• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1#!/bin/sh
2# Extracts the ISO_639-2 file from http://www.loc.gov/standards/iso639-2/code_list.html
3# Usage: iso-639-2-table.sh < code_list.html
4LC_ALL=C
5export LC_ALL
6tr '\012' ' ' |
7sed -e 's,<tr ,\
8<tr ,g' |
9sed -n -e 's,^<tr [^>]*>[^<>]*<td[^>]*>\([^<>]*\)</td>[^<>]*<td>&nbsp;</td>[^<>]*<td>\([^<>]*\)</td>.*$,\1   \2,p' |
10iconv -f ISO-8859-1 -t UTF-8
11