lib/plugin/SemanticSearch.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id: SemanticSearch.php,v 1.2 2007-01-02 13:23:06 rurban Exp $');
   3 /*
   4  Copyright 2007 Reini Urban
   5
   6  This file is part of PhpWiki.
   7
   8  PhpWiki is free software; you can redistribute it and/or modify
   9  it under the terms of the GNU General Public License as published by
  10  the Free Software Foundation; either version 2 of the License, or
  11  (at your option) any later version.
  12
  13  PhpWiki is distributed in the hope that it will be useful,
  14  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  GNU General Public License for more details.
  17
  18  You should have received a copy of the GNU General Public License
  19  along with PhpWiki; if not, write to the Free Software
  20  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21  */
  22
  23 require_once('lib/TextSearchQuery.php');
  24 require_once('lib/PageList.php');
  25
  26 /**
  27  * Search for relations/attributes and its values.
  28  * page - relation::object. e.g list all cities: is_a::city => relation=is_a&s=city
  29  *
  30  * An attribute has just a value, which is a number, and which is for sure no pagename,
  31  * and its value goes through some units unification. (not yet)
  32  * We can also do numerical comparison (e.g. range searching) with attributes.
  33  *   population>1000000 (not yet)
  34  *
  35  * A more generic <ask> feature will use multiple comparison and nesting.
  36  *   <ask is_a::city and population &gt; 1.000.000 and population &lt; 10.000.000>
  37  *   <ask (is_a::city or is_a::country) and population &lt; 10.000.000>
  38  *
  39  * @author: Reini Urban
  40  */
  41 class WikiPlugin_SemanticSearch
  42 extends WikiPlugin
  43 {
  44     function getName() {
  45         return _("SemanticSearch");
  46     }
  47     function getDescription() {
  48         return _("Search relations and attributes");
  49     }
  50     function getVersion() {
  51         return preg_replace("/[Revision: $]/", '',
  52                             "\$Revision: 1.2 $");
  53     }
  54     function getDefaultArguments() {
  55         return array_merge
  56             (
  57              PageList::supportedArgs(),  // paging and more.
  58              array(
  59                    's'          => "*",  // linkvalue query string
  60                    'page'       => "*",  // which pages (glob allowed), default: all
  61                    'relation'   => '',   // linkname. which relations. default all
  62                    'attribute'  => '',   // linkname. which attributes. default all
  63                    'attr_op'    => ':=', // a funny written way for equality for pure aesthetic pleasure
  64                                          // "All attributes which have this value set"
  65                    'units'      => '',   // ?
  66                    'case_exact' => false,
  67                    'regex'      => 'auto',
  68                    'noform'     => false, // don't show form with results.
  69                    'noheader'   => false  // no caption
  70                    ));
  71     }
  72
  73     function showForm (&$dbi, &$request, $args) {
  74         global $WikiTheme;
  75         $action = $request->getPostURL();
  76         $hiddenfield = HiddenInputs($request->getArgs(),'',
  77                                     array('action','page','s','relation','attribute'));
  78         $pagefilter = HTML::input(array('name' => 'page',
  79                                         'value' => $args['page'],
  80                                         'title' => _("Search only in these pages. With autocompletion."),
  81                                         'class' => 'dropdown',
  82                                         'acdropdown' => 'true',
  83                                         'autocomplete_complete' => 'true',
  84                                         'autocomplete_matchsubstring' => 'false',
  85                                         'autocomplete_list' => 'xmlrpc:wiki.titleSearch ^[S] 4'
  86                                         ), '');
  87         $allrelations = $dbi->listRelations(false,false,true);
  88         $svalues = empty($allrelations) ? "" : join("','", $allrelations);
  89         $reldef = JavaScript("var semsearch_relations = new Array('".$svalues."')");
  90         $relation = HTML::input(array('name' => 'relation',
  91                                       'value' => $args['relation'],
  92                                       'title' => _("Filter by this relation. With autocompletion."),
  93                                       'class' => 'dropdown',
  94                                       'style' => 'width:10em',
  95                                       'acdropdown' => 'true',
  96                                       'autocomplete_assoc' => 'false',
  97                                       'autocomplete_complete' => 'true',
  98                                       'autocomplete_matchsubstring' => 'true',
  99                                       'autocomplete_list' => 'array:semsearch_relations'
 100                                       ), '');
 101         $queryrel = HTML::input(array('name' => 's',
 102                                       'value' => $args['s'],
 103                                       'title' => _("Filter by this link. These are pagenames. With autocompletion."),
 104                                       'class' => 'dropdown',
 105                                       'acdropdown' => 'true',
 106                                       'autocomplete_complete' => 'true',
 107                                       'autocomplete_matchsubstring' => 'true',
 108                                       'autocomplete_list' => 'xmlrpc:wiki.titleSearch ^[S] 4'
 109                                       ), '');
 110         $relsubmit = Button('submit:semsearch[relations]',  _("Relations"), false);
 111         // just testing some dhtml... not yet done
 112         $enhancements = HTML();
 113         $nbsp = HTML::raw('&nbsp;');
 114         $this_uri = $_SERVER['REQUEST_URI'].'#';
 115         $andbutton = new Button(_("AND"),$this_uri,'wikiaction',
 116                                 array(
 117                                       'onclick' => "addquery('rel', 'and')",
 118                                       'title' => _("Add an AND query")));
 119         $orbutton = new Button(_("OR"),$this_uri,'wikiaction',
 120                                 array(
 121                                       'onclick' => "addquery('rel', 'or')",
 122                                       'title' => _("Add an OR query")));
 123         if (DEBUG)
 124             $enhancements = HTML::span($andbutton, $nbsp, $orbutton);
 125         $instructions = _("Search in pages for a relation with that value, which is a pagename.");
 126         $form1 = HTML::form(array('action' => $action,
 127                                   'method' => 'post',
 128                                   'accept-charset' => $GLOBALS['charset']),
 129                             $reldef,
 130                             $hiddenfield, HiddenInputs(array('attribute'=>'')),
 131                             $instructions, HTML::br(),
 132                             HTML::table
 133                             (array('border' => 0,'cellspacing' => 2),
 134                              HTML::colgroup(array('span' => 6)),
 135                              HTML::thead
 136                              (HTML::th(''),HTML::th('Pagefilter'),HTML::th('Relation'),
 137                               HTML::th(''),HTML::th(array('span' => 2),'Links')),
 138                              HTML::tbody
 139                              (HTML::tr(
 140                                        HTML::td($nbsp,$nbsp,$nbsp),
 141                                        HTML::td($pagefilter, ": "),
 142                                        HTML::td($relation),
 143                                        HTML::td(HTML::strong(HTML::tt('  ::  '))),
 144                                        HTML::td($queryrel),
 145                                        HTML::td($nbsp, $relsubmit,
 146                                                 $nbsp, $enhancements)))));
 147
 148         $allattrs = $dbi->listRelations(false,true,true);
 149         if (empty($allrelations) and empty($allattrs)) // be nice to the dummy.
 150             $this->_norelations_warning = _("No relations nor attributes in the whole wikidb defined!");
 151         $svalues = empty($allattrs) ? "" : join("','", $allattrs);
 152         $attdef = JavaScript("var semsearch_attributes = new Array('".$svalues."')\n"
 153                             ."var semsearch_op = new Array('"
 154                                   .join("','", $this->_supported_operators)
 155                                   ."')");
 156         $attribute = HTML::input(array('name' => 'attribute',
 157                                        'value' => $args['attribute'],
 158                                        'title' => _("Filter by this attribute name. With autocompletion."),
 159                                        'class' => 'dropdown',
 160                                        'style' => 'width:10em',
 161                                        'acdropdown' => 'true',
 162                                        'autocomplete_complete' => 'true',
 163                                        'autocomplete_matchsubstring' => 'true',
 164                                        'autocomplete_assoc' => 'false',
 165                                        'autocomplete_list' => 'array:semsearch_attributes'
 166                                       ), '');
 167         $attr_op = HTML::input(array('name' => 'attr_op',
 168                                         'value' => $args['attr_op'],
 169                                         'title' => _("Logical operator. With autocompletion."),
 170                                         'class' => 'dropdown',
 171                                         'style' => 'width:2em',
 172                                         'acdropdown' => 'true',
 173                                         'autocomplete_complete' => 'true',
 174                                         'autocomplete_matchsubstring' => 'true',
 175                                         'autocomplete_assoc' => 'false',
 176                                         'autocomplete_list' => 'array:semsearch_op'
 177                                       ), '');
 178         $queryatt = HTML::input(array('name' => 's',
 179                                       'value' => $args['s'],
 180                                       'title' => _("Filter by this numeric attribute value. With autocompletion."), //?
 181                                       'class' => 'dropdown',
 182                                       'acdropdown' => 'false',
 183                                       'autocomplete_complete' => 'true',
 184                                       'autocomplete_matchsubstring' => 'false',
 185                                       'autocomplete_assoc' => 'false',
 186                                       'autocomplete_list' => 'plugin:SemanticSearch page='.$args['page'].' attribute=^[S]'
 187                                       ), '');
 188         $andbutton = new Button(_("AND"),$this_uri,'wikiaction',
 189                                 array(
 190                                       'onclick' => "addquery('attr', 'and')",
 191                                       'title' => _("Add an AND query")));
 192         $orbutton = new Button(_("OR"),$this_uri,'wikiaction',
 193                                 array(
 194                                       'onclick' => "addquery('attr', 'or')",
 195                                       'title' => _("Add an OR query")));
 196         if (DEBUG)
 197             $enhancements = HTML::span($andbutton, $nbsp, $orbutton);
 198         $attsubmit = Button('submit:semsearch[attributes]', _("Attributes"), false);
 199         $instructions = HTML::span(_("Search in pages for an attribute with that numeric value."),"\n");
 200         if (DEBUG)
 201             $instructions->pushContent
 202               (HTML(" ", new Button(_("Advanced..."),$this_uri)));
 203         $form2 = HTML::form(array('action' => $action,
 204                                   'method' => 'post',
 205                                   'accept-charset' => $GLOBALS['charset']),
 206                             $attdef,
 207                             $hiddenfield, HiddenInputs(array('relation'=>'')),
 208                             $instructions, HTML::br(),
 209                             HTML::table
 210                             (array('border' => 0,'cellspacing' => 2),
 211                              HTML::colgroup(array('span' => 6)),
 212                              HTML::thead
 213                              (HTML::th(''),HTML::th('Pagefilter'),HTML::th('Attribute'),
 214                               HTML::th('Op'),HTML::th(array('span' => 2),'Value')),
 215                              HTML::tbody
 216                              (HTML::tr(
 217                                        HTML::td($nbsp,$nbsp,$nbsp),
 218                                        HTML::td($pagefilter, ": "),
 219                                        HTML::td($attribute),
 220                                        HTML::td($attr_op),
 221                                        HTML::td($queryatt),
 222                                        HTML::td($nbsp, $attsubmit,
 223                                                 $nbsp, $enhancements)))));
 224
 225         return HTML($form1, $form2);
 226     }
 227
 228     function run ($dbi, $argstr, &$request, $basepage) {
 229         global $WikiTheme;
 230
 231         $this->_supported_operators = array(':=','<','<=','>','>=','!=','==','=~');
 232         $args = $this->getArgs($argstr, $request);
 233         if (empty($args['page']))
 234             $args['page'] = "*";
 235         if (!isset($args['s'])) // it might be (integer) 0
 236             $args['s'] = "*";
 237         $form = $this->showForm($dbi, $request, $args);
 238         if (isset($this->_norelations_warning))
 239             $form->pushContent(HTML::div(array('class' => 'warning'),
 240                                          _("Warning:").$this->_norelations_warning));
 241         extract($args);
 242         // for convenience and harmony we allow GET requests also.
 243         if (!$request->isPost()) {
 244             if ($relation or $attribute) // check for good GET request
 245                 ;
 246             else
 247                 return $form; // nobody called us, so just display our supadupa form
 248         }
 249         $pagequery = new TextSearchQuery($page, $args['case_exact'], $args['regex']);
 250         // we might want to check for semsearch['relations'] and semsearch['attributes'] also
 251         if (empty($relation) and empty($attribute)) {
 252             // so we just clicked without selecting any relation.
 253             // hmm. check which button we clicked, before we do the massive alltogether search.
 254             $posted = $request->getArg("semsearch");
 255             if (isset($semsearch['relations']))
 256                 $relation = '*';
 257             elseif (isset($posted['attributes']))
 258                 $attribute = '*';
 259         }
 260         $searchtype = "Text";
 261         if (!empty($relation)) {
 262             $querydesc = $relation."::".$s;
 263             $linkquery = new TextSearchQuery($s, $args['case_exact'], $args['regex']);
 264             $relquery = new TextSearchQuery($relation, $args['case_exact'], $args['regex']);
 265             $links = $dbi->linkSearch($pagequery, $linkquery, 'relation', $relquery);
 266             $pagelist = new PageList($args['info'], $args['exclude'], $args);
 267             $pagelist->_links = array();
 268             while ($link = $links->next()) {
 269                 $pagelist->addPage($link['pagename']);
 270                 $pagelist->_links[] = $link;
 271             }
 272             $pagelist->addColumnObject
 273                 (new _PageList_Column_SemanticSearch_relation('relation', _("Relation"), $pagelist));
 274             $pagelist->addColumnObject
 275                 (new _PageList_Column_SemanticSearch_link('link', _("Link"), $pagelist));
 276         }
 277         // can we merge two different pagelist?
 278         if (!empty($attribute)) {
 279             $relquery = new TextSearchQuery($attribute, $args['case_exact'], $args['regex']);
 280             if (!in_array($attr_op, $this->_supported_operators)) {
 281                 return HTML($form,$this->error(fmt("Illegal operator: %s",
 282                                                    HTML::tt($attr_op))));
 283             }
 284             // TODO: support unit suffixes
 285             if (preg_match('/^\d+$/', $s)) { // do comparison only with numbers
 286                 //include_once("lib/SemanticWeb.php");
 287                 // TODO: Unify units to some format before comparison.
 288                 /* Sooner or later we want logical expressions also:
 289                  *  population < 1million AND area > 50km2
 290                  * Here we check only for one attribute per page.
 291                  */
 292                 // it might not be the best idea to use '*' as variable to expand
 293                 if ($attribute == '*') $attribute = '_x';
 294                 $querydesc = $attribute." ".$attr_op." ".$s;
 295                 $searchtype = "Numeric";
 296                 $linkquery = new NumericSearchQuery($querydesc, $attribute);
 297                 if ($attribute == '_x') $attribute = '*';
 298                 $querydesc = $attribute." ".$attr_op." ".$s;
 299             // text matcher or '*' MATCH_ALL
 300             } elseif (in_array($attr_op, array(':=','==','=~'))) {
 301                 if ($attr_op == '=~') {
 302                     if ($s == '*') $s = '.*'; // help the poor user. we need pcre syntax.
 303                     $linkquery = new TextSearchQuery("$s", $args['case_exact'], 'pcre');
 304                 }
 305                 else
 306                     $linkquery = new TextSearchQuery("$s", $args['case_exact'], $args['regex']);
 307                 $querydesc = "$attribute $attr_op $s";
 308             } else {
 309                 $querydesc = $attribute." ".$attr_op." ".$s;
 310                 return HTML($form, $this->error(fmt("Only text operators can used with strings: %s",
 311                                                     HTML::tt($querydesc))));
 312             }
 313             $links = $dbi->linkSearch($pagequery, $linkquery, 'attribute', $relquery);
 314             if (empty($relation)) {
 315                 $pagelist = new PageList($args['info'], $args['exclude'], $args);
 316                 $pagelist->_links = array();
 317             }
 318             while ($link = $links->next()) {
 319                 $pagelist->addPage($link['pagename']);
 320                 $pagelist->_links[] = $link;
 321             }
 322             $pagelist->addColumnObject
 323                 (new _PageList_Column_SemanticSearch_relation('attribute', _("Attribute"), $pagelist));
 324             $pagelist->addColumnObject
 325                 (new _PageList_Column_SemanticSearch_link('value', _("Value"), $pagelist));
 326         }
 327         if (!isset($pagelist)) {
 328             $querydesc = _("<empty>");
 329             $pagelist = new PageList();
 330         }
 331         if (!$noheader) {
 332         // We put the form into the caption just to be able to return one pagelist object,
 333         // and to still have the convenience form at the top. we could workaround this by
 334         // putting the form as WikiFormRich into the actionpage. but thid doesnt look as
 335         // nice as this here.
 336             $pagelist->setCaption
 337             (   // on mozilla the form doesn't fit into the caption very well.
 338                 HTML($noform ? '' : HTML($form,HTML::hr()),
 339                      fmt("Semantic %s Search Result for \"%s\" in pages \"%s\"",$searchtype,$querydesc,$page)));
 340         }
 341         return $pagelist;
 342     }
 343 };
 344
 345 class _PageList_Column_SemanticSearch_relation
 346 extends _PageList_Column
 347 {
 348     function _PageList_Column_SemanticSearch_relation ($field, $heading, &$pagelist) {
 349         $this->_field = $field;
 350         $this->_heading = $heading;
 351         $this->_need_rev = false;
 352         $this->_iscustom = true;
 353         $this->_pagelist =& $pagelist;
 354     }
 355     function _getValue(&$page, $revision_handle) {
 356         if (is_object($page)) $text = $page->getName();
 357         else $text = $page;
 358         $link = $this->_pagelist->_links[$this->current_row];
 359         return WikiLink($link['linkname'],'if_known');
 360     }
 361 }
 362 class _PageList_Column_SemanticSearch_link
 363 extends _PageList_Column_SemanticSearch_relation
 364 {
 365     function _getValue(&$page, $revision_handle) {
 366         if (is_object($page)) $text = $page->getName();
 367         else $text = $page;
 368         $link = $this->_pagelist->_links[$this->current_row];
 369         if ($this->_field != 'value')
 370             return WikiLink($link['linkvalue'],'if_known');
 371         else
 372             return $link['linkvalue'];
 373     }
 374 }
 375
 376 // $Log: not supported by cvs2svn $
 377 // Revision 1.1  2006/03/07 20:52:01  rurban
 378 // not yet working good enough
 379 //
 380
 381 // Local Variables:
 382 // mode: php
 383 // tab-width: 8
 384 // c-basic-offset: 4
 385 // c-hanging-comment-ender-p: nil
 386 // indent-tabs-mode: nil
 387 // End:
 388 ?>