X-Git-Url: https://git.plutz.net/?a=blobdiff_plain;f=parsers%2F40_indexer.sh;h=7190b0b84f02c6a62f3a91b83f8887b767edac2f;hb=4e89a8d6ab962a2582f78fbf47b1f49c735662cd;hp=d5783f4ea9c8e79cb78dbf706ab75888ec0fd78f;hpb=feca7d3f5bff6cd8b1437adb95686a77ab8e9367;p=shellwiki

diff --git a/parsers/40_indexer.sh b/parsers/40_indexer.sh
index d5783f4..7190b0b 100755
--- a/parsers/40_indexer.sh
+++ b/parsers/40_indexer.sh
@@ -14,10 +14,9 @@
 # ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR
 # IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 
-DOC="${PATH_INFO%/}/" DOC="${DOC%/\[*\]/}" DOC="${DOC%/}/"
-P="$_DATA/pages${DOC}" I="$_DATA/index/"
+DOC="${PATH_INFO%/}/" P="$_DATA/pages${DOC}" I="$_DATA/index/"
 
-if [ -f "$P/#index.flag" -a ! "$P/#page.md" -nt "$P/#index.flag" ]; then
+if [ -f "$P/#index.flag" -a ! "$P/#page.md" -nt "$P/#index.flag" ] || [ ! -d "$P" ]; then
   cat
   exit 0
 fi
@@ -26,7 +25,7 @@ fi
 
 exec 3>&1
 
-touch "$P/#index.flag"
+touch -d "@$_DATE" "$P/#index.flag"
 mkdir -p "$I"
 
 { cat; printf \\n; } \
@@ -35,26 +34,23 @@ mkdir -p "$I"
   printf '%s\n' "$line"
 done \
 | awk '
-  BEGIN { FS = "[] \t\n\r!\"#'\''()*+,./:;<=>?\\^_`{|}~[-]+" }
-        { for (n = 1; n <= NF; n++) if ( $n != "" ) { words[tolower($n)]++; total++; } }
+  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}|'"$(printf '[\342\200\200-\342\201\257]')"')+";
+          else                     # UTF-8 Hack
+          FS = "([] \\t\\n\\r!\"#'\''()*+,./:;<=>?\\\\^_`{|}~[-]|%[0-9A-Fa-f]{2}|'"$(printf '\342\200[\200-\277]|\342\201[\201-\257]')"')+";
+          fi
+        }
+        { for (n = 1; n <= NF; n++) {
+            if ( $n != "" && length($n) <= 128 ) {
+              words[tolower($n)]++; total++;
+        } } }
     END { for (w in words) printf "%i %i %f %s\n", words[w], total, words[w] / total, w; }
 ' \
-| while read num total freq word; do
+| while read -r num total freq word; do
   [ "$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