X-Git-Url: https://git.plutz.net/?a=blobdiff_plain;f=parsers%2F40_indexer.sh;h=7190b0b84f02c6a62f3a91b83f8887b767edac2f;hb=4e89a8d6ab962a2582f78fbf47b1f49c735662cd;hp=e3065c14aa11b8c5c82b357188f008ce48223e3a;hpb=e16cbbded1805c1cd2256b7679bd299dc4746579;p=shellwiki

diff --git a/parsers/40_indexer.sh b/parsers/40_indexer.sh
index e3065c1..7190b0b 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++) {
@@ -47,22 +47,10 @@ 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 "$DOC")"
-
-  if LOCK "$F"; then
-    touch "$F"
-    { while read d l f n t; do
-      [ "$l" = "$L" ] \
-      || printf "%i	%s	%f	%i	%i\n" \
-                "$d" "$l" "$f" "$n" "$t"
-      done <"$F"
-      printf "%i	%s	%f	%i	%i\n" \
-             "$_DATE" "$L" "$freq" "$num" "$total"
-    } >"$F.$$"
-    mv -- "$F.$$" "$F"
-    RELEASE "$F"
-  fi
+  printf "%i	%s	%f	%i	%i\n" \
+         "$_DATE" "$(STRING "$DOC")" \
+         "$freq" "$num" "$total" \
+  >>"$I/$word"
 done