Add pt download, make curl follow "moved permanently" redirects.

[DictionaryPC.git] / data / downloadInputs.sh

diff --git a/data/downloadInputs.sh b/data/downloadInputs.sh
old mode 100644 (file)
new mode 100755 (executable)
index 89e584a..b5640b7
--- a/data/downloadInputs.sh
+++ b/data/downloadInputs.sh
@@ -2,22 +2,58 @@
 
 OLD_DIR=`pwd`
 DIR=`dirname $0`
-
 cd $DIR
 
-echo "Downloading from: http://dumps.wikimedia.org/enwiktionary/"
-WIKI=enwiktionary-20111224-pages-articles.xml
-curl --remote-name http://dumps.wikimedia.org/enwiktionary/20111224/${WIKI}.bz2
-echo "Unzipping... this is slow."
+echo "Note that unzipping is slow."
+
+L=en
+echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
+WIKI=${L}wiktionary-latest-pages-articles.xml
+curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
 bunzip2 ${WIKI}.bz2
-mv ${WIKI} inputs/
+mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
 
 echo "Downloading from: http://ftp.tu-chemnitz.de/pub/Local/urz/ding/de-en-devel/"
 CHEMNITZ=de-en.txt
-curl --remote-name http://ftp.tu-chemnitz.de/pub/Local/urz/ding/de-en-devel/${CHEMNITZ}.gz
+curl -L --remote-name http://ftp.tu-chemnitz.de/pub/Local/urz/ding/de-en-devel/${CHEMNITZ}.gz
 gunzip ${CHEMNITZ}.gz
-mv ${CHEMNITZ} inputs/
+mv ${CHEMNITZ} inputs/de-en_chemnitz.txt
+
+L=fr
+echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
+WIKI=${L}wiktionary-latest-pages-articles.xml
+curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
+bunzip2 --force ${WIKI}.bz2
+mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
+
+L=it
+echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
+WIKI=${L}wiktionary-latest-pages-articles.xml
+curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
+bunzip2 ${WIKI}.bz2
+mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
+
+L=de
+echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
+WIKI=${L}wiktionary-latest-pages-articles.xml
+curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
+bunzip2 ${WIKI}.bz2
+mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
+
+L=es
+echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
+WIKI=${L}wiktionary-latest-pages-articles.xml
+curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
+bunzip2 ${WIKI}.bz2
+mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
+
+L=pt
+echo "Downloading from: http://dumps.wikimedia.org/${L}wiktionary/"
+WIKI=${L}wiktionary-latest-pages-articles.xml
+curl -L --remote-name http://dumps.wikimedia.org/${L}wiktionary/latest/${WIKI}.bz2
+bunzip2 ${WIKI}.bz2
+mv ${WIKI} inputs/${L}wiktionary-pages-articles.xml
 
-echo "Done.  Now run WiktionarySplitter to spit apart enwiktionary."
+echo "Done.  Now run WiktionarySplitter to split apart enwiktionary."
 
 cd $OLD_DIR