To do

• update Tools/Sphinxsearch and the public repository source
• List the non working queries and debug them (use = keyword_to_test for comparison)
• full text search should return #FullTextSearch
• facebook
• google
• apple
• Brooks
• ... (cf log analysis "?bug&keyword=problematickeyword")
• see also ProblematicQueries
• make it more generalist via e.g. reverse dictionary
• better split documents, e.g. logs per day
• handle the histories (wiki and repository diffs)
• make URLs work, consider specific stemming
• improving weightning and thus ranking
• properly handle remote URLs
• leverage .pageindex of each wiki
• work on snippets
• replace WikiName: by their icon, cf localmap
• add pagerank and other topology metrics as attributes
• use delta or RT indexing, requires better ID generation first though
• add visualization, e.g. network or linked document with the same color
• provide links to specific interesting queries, e.g. only this week
• "internet brain"~10
• LearningSearch
• integrate Social PIM, especially if there are 0 results
• integrate external search, e.g. Seeks, especially if there are 0 results including through Social PIM

"; $star_img = ""; // this could use some Javascript effect $alt_img = ""; $bug_img = ""; if ($words == ""){ print "You need to make an actual search, use ?query=keyword
"; print "

"; return; } if ($startingitem == "" || $startingitem < 0){ $startingitem=0; } $itemspan=20; $cl = new SphinxClient (); $cl->SetLimits($startingitem, $startingitem+$itemspan); $cl->SetMatchMode(SPH_MATCH_EXTENDED2); $cl->SetRankingMode(SPH_RANK_PROXIMITY_BM25); # consider other ranking e.g. pagerank, weighted with freshness and hybrid $cl->SetSortMode(SPH_SORT_RELEVANCE); $res = $cl->Query($words,$index); #cf http://sphinxsearch.com/docs/manual-2.0.1.html#api-funcgroup-querying #$res = $cl->BuildExcerpts ( $docs, $index, $words, $opts ); if ( !$res ) { die ( "ERROR: " . $cl->GetLastError() . ".\n" ); } else { // XXX should test for empty result before doing the assigment // if (0 results) { $socialsearch = shell_exec("pmwiki_social_search "$keyword"); } // ideally this would be done ascynhroneously as it does take some time (few seconds!) // yet still always offer it as an option (since it's "costly") #var_dump($res); $IDs = array_keys($res["matches"]); // note that $IDs = array_unique(array_keys($res["matches"])); does not solve the multiple page issue // thus probably comes from a duplicate indexing print "

Query$bug_img

"; if (isset( $res["words"]["$words"]["hits"]) ) $hits = $res["words"]["$words"]["hits"]." time(s) "; else $hits = ""; print "\"$words\" found ".$hits." in ".$res["total_found"]." document(s):
"; print "

"; foreach ($IDs as $i) { #BuildExcerpts() #consider here adding snippet, would require few disk access, MySQL storage could be faster... print "
• "; $file=$idx["$i"]; if (preg_match('|/home/fabien/irclogs/|',$file)){ $target = preg_replace('|/home/fabien/irclogs/(.*)|','$1',$file); $url = urlencode($target); print "IRClogs:$target"; } elseif (preg_match('|/home/fabien/www/mirrors/|',$file)){ $page = preg_replace('|/home/fabien/www/mirrors/(.*)/wiki.d/(.*)\.(.*)|','$2/$3',$file); $source = preg_replace('|/home/fabien/www/mirrors/(.*)/wiki.d/(.*)\.(.*)|','$1',$file); switch ( $source ) { case "saint-maur": $url="http://saint-maur.benetou.fr"; break; case "fabien": $url="http://fabien.benetou.fr"; break; case "pim": $url="http://www.ourp.im/"; break; case "agiwiki": $url="http://www.agi-wiki.org/"; break; case "wiki": $url="http://fabien.benetou.fr/innovativ.it/www/HistoricalArchives/Seedea"; break; default: $url=$source; } print "$source:$page"; print preg_replace('|/home/fabien/www/mirrors/(.*)/wiki.d/(.*)\.(.*)|',''.$alt_img.'',$file); print "$copy_img"; } elseif (preg_match('|/home/fabien/repository/|',$file)){ print preg_replace('|/home/fabien/repository/(.*)|','Repository:$1',$file); } else { print preg_replace('|/home/fabien/www/(.*)/wiki.d/(.*)\.(.*)|','$1:$2/$3',$file); print preg_replace('|/home/fabien/www/(.*)/wiki.d/(.*)\.(.*)|',''.$alt_img.'',$file); } print "$star_img"; print "
"; } $previousitems=$startingitem-$itemspan; $nextitems=$startingitem+$itemspan; print "

"; print ""; $indexage = "unknown"; $targetindexfile = "/var/lib/sphinxsearch/data/pmwikis.spd"; $updatedindex = filemtime($targetindexfile); $indexage = date("c",$updatedindex); print "

---

$indexage index
(if bug check indexer errors)."; } ?>