mirror of
https://github.com/ezyang/htmlpurifier.git
synced 2025-01-29 12:38:11 +01:00
Initial implementation of XHTMLDefinition, you can see it in action at the smoketest printDefinition.php?x (add the x at the end).
git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@707 48356398-32a2-884e-a903-53898d9a118a
This commit is contained in:
Edward Z. Yang
parent
806901cfd2
commit
f6b50d4bfd
80
library/HTMLPurifier/AttrCollection.php
Normal file
80
library/HTMLPurifier/AttrCollection.php
Normal file
@ -0,0 +1,80 @@
|
||||
<?php
|
||||
|
||||
require_once 'HTMLPurifier/AttrTypes.php';
|
||||
require_once 'HTMLPurifier/AttrDef/Lang.php';
|
||||
|
||||
/**
|
||||
* Defines common attribute collections that modules reference
|
||||
*/
|
||||
|
||||
class HTMLPurifier_AttrCollection
|
||||
{
|
||||
|
||||
var $info = array(
|
||||
'Core' => array(
|
||||
// 'xml:space' => false,
|
||||
'class' => 'NMTOKENS',
|
||||
'id' => 'ID',
|
||||
'title' => 'CDATA',
|
||||
),
|
||||
'I18N' => array(
|
||||
'xml:lang' => false, // see constructor
|
||||
'lang' => false, // see constructor
|
||||
),
|
||||
'Events' => array(),
|
||||
'Style' => array(),
|
||||
'Common' => array(
|
||||
0 => array('Core', 'Events', 'I18N', 'Style')
|
||||
)
|
||||
);
|
||||
|
||||
function HTMLPurifier_AttrCollection() {
|
||||
// setup direct objects
|
||||
$this->info['I18N']['xml:lang'] =
|
||||
$this->info['I18N']['lang'] = new HTMLPurifier_AttrDef_Lang();
|
||||
}
|
||||
|
||||
function setup($attr_types, $modules) {
|
||||
$info =& $this->info;
|
||||
|
||||
// replace string identifiers with actual attribute objects
|
||||
foreach ($info as $collection_i => $collection) {
|
||||
foreach ($collection as $attr_i => $attr) {
|
||||
if ($attr_i === 0) continue;
|
||||
if (!is_string($attr)) continue;
|
||||
if (isset($attr_types->info[$attr])) {
|
||||
$info[$collection_i][$attr_i] = $attr_types->info[$attr];
|
||||
} else {
|
||||
unset($info[$collection_i][$attr_i]);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// merge attribute collections that include others
|
||||
foreach ($info as $name => $attr) {
|
||||
$this->performInclusions($info[$name]);
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
function performInclusions(&$attr) {
|
||||
if (!isset($attr[0])) return;
|
||||
$merge = $attr[0];
|
||||
// loop through all the inclusions
|
||||
for ($i = 0; isset($merge[$i]); $i++) {
|
||||
// foreach attribute of the inclusion, copy it over
|
||||
foreach ($this->info[$merge[$i]] as $key => $value) {
|
||||
if (isset($attr[$key])) continue; // also catches more inclusions
|
||||
$attr[$key] = $value;
|
||||
}
|
||||
if (isset($info[$merge[$i]][0])) {
|
||||
// recursion
|
||||
$merge = array_merge($merge, isset($info[$merge[$i]][0]));
|
||||
}
|
||||
}
|
||||
unset($attr[0]);
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
?>
|
||||
21
library/HTMLPurifier/AttrTypes.php
Normal file
21
library/HTMLPurifier/AttrTypes.php
Normal file
@ -0,0 +1,21 @@
|
||||
<?php
|
||||
|
||||
require_once 'HTMLPurifier/AttrDef/Nmtokens.php';
|
||||
require_once 'HTMLPurifier/AttrDef/Text.php';
|
||||
require_once 'HTMLPurifier/AttrDef/ID.php';
|
||||
|
||||
/**
|
||||
* Provides lookup array of attribute types to HTMLPurifier_AttrDef objects
|
||||
*/
|
||||
class HTMLPurifier_AttrTypes
|
||||
{
|
||||
var $info = array();
|
||||
function HTMLPurifier_AttrTypes() {
|
||||
$this->info['NMTOKENS'] = new HTMLPurifier_AttrDef_Nmtokens();
|
||||
$this->info['CDATA'] = new HTMLPurifier_AttrDef_Text();
|
||||
$this->info['ID'] = new HTMLPurifier_AttrDef_ID();
|
||||
$this->info['URI'] = new HTMLPurifier_AttrDef_URI();
|
||||
}
|
||||
}
|
||||
|
||||
?>
|
||||
27
library/HTMLPurifier/HTMLModule.php
Normal file
27
library/HTMLPurifier/HTMLModule.php
Normal file
@ -0,0 +1,27 @@
|
||||
<?php
|
||||
|
||||
/**
|
||||
* Represents an XHTML 1.1 module, with information on elements, tags
|
||||
* and attributes.
|
||||
* @note Even though this is technically XHTML 1.1, it is also used for
|
||||
* regular HTML parsing. We are using modulization as a convenient
|
||||
* way to represent the internals of HTMLDefinition, and our
|
||||
* implementation is by no means conforming and does not directly
|
||||
* use the normative DTDs or XML schemas.
|
||||
*/
|
||||
|
||||
class HTMLPurifier_HTMLModule
|
||||
{
|
||||
var $elements = array();
|
||||
var $info = array();
|
||||
var $content_sets = array();
|
||||
}
|
||||
|
||||
class HTMLPurifier_HTMLModuleElement
|
||||
{
|
||||
var $attr = array();
|
||||
var $content_model;
|
||||
var $content_model_type = 'optional';
|
||||
}
|
||||
|
||||
?>
|
||||
51
library/HTMLPurifier/HTMLModule/Text.php
Normal file
51
library/HTMLPurifier/HTMLModule/Text.php
Normal file
@ -0,0 +1,51 @@
|
||||
<?php
|
||||
|
||||
require_once 'HTMLPurifier/HTMLModule.php';
|
||||
|
||||
/**
|
||||
* XHTML 1.1 Text Module, defines basic text containers. Core module.
|
||||
*/
|
||||
class HTMLPurifier_HTMLModule_Text extends HTMLPurifier_HTMLModule
|
||||
{
|
||||
|
||||
var $elements = array('abbr', 'acronym', 'address', 'blockquote',
|
||||
'br', 'cite', 'code', 'dfn', 'div', 'em', 'h1', 'h2', 'h3',
|
||||
'h4', 'h5', 'h6', 'kbd', 'p', 'pre', 'q', 'samp', 'span', 'strong',
|
||||
'var');
|
||||
|
||||
var $info = array();
|
||||
|
||||
var $content_sets = array(
|
||||
'Heading' => 'h1 | h2 | h3 | h4 | h5 | h6',
|
||||
'Block' => 'address | blockquote | div | p | pre',
|
||||
'Inline' => 'abbr | acronym | br | cite | code | dfn | em | kbd | q | samp | span | strong | var',
|
||||
'Flow' => 'Heading | Block | Inline'
|
||||
);
|
||||
|
||||
function HTMLPurifier_HTMLModule_Text() {
|
||||
foreach ($this->elements as $element) {
|
||||
$this->info[$element] = new HTMLPurifier_HTMLModuleElement();
|
||||
// attributes
|
||||
if ($element == 'br') {
|
||||
$this->info[$element]->attr = array(0 => array('Core'));
|
||||
} elseif ($element == 'blockquote' || $element == 'q') {
|
||||
$this->info[$element]->attr = array(0 => array('Common'), 'cite' => 'URI');
|
||||
} else {
|
||||
$this->info[$element]->attr = array(0 => array('Common'));
|
||||
}
|
||||
// content models
|
||||
if ($element == 'br') {
|
||||
$this->info[$element]->content_model_type = 'empty';
|
||||
} elseif ($element == 'blockquote') {
|
||||
$this->info[$element]->content_model_type = 'strictblockquote';
|
||||
} elseif ($element == 'div') {
|
||||
$this->info[$element]->content_model = '#PCDATA | Flow';
|
||||
} else {
|
||||
$this->info[$element]->content_model = '#PCDATA | Inline';
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
?>
|
||||
@ -1,6 +1,7 @@
|
||||
<?php
|
||||
|
||||
require_once 'HTMLPurifier/Printer.php';
|
||||
require_once 'HTMLPurifier/XHTMLDefinition.php';
|
||||
|
||||
class HTMLPurifier_Printer_HTMLDefinition extends HTMLPurifier_Printer
|
||||
{
|
||||
@ -13,7 +14,14 @@ class HTMLPurifier_Printer_HTMLDefinition extends HTMLPurifier_Printer
|
||||
function render($config) {
|
||||
$ret = '';
|
||||
$this->config =& $config;
|
||||
$this->def = $config->getHTMLDefinition();
|
||||
|
||||
if (isset($_GET['x'])) { // hidden settings
|
||||
$this->def = new HTMLPurifier_XHTMLDefinition();
|
||||
$this->def->initialize($config);
|
||||
$this->def->setup($config);
|
||||
} else {
|
||||
$this->def = $config->getHTMLDefinition();
|
||||
}
|
||||
$def =& $this->def;
|
||||
|
||||
$ret .= $this->start('div', array('class' => 'HTMLPurifier_Printer'));
|
||||
|
||||
111
library/HTMLPurifier/XHTMLDefinition.php
Normal file
111
library/HTMLPurifier/XHTMLDefinition.php
Normal file
@ -0,0 +1,111 @@
|
||||
<?php
|
||||
|
||||
require_once 'HTMLPurifier/HTMLDefinition.php';
|
||||
|
||||
require_once 'HTMLPurifier/AttrTypes.php';
|
||||
require_once 'HTMLPurifier/AttrCollection.php';
|
||||
require_once 'HTMLPurifier/HTMLModule.php';
|
||||
require_once 'HTMLPurifier/HTMLModule/Text.php';
|
||||
|
||||
/**
|
||||
* Next-generation HTML definition that will supplant HTMLPurifier_HTMLDefinition
|
||||
*/
|
||||
class HTMLPurifier_XHTMLDefinition extends HTMLPurifier_HTMLDefinition
|
||||
{
|
||||
|
||||
var $modules = array();
|
||||
var $attr_types;
|
||||
var $attr_collection;
|
||||
|
||||
function initialize($config) {
|
||||
|
||||
$this->modules['Text'] = new HTMLPurifier_HTMLModule_Text();
|
||||
|
||||
$this->attr_types = new HTMLPurifier_AttrTypes();
|
||||
$this->attr_collection = new HTMLPurifier_AttrCollection();
|
||||
|
||||
}
|
||||
|
||||
function setup($config) {
|
||||
|
||||
// perform attribute collection substitutions
|
||||
$this->attr_collection->setup($this->attr_types, $this->modules);
|
||||
|
||||
// populate content_sets based on module hints
|
||||
$content_sets = array();
|
||||
foreach ($this->modules as $module_i => $module) {
|
||||
foreach ($module->content_sets as $key => $value) {
|
||||
if (isset($content_sets[$key])) {
|
||||
// add it into the existing content set
|
||||
$content_sets[$key] = $content_sets[$key] . ' | ' . $value;
|
||||
} else {
|
||||
$content_sets[$key] = $value;
|
||||
}
|
||||
}
|
||||
}
|
||||
// perform content_set expansions
|
||||
foreach ($content_sets as $i => $set) {
|
||||
// only performed once, so infinite recursion is not
|
||||
// a problem, you'll just have a stray $Set lying around
|
||||
// at the end
|
||||
$content_sets[$i] =
|
||||
str_replace(
|
||||
array_keys($content_sets),
|
||||
array_values($content_sets),
|
||||
$set);
|
||||
}
|
||||
$content_sets_keys = array_keys($content_sets);
|
||||
$content_sets_values = array_values($content_sets);
|
||||
|
||||
foreach ($this->modules as $module_i => $module) {
|
||||
foreach ($module->info as $element_i => $element) {
|
||||
$element =& $this->modules[$module_i]->info[$element_i];
|
||||
|
||||
// attribute value expansions
|
||||
$this->attr_collection->performInclusions($element->attr);
|
||||
|
||||
// perform content model expansions
|
||||
$content_model = $element->content_model;
|
||||
if (is_string($content_model)) {
|
||||
$element->content_model = str_replace(
|
||||
$content_sets_keys, $content_sets_values, $content_model);
|
||||
}
|
||||
|
||||
// setup info
|
||||
$this->info[$element_i] = new HTMLPurifier_ElementDef();
|
||||
$this->info[$element_i]->attr = $element->attr;
|
||||
$this->info[$element_i]->child = $this->getChildDef($element);
|
||||
if ($this->info_parent == $element_i) {
|
||||
$this->info_parent_def = $this->info[$element_i];
|
||||
}
|
||||
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
function getChildDef($element) {
|
||||
$value = $element->content_model;
|
||||
$type = $element->content_model_type;
|
||||
switch ($type) {
|
||||
case 'required':
|
||||
return new HTMLPurifier_ChildDef_Required($value);
|
||||
case 'optional':
|
||||
return new HTMLPurifier_ChildDef_Optional($value);
|
||||
case 'empty':
|
||||
return new HTMLPurifier_ChildDef_Empty();
|
||||
case 'strictblockquote':
|
||||
return new HTMLPurifier_ChildDef_StrictBlockquote();
|
||||
case 'table':
|
||||
return new HTMLPurifier_ChildDef_Table();
|
||||
case 'chameleon':
|
||||
return new HTMLPurifier_ChildDef_Chameleon($value[0], $value[1]);
|
||||
case 'custom':
|
||||
return new HTMLPurifier_ChildDef_Custom($value);
|
||||
}
|
||||
return HTMLPurifier_ChildDef_Empty();
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
?>
|
||||
Loading…
x
Reference in New Issue
Block a user