up to date language codes

git-svn-id: https://vdr-muggle.svn.sourceforge.net/svnroot/vdr-muggle/branches/0.1.3-wr@530 e10066b5-e1e2-0310-b819-94efdf66514b

3 files changed, 89 insertions, 27 deletions

diff --git a/scripts/COPYRIGHT b/scripts/COPYRIGHT
index 9b0e8cb..e41ca51 100644
--- a/scripts/COPYRIGHT
+++ b/scripts/COPYRIGHT
@@ -1,16 +1,15 @@
-# the content of languages.txt is taken from the file
-# iso_639.tab which contains this copyright:
-
-
-## iso-639.tab
-##
-## Copyright (C) 2004 Alastair McKinstry   <mckinstry@computer.org> 
-## Released under the GNU License; see file COPYING for details
-##
-## Last update: 2004-03-29
-##
-## This file gives a list of all languages in the ISO-639
-## standard, and is used to provide translations (via gettext)
-##
-## Status: ISO 639-2:1998 + additions and changes until 2003-03-05
-## Source: http://lcweb.loc.gov/standards/iso639-2/englangn.html
+# the content of languages.txt is generated by iso639tab.py from
+# the file iso_639.xml. iso_639.xml contains this copyright note:
+#
+# <!-- iso-639.tab                                                        -->
+# <!--                                                                    -->
+# <!-- Copyright (C) 2004 Alastair McKinstry   <mckinstry@computer.org>   -->
+# <!-- Released under the GNU License; see file COPYING for details       -->
+# <!--                                                                    -->
+# <!-- Last update: 2004-10-27                                            -->
+# <!--                                                                    -->
+# <!-- This file gives a list of all languages in the ISO-639             -->
+# <!-- standard, and is used to provide translations (via gettext)        -->
+# <!--                                                                    -->
+# <!-- Status: ISO 639-2:1998 + additions and changes until 2004-03-05    -->
+# <!-- Source: http://lcweb.loc.gov/standards/iso639-2/englangn.html      -->
diff --git a/scripts/iso639tab.py b/scripts/iso639tab.py
new file mode 100755
index 0000000..d9d0b86
--- /dev/null
+++ b/scripts/iso639tab.py
@@ -0,0 +1,60 @@
+#!/usr/bin/python
+#
+# Read iso-codes iso_639.xml data file and output a .tab file
+# 
+# Copyright (C) 2005 Alastair McKinstry <mckinstry@debian.org>
+# Released under the GPL.
+# $Id: iso639tab.py,v 1.1 2005/03/02 07:24:51 mckinstry Exp $
+
+# modified by Wolfgang Rohdewald such that it only converts the 
+#   iso-639-2 bibliography code and the name without comments
+
+from xml.sax import saxutils, make_parser, saxlib, saxexts, ContentHandler
+from xml.sax.handler import feature_namespaces
+import sys, os, getopt, urllib2
+
+class printLines(saxutils.DefaultHandler):
+	def __init__(self, ofile):
+		self.ofile = ofile
+
+	def startElement(self, name, attrs):
+		if name != 'iso_639_entry':
+			return
+		t_code = attrs.get('iso_639_2T_code', None)
+		if t_code == None:
+			raise RunTimeError, "Bad file"	
+		if type(t_code) == unicode:
+			t_code = t_code.encode('UTF-8')
+		b_code = attrs.get('iso_639_2B_code', None)
+		if b_code == None:
+			raise RunTimeError, "Bad file"	
+		if type(b_code) == unicode:
+			b_code = b_code.encode('UTF-8')
+		name = attrs.get('name', None)
+		if name == None:
+			raise RunTimeError, " BadFile"
+		short_code=attrs.get('iso_639_1_code','XX')
+		short_code=short_code.encode('UTF-8')
+		if type(name) == unicode:
+			name = name.encode('UTF-8')
+		self.ofile.write (b_code + '\t' + name + '\n')
+
+
+## 
+## MAIN
+##
+
+
+ofile = sys.stdout
+p = make_parser()
+p.setErrorHandler(saxutils.ErrorPrinter())
+try:
+	dh = printLines(ofile)
+	p.setContentHandler(dh)
+	p.parse(sys.argv[1])
+except IOError,e:
+	print in_sysID+": "+str(e)
+except saxlib.SAXException,e:
+	print str(e)
+
+ofile.close()
diff --git a/scripts/languages.txt b/scripts/languages.txt
index 69ab08c..6378dcc 100755
--- a/scripts/languages.txt
+++ b/scripts/languages.txt
@@ -10,8 +10,6 @@ afr	Afrikaans
 aka	Akan
 akk	Akkadian
 alb	Albanian
-ale	Aleut
-alg	Algonquian languages
 amh	Amharic
 ang	English, Old (ca.450-1100)
 apa	Apache languages
@@ -99,7 +97,6 @@ cze	Czech
 dak	Dakota
 dan	Danish
 dar	Dargwa
-dsb	Lower Sorbian
 del	Delaware
 den	Slave (Athapascan)
 dgr	Dogrib
@@ -107,6 +104,7 @@ din	Dinka
 div	Divehi
 doi	Dogri
 dra	Dravidian (Other)
+dsb	Lower Sorbian
 dua	Duala
 dum	Dutch, Middle (ca. 1050-1350)
 dut	Dutch
@@ -126,6 +124,7 @@ fan	Fang
 fao	Faroese
 fat	Fanti
 fij	Fijian
+fil	Filipino; Pilipino
 fin	Finnish
 fiu	Finno-Ugrian (Other)
 fon	Fon
@@ -261,11 +260,12 @@ mao	Maori
 map	Austronesian (Other)
 mar	Marathi
 mas	Masai
+may	Malay
 mdf	Moksha
 mdr	Mandar
 men	Mende
 mga	Irish, Middle (900-1200)
-mic	Micmac
+mic	Mi'kmaq; Micmac
 min	Minangkabau
 mis	Miscellaneous languages
 mkh	Mon-Khmer (Other)
@@ -280,6 +280,7 @@ mos	Mossi
 mul	Multiple languages
 mun	Munda languages
 mus	Creek
+mwl	Mirandese
 mwr	Marwari
 myn	Mayan languages
 myv	Erzya
@@ -298,15 +299,16 @@ nia	Nias
 nic	Niger-Kordofanian (Other)
 niu	Niuean
 nno	Norwegian Nynorsk
-nob	Bøkmal, Norwegian
+nob	Bokmål, Norwegian
 nog	Nogai
 non	Norse, Old
 nor	Norwegian
-nso	Sotho, Northern
+nso	Northern Sotho; Pedi; Sepedi
 nub	Nubian languages
+nym	Nyamwezi
 nwc	Classical Newari; Old Newari
 nya	Chewa; Chichewa; Nyanja
-nym	Nyankole
+nyn	Nyankole
 nyo	Nyoro
 nzi	Nzima
 oci	Occitan (post 1500)
@@ -333,7 +335,7 @@ pol	Polish
 por	Portuguese
 pon	Pohnpeian
 pra	Prakrit languages
-pro	Proveçal, Old (to 1500)
+pro	Provençal, Old (to 1500)
 pus	Pushto
 que	Quechua
 raj	Rajasthani
@@ -355,6 +357,7 @@ san	Sanskrit
 sas	Sasak
 sat	Santali
 scc	Serbian
+scn	Sicilian
 sco	Scots
 scr	Croatian
 sel	Selkup
@@ -363,7 +366,7 @@ sga	Irish, Old (to 900)
 sgn	Sign languages
 shn	Shan
 sid	Sidamo
-sin	Sinhalese
+sin	Sinhala; Sinhalese
 sio	Siouan languages
 sit	Sino-Tibetan (Other)
 sla	Slavic (Other)
@@ -383,7 +386,7 @@ sog	Sogdian
 som	Somali
 son	Songhai
 sot	Sotho, Southern
-spa	Spanish (Castilian)
+spa	Spanish
 srd	Sardinian
 srr	Serer
 ssa	Nilo-Saharan (Other)
@@ -411,8 +414,8 @@ tib	Tibetan
 tig	Tigre
 tir	Tigrinya
 tiv	Tiv
-tkl	Tokelau
 tlh	Klingon; tlhIngan-Hol
+tkl	Tokelau
 tli	Tlinglit
 tmh	Tamashek
 tog	Tonga (Nyasa)