fixed a bug in grep, make sure words starting with - are not valid.

nifgraup · Mar 11, 2012 · 0dfad19 · 0dfad19
1 parent f96e7f1
commit 0dfad19
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/makedict.sh b/makedict.sh
@@ -91,7 +91,7 @@ elif [ "$1" = "packages" ]; then
 elif [ "$1" != "" ]; then
   echo "Downloading files..."
   test -e ${TMP}/${1}wiktionary-latest-pages-articles.xml || ( wget http://dumps.wikimedia.org/${1}wiktionary/latest/${1}wiktionary-latest-pages-articles.xml.bz2 -O ${TMP}/${1}wiktionary-latest-pages-articles.xml.bz2 && bunzip2 ${TMP}/${1}wiktionary-latest-pages-articles.xml.bz2 )
-  test -e ${TMP}/${1}wiktionary-latest-pages-articles.xml.texts || grep -o "{{[^.]*|[^-.][^}]*" ${TMP}/iswiktionary-latest-pages-articles.xml | grep -v "{{.*|.*[ =]" | sort | uniq > ${TMP}/iswiktionary-latest-pages-articles.xml.texts
+  test -e ${TMP}/${1}wiktionary-latest-pages-articles.xml.texts || grep -o "{{[^.|]*|[^-.][^}]*" ${TMP}/iswiktionary-latest-pages-articles.xml | grep -v "{{.*|.*[ =]" | sort | uniq > ${TMP}/iswiktionary-latest-pages-articles.xml.texts
 
   echo "Extracting valid words from the wiktionary dump..."
   rm -f ${TMP}/wiktionary.extracted