git.oblomov.eu Git - ikiwiki/blob - doc/todo/xapian_omega_same_lang_when_indexing_and_searching.mdwn

   1 Hi, by default xapian/omega use locate param from blog.setup to set stemmer language when wiki is indexing.
   2
   3 But, when you search, we use omega cgi, and we not set language, so if you indexing in french, but search in english, you have a bad result.
   4
   5 I propose to set a new param omega_stemmer in blog.setup, to fix the same language when we indexing, and searching. And if omega_stemmer is not set, we use LANG env param.
   6
   7 Bellow, you can find the patch.
   8
   9
  10
  11         diff --git a/IkiWiki/Plugin/search.pm b/IkiWiki/Plugin/search.pm
  12         index 42d2e0d..08a0a01 100644
  13         --- a/IkiWiki/Plugin/search.pm
  14         +++ b/IkiWiki/Plugin/search.pm
  15         @@ -33,6 +33,13 @@ sub getsetup () {
  16                                 safe => 0, # external program
  17                                 rebuild => 0,
  18                         },
  19         +               omega_stemmer => {
  20         +                       type => "string",
  21         +                       example => "en",
  22         +                       description => "language used for indexing and searching",
  23         +                       safe => 0, # external program
  24         +                       rebuild => 0,
  25         +               },
  26          }
  27
  28          sub checkconfig () {
  29         @@ -136,7 +143,7 @@ sub indexhtml (@) {
  30                 # Index document and add terms for other metadata.
  31                 my $tg = Search::Xapian::TermGenerator->new();
  32                 if (! $stemmer) {
  33         -               my $langcode=$ENV{LANG} || "en";
  34         +               my $langcode=$config{omega_stemmer} || $ENV{LANG} || "en";
  35                         $langcode=~s/_.*//;
  36
  37                         # This whitelist is here to work around a xapian bug (#486138)
  38         @@ -183,6 +190,18 @@ sub cgi ($) {
  39                         IkiWiki::loadindex();
  40                         $ENV{HELPLINK}=htmllink("", "", "ikiwiki/searching",
  41                                 noimageinline => 1, linktext => "Help");
  42         +               my $langcode=$config{omega_stemmer} || $ENV{LANG} || "en";
  43         +               $langcode=~s/_.*//;
  44         +
  45         +               # This whitelist is here to work around a xapian bug (#486138)
  46         +               my @whitelist=qw{da de en es fi fr hu it no pt ru ro sv tr};
  47         +
  48         +               if (grep { $_ eq $langcode } @whitelist) {
  49         +                        $ENV{STEMMER}=$langcode;
  50         +               }
  51         +               else {
  52         +                        $ENV{STEMMER}="en";
  53         +               }
  54                         exec($config{omega_cgi}) || error("$config{omega_cgi} failed: $!");
  55                 }
  56          }
  57         diff --git a/templates/searchquery.tmpl b/templates/searchquery.tmpl
  58         index 15bc78e..4742460 100644
  59         --- a/templates/searchquery.tmpl
  60         +++ b/templates/searchquery.tmpl
  61         @@ -1,6 +1,6 @@
  62          $setmap{prefix,title,S}
  63          $setmap{prefix,link,XLINK}
  64         -$set{thousand,$.}$set{decimal,.}$setmap{BN,,Any Country,uk,England,fr,France}
  65         +$set{thousand,$.}$set{decimal,.}$setmap{BN,,Any Country,uk,England,fr,France}$set{stemmer,$env{STEMMER}}
  66          ${
  67          $def{PREV,
  68          $if{$ne{$topdoc,0},<INPUT TYPE=image NAME="&lt;" ALT="&lt;"
  69
  70 Regards,
  71
  72 [[!tag patch]]