnish"), fra = _("French"), fry = _("Western Frisian"), @@ -42,6 +48,7 @@ local locale_lang_map = { hbs = _("Serbo-Croatian"), -- macrolanguage heb = _("Hebrew"), hin = _("Hindi"), + hrv = _("Croatian"), hun = _("Hungarian"), hye = _("Armenian"), ido = _("Ido"), -- constructed language @@ -82,6 +89,7 @@ local locale_lang_map = { nob = _("Norwegian Bokmål"), nor = _("Norwegian"), -- macrolanguage oci = _("Occitan"), -- (post 1500) + ori = _("Oriya"), orm = _("Oromo"), -- macrolanguage pan = _("Panjabi"), pli = _("Pali"), @@ -93,8 +101,10 @@ local locale_lang_map = { ron = _("Romanian"), rus = _("Russian"), san = _("Sanskrit"), + sin = _("Sinhala"), slk = _("Slovak"), slv = _("Slovenian"), + snd = _("Sindhi"), spa = _("Spanish"), sqi = _("Albanian"), -- macrolanguage srp = _("Serbian"), @@ -102,12 +112,15 @@ local locale_lang_map = { sun = _("Sundanese"), swa = _("Swahili"), -- macrolanguage swe = _("Swedish"), + syr = _("Syriac"), tam = _("Tamil"), tat = _("Tatar"), tel = _("Telugu"), tgk = _("Tajik"), tgl = _("Tagalog"), tha = _("Thai"), + tir = _("Tigrinya"), + ton = _("Tonga"), tur = _("Turkish"), uig = _("Uighur"), ukr = _("Ukrainian"), @@ -204,6 +217,21 @@ local locale_lang_map = { xxpw = _("Proto-West Germanic"), xxps = _("Proto-Sami"), xxsl = _("Proto-Slavic"), + + -- These are Tesseract-specific extensions. + -- See . + aze_cyrl = _("Azerbaijani - Cyrillic"), + chi_sim = _("Chinese - Simplified"), + chi_tra = _("Chinese - Traditional"), + dan_frak = _("Danish - Fraktur"), + deu_frak = _("German - Fraktur"), + equ = _("Math equation"), + ita_old = _("Old Italian"), + kat_old = _("Old Georgian"), + slk_frak = _("Slovak - Fraktur"), + spa_old = _("Old Spanish"), + srp_latn = _("Serbian - Latin"), + uzb_cyrl = _("Uzbek -Cyrillic"), } local iso_bcp47_map = {