aboutsummaryrefslogtreecommitdiffstats
path: root/library/HTMLPurifier/ChildDef/Table.php
diff options
context:
space:
mode:
authorfriendica <info@friendica.com>2015-01-01 22:18:27 -0800
committerfriendica <info@friendica.com>2015-01-01 22:18:27 -0800
commita0052f0176bd079e6a94baec59fea2ec5a8d651e (patch)
treec323edd823681bc2e8ca757e7eaf8354d42c7b51 /library/HTMLPurifier/ChildDef/Table.php
parent545e47933a0816699c68d98a7742a03260d6a54f (diff)
downloadvolse-hubzilla-a0052f0176bd079e6a94baec59fea2ec5a8d651e.tar.gz
volse-hubzilla-a0052f0176bd079e6a94baec59fea2ec5a8d651e.tar.bz2
volse-hubzilla-a0052f0176bd079e6a94baec59fea2ec5a8d651e.zip
htmlpurifier update - compatibility issue with language library autoloader
Diffstat (limited to 'library/HTMLPurifier/ChildDef/Table.php')
-rw-r--r--library/HTMLPurifier/ChildDef/Table.php302
1 files changed, 192 insertions, 110 deletions
diff --git a/library/HTMLPurifier/ChildDef/Table.php b/library/HTMLPurifier/ChildDef/Table.php
index 34f0227dd..3e4a0f218 100644
--- a/library/HTMLPurifier/ChildDef/Table.php
+++ b/library/HTMLPurifier/ChildDef/Table.php
@@ -1,140 +1,222 @@
<?php
/**
- * Definition for tables
+ * Definition for tables. The general idea is to extract out all of the
+ * essential bits, and then reconstruct it later.
+ *
+ * This is a bit confusing, because the DTDs and the W3C
+ * validators seem to disagree on the appropriate definition. The
+ * DTD claims:
+ *
+ * (CAPTION?, (COL*|COLGROUP*), THEAD?, TFOOT?, TBODY+)
+ *
+ * But actually, the HTML4 spec then has this to say:
+ *
+ * The TBODY start tag is always required except when the table
+ * contains only one table body and no table head or foot sections.
+ * The TBODY end tag may always be safely omitted.
+ *
+ * So the DTD is kind of wrong. The validator is, unfortunately, kind
+ * of on crack.
+ *
+ * The definition changed again in XHTML1.1; and in my opinion, this
+ * formulation makes the most sense.
+ *
+ * caption?, ( col* | colgroup* ), (( thead?, tfoot?, tbody+ ) | ( tr+ ))
+ *
+ * Essentially, we have two modes: thead/tfoot/tbody mode, and tr mode.
+ * If we encounter a thead, tfoot or tbody, we are placed in the former
+ * mode, and we *must* wrap any stray tr segments with a tbody. But if
+ * we don't run into any of them, just have tr tags is OK.
*/
class HTMLPurifier_ChildDef_Table extends HTMLPurifier_ChildDef
{
+ /**
+ * @type bool
+ */
public $allow_empty = false;
+
+ /**
+ * @type string
+ */
public $type = 'table';
- public $elements = array('tr' => true, 'tbody' => true, 'thead' => true,
- 'tfoot' => true, 'caption' => true, 'colgroup' => true, 'col' => true);
- public function __construct() {}
- public function validateChildren($tokens_of_children, $config, $context) {
- if (empty($tokens_of_children)) return false;
- // this ensures that the loop gets run one last time before closing
- // up. It's a little bit of a hack, but it works! Just make sure you
- // get rid of the token later.
- $tokens_of_children[] = false;
+ /**
+ * @type array
+ */
+ public $elements = array(
+ 'tr' => true,
+ 'tbody' => true,
+ 'thead' => true,
+ 'tfoot' => true,
+ 'caption' => true,
+ 'colgroup' => true,
+ 'col' => true
+ );
+
+ public function __construct()
+ {
+ }
+
+ /**
+ * @param array $children
+ * @param HTMLPurifier_Config $config
+ * @param HTMLPurifier_Context $context
+ * @return array
+ */
+ public function validateChildren($children, $config, $context)
+ {
+ if (empty($children)) {
+ return false;
+ }
// only one of these elements is allowed in a table
$caption = false;
- $thead = false;
- $tfoot = false;
+ $thead = false;
+ $tfoot = false;
+
+ // whitespace
+ $initial_ws = array();
+ $after_caption_ws = array();
+ $after_thead_ws = array();
+ $after_tfoot_ws = array();
// as many of these as you want
- $cols = array();
+ $cols = array();
$content = array();
- $nesting = 0; // current depth so we can determine nodes
- $is_collecting = false; // are we globbing together tokens to package
- // into one of the collectors?
- $collection = array(); // collected nodes
- $tag_index = 0; // the first node might be whitespace,
- // so this tells us where the start tag is
-
- foreach ($tokens_of_children as $token) {
- $is_child = ($nesting == 0);
-
- if ($token === false) {
- // terminating sequence started
- } elseif ($token instanceof HTMLPurifier_Token_Start) {
- $nesting++;
- } elseif ($token instanceof HTMLPurifier_Token_End) {
- $nesting--;
- }
+ $tbody_mode = false; // if true, then we need to wrap any stray
+ // <tr>s with a <tbody>.
- // handle node collection
- if ($is_collecting) {
- if ($is_child) {
- // okay, let's stash the tokens away
- // first token tells us the type of the collection
- switch ($collection[$tag_index]->name) {
- case 'tr':
- case 'tbody':
- $content[] = $collection;
- break;
- case 'caption':
- if ($caption !== false) break;
- $caption = $collection;
- break;
- case 'thead':
- case 'tfoot':
- // access the appropriate variable, $thead or $tfoot
- $var = $collection[$tag_index]->name;
- if ($$var === false) {
- $$var = $collection;
- } else {
- // transmutate the first and less entries into
- // tbody tags, and then put into content
- $collection[$tag_index]->name = 'tbody';
- $collection[count($collection)-1]->name = 'tbody';
- $content[] = $collection;
- }
- break;
- case 'colgroup':
- $cols[] = $collection;
- break;
- }
- $collection = array();
- $is_collecting = false;
- $tag_index = 0;
+ $ws_accum =& $initial_ws;
+
+ foreach ($children as $node) {
+ if ($node instanceof HTMLPurifier_Node_Comment) {
+ $ws_accum[] = $node;
+ continue;
+ }
+ switch ($node->name) {
+ case 'tbody':
+ $tbody_mode = true;
+ // fall through
+ case 'tr':
+ $content[] = $node;
+ $ws_accum =& $content;
+ break;
+ case 'caption':
+ // there can only be one caption!
+ if ($caption !== false) break;
+ $caption = $node;
+ $ws_accum =& $after_caption_ws;
+ break;
+ case 'thead':
+ $tbody_mode = true;
+ // XXX This breaks rendering properties with
+ // Firefox, which never floats a <thead> to
+ // the top. Ever. (Our scheme will float the
+ // first <thead> to the top.) So maybe
+ // <thead>s that are not first should be
+ // turned into <tbody>? Very tricky, indeed.
+ if ($thead === false) {
+ $thead = $node;
+ $ws_accum =& $after_thead_ws;
} else {
- // add the node to the collection
- $collection[] = $token;
+ // Oops, there's a second one! What
+ // should we do? Current behavior is to
+ // transmutate the first and last entries into
+ // tbody tags, and then put into content.
+ // Maybe a better idea is to *attach
+ // it* to the existing thead or tfoot?
+ // We don't do this, because Firefox
+ // doesn't float an extra tfoot to the
+ // bottom like it does for the first one.
+ $node->name = 'tbody';
+ $content[] = $node;
+ $ws_accum =& $content;
}
- }
-
- // terminate
- if ($token === false) break;
-
- if ($is_child) {
- // determine what we're dealing with
- if ($token->name == 'col') {
- // the only empty tag in the possie, we can handle it
- // immediately
- $cols[] = array_merge($collection, array($token));
- $collection = array();
- $tag_index = 0;
- continue;
+ break;
+ case 'tfoot':
+ // see above for some aveats
+ $tbody_mode = true;
+ if ($tfoot === false) {
+ $tfoot = $node;
+ $ws_accum =& $after_tfoot_ws;
+ } else {
+ $node->name = 'tbody';
+ $content[] = $node;
+ $ws_accum =& $content;
}
- switch($token->name) {
- case 'caption':
- case 'colgroup':
- case 'thead':
- case 'tfoot':
- case 'tbody':
- case 'tr':
- $is_collecting = true;
- $collection[] = $token;
- continue;
- default:
- if (!empty($token->is_whitespace)) {
- $collection[] = $token;
- $tag_index++;
- }
- continue;
+ break;
+ case 'colgroup':
+ case 'col':
+ $cols[] = $node;
+ $ws_accum =& $cols;
+ break;
+ case '#PCDATA':
+ // How is whitespace handled? We treat is as sticky to
+ // the *end* of the previous element. So all of the
+ // nonsense we have worked on is to keep things
+ // together.
+ if (!empty($node->is_whitespace)) {
+ $ws_accum[] = $node;
}
+ break;
}
}
- if (empty($content)) return false;
-
- $ret = array();
- if ($caption !== false) $ret = array_merge($ret, $caption);
- if ($cols !== false) foreach ($cols as $token_array) $ret = array_merge($ret, $token_array);
- if ($thead !== false) $ret = array_merge($ret, $thead);
- if ($tfoot !== false) $ret = array_merge($ret, $tfoot);
- foreach ($content as $token_array) $ret = array_merge($ret, $token_array);
- if (!empty($collection) && $is_collecting == false){
- // grab the trailing space
- $ret = array_merge($ret, $collection);
+ if (empty($content)) {
+ return false;
+ }
+
+ $ret = $initial_ws;
+ if ($caption !== false) {
+ $ret[] = $caption;
+ $ret = array_merge($ret, $after_caption_ws);
+ }
+ if ($cols !== false) {
+ $ret = array_merge($ret, $cols);
+ }
+ if ($thead !== false) {
+ $ret[] = $thead;
+ $ret = array_merge($ret, $after_thead_ws);
+ }
+ if ($tfoot !== false) {
+ $ret[] = $tfoot;
+ $ret = array_merge($ret, $after_tfoot_ws);
}
- array_pop($tokens_of_children); // remove phantom token
+ if ($tbody_mode) {
+ // we have to shuffle tr into tbody
+ $current_tr_tbody = null;
+
+ foreach($content as $node) {
+ switch ($node->name) {
+ case 'tbody':
+ $current_tr_tbody = null;
+ $ret[] = $node;
+ break;
+ case 'tr':
+ if ($current_tr_tbody === null) {
+ $current_tr_tbody = new HTMLPurifier_Node_Element('tbody');
+ $ret[] = $current_tr_tbody;
+ }
+ $current_tr_tbody->children[] = $node;
+ break;
+ case '#PCDATA':
+ assert($node->is_whitespace);
+ if ($current_tr_tbody === null) {
+ $ret[] = $node;
+ } else {
+ $current_tr_tbody->children[] = $node;
+ }
+ break;
+ }
+ }
+ } else {
+ $ret = array_merge($ret, $content);
+ }
- return ($ret === $tokens_of_children) ? true : $ret;
+ return $ret;
}
}