diff options
author | Andreas Gohr <andi@splitbrain.org> | 2009-01-19 20:09:20 +0100 |
---|---|---|
committer | Andreas Gohr <andi@splitbrain.org> | 2009-01-19 20:09:20 +0100 |
commit | d3fb321928ab89178db9ff1482e6fa7f20c25f91 (patch) | |
tree | d5193f6bfdb3271ce1363884de3f33746bb6faad | |
parent | a57c7caf07c9f9df0792f7636700528fa1e1d561 (diff) | |
download | rpg-d3fb321928ab89178db9ff1482e6fa7f20c25f91.tar.gz rpg-d3fb321928ab89178db9ff1482e6fa7f20c25f91.tar.bz2 |
Changed minimum word length for fulltext index to 2
darcs-hash:20090119190920-7ad00-5409285ea5c44379fec906d08f5ccb710eac5b6d.gz
-rw-r--r-- | inc/indexer.php | 2 | ||||
-rw-r--r-- | inc/lang/de/stopwords.txt | 3 | ||||
-rw-r--r-- | inc/lang/en/stopwords.txt | 10 |
3 files changed, 14 insertions, 1 deletions
diff --git a/inc/indexer.php b/inc/indexer.php index 08ad03965..4b59684cd 100644 --- a/inc/indexer.php +++ b/inc/indexer.php @@ -12,7 +12,7 @@ require_once(DOKU_INC.'inc/utf8.php'); require_once(DOKU_INC.'inc/parserutils.php'); // set the minimum token length to use in the index (note, this doesn't apply to numeric tokens) -if (!defined('IDX_MINWORDLENGTH')) define('IDX_MINWORDLENGTH',3); +if (!defined('IDX_MINWORDLENGTH')) define('IDX_MINWORDLENGTH',2); // Asian characters are handled as words. The following regexp defines the // Unicode-Ranges for Asian characters diff --git a/inc/lang/de/stopwords.txt b/inc/lang/de/stopwords.txt index 78261b7c9..443b17723 100644 --- a/inc/lang/de/stopwords.txt +++ b/inc/lang/de/stopwords.txt @@ -50,6 +50,8 @@ hinter ich ihr ihre +in +im ist jede jedem @@ -98,6 +100,7 @@ unter vom von vor +um wann warum was diff --git a/inc/lang/en/stopwords.txt b/inc/lang/en/stopwords.txt index bc6eb48ae..afc301659 100644 --- a/inc/lang/en/stopwords.txt +++ b/inc/lang/en/stopwords.txt @@ -4,6 +4,8 @@ # This list is based upon the ones found at http://www.ranks.nl/stopwords/ about are +as +an and you your @@ -13,10 +15,18 @@ com for from into +if +in +is +it how +of +on +or that the this +to was what when |