data/downloadInputs.sh

   1 #!/bin/bash -e
   2
   3 OLD_DIR=`pwd`
   4 DIR=`dirname $0`
   5 cd $DIR
   6
   7 echo "Note that unzipping is slow."
   8
   9 L=en
  10 echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
  11 WIKI=${L}wiktionary-latest-pages-articles.xml
  12 curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
  13 bunzip2 ${WIKI}.bz2
  14 mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
  15
  16 echo "Downloading from: http://ftp.tu-chemnitz.de/pub/Local/urz/ding/de-en-devel/"
  17 CHEMNITZ=de-en.txt
  18 curl -L --remote-name http://ftp.tu-chemnitz.de/pub/Local/urz/ding/de-en-devel/${CHEMNITZ}.gz
  19 gunzip ${CHEMNITZ}.gz
  20 mv ${CHEMNITZ} inputs/de-en_chemnitz.txt
  21
  22 L=fr
  23 echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
  24 WIKI=${L}wiktionary-latest-pages-articles.xml
  25 curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
  26 bunzip2 --force ${WIKI}.bz2
  27 mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
  28
  29 L=it
  30 echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
  31 WIKI=${L}wiktionary-latest-pages-articles.xml
  32 curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
  33 bunzip2 ${WIKI}.bz2
  34 mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
  35
  36 L=de
  37 echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
  38 WIKI=${L}wiktionary-latest-pages-articles.xml
  39 curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
  40 bunzip2 ${WIKI}.bz2
  41 mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
  42
  43 L=es
  44 echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
  45 WIKI=${L}wiktionary-latest-pages-articles.xml
  46 curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
  47 bunzip2 ${WIKI}.bz2
  48 mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
  49
  50 L=pt
  51 echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
  52 WIKI=${L}wiktionary-latest-pages-articles.xml
  53 curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
  54 bunzip2 ${WIKI}.bz2
  55 mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
  56
  57 echo "Done.  Now run WiktionarySplitter to split apart enwiktionary."
  58
  59 cd $OLD_DIR