Vérifie la validité d'un code HTML

                    

                    

                    
<?php
/*------------------------------*/
/*
 Titre : Vérifie la validité d'un code HTML                                                                         
                                                                                                                          
 Date édition     : 14 Fev 2019                                                                                        
 Date mise a jour : 13 Aout 2019
                                                                                         
 Rapport de la maj:                                                                                                       
  - fonctionnement du code vérifié                                                                                  
  - maintenance du code                                                                                                   
*/
/*------------------------------*/

     function html_check($url){
     /* Parametre : $url : URL ou chemin d'acces du fichier a tester 
      * La variable retournée est 1 tableau associatif contenant des
 tableaux:
      * "ouvertures multiples" => tableau des balises concernées,

      * "non ouverte" => tableau a 2 dimensions (balise -> nombre)
      * "non fermee" => tableau a 2 dimensions (balise -> nombre)
      * "mal placee" => tableau a 2 dimensions (balise -> nombre)
      * "mal emboitee" => tableau a 2 dimensions 
      * (balise fermante -> balise ouvrante en conflit) 
      * "inconnue" => tableau des balises inconnues 
      */ 
      
      $notonlyonce = array("TABLE", "FRAMESET"); 
      $openclosetags = array( 
      "A" => 0, 
      "ADDRESS" => 0, 
      "APPLET" => 0, 
      "B" => 0, 
      "BIG" => 0, 
      "BLINK" => 0, 
      "BLOCKQUOTE" => 0, 
      "BODY" => 0, 
      "CAPTION" => 0, 
      "CENTER" => 0, 
      "CITE" => 0, 
      "CODE" => 0, 
      "COMMENT" => 0, 
      "DD" => 0, 
      "DFN" => 0, 
      "DIR" => 0, 
      "DIV" => 0, 
      "DL" => 0, 
      "DT" => 0, 
      "DT" => 0, 
      "EM" => 0, 
      "EMBED" => 0, 
      "FONT" => 0, 
      "FORM" => 0, 
      "FRAME" => 0, 
      "FRAMESET" => 0, 
      "H1" => 0, "H2" => 0, "H3" => 0, "H4" => 0, "H5" => 0, "H6" => 0,
      "HEAD" => 0, 
      "HTML" => 0, 
      "I" => 0, 
      "IFRAME" => 0, 
      "ILAYER" => 0, 
      "KBD" => 0, 
      "LAYER" => 0, 
      "LISTING" => 0, 
      "MAP" => 0, 
      "MARQUEE" => 0, 
      "MENU" => 0, 
      "MULTICOL" => 0, 
      "NOBR" => 0, 
      "NOEMBED" => 0, 
      "NOFRAMES" => 0, 
      "NOLAYER" => 0, 
      "NOSCRIPT" => 0, 
      "OBJECT" => 0, 
      "OL" => 0, 
      "OPTION" => 0, 
      "PARAM" => 0, 
      "PRE" => 0, 
      "S" => 0, 
      "SAMP" => 0, 
      "SCRIPT" => 0, 
      "SELECT" => 0, 
      "SERVER" => 0, 
      "SMALL" => 0, 
      "SPAN" => 0, 
      "STRIKE" => 0, 
      "STRONG" => 0, 
      "STYLE" => 0, 
      "SUB" => 0, 
      "SUP" => 0, 
      "TABLE" => 0, 
      "TD" => 0, 
      "TEXTAREA" => 0, 
      "TFOOT" => 0, 
      "TH" => 0, 
      "THEAD" => 0, 
      "TITLE" => 0, 
      "TR" => 0, 
      "TT" => 0, 
      "U" => 0, 
      "UL" => 0, 
      "VAR" => 0); 
      
      $alltags = array( 
      "!--", 
      "!DOCTYPE", 
      "A", "ADDRESS", "APPLET", "AREA", 
      "B", "BASE", "BASEFONT", "BGSOUND", "BIG", "BLINK", "BLOCKQUOTE", "BODY", 
"BR",
      "CAPTION", "CENTER", "CITE", "CODE", "COL", "COLGROUP", "COMMENT",
      "DD", "DFN", "DIR", "DIV", "DL", "DT", "DD", "DT", 
      "EM", "EMBED", 
      "FONT", "FORM", "FRAME", "FRAMESET", 
      "H1", "H2", "H3", "H4", "H5", "H6", "HEAD", "HR", "HTML", 
      "I", "IFRAME", "ILAYER", "IMG", "INPUT", "ISINDEX", 
      "KBD", "KEYGEN", 
      "LAYER", "LI", "LINK", "LISTING", 
      "MAP", "MARQUEE", "MENU", "META", "MULTICOL", 
      "NOBR", "NOEMBED", "NOFRAMES", "NOLAYER", "NOSCRIPT", 
      "OBJECT", "OL", "OPTION", 
      "P", "PARAM", "PLAINTEXT", "PRE", 
      "S", "STRIKE", "SAMP", "SCRIPT", "SELECT", "SELECT", "SERVER", "SMALL", 
"SPACER",
      "SPAN", "STRIKE", "STRONG", "STYLE", "SUB", "SUP", 
      "TABLE", "TBODY", "TD", "TEXTAREA", "TFOOT", "TH", "THEAD", "TITLE", "TR",
  "TT",
      "U", "UL", 
      "VAR", 
      "WBR"); 
      
      $require = array("THEAD" => "TABLE", 
      "TH" => "TABLE", 
      "TFOOT" => "TABLE", 
      "TBODY" => "TABLE", 
      "TD" => "TABLE", 
      "TR" => "TABLE", 
      "LI" => array("OL", "UL", "DIR", "MENU"), 
      "FRAME" => "FRAMSET", 
      "INPUT" => "FORM", 
      "SELECT" => "FORM", 
      "TEXTAREA" => "FORM", 
      "PARAM" => array("APPLET", "EMBED", "OBJECT"), 
      "WBR" => "NOBR", 
      "TITLE" => "HEAD", 
      "LINK" => "HEAD", 
      "OPTION" => "SELECT", 
      "DT" => "DL", 
      "DD" => "DL", 
      "HEAD" => "HTML", 
      "BODY" => "HTML"); 
      
      $mal_place = array("THEAD" => 0, 
      "TH" => 0, 
      "TFOOT" => 0, 
      "TBODY" => 0, 
      "TD" => 0, 
      "TR" => 0, 
      "LI" => 0, 
      "FRAME" => 0, 
      "INPUT" => 0, 
      "SELECT" => 0, 
      "TEXTAREA" => 0, 
      "PARAM" => 0, 
      "WBR" => 0, 
      "TITLE" => 0, 
      "LINK" => 0, 
      "OPTION" => 0, 
      "DT" => 0, 
      "DD" => 0, 
      "HEAD" => 0, 
      "BODY" => 0); 
      
      
      $retour = array("ouvertures multiples" => array(), 
      "non ouverte" => $openclosetags, 
      "non fermee" => $openclosetags, 
      "mal placee" => $mal_place, 
      "mal emboitee" => array(), 
      "inconnue" => array()); 
      $lines = @file($url); 
      
      if (!$lines) 
      return FALSE; 
      
      $file = implode(" ", $lines); 
      
      $file = preg_replace("/\s+/", " ", $file); 
      $file = preg_replace("/<!--[^>]+>/", "", $file); 
      
      preg_match_all("/<(\/?[^> ]+)[^>]*>/i", $file, $balises); 
      $pile_balise = array(); 
      for ($i=0;$i<sizeof($balises[0]);$i++) 
      { 
      $balise = strtoupper($balises[1][$i]); 
      if (!in_array($balise, $alltags) &&  !in_array(substr($balise, 1, strlen(
$balise)), $alltags) && !in_array($balise, $retour["inconnue"]))
      { 
      $retour["inconnue"][] = $balise; 
      } 
      else 
      { 
      /* Fermeture d'une balise */ 
      if (substr($balise, 0, 1) == "/") 
      { 
      $balise = substr($balise, 1, strlen($balise)); 
      if (isset($openclosetags[$balise])) 
      { 
      $openclosetags[$balise]--; 
      if (($parent = array_pop($pile_balise)) != $balise) 
      $retour["mal emboitee"][$balise] = $parent; 
      if ($openclosetags[$balise] < 0) 
      { 
      $openclosetags[$balise] = 0; 
      $retour["non ouverte"][$balise]++; 
      } 
      } 
      } 
      /* Ouverture d'une balise */ 
      else 
      { 
      if (isset($require[$balise])) 
      { 
      if (is_array($require[$balise])) 
      { 
      $ok = FALSE; 
      $j=0; 
      while ($j<sizeof($require[$balise]) && !$ok) 
      { 
      if ($openclosetags[$require[$balise][$j]] > 0) 
      $ok = TRUE; 
      $j++; 
      } 
      if (!$ok) 
      { 
      $retour["mal placee"][$balise]++; 
      } 
      } 
      else 
      { 
      if ($openclosetags[$require[$balise]] <= 0) 
      { 
      $retour["mal placee"][$balise]++; 
      } 
      } 
      } 
      if (isset($openclosetags[$balise])) 
      { 
      if (!in_array($balise, $notonlyonce) && ($openclosetags[$balise] > 0) &&  
!in_array($balise, $retour["ouvertures multiples"]))
      { 
      $retour["ouvertures multiples"][] = $balise; 
      } 
      array_push($pile_balise, $balise); 
      $openclosetags[$balise]++; 
      } 
      } 
      } 
      } 
      
      foreach ($openclosetags as $key => $value) 
      { 
      if ($value > 0) 
      $retour["non fermee"][$key] = $value; 
      } 
      return $retour; 
    }
?>