include/exalead/exalead.parser.inc.php

   1 <?php
   2
   3 require_once('exalead.class.php');
   4
   5 function convert_url($string){
   6   return str_replace('+', '%2B', $string);
   7 }
   8
   9 $GLOBALS['query_all'] = 'a*';
  10
  11 class Exalead{
  12
  13   var $parserId;
  14
  15   var $data;
  16
  17   var $currentGroup;
  18   var $currentCategories = array();
  19   var $currentSpelling;
  20   var $currentHit;
  21   var $currentHitField;
  22   var $currentHitGroup;
  23   var $currentHitCategory;
  24   var $currentAction;
  25   var $currentTextSegment;
  26   var $currentQuery;
  27   var $currentQueryTerm;
  28   var $currentQueryParameter;
  29   var $currentKeyword;
  30
  31   //url de base du produit Exalead
  32   var $base_cgi = '';
  33
  34   // Query to dump indexed database
  35   var $query_all= '';
  36
  37 /****    Constructeur     *********/
  38
  39
  40   function Exalead($base_cgi = '', $override_query_all = ''){
  41      $this->data = new ExaleadData();
  42      $this->currentGroup = new ExaleadGroup();
  43      $this->currentCategories = array();
  44      $this->currentSpelling = new ExaleadSpelling();
  45      $this->currentHit = new ExaleadHit();
  46      $this->currentHitField = new ExaleadHitField();
  47      $this->currentHitGroup = new ExaleadHitGroup();
  48      $this->currentHitCategory = new ExaleadHitCategory();
  49      $this->currentAction = new ExaleadAction();
  50      $this->currentTextSegment = new ExaleadTextSegment();
  51      $this->currentQuery = new ExaleadQuery();
  52      $this->currentQueryTerm = new ExaleadQueryTerm();
  53      $this->currentQueryParameter = new ExaleadQueryParameter();
  54      $this->currentKeyword = new ExaleadKeyword();
  55
  56      //url de base du produit Exalead
  57      $this->base_cgi = $base_cgi;
  58      if(!empty($override_query_all)){
  59        $this->query_all = $override_query_all;
  60      }
  61      else{
  62        $this->query_all = $GLOBALS['query_all'];
  63      }
  64   }
  65
  66 /****  Fonctions d'interface avec le cgi d'Exalead Corporate   ******/
  67
  68   function set_base_cgi($base_cgi){
  69     $this->base_cgi = $base_cgi;
  70   }
  71
  72   //retourne vrai si une requete a été faite, faux sinon
  73   function query($varname = 'query'){
  74     if(!empty($_REQUEST[$varname])){
  75
  76       $this->first_query(stripslashes($_REQUEST[$varname]));
  77       return true;
  78     }
  79     elseif(isset($_REQUEST['_C'])){
  80
  81       $this->handle_request();
  82       return true;
  83     }
  84     return false;
  85   }
  86
  87   //a appeller pour faire la premiere requete
  88   function first_query($query, $offset = 0){
  89     if(empty($this->base_cgi)) return false;
  90
  91     //$tmp = parse_url($this->base_cgi);
  92     //$view_name = substr($tmp['path'], 5);
  93     //$query_exa = $this->base_cgi."?_q=".urlencode($query)."&_f=xml2&A=-1&_vn=".$view_name;
  94     $query_exa = $this->base_cgi."?_q=".urlencode($query)."&_f=xml2";
  95     if($offset > 0){
  96       $query_exa .= "&_s=".$offset;
  97     }
  98
  99     $xml_response = file_get_contents($query_exa);
 100     /*$xml_response = '';
 101     $query_explode = parse_url($query_exa);
 102
 103     $fp = fsockopen("murphy.m4x.org", 10000, $errno, $errstr, 30);
 104     if (!$fp) {
 105       echo "$errstr ($errno)<br />\n";
 106     } else {
 107       $out = "GET {$query_explode['path']}?{$query_explode['query']} HTTP/1.1\r\n";
 108       $out .= "Host: murphy.m4x.org:10000\r\n";
 109       $out .= "Accept: text/xml\r\n";
 110       $out .= "Accept-Charset: utf-8\r\n";
 111       $out .= "Connection: Close\r\n\r\n";
 112
 113       fwrite($fp, $out);
 114       $body = false;
 115       while (!feof($fp)) {
 116        $s = fgets($fp, 1024);
 117         if ( $body )
 118           $xml_response .= $s;
 119         if ( $s == "\r\n" )
 120           $body = true;
 121       }
 122       fclose($fp);
 123     }*/
 124     //echo $xml_response;exit;
 125     $this->parse($xml_response);
 126     //var_dump($this);
 127   }
 128
 129   //pour recuperer tous les résultats d'une base indexée
 130   function get_db_dump(){
 131     $this->first_query($this->query_all);
 132   }
 133
 134   function handle_request(){
 135     if(empty($this->base_cgi)) return false;
 136     if(empty($_REQUEST['_C'])) return false;// _C est le contexte Exalead
 137     $query_exa = $this->base_cgi.'/_C='.str_replace(' ', '%20', $_REQUEST['_C']).'&_f=xml2';
 138     if(!empty($_REQUEST['_s'])){
 139       $query_exa .= "&_s=".((int) $_REQUEST['_s']);
 140     }
 141     $xml_response = file_get_contents($query_exa);
 142     $this->parse($xml_response);
 143   }
 144
 145 /********      Fonctions annexes relatives au parser     ********/
 146
 147   function createParser(){
 148     $this->parserId = xml_parser_create();
 149     xml_set_element_handler($this->parserId, array(&$this, "startElement"), array(&$this, "endElement"));
 150     xml_set_character_data_handler($this->parserId, array(&$this, "parsePCData"));
 151   }
 152
 153   function freeParser(){
 154     xml_parser_free($this->parserId);
 155   }
 156
 157   function parseString($string){
 158     if (!xml_parse($this->parserId, $string, true)) {
 159        die(sprintf("XML error: %s at line %d",
 160              xml_error_string(xml_get_error_code($this->parserId)),
 161              xml_get_current_line_number($this->parserId)));
 162     }
 163   }
 164
 165
 166
 167 /********        Méthode qui lance le parser           ***********/
 168
 169   function parse($string){
 170     $this->createParser();
 171     $this->parseString($string);
 172     $this->freeParser();
 173   }
 174
 175 /*********      fonctions spécifiques à chaque balise     ******/
 176
 177 //Ces méthodes peuvent être surchargées
 178
 179   function startQuery(&$attrs){
 180      $this->currentQuery->query = utf8_decode($attrs['QUERY']);
 181      $this->currentQuery->context = $attrs['CONTEXT'];
 182      $this->currentQuery->time = $attrs['TIME'];
 183      if(isset($attrs['INTERRUPTED'])) $this->currentQuery->interrupted = $attrs['INTERRUPTED'];
 184      if(isset($attrs['BROWSED'])) $this->currentQuery->browsed = $attrs['BROWSED'];
 185   }
 186
 187   function StartQueryTerm(&$attrs){
 188      $this->currentQueryTerm->level = $attrs['LEVEL'];
 189      $this->currentQueryTerm->regexp = utf8_decode($attrs['REGEXP']);
 190   }
 191
 192   function startQueryParameter(&$attrs){
 193      $this->currentQueryParameter->name = $attrs['NAME'];
 194      if(isset($attrs['VALUE'])) $this->currentQueryParameter->value = utf8_decode($attrs['VALUE']);
 195   }
 196
 197   function startKeyword(&$attrs){
 198      if(isset($attrs['NAME'])) $this->currentKeyword->name = $attrs['NAME'];
 199      $this->currentKeyword->display = utf8_decode( $attrs['DISPLAY'] );
 200      $this->currentKeyword->count = $attrs['COUNT'];
 201      $this->currentKeyword->automatic = $attrs['AUTOMATIC'];
 202      if(isset($attrs['REFINEHREF'])) $this->currentKeyword->refine_href = convert_url($attrs['REFINEHREF']);
 203      if(isset($attrs['EXCLUDEHREF'])) $this->currentKeyword->exclude_href = $attrs['EXCLUDEHREF'];
 204      if(isset($attrs['RESETHREF'])) $this->currentKeyword->reset_href = $attrs['RESETHREF'];
 205   }
 206
 207   function startHits(&$attrs){
 208      $this->data->nmatches = $attrs['NMATCHES'];
 209      $this->data->nhits = $attrs['NHITS'];
 210      if(isset($attrs['INTERRUPTED'])) $this->data->interrupted = $attrs['INTERRUPTED'];
 211      $this->data->last = $attrs['LAST'];
 212      $this->data->end = $attrs['END'];
 213      $this->data->start = $attrs['START'];
 214   }
 215
 216   function startHit(&$attrs){
 217      $this->currentHit->url = $attrs['URL'];
 218      $this->currentHit->score = $attrs['SCORE'];
 219   }
 220
 221   function startHitGroup(&$attrs){
 222      $this->currentHitGroup->title = utf8_decode($attrs['TITLE']);
 223      $this->currentHitGroup->gid = $attrs['GID'];
 224   }
 225
 226   function startHitCategory(&$attrs){
 227      $this->currentHitCategory->name = $attrs['NAME'];
 228      $this->currentHitCategory->display = utf8_decode($attrs['DISPLAY']);
 229      $this->currentHitCategory->cref = $attrs['CREF'];
 230      $this->currentHitCategory->gid = $attrs['GID'];
 231      if(isset($attrs['BROWSEHREF'])) $this->currentHitCategory->browsehref = $attrs['BROWSEHREF'];
 232   }
 233
 234   function startAction(&$attrs){
 235      $this->currentAction->display = $attrs['DISPLAY'];
 236      $this->currentAction->kind = $attrs['KIND'];
 237      if(isset($attrs['EXECHREF']))$this->currentAction->execHref = $attrs['EXECHREF'];
 238   }
 239
 240   function startHitField(&$attrs){
 241      $this->currentHitField->name = $attrs['NAME'];
 242      if(isset($attrs['VALUE'])) $this->currentHitField->value = utf8_decode($attrs['VALUE']);
 243   }
 244
 245   function startTextSeg(&$attrs){
 246     $this->currentTextSegment->setHighlighted($attrs['HIGHLIGHTED']);
 247   }
 248   function startTextCut(&$attrs){}
 249
 250   function startSpellingSuggestionVariant(&$attrs){
 251      $this->currentSpelling->setDisplay($attrs['DISPLAY']);
 252      $this->currentSpelling->setQueryHref($attrs['QUERY']);
 253   }
 254
 255   function startGroup(&$attrs){
 256      $this->currentGroup->setGid(utf8_decode($attrs['GID']));
 257      $this->currentGroup->setTitle(utf8_decode($attrs['TITLE']));
 258      $this->currentGroup->setClipped($attrs['CLIPPED']);
 259      $this->currentGroup->setCount($attrs['COUNT']);
 260      $this->currentGroup->setBrowsed($attrs['BROWSED']);
 261      if(isset($attrs['CLIPHREF'])) $this->currentGroup->setClipHref($attrs['CLIPHREF']);
 262      if(isset($attrs['RESETHREF'])) $this->currentGroup->setResetHref($attrs['RESETHREF']);
 263   }
 264
 265   function startCategory(&$attrs){
 266      $currentCategory = new ExaleadCategory();
 267      $currentCategory->name = utf8_decode($attrs['NAME']);
 268      $currentCategory->display = utf8_decode($attrs['DISPLAY']);
 269      $currentCategory->count = $attrs['COUNT'];
 270      $currentCategory->automatic = $attrs['AUTOMATIC'];
 271      $currentCategory->cref = $attrs['CREF'];
 272      if(isset($attrs['REFINEHREF'])) $currentCategory->refine_href = convert_url($attrs['REFINEHREF']);
 273      $currentCategory->exclude_href = '_c=-'.$currentCategory->cref;
 274      if(isset($attrs['RESETHREF'])){
 275        $currentCategory->reset_href = $attrs['RESETHREF'];
 276      }
 277      $currentCategory->gid = $attrs['GID'];
 278      $currentCategory->gcount = $attrs['GCOUNT'];
 279      $this->currentCategories[] = $currentCategory;
 280   }
 281
 282   function startSearch(&$attrs){}
 283
 284   function startElement($parser, $name, $attrs) {
 285    //echo "start $name<br />";
 286    //recupération des paramètres de query
 287    if($name == 'QUERY'){
 288      $this->startQuery($attrs);
 289    }
 290    elseif($name == 'QUERYTERM'){
 291      $this->startQueryTerm($attrs);
 292    }
 293    elseif($name == 'QUERYPARAMETER'){
 294      $this->startQueryParameter($attrs);
 295    }
 296    //gestion des mots-clés
 297    elseif($name == 'KEYWORD'){
 298      $this->startKeyword($attrs);
 299    }
 300    //gestion des resultats
 301    elseif($name == 'HITS'){
 302      $this->startHits($attrs);
 303    }
 304    elseif($name == 'HIT'){
 305      $this->startHit($attrs);
 306    }
 307    elseif($name == 'HITFIELD'){
 308      $this->startHitField($attrs);
 309    }
 310    elseif($name == 'HITGROUP'){
 311      $this->startHitGroup($attrs);
 312    }
 313    elseif($name == 'HITCATEGORY'){
 314      $this->startHitCategory($attrs);
 315    }
 316    elseif($name == 'ACTION'){
 317      $this->startAction($attrs);
 318    }
 319    elseif($name == 'TEXTSEG'){
 320      $this->startTextSeg($attrs);
 321    }
 322    elseif($name == 'TEXTCUT'){
 323      $this->startTextCut($attrs);
 324    }
 325    //gestion suggestions d'orthographe
 326    elseif($name == 'SPELLINGSUGGESTIONVARIANT'){
 327      $this->startSpellingSuggestionVariant($attrs);
 328    }
 329    //gestion des categories pour raffiner
 330    elseif($name == 'GROUP'){
 331      $this->startGroup($attrs);
 332    }
 333    elseif($name == 'CATEGORY'){
 334      $this->startCategory($attrs);
 335    }
 336    elseif($name == 'SEARCH'){
 337      $this->startSearch($attrs);
 338    }
 339   }
 340
 341   function endQuery(){
 342      $this->data->query = $this->currentQuery;
 343      $this->currentQuery->clear();
 344   }
 345   function endQueryTerm(){
 346      $this->currentQuery->addTerm($this->currentQueryTerm);
 347      $this->currentQueryTerm->clear();
 348   }
 349   function endQueryParameter(){
 350      $this->currentQuery->addParameter($this->currentQueryParameter);
 351      $this->currentQueryParameter->clear();
 352   }
 353   function endKeyword(){
 354      $this->data->addKeyword($this->currentKeyword);
 355      $this->currentKeyword->clear();
 356   }
 357   function endHits(){
 358   }
 359   function endHit(){
 360      $this->data->addHit($this->currentHit);
 361      $this->currentHit->clear();
 362   }
 363   function endHitGroup(){
 364     $this->currentHit->addHitGroup($this->currentHitGroup);
 365     $this->currentHitGroup->clear();
 366   }
 367   function endHitCategory(){
 368     $this->currentHitGroup->addHitCategory($this->currentHitCategory);
 369     $this->currentHitCategory->clear();
 370   }
 371   function endAction(){
 372     $this->currentHit->addAction($this->currentAction);
 373     $this->currentAction->clear();
 374   }
 375   function endHitField(){
 376      $this->currentHit->addHitField($this->currentHitField);
 377      $this->currentHitField->clear();
 378   }
 379   function endTextSeg(){
 380      $this->currentHitField->addTextSegment($this->currentTextSegment);
 381      $this->currentTextSegment->clear();
 382   }
 383   function endTextCut(){
 384      $this->currentHitField->setHasTextCut(true);
 385   }
 386   function endSpellingSuggestionVariant(){
 387      $this->data->addSpelling($this->currentSpelling);
 388      $this->currentSpelling->clear();
 389   }
 390   function endGroup(){
 391      $this->data->addGroup($this->currentGroup);
 392      $this->currentGroup->clear();
 393   }
 394   function endCategory(){
 395      //the parent element is a Group element ?
 396      if(count($this->currentCategories) == 1){
 397        $this->currentGroup->addCategory(array_pop($this->currentCategories));
 398      }
 399      else{
 400        //var_dump($this->currentCategories);
 401        $category = array_pop($this->currentCategories);
 402        //reset($this->currentCategories);
 403        end($this->currentCategories);
 404        //var_dump($this->currentCategories);
 405        $parentCategory = &$this->currentCategories[key($this->currentCategories)];
 406        //var_dump($parentCategory);
 407        $parentCategory->addCategory($category);
 408      }
 409   }
 410   function endSearch(){
 411   }
 412
 413
 414   function endElement($parser, $name) {
 415    //echo "end $name<br >";
 416    if($name == 'QUERY'){
 417      $this->endQuery();
 418    }
 419    elseif($name == 'QUERYTERM'){
 420      $this->endQueryTerm();
 421    }
 422    elseif($name == 'QUERYPARAMETER'){
 423      $this->endQueryParameter();
 424    }
 425    elseif($name == 'KEYWORD'){
 426      $this->endKeyword();
 427    }
 428    elseif($name == 'HITS'){
 429      $this->endHits();
 430    }
 431    elseif($name == 'HIT'){
 432      $this->endHit();
 433    }
 434    elseif($name == 'HITFIELD'){
 435      $this->endHitField();
 436    }
 437    elseif($name == 'HITGROUP'){
 438      $this->endHitGroup();
 439    }
 440    elseif($name == 'HITCATEGORY'){
 441      $this->endHitCategory();
 442    }
 443    elseif($name == 'ACTION'){
 444      $this->endAction();
 445    }
 446    elseif($name == 'TEXTSEG'){
 447      $this->endTextSeg();
 448    }
 449    elseif($name == 'TEXTCUT'){
 450      $this->endTextCut();
 451    }
 452    //gestion suggestions d'orthographe
 453    elseif($name == 'SPELLINGSUGGESTIONVARIANT'){
 454      $this->endSpellingSuggestionVariant();
 455    }
 456    //gestion des categories pour raffiner
 457    elseif($name == 'GROUP'){
 458      $this->endGroup();
 459    }
 460    elseif($name == 'CATEGORY'){
 461      $this->endCategory();
 462    }
 463    elseif($name == 'SEARCH'){
 464      $this->endSearch();
 465    }
 466   }
 467
 468   function parsePCData($parser, $text){
 469     $this->currentTextSegment->append(utf8_decode($text));
 470   }
 471
 472 }
 473
 474 ?>