27 lines
505 B
Bash
27 lines
505 B
Bash
# Variables
|
|
DICT=Diccionario.jar
|
|
URL=http://www.redeletras.com/diccionario/$DICT
|
|
TMP=extracted
|
|
LIST=lexicon.txt
|
|
|
|
# Goes to repo root
|
|
cd $(dirname $0)/..
|
|
|
|
# Downloads dictionary
|
|
wget -q --show-progress -O $DICT $URL && echo "Downloaded $DICT" || rm $DICT
|
|
|
|
# Extracts dictionary
|
|
unzip -q -d $TMP $DICT
|
|
|
|
# Extracts list
|
|
for txt in $TMP/*.txt; do
|
|
iconv -f ISO-8859-1 -t UTF-8 $txt > $txt.utf8
|
|
cat $txt.utf8 >> $LIST.tmp
|
|
done
|
|
|
|
# Sorts and uniqs
|
|
cat $LIST.tmp | sort > $LIST
|
|
|
|
# Cleans
|
|
rm -rf $TMP $LIST.tmp
|