<?php
/* Class to verify that HTML output conforms to strict HTML v4.01 spec.
 *
 * Written by: Chris Studholme
 * Copyright:  GPL (http://www.fsf.org/copyleft/gpl.html)
 * $Id: html_verifier.php,v 1.1 2003/12/23 06:07:15 cstudhol Exp $
 */

class Tag {
  var $singleton;   // (boolean) no end tag?
  var $attributes;  // (array of strings) value attributes
  var $max_nesting; // (int) maximum nesting level (0=infinite)
  //    var $requires;    // (string[]) 
  //    var $requires_immediate;
  //    var $allows;
  
  var $nesting;     // (int) current nesting level
  
  function Tag($attributes,
	       $singleton=false,
	       $max_nesting=1) {
    $this->singleton = $singleton;
    $this->max_nesting = $max_nesting;
    $this->nesting = 0;
    $this->attributes = array();
    reset($attributes);
    while (list($key,$value)=each($attributes))
      $this->attributes[$value]=true;
  }
}


class HTMLVerifier {
  // master tag list
  var $tag_list = array();

  // attributes valid in all tags
  var $global_attributes = array();
	      
  // stack to keep track of html elements
  var $tag_stack = array();
  var $tag_stack_top = -1;
  
  var $errors = array();
  
  function HTMLVerifier() {
    $this->global_attributes =
      array("style"=>true,
	    "id"=>true);

    $this->tag_list =
      array("a" => new Tag(array("href","id","name","onMouseOver")),
	    "b" => new Tag(array()),
	    "blockquote" => new Tag(array()),
	    "body" => new Tag(array()),
	    "div" => new Tag(array()),
	    "form" => new Tag(array("action","method","name")),
	    "h1" => new Tag(array()),
	    "h2" => new Tag(array()),
	    "h3" => new Tag(array()),
	    "head" => new Tag(array()),
	    "html" => new Tag(array()),
	    "i" => new Tag(array()),
	    "li" => new Tag(array()),
	    "nobr" => new Tag(array()),
	    "ol" => new Tag(array()),
	    "option" => new Tag(array("value","selected")),
	    "p" => new Tag(array()),
	    "script" => new Tag(array("type")),
	    "select" => new Tag(array("multiple","name","onChange","size")),
	    "span" => new Tag(array()),
	    "strong" => new Tag(array()),
	    "table" => new Tag(array("border","cellspacing","cellpadding","width"),false,0),
	    "td" => new Tag(array("align","colspan","valign")),
	    "textarea" => new Tag(array("cols","name","rows")),
	    "th" => new Tag(array("align","colspan","valign")),
	    "title" => new Tag(array()),
	    "tr" => new Tag(array()),
	    "ul" => new Tag(array()),
	    
	    // singleton tags
	    "br" => new Tag(array(),true),
	    "hr" => new Tag(array(),true),
	    "img" => new Tag(array("alt","height","src","width"),true),
	    "input" => new Tag(array("checked","maxlength","name","onClick","src","size","type","value"),true),
	    "link" => new Tag(array("href","rel","type"),true),
	    );
  }
  

  function tag_check($tag,$attributes,$singleton=false,$result="") {
    $result=ereg_replace("[<]","&lt;",ereg_replace("[>]","&gt;",$result));
    $tagobj=$this->tag_list[$tag];
    // check tag
    if (!$tagobj) {
      $this->errors[]="Invalid tag ".$result;
      return;
    }
    if ($tagobj->singleton!=$singleton) {
      $this->errors[]="Tag is incorrect type ".$result;
    }
    // check nesting level
    if ($tagobj->max_nesting&&($tagobj->max_nesting==$tagobj->nesting)) {
      $this->errors[]="Tag nested too deep ".$result;
    }
    // check attributes
    if (is_array($attributes)) {
      reset($attributes);
      while (list($key,$value)=each($attributes)) {
	if (!$tagobj->attributes[$key]&&!$this->global_attributes[$key])
	  $this->errors[]="Invalid attribute '".$key."=".$value.
	    "' for tag ".$result;
	else if (ereg("[<>\]",$value))
	  $this->errors[]="Invalid attribute '".$key."=".$value.
	    "' for tag ".$result;
	else if (ereg("\"",$value)) {
	  if (ereg("'",$value))
	    $this->errors[]="Too many quote types in value '".$value.
	      "' for tag ".$result;
	  else
	    $this->errors[]="WARNING: double quote used in value '".$value.
	      "' for tag ".$result;
	}
      }
    }
  }
  
  function check_tag($tag,$attributes,$content,$result="") {
    $this->tag_check($tag,$attributes,false,$result);
  }
  
  function check_stag($tag,$attributes,$result="") {
    $this->tag_check($tag,$attributes,true,$result);
  }
  
  function check_btag($tag,$attributes,$result="") {
    $this->tag_check($tag,$attributes,false,$result);
    $this->tag_stack[++$this->tag_stack_top]=$tag;
    $tagobj=$this->tag_list[$tag];
    if ($tagobj)
      $tagobj->nesting++;
  }
  
  function check_etag($tag,$result="") {
    if ($this->tag_stack_top<0)
      $this->errors[]="Tag stack is empty";
    else if ($this->tag_stack[$this->tag_stack_top--]!=$tag) {
      $result = "Overlapping or missing tags: expected '".$tag.
	"', current stack:";
      for ($i=$this->tag_stack_top+1; $i>=0; --$i)
	$result .= " ".$this->tag_stack[$i];
      $this->errors[]=$result;
      // remove tags until we find the one we want
      while (($this->tag_stack[$this->tag_stack_top--]!=$tag)&&
	     ($this->tag_stack_top>=0)) ;
    }
    $tagobj=$this->tag_list[$tag];
    if ($tagobj)
      $tagobj->nesting--;
  }
  
  function report() {
    echo "Maximum stack depth: ".count($this->tag_stack).stag("br")."\n";
    
    // check for non-empy html stack
    if ($this->tag_stack_top>=0) {
      $err="Tags left on stack:";
      for ($i=$this->tag_stack_top; $i>=0; --$i)
	$err.=" ".$this->tag_stack[$i];
      $this->errors[] = $err;
    }
    
    // display html errors
    if ($this->errors) {
      reset($this->errors);
      while (list($key,$value)=each($this->errors)) {
	echo tag("strong",false,$value).stag("br")."\n";
      }
    }
  }
}
?>