include/exalead/exalead.parser.inc.php

   1 <?php
   2
   3 require_once('exalead/exalead.class.php');
   4
   5 function convert_url($string)
   6 {
   7     return str_replace('+', '%2B', $string);
   8 }
   9
  10 $GLOBALS['query_all'] = 'a*';
  11
  12 class Exalead
  13 {
  14
  15     var $parserId;
  16
  17     var $data;
  18
  19     var $currentGroup;
  20     var $currentCategories = array();
  21     var $currentSpelling;
  22     var $currentHit;
  23     var $currentHitField;
  24     var $currentHitGroup;
  25     var $currentHitCategory;
  26     var $currentAction;
  27     var $currentTextSegment;
  28     var $currentQuery;
  29     var $currentQueryTerm;
  30     var $currentQueryParameter;
  31     var $currentKeyword;
  32
  33     //url de base du produit Exalead
  34     var $base_cgi = '';
  35
  36     // Query to dump indexed database
  37     var $query_all= '';
  38
  39     /****    Constructeur     *********/
  40
  41
  42     function Exalead($base_cgi = '', $override_query_all = '')
  43     {
  44         $this->data = new ExaleadData();
  45         $this->currentGroup = new ExaleadGroup();
  46         $this->currentCategories = array();
  47         $this->currentSpelling = new ExaleadSpelling();
  48         $this->currentHit = new ExaleadHit();
  49         $this->currentHitField = new ExaleadHitField();
  50         $this->currentHitGroup = new ExaleadHitGroup();
  51         $this->currentHitCategory = new ExaleadHitCategory();
  52         $this->currentAction = new ExaleadAction();
  53         $this->currentTextSegment = new ExaleadTextSegment();
  54         $this->currentQuery = new ExaleadQuery();
  55         $this->currentQueryTerm = new ExaleadQueryTerm();
  56         $this->currentQueryParameter = new ExaleadQueryParameter();
  57         $this->currentKeyword = new ExaleadKeyword();
  58
  59         //url de base du produit Exalead
  60         $this->base_cgi = $base_cgi;
  61         if (!empty($override_query_all)) {
  62             $this->query_all = $override_query_all;
  63         } else {
  64             $this->query_all = $GLOBALS['query_all'];
  65         }
  66     }
  67
  68     /****  Fonctions d'interface avec le cgi d'Exalead Corporate   ******/
  69
  70     function set_base_cgi($base_cgi)
  71     {
  72         $this->base_cgi = $base_cgi;
  73     }
  74
  75     //retourne vrai si une requete a été faite, faux sinon
  76     function query($varname = 'query')
  77     {
  78         if (!empty($_REQUEST[$varname])) {
  79
  80             $this->first_query(stripslashes($_REQUEST[$varname]));
  81             return true;
  82
  83         } elseif (isset($_REQUEST['_C'])) {
  84
  85             $this->handle_request();
  86             return true;
  87         }
  88         return false;
  89     }
  90
  91     //a appeller pour faire la premiere requete
  92     function first_query($query, $offset = 0)
  93     {
  94         if (empty($this->base_cgi)) return false;
  95
  96         //$tmp = parse_url($this->base_cgi);
  97         //$view_name = substr($tmp['path'], 5);
  98         //$query_exa = $this->base_cgi."?_q=".urlencode($query)."&_f=xml2&A=-1&_vn=".$view_name;
  99         $query_exa = $this->base_cgi."?_q=".urlencode($query)."&_f=xml2";
 100         if ($offset > 0) {
 101             $query_exa .= "&_s=".$offset;
 102         }
 103
 104         $xml_response = file_get_contents($query_exa);
 105         $this->parse($xml_response);
 106     }
 107
 108     //pour recuperer tous les résultats d'une base indexée
 109     function get_db_dump()
 110     {
 111         $this->first_query($this->query_all);
 112     }
 113
 114     function handle_request()
 115     {
 116         if (empty($this->base_cgi)) return false;
 117         if (empty($_REQUEST['_C'])) return false;// _C est le contexte Exalead
 118         $query_exa = $this->base_cgi.'/_C='.str_replace(' ', '%20', $_REQUEST['_C']).'&_f=xml2';
 119         if (!empty($_REQUEST['_s'])) {
 120             $query_exa .= "&_s=".((int) $_REQUEST['_s']);
 121         }
 122         $xml_response = file_get_contents($query_exa);
 123         $this->parse($xml_response);
 124     }
 125
 126     /********      Fonctions annexes relatives au parser     ********/
 127
 128     function createParser()
 129     {
 130         $this->parserId = xml_parser_create();
 131         xml_set_element_handler($this->parserId, array(&$this, "startElement"), array(&$this, "endElement"));
 132         xml_set_character_data_handler($this->parserId, array(&$this, "parsePCData"));
 133     }
 134
 135     function freeParser()
 136     {
 137         xml_parser_free($this->parserId);
 138     }
 139
 140     function parseString($string)
 141     {
 142         if (!xml_parse($this->parserId, $string, true)) {
 143             die(sprintf("XML error: %s at line %d",
 144             xml_error_string(xml_get_error_code($this->parserId)),
 145             xml_get_current_line_number($this->parserId)));
 146         }
 147     }
 148
 149
 150
 151     /********        Méthode qui lance le parser           ***********/
 152
 153     function parse($string)
 154     {
 155         $this->createParser();
 156         $this->parseString($string);
 157         $this->freeParser();
 158     }
 159
 160     /*********      fonctions spécifiques à chaque balise     ******/
 161
 162     //Ces méthodes peuvent être surchargées
 163
 164     function startQuery(&$attrs)
 165     {
 166         $this->currentQuery->query = utf8_decode($attrs['QUERY']);
 167         $this->currentQuery->context = $attrs['CONTEXT'];
 168         $this->currentQuery->time = $attrs['TIME'];
 169         if (isset($attrs['INTERRUPTED'])) $this->currentQuery->interrupted = $attrs['INTERRUPTED'];
 170         if (isset($attrs['BROWSED'])) $this->currentQuery->browsed = $attrs['BROWSED'];
 171     }
 172
 173     function StartQueryTerm(&$attrs)
 174     {
 175         $this->currentQueryTerm->level = $attrs['LEVEL'];
 176         $this->currentQueryTerm->regexp = utf8_decode($attrs['REGEXP']);
 177     }
 178
 179     function startQueryParameter(&$attrs)
 180     {
 181         $this->currentQueryParameter->name = $attrs['NAME'];
 182         if (isset($attrs['VALUE'])) $this->currentQueryParameter->value = utf8_decode($attrs['VALUE']);
 183     }
 184
 185     function startKeyword(&$attrs)
 186     {
 187         if (isset($attrs['NAME'])) $this->currentKeyword->name = $attrs['NAME'];
 188         $this->currentKeyword->display = utf8_decode( $attrs['DISPLAY'] );
 189         $this->currentKeyword->count = $attrs['COUNT'];
 190         $this->currentKeyword->automatic = $attrs['AUTOMATIC'];
 191         if (isset($attrs['REFINEHREF'])) $this->currentKeyword->refine_href = convert_url($attrs['REFINEHREF']);
 192         if (isset($attrs['EXCLUDEHREF'])) $this->currentKeyword->exclude_href = $attrs['EXCLUDEHREF'];
 193         if (isset($attrs['RESETHREF'])) $this->currentKeyword->reset_href = $attrs['RESETHREF'];
 194     }
 195
 196     function startHits(&$attrs)
 197     {
 198         $this->data->nmatches = $attrs['NMATCHES'];
 199         $this->data->nhits = $attrs['NHITS'];
 200         if (isset($attrs['INTERRUPTED'])) $this->data->interrupted = $attrs['INTERRUPTED'];
 201         $this->data->last = $attrs['LAST'];
 202         $this->data->end = $attrs['END'];
 203         $this->data->start = $attrs['START'];
 204     }
 205
 206     function startHit(&$attrs)
 207     {
 208         $this->currentHit->url = $attrs['URL'];
 209         $this->currentHit->score = $attrs['SCORE'];
 210     }
 211
 212     function startHitGroup(&$attrs)
 213     {
 214         $this->currentHitGroup->title = utf8_decode($attrs['TITLE']);
 215         $this->currentHitGroup->gid = $attrs['GID'];
 216     }
 217
 218     function startHitCategory(&$attrs)
 219     {
 220         $this->currentHitCategory->name = $attrs['NAME'];
 221         $this->currentHitCategory->display = utf8_decode($attrs['DISPLAY']);
 222         $this->currentHitCategory->cref = $attrs['CREF'];
 223         $this->currentHitCategory->gid = $attrs['GID'];
 224         if (isset($attrs['BROWSEHREF'])) $this->currentHitCategory->browsehref = $attrs['BROWSEHREF'];
 225     }
 226
 227     function startAction(&$attrs)
 228     {
 229         $this->currentAction->display = $attrs['DISPLAY'];
 230         $this->currentAction->kind = $attrs['KIND'];
 231         if (isset($attrs['EXECHREF']))$this->currentAction->execHref = $attrs['EXECHREF'];
 232     }
 233
 234     function startHitField(&$attrs)
 235     {
 236         $this->currentHitField->name = $attrs['NAME'];
 237         if (isset($attrs['VALUE'])) $this->currentHitField->value = utf8_decode($attrs['VALUE']);
 238     }
 239
 240     function startTextSeg(&$attrs)
 241     {
 242         $this->currentTextSegment->setHighlighted($attrs['HIGHLIGHTED']);
 243     }
 244     function startTextCut(&$attrs) { }
 245
 246     function startSpellingSuggestionVariant(&$attrs)
 247     {
 248         $this->currentSpelling->setDisplay($attrs['DISPLAY']);
 249         $this->currentSpelling->setQueryHref($attrs['QUERY']);
 250     }
 251
 252     function startGroup(&$attrs)
 253     {
 254         $this->currentGroup->setGid(utf8_decode($attrs['GID']));
 255         $this->currentGroup->setTitle(utf8_decode($attrs['TITLE']));
 256         $this->currentGroup->setClipped($attrs['CLIPPED']);
 257         $this->currentGroup->setCount($attrs['COUNT']);
 258         $this->currentGroup->setBrowsed($attrs['BROWSED']);
 259         if (isset($attrs['CLIPHREF'])) $this->currentGroup->setClipHref($attrs['CLIPHREF']);
 260         if (isset($attrs['RESETHREF'])) $this->currentGroup->setResetHref($attrs['RESETHREF']);
 261     }
 262
 263     function startCategory(&$attrs)
 264     {
 265         $currentCategory = new ExaleadCategory();
 266         $currentCategory->name = utf8_decode($attrs['NAME']);
 267         $currentCategory->display = utf8_decode($attrs['DISPLAY']);
 268         $currentCategory->count = $attrs['COUNT'];
 269         $currentCategory->automatic = $attrs['AUTOMATIC'];
 270         $currentCategory->cref = $attrs['CREF'];
 271         if (isset($attrs['REFINEHREF'])) $currentCategory->refine_href = convert_url($attrs['REFINEHREF']);
 272         $currentCategory->exclude_href = '_c=-'.$currentCategory->cref;
 273         if (isset($attrs['RESETHREF'])) {
 274             $currentCategory->reset_href = $attrs['RESETHREF'];
 275         }
 276         $currentCategory->gid = $attrs['GID'];
 277         $currentCategory->gcount = $attrs['GCOUNT'];
 278         $this->currentCategories[] = $currentCategory;
 279     }
 280
 281     function startSearch(&$attrs) { }
 282
 283     function startElement($parser, $name, $attrs)
 284     {
 285         //recupération des paramètres de query
 286         switch ($name) {
 287             case 'QUERY':
 288                 $this->startQuery($attrs);
 289                 break;
 290             case 'QUERYTERM':
 291                 $this->startQueryTerm($attrs);
 292                 break;
 293             case 'QUERYPARAMETER':
 294                 $this->startQueryParameter($attrs);
 295                 break;
 296             case 'KEYWORD':
 297                 $this->startKeyword($attrs);
 298                 break;
 299             case 'HITS':
 300                 $this->startHits($attrs);
 301                 break;
 302             case 'HIT':
 303                 $this->startHit($attrs);
 304                 break;
 305             case 'HITFIELD':
 306                 $this->startHitField($attrs);
 307                 break;
 308             case 'HITGROUP':
 309                 $this->startHitGroup($attrs);
 310                 break;
 311             case 'HITCATEGORY':
 312                 $this->startHitCategory($attrs);
 313                 break;
 314             case 'ACTION':
 315                 $this->startAction($attrs);
 316                 break;
 317             case 'TEXTSEG':
 318                 $this->startTextSeg($attrs);
 319                 break;
 320             case 'TEXTCUT':
 321                 $this->startTextCut($attrs);
 322                 break;
 323             case 'SPELLINGSUGGESTIONVARIANT':
 324                 $this->startSpellingSuggestionVariant($attrs);
 325                 break;
 326             case 'GROUP':
 327                 $this->startGroup($attrs);
 328                 break;
 329             case 'CATEGORY':
 330                 $this->startCategory($attrs);
 331                 break;
 332             case 'SEARCH':
 333                 $this->startSearch($attrs);
 334                 break;
 335         }
 336     }
 337
 338     function endQuery()
 339     {
 340         $this->data->query = $this->currentQuery;
 341         $this->currentQuery->clear();
 342     }
 343     function endQueryTerm()
 344     {
 345         $this->currentQuery->addTerm($this->currentQueryTerm);
 346         $this->currentQueryTerm->clear();
 347     }
 348     function endQueryParameter()
 349     {
 350         $this->currentQuery->addParameter($this->currentQueryParameter);
 351         $this->currentQueryParameter->clear();
 352     }
 353     function endKeyword()
 354     {
 355         $this->data->addKeyword($this->currentKeyword);
 356         $this->currentKeyword->clear();
 357     }
 358     function endHits()
 359     {
 360     }
 361     function endHit()
 362     {
 363         $this->data->addHit($this->currentHit);
 364         $this->currentHit->clear();
 365     }
 366     function endHitGroup()
 367     {
 368         $this->currentHit->addHitGroup($this->currentHitGroup);
 369         $this->currentHitGroup->clear();
 370     }
 371     function endHitCategory()
 372     {
 373         $this->currentHitGroup->addHitCategory($this->currentHitCategory);
 374         $this->currentHitCategory->clear();
 375     }
 376     function endAction()
 377     {
 378         $this->currentHit->addAction($this->currentAction);
 379         $this->currentAction->clear();
 380     }
 381     function endHitField()
 382     {
 383         $this->currentHit->addHitField($this->currentHitField);
 384         $this->currentHitField->clear();
 385     }
 386     function endTextSeg()
 387     {
 388         $this->currentHitField->addTextSegment($this->currentTextSegment);
 389         $this->currentTextSegment->clear();
 390     }
 391     function endTextCut()
 392     {
 393         $this->currentHitField->setHasTextCut(true);
 394     }
 395     function endSpellingSuggestionVariant()
 396     {
 397         $this->data->addSpelling($this->currentSpelling);
 398         $this->currentSpelling->clear();
 399     }
 400     function endGroup()
 401     {
 402         $this->data->addGroup($this->currentGroup);
 403         $this->currentGroup->clear();
 404     }
 405     function endCategory()
 406     {
 407         //the parent element is a Group element ?
 408         if (count($this->currentCategories) == 1) {
 409             $this->currentGroup->addCategory(array_pop($this->currentCategories));
 410         } else {
 411             //var_dump($this->currentCategories);
 412             $category = array_pop($this->currentCategories);
 413             //reset($this->currentCategories);
 414             end($this->currentCategories);
 415             //var_dump($this->currentCategories);
 416             $parentCategory = &$this->currentCategories[key($this->currentCategories)];
 417             //var_dump($parentCategory);
 418             $parentCategory->addCategory($category);
 419         }
 420     }
 421     function endSearch() { }
 422
 423
 424     function endElement($parser, $name)
 425     {
 426         switch ($name) {
 427             case 'QUERY':
 428                 $this->endQuery();
 429                 break;
 430             case 'QUERYTERM':
 431                 $this->endQueryTerm();
 432                 break;
 433             case 'QUERYPARAMETER':
 434                 $this->endQueryParameter();
 435                 break;
 436             case 'KEYWORD':
 437                 $this->endKeyword();
 438                 break;
 439             case 'HITS':
 440                 $this->endHits();
 441                 break;
 442             case 'HIT':
 443                 $this->endHit();
 444                 break;
 445             case 'HITFIELD':
 446                 $this->endHitField();
 447                 break;
 448             case 'HITGROUP':
 449                 $this->endHitGroup();
 450                 break;
 451             case 'HITCATEGORY':
 452                 $this->endHitCategory();
 453                 break;
 454             case 'ACTION':
 455                 $this->endAction();
 456                 break;
 457             case 'TEXTSEG':
 458                 $this->endTextSeg();
 459                 break;
 460             case 'TEXTCUT':
 461                 $this->endTextCut();
 462                 break;
 463                 //gestion suggestions d'orthographe
 464             case 'SPELLINGSUGGESTIONVARIANT':
 465                 $this->endSpellingSuggestionVariant();
 466                 break;
 467                 //gestion des categories pour raffiner
 468             case 'GROUP':
 469                 $this->endGroup();
 470                 break;
 471             case 'CATEGORY':
 472                 $this->endCategory();
 473                 break;
 474             case 'SEARCH':
 475                 $this->endSearch();
 476                 break;
 477         }
 478     }
 479
 480     function parsePCData($parser, $text)
 481     {
 482         $this->currentTextSegment->append(utf8_decode($text));
 483     }
 484 }
 485
 486 ?>