X-Git-Url: https://git.plutz.net/?a=blobdiff_plain;f=parsers%2F40_indexer.sh;fp=parsers%2F40_indexer.sh;h=08060f1637ba6aa4b83a1d686fab601f09751ffd;hb=634d3aa5daf5404f2d209512acf70c601ece5697;hp=e3065c14aa11b8c5c82b357188f008ce48223e3a;hpb=22724fcc85fc43d40da86fedd6f11a7b32289f46;p=shellwiki

diff --git a/parsers/40_indexer.sh b/parsers/40_indexer.sh
index e3065c1..08060f1 100755
--- a/parsers/40_indexer.sh
+++ b/parsers/40_indexer.sh
@@ -36,9 +36,9 @@ done \
 | awk '
   BEGIN { # Field separator FS should include punctuation, including Unicode Block U+2000 - U+206F
           if ( length("¡") == 1 )  # Utf-8 aware AWK
-          FS = "([] \t\n\r!\"#'\''()*+,./:;<=>?\\^_`{|}~[-]|%[0-9A-Fa-f]{2}|'"$(printf '[\342\200\200-\342\201\257]')"')+";
+          FS = "([] \\t\\n\\r!\"#'\''()*+,./:;<=>?\\\\^_`{|}~[-]|%[0-9A-Fa-f]{2}|'"$(printf '[\342\200\200-\342\201\257]')"')+";
           else                     # UTF-8 Hack
-          FS = "([] \t\n\r!\"#'\''()*+,./:;<=>?\\^_`{|}~[-]|%[0-9A-Fa-f]{2}|'"$(printf '\342\200[\200-\277]|\342\201[\201-\257]')"')+";
+          FS = "([] \\t\\n\\r!\"#'\''()*+,./:;<=>?\\\\^_`{|}~[-]|%[0-9A-Fa-f]{2}|'"$(printf '\342\200[\200-\277]|\342\201[\201-\257]')"')+";
           fi
         }
         { for (n = 1; n <= NF; n++) {
@@ -47,7 +47,7 @@ done \
         } } }
     END { for (w in words) printf "%i %i %f %s\n", words[w], total, words[w] / total, w; }
 ' \
-| while read num total freq word; do
+| while read -r num total freq word; do
   [ "$word" ] || continue
   F="$I/$word"
   L="$(STRING "$DOC")"