separate word index by unicode punctuation

[shellwiki] / handlers / 40_search.sh

diff --git a/handlers/40_search.sh b/handlers/40_search.sh
index c558f5394b9cd01b8a86063d69a291de094b7fbc..c9bc62738c985f8db46b8a4326f5eac018c08613 100644 (file)
--- a/handlers/40_search.sh
+++ b/handlers/40_search.sh
@@ -5,9 +5,15 @@
 . "$_EXEC/cgilite/storage.sh"
 
 I="$_DATA/index"
-words="$( GET q | sed -E '
-  :X $bY; N; bX; :Y
-  s;([] \t\n\r!\"#'\''()*+,./:;<=>?\\^_`{|}~[-]|%[1-9A-Fa-f]{2})+;     ;g
+words="$( GET q | 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}|'"$(printf '[\342\200\200-\342\201\257]')"')+";
+          else                     # UTF-8 Hack
+          FS = "([] \t\n\r!\"#'\''()*+,./:;<=>?\\^_`{|}~[-]|%[0-9A-Fa-f]{2}|'"$(printf '\342\200[\200-\277]|\342\201[\201-\257]')"')+";
+          fi
+        }
+       { for (n = 1; n <= NF; n++) printf "%s  ", $n; }
 ')"
 
 for w in ${words}; do