protected function setUp()
{
parent::setUp();
if (!MWTidy::isEnabled()) {
$this->markTestSkipped('Tidy not found');
}
}
/**
* Helper function for parse() that transforms half-parsed HTML into fully
* parsed HTML.
*
* @param string $text
* @param bool $isMain
* @param bool $linestart
* @return string
*/
private function internalParseHalfParsed($text, $isMain = true, $linestart = true)
{
$text = $this->mStripState->unstripGeneral($text);
if ($isMain) {
Hooks::run('ParserAfterUnstrip', array(&$this, &$text));
}
# Clean up special characters, only run once, next-to-last before doBlockLevels
$fixtags = array('/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1 ', '/(\\302\\253) /' => '\\1 ', '/ (!\\s*important)/' => ' \\1');
$text = preg_replace(array_keys($fixtags), array_values($fixtags), $text);
$text = $this->doBlockLevels($text, $linestart);
$this->replaceLinkHolders($text);
/**
* The input doesn't get language converted if
* a) It's disabled
* b) Content isn't converted
* c) It's a conversion table
* d) it is an interface message (which is in the user language)
*/
if (!($this->mOptions->getDisableContentConversion() || isset($this->mDoubleUnderscores['nocontentconvert']))) {
if (!$this->mOptions->getInterfaceMessage()) {
# The position of the convert() call should not be changed. it
# assumes that the links are all replaced and the only thing left
# is the <nowiki> mark.
$text = $this->getConverterLanguage()->convert($text);
}
}
$text = $this->mStripState->unstripNoWiki($text);
if ($isMain) {
Hooks::run('ParserBeforeTidy', array(&$this, &$text));
}
$text = $this->replaceTransparentTags($text);
$text = $this->mStripState->unstripGeneral($text);
$text = Sanitizer::normalizeCharReferences($text);
if (MWTidy::isEnabled() && $this->mOptions->getTidy()) {
$text = MWTidy::tidy($text);
} else {
# attempt to sanitize at least some nesting problems
# (bug #2702 and quite a few others)
$tidyregs = array('/(<([bi])>)(<([bi])>)?([^<]*)(<\\/?a[^<]*>)([^<]*)(<\\/\\4>)?(<\\/\\2>)/' => '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9', '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\\/a>(.*)<\\/a>/' => '\\1\\2</a>\\3</a>\\1\\4</a>', '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\\/div>)([^<]*)(<\\/\\2>)/' => '\\1\\3<div\\5>\\6</div>\\8\\9', '/<([bi])><\\/\\1>/' => '');
$text = preg_replace(array_keys($tidyregs), array_values($tidyregs), $text);
}
if ($isMain) {
Hooks::run('ParserAfterTidy', array(&$this, &$text));
}
return $text;
}
/**
* Cleans up HTML, removes dangerous tags and attributes, and
* removes HTML comments
* @param string $text
* @param callable $processCallback Callback to do any variable or parameter
* replacements in HTML attribute values
* @param array|bool $args Arguments for the processing callback
* @param array $extratags For any extra tags to include
* @param array $removetags For any tags (default or extra) to exclude
* @return string
*/
public static function removeHTMLtags($text, $processCallback = null, $args = array(), $extratags = array(), $removetags = array())
{
extract(self::getRecognizedTagData($extratags, $removetags));
# Remove HTML comments
$text = Sanitizer::removeHTMLcomments($text);
$bits = explode('<', $text);
$text = str_replace('>', '>', array_shift($bits));
if (!MWTidy::isEnabled()) {
$tagstack = $tablestack = array();
foreach ($bits as $x) {
$regs = array();
# $slash: Does the current element start with a '/'?
# $t: Current element name
# $params: String between element name and >
# $brace: Ending '>' or '/>'
# $rest: Everything until the next element of $bits
if (preg_match(self::ELEMENT_BITS_REGEX, $x, $regs)) {
list(, $slash, $t, $params, $brace, $rest) = $regs;
} else {
$slash = $t = $params = $brace = $rest = null;
}
$badtag = false;
if (isset($htmlelements[$t = strtolower($t)])) {
# Check our stack
if ($slash && isset($htmlsingleonly[$t])) {
$badtag = true;
} elseif ($slash) {
# Closing a tag... is it the one we just opened?
MediaWiki\suppressWarnings();
$ot = array_pop($tagstack);
MediaWiki\restoreWarnings();
if ($ot != $t) {
if (isset($htmlsingleallowed[$ot])) {
# Pop all elements with an optional close tag
# and see if we find a match below them
$optstack = array();
array_push($optstack, $ot);
MediaWiki\suppressWarnings();
$ot = array_pop($tagstack);
MediaWiki\restoreWarnings();
while ($ot != $t && isset($htmlsingleallowed[$ot])) {
array_push($optstack, $ot);
MediaWiki\suppressWarnings();
$ot = array_pop($tagstack);
MediaWiki\restoreWarnings();
}
if ($t != $ot) {
# No match. Push the optional elements back again
$badtag = true;
MediaWiki\suppressWarnings();
$ot = array_pop($optstack);
MediaWiki\restoreWarnings();
while ($ot) {
array_push($tagstack, $ot);
MediaWiki\suppressWarnings();
$ot = array_pop($optstack);
MediaWiki\restoreWarnings();
}
}
} else {
MediaWiki\suppressWarnings();
array_push($tagstack, $ot);
MediaWiki\restoreWarnings();
# <li> can be nested in <ul> or <ol>, skip those cases:
if (!isset($htmllist[$ot]) || !isset($listtags[$t])) {
$badtag = true;
}
}
} else {
if ($t == 'table') {
$tagstack = array_pop($tablestack);
}
}
$newparams = '';
} else {
# Keep track for later
if (isset($tabletags[$t]) && !in_array('table', $tagstack)) {
$badtag = true;
} elseif (in_array($t, $tagstack) && !isset($htmlnest[$t])) {
$badtag = true;
# Is it a self closed htmlpair ? (bug 5487)
} elseif ($brace == '/>' && isset($htmlpairs[$t])) {
$badtag = true;
} elseif (isset($htmlsingleonly[$t])) {
# Hack to force empty tag for unclosable elements
$brace = '/>';
} elseif (isset($htmlsingle[$t])) {
# Hack to not close $htmlsingle tags
$brace = null;
# Still need to push this optionally-closed tag to
# the tag stack so that we can match end tags
# instead of marking them as bad.
array_push($tagstack, $t);
} elseif (isset($tabletags[$t]) && in_array($t, $tagstack)) {
// New table tag but forgot to close the previous one
$text .= "</{$t}>";
} else {
if ($t == 'table') {
array_push($tablestack, $tagstack);
$tagstack = array();
}
array_push($tagstack, $t);
}
# Replace any variables or template parameters with
# plaintext results.
if (is_callable($processCallback)) {
call_user_func_array($processCallback, array(&$params, $args));
}
if (!Sanitizer::validateTag($params, $t)) {
$badtag = true;
}
# Strip non-approved attributes from the tag
$newparams = Sanitizer::fixTagAttributes($params, $t);
}
if (!$badtag) {
$rest = str_replace('>', '>', $rest);
$close = $brace == '/>' && !$slash ? ' /' : '';
$text .= "<{$slash}{$t}{$newparams}{$close}>{$rest}";
continue;
}
}
$text .= '<' . str_replace('>', '>', $x);
}
# Close off any remaining tags
while (is_array($tagstack) && ($t = array_pop($tagstack))) {
$text .= "</{$t}>\n";
if ($t == 'table') {
$tagstack = array_pop($tablestack);
}
}
} else {
# this might be possible using tidy itself
foreach ($bits as $x) {
if (preg_match(self::ELEMENT_BITS_REGEX, $x, $regs)) {
list(, $slash, $t, $params, $brace, $rest) = $regs;
$badtag = false;
if (isset($htmlelements[$t = strtolower($t)])) {
if (is_callable($processCallback)) {
call_user_func_array($processCallback, array(&$params, $args));
}
if (!Sanitizer::validateTag($params, $t)) {
$badtag = true;
}
$newparams = Sanitizer::fixTagAttributes($params, $t);
if (!$badtag) {
$rest = str_replace('>', '>', $rest);
$text .= "<{$slash}{$t}{$newparams}{$brace}{$rest}";
continue;
}
}
}
$text .= '<' . str_replace('>', '>', $x);
}
}
return $text;
}
/**
* Asserts that the given string is valid HTML document.
*
* @since 1.23
*
* @note Will mark the test as skipped if the "tidy" module is not installed.
* @note This ignores $wgUseTidy, so we can check for valid HTML even (and especially)
* when automatic tidying is disabled.
*
* @param string $html A complete HTML document
*/
protected function assertValidHtmlDocument($html)
{
// Note: we only validate if the tidy PHP extension is available.
// In case wgTidyInternal is false, MWTidy would fall back to the command line version
// of tidy. In that case however, we can not reliably detect whether a failing validation
// is due to malformed HTML, or caused by tidy not being installed as a command line tool.
// That would cause all HTML assertions to fail on a system that has no tidy installed.
if (!$GLOBALS['wgTidyInternal'] || !MWTidy::isEnabled()) {
$this->markTestSkipped('Tidy extension not installed');
}
$errorBuffer = '';
MWTidy::checkErrors($html, $errorBuffer);
$allErrors = preg_split('/[\\r\\n]+/', $errorBuffer);
// Filter Tidy warnings which aren't useful for us.
// Tidy eg. often cries about parameters missing which have actually
// been deprecated since HTML4, thus we should not care about them.
$errors = preg_grep('/^(.*Warning: (trimming empty|.* lacks ".*?" attribute).*|\\s*)$/m', $allErrors, PREG_GREP_INVERT);
$this->assertEmpty($errors, implode("\n", $errors));
}
