function tidy_html($html)
{
$tidy_config = array('output-xhtml' => true, 'show-body-only' => true);
$tidy = tidy_parse_string($html, $tidy_config, 'UTF8');
$tidy->cleanRepair();
return tidy_get_output($tidy);
}
/**
* Filter a content item's content
*
* @return string
*/
function filter($item, $field = "content", $length = 0)
{
$nodefilters = array();
if (is_a($item, 'Zoo_Content_Interface')) {
$txt = $item->{$field};
$nodefilters = Zoo::getService('content')->getFilters($item);
} else {
$txt = $item;
}
if ($length > 0) {
$txt = substr($txt, 0, $length);
}
if (count($nodefilters)) {
$ids = array();
foreach ($nodefilters as $nodefilter) {
$ids[] = $nodefilter->filter_id;
}
$filters = Zoo::getService('filter')->getFilters($ids);
foreach ($filters as $filter) {
$txt = $filter->filter($txt);
}
if (extension_loaded('tidy')) {
$config = array('indent' => TRUE, 'show-body-only' => TRUE, 'output-xhtml' => TRUE, 'wrap' => 0);
$tidy = tidy_parse_string($txt, $config, 'UTF8');
$tidy->cleanRepair();
$txt = tidy_get_output($tidy);
}
} else {
$txt = htmlspecialchars($txt);
}
return $txt;
}
/**
* Use the HTML tidy extension to use the tidy library in-process,
* saving the overhead of spawning a new process.
*
* @param string $text HTML to check
* @param bool $stderr Whether to read result from error status instead of output
* @param int &$retval Exit code (-1 on internal error)
* @return string|null
*/
protected function cleanWrapped($text, $stderr = false, &$retval = null)
{
if (!class_exists('tidy')) {
wfWarn("Unable to load internal tidy class.");
$retval = -1;
return null;
}
$tidy = new \tidy();
$tidy->parseString($text, $this->config['tidyConfigFile'], 'utf8');
if ($stderr) {
$retval = $tidy->getStatus();
return $tidy->errorBuffer;
}
$tidy->cleanRepair();
$retval = $tidy->getStatus();
if ($retval == 2) {
// 2 is magic number for fatal error
// http://www.php.net/manual/en/function.tidy-get-status.php
$cleansource = null;
} else {
$cleansource = tidy_get_output($tidy);
if (!empty($this->config['debugComment']) && $retval > 0) {
$cleansource .= "<!--\nTidy reports:\n" . str_replace('-->', '-->', $tidy->errorBuffer) . "\n-->";
}
}
return $cleansource;
}
/**
* Filter through Tidy
*
* @param array
* @param string
* @param bool
* @return bool
*/
public static function tidy(array &$headers, &$body, $uncached)
{
$tidy = new tidy();
$tidy->parseString($body, array('clean' => 1, 'bare' => 1, 'hide-comments' => 1, 'doctype' => 'omit', 'indent-spaces' => 0, 'tab-size' => 0, 'wrap' => 0, 'quote-ampersand' => 0, 'output-xhtml' => true, 'quiet' => 1), 'utf8');
$tidy->cleanRepair();
$body = tidy_get_output($tidy);
}
function run_loop($aHtml, $aCount)
{
for ($i = 0; $i < $aCount; $i++) {
$tidy = new tidy();
$tidy->parseString($aHtml, array('output-xml' => true, 'clean' => true, 'numeric-entities' => true), 'utf8');
$tidy->cleanRepair();
//echo tidy_get_output( $tidy );
$xml = simplexml_load_string(tidy_get_output($tidy));
if ($xml === false) {
file_put_contents('php://stderr', 'Unable to parse file');
return;
}
unset($xml);
unset($tidy);
}
}
/**
* Turn a string or array into valid, standards-compliant (x)HTML
*
* Uses configuraton options in tidy.conf - which should minimally have show-body-only set to yes
*
* @param mixed $text The data to be tidied up
* @return mixed $result Tidied data
*/
function tidy($text)
{
static $tidy_funcs;
static $tidy_conf;
if (!isset($tidy_conf)) {
$tidy_conf = SETTINGS_INC . 'tidy.conf';
}
if (is_array($text)) {
$result = array();
foreach (array_keys($text) as $key) {
$result[$key] = tidy($text[$key]);
}
return $result;
}
// determine what tidy libraries are available
if (empty($tidy_funcs)) {
$tidy_funcs = get_extension_funcs('tidy');
}
$tidy_1_lib_available = !empty($tidy_funcs) && array_search('tidy_setopt', $tidy_funcs) !== false;
$tidy_2_lib_available = !empty($tidy_funcs) && array_search('tidy_setopt', $tidy_funcs) === false;
$tidy_command_line_available = TIDY_EXE ? file_exists(TIDY_EXE) : false;
$text = protect_string_from_tidy($text);
$text = '<html><body>' . $text . '</body></html>';
if ($tidy_2_lib_available) {
$tidy = new tidy();
$tidy->parseString($text, $tidy_conf, 'utf8');
$tidy->cleanRepair();
$result = $tidy;
} elseif ($tidy_1_lib_available) {
tidy_load_config($tidy_conf);
tidy_set_encoding('utf8');
tidy_parse_string($text);
tidy_clean_repair();
$result = tidy_get_output();
} elseif ($tidy_command_line_available) {
$arg = escapeshellarg($text);
// escape the bad stuff in the text
$cmd = 'echo ' . $arg . ' | ' . TIDY_EXE . ' -q -config ' . $tidy_conf . ' 2> /dev/null';
// the actual command - pipes the input to tidy which diverts its output to the random file
$result = shell_exec($cmd);
// execute the command
} else {
trigger_error('tidy does not appear to be available within php or at the command line - no tidying is taking place.');
$result = $text;
}
return trim($result);
}
public function process($html)
{
if (!$this->enabled) {
return $html;
}
if ($this->addTimeMark) {
Debugger::timer('tidy');
}
$this->parseString($html, $this->config, 'utf8');
if ($this->runCleanRepair) {
$this->cleanRepair();
}
$output = tidy_get_output($this);
if ($this->addTimeMark) {
$elapsed = Debugger::timer('tidy');
$output .= "\n\n<!-- Tidy formatting took: " . number_format($elapsed * 1000, 2) . " ms -->";
}
return $output;
}
/**
* Reads input and returns Tidy-filtered output.
*
* @param null $len
*
* @throws BuildException
* @return the resulting stream, or -1 if the end of the resulting stream has been reached
*
*/
public function read($len = null)
{
if (!class_exists('Tidy')) {
throw new BuildException("You must enable the 'tidy' extension in your PHP configuration in order to use the Tidy filter.");
}
if (!$this->getInitialized()) {
$this->_initialize();
$this->setInitialized(true);
}
$buffer = $this->in->read($len);
if ($buffer === -1) {
return -1;
}
$config = $this->getDistilledConfig();
$tidy = new Tidy();
$tidy->parseString($buffer, $config, $this->encoding);
$tidy->cleanRepair();
return tidy_get_output($tidy);
}
public function executeFilter(HttpRequestInterface $request, HttpResponseInterface $response)
{
// htmltidy must be enabled in configuration
if ($this->config['htmltidy']['enabled'] === 1 and extension_loaded('tidy')) {
// bypass
return;
}
// get output from response
$content = $response->getContent();
// init tidy
$tidy = new tidy();
/*
$tidyoptions = array(
'indent-spaces' => 4,
'wrap' => 120,
'indent' => auto,
'tidy-mark' => true,
'show-body-only' => true,
'force-output' => true,
'output-xhtml' => true,
'clean' => true,
'hide-comments' => false,
'join-classes' => false,
'join-styles' => false,
'doctype' => 'strict',
'lower-literals' => true,
'quote-ampersand' => true,
'wrap' => 0,
'drop-font-tags' => true,
'drop-empty-paras' => true,
'drop-proprietary-attributes' => true);
*/
$tidyoptions = ['clean' => true, 'doctype' => 'transitional', 'output-xhtml' => true, 'drop-proprietary-attributes' => true, 'lower-literals' => true, 'show-body-only' => false, 'indent-spaces' => 4, 'wrap' => 130, 'indent' => 'auto'];
// tidy the output
$tidy->parseString($content, $tidyoptions, 'utf8');
$tidy->cleanRepair();
// @todo diagnose? errorreport?
// set output to response
$response->setContent(tidy_get_output($tidy), true);
}
/**
* tidy the data
*
* @access public
* @param string data
* @return string compressed data
*/
function apply($data)
{
if (!function_exists('tidy_parse_string')) {
return $data;
}
/**
* tidy 1.0
*/
if (function_exists('tidy_setopt') && is_array($this->_params)) {
foreach ($this->_params as $opt => $value) {
tidy_setopt($opt, $value);
}
tidy_parse_string($data);
tidy_clean_repair();
$data = tidy_get_output();
} else {
$tidy = tidy_parse_string($data, $this->_params);
tidy_clean_repair($tidy);
$data = tidy_get_output($tidy);
}
return $data;
}
function return_parsed_bbcode($message, $nowrap = false)
{
// never strip_tags here, see Page.Talks for details
$message = str_replace("[b]", "<b>", $message);
$message = str_replace("[/b]", "</b>", $message);
$message = str_replace("[i]", "<i>", $message);
$message = str_replace("[/i]", "</i>", $message);
$message = str_replace("[u]", "<u>", $message);
$message = str_replace("[/u]", "</u>", $message);
$message = str_replace("[center]", "<div align=\"center\">", $message);
$message = str_replace("[/center]", "</div>", $message);
$message = str_replace("[left]", "<div align=\"left\">", $message);
$message = str_replace("[/left]", "</div>", $message);
$message = str_replace("[right]", "<div align=\"right\">", $message);
$message = str_replace("[/right]", "</div>", $message);
$message = str_replace("[ol]", "<ol>", $message);
$message = str_replace("[ul]", "<ul>", $message);
$message = str_replace("[li]", "<li>", $message);
$message = str_replace("[/ol]", "</ol>", $message);
$message = str_replace("[/ul]", "</ul>", $message);
$message = str_replace("[br]", "<br>", $message);
$message = eregi_replace("\\[img\\]([^\\[]*)\\[/img\\]", "<img src=\"\\1\" border=\"0\">", $message);
$message = eregi_replace("\\[url\\](https?://[^\\[]*)\\[/url\\]", "<a href=\"\\1\">\\1</a>", $message);
if (function_exists("tidy_get_output")) {
if (!$nowrap) {
$config = array('indent' => FALSE, 'output-xhtml' => TRUE, 'show-body-only' => TRUE, 'wrap' => 80);
} else {
$config = array('indent' => FALSE, 'output-xhtml' => TRUE, 'show-body-only' => TRUE);
}
tidy_set_encoding('UTF8');
foreach ($config as $key => $value) {
tidy_setopt($key, $value);
}
tidy_parse_string($message);
tidy_clean_repair();
$message = tidy_get_output();
}
return $message;
}
function TidyClean()
{
if (!class_exists('tidy')) {
if (function_exists('tidy_parse_string')) {
//use procedural style for compatibility with PHP 4.3
tidy_set_encoding($this->Encoding);
foreach ($this->TidyConfig as $key => $value) {
tidy_setopt($key, $value);
}
tidy_parse_string($this->html);
tidy_clean_repair();
$this->html = tidy_get_output();
} else {
print "<b>No tidy support. Please enable it in your php.ini.\r\nOnly basic cleaning is beeing applied\r\n</b>";
}
} else {
//PHP 5 only !!!
$tidy = new tidy();
$tidy->parseString($this->html, $this->TidyConfig, $this->Encoding);
$tidy->cleanRepair();
$this->html = $tidy;
}
}
/**
* Tidyfication of the strings
*
* @param string $str
* @return string
*/
public function tidyCleaner($str)
{
eZDebug::accumulatorStart('eztidytemplateoperator', 'Tidy', 'Tidy template operator');
if (!class_exists('tidy')) {
eZDebug::writeError("phpTidy isn't installed", 'eZTidy::tidyCleaner()');
return $str;
}
$str = trim($str);
if ($str == "") {
return "";
}
$this->tidy = new tidy();
$this->tidy->parseString($str, $this->config, $this->options['charset']);
$this->tidy->cleanRepair();
$this->isTidyfied = true;
$this->reportWarning();
$output = tidy_get_output($this->tidy);
if (strtolower($this->options['showTidyElement']) == 'enabled') {
return "<!-- Tidy - Begin -->\n" . $output . "\n<!-- Tidy - End -->";
}
eZDebug::accumulatorStop('eztidytemplateoperator');
return $output;
}
/**
* Delivers a PDF file from XHTML
*
* @param string $html The XHTML string
* @access public
*/
public function deliverPDFfromHTML($content, $title = NULL)
{
$content = preg_replace("/href=\".*?\"/", "", $content);
$printbody = new ilTemplate("tpl.il_as_tst_print_body.html", TRUE, TRUE, "Modules/Test");
$printbody->setVariable("TITLE", ilUtil::prepareFormOutput($this->getTitle()));
$printbody->setVariable("ADM_CONTENT", $content);
$printbody->setCurrentBlock("css_file");
$printbody->setVariable("CSS_FILE", $this->getTestStyleLocation("filesystem"));
$printbody->parseCurrentBlock();
$printbody->setCurrentBlock("css_file");
$printbody->setVariable("CSS_FILE", ilUtil::getStyleSheetLocation("filesystem", "delos.css"));
$printbody->parseCurrentBlock();
$printoutput = $printbody->get();
$html = str_replace("href=\"./", "href=\"" . ILIAS_HTTP_PATH . "/", $printoutput);
$html = preg_replace("/<div id=\"dontprint\">.*?<\\/div>/ims", "", $html);
if (extension_loaded("tidy")) {
$config = array("indent" => false, "output-xml" => true, "numeric-entities" => true);
$tidy = new tidy();
$tidy->parseString($html, $config, 'utf8');
$tidy->cleanRepair();
$html = tidy_get_output($tidy);
$html = preg_replace("/^.*?(<html)/", "\\1", $html);
} else {
$html = str_replace(" ", " ", $html);
$html = str_replace("⊗", "X", $html);
}
$html = preg_replace("/src=\".\\//ims", "src=\"" . ILIAS_HTTP_PATH . "/", $html);
$this->deliverPDFfromFO($this->processPrintoutput2FO($html), $title);
}
$node->appendChild($dom->createTextNode('.1'));
//hack to trigger an update for 4.1.10 release due translations not being loaded on updates
}
}
$results = $xpath->query('//*[@locale]');
for ($i = 0; $i < $results->length; $i++) {
$results->item($i)->setAttribute('locale', $locale);
}
$out = $dom->saveXML();
if (function_exists('tidy_get_output')) {
$tidy = new tidy();
$tidy_config = array('input-xml' => true, 'output-xml' => true, 'indent' => true, 'wrap' => 0);
$tidy->isXML();
$tidy->parseString($out, $tidy_config, 'UTF8');
$tidy->cleanRepair();
$out = tidy_get_output($tidy);
}
if ($booleans['create_archive']) {
$cwd = realpath(getcwd());
$c_path = realpath($configs[$module]);
if (strpos($c_path, $cwd) !== 0) {
I2CE::raiseError("Cannot determine module sub-directory structure for {$module}", E_USER_ERROR);
}
$target_dir = $archive_dir . DIRECTORY_SEPARATOR . 'files' . DIRECTORY_SEPARATOR . $locale . DIRECTORY_SEPARATOR . substr($c_path, strlen($cwd)) . DIRECTORY_SEPARATOR . $locale . DIRECTORY_SEPARATOR;
} else {
$target_dir = $configs[$module] . DIRECTORY_SEPARATOR . $locale . DIRECTORY_SEPARATOR;
}
if (!is_dir($target_dir)) {
if (!mkdir($target_dir, 0775, true)) {
I2CE::raiseError("Could not created {$target_dir}", E_USER_ERROR);
}
function tidyToXml($htmlTagSoup)
{
// Create the Tidy object
$tidy = new Tidy();
// Parse the HTML into memory, turning on the option to convert to
// XHTML as part of the tidying process
$tidy->parseString($htmlTagSoup, array('output-xhtml' => true));
// Do the tidying
$tidy->cleanRepair();
// And get the tidied version as a string
$tidied_xml = tidy_get_output($tidy);
// Opinions seem to differ as to whether the non-breaking space
// entity ' ' is predeclared as part of XHTML. Tidy thinks it
// is, and so leaves it alone, while the XML parser we're about to
// use on this string thinks otherwise. So replace any occurrences
// of it with its numeric equivalent (which doesn't need to be
// declared).
return str_replace(' ', ' ', $tidied_xml);
}
fwrite($pipes[0], $source);
fclose($pipes[0]);
// Read clean source out to the browser
while (!feof($pipes[1])) {
//echo fgets($pipes[1], 1024);
$newsrc .= fgets($pipes[1], 1024);
}
fclose($pipes[1]);
// Clean up after ourselves
proc_close($process);
} else {
/* Use tidy if it's available from PECL */
if (function_exists('tidy_parse_string')) {
$tempsrc = tidy_parse_string($source);
tidy_clean_repair();
$newsrc = tidy_get_output();
} else {
// Better give them back what they came with, so they don't lose it all...
$newsrc = "<body>\n" . $source . "\n</body>";
}
}
// Split our source into an array by lines
$srcLines = preg_split("/\n/", $newsrc, -1, PREG_SPLIT_NO_EMPTY);
// Get only the lines between the body tags
$startLn = 0;
while (strpos($srcLines[$startLn++], '<body') === false && $startLn < sizeof($srcLines)) {
}
$endLn = $startLn;
while (strpos($srcLines[$endLn++], '</body') === false && $endLn < sizeof($srcLines)) {
}
$srcLines = array_slice($srcLines, $startLn, $endLn - $startLn - 1);
function tidy_html($input_string)
{
// Detect if Tidy is in configured
if (function_exists('tidy_get_release')) {
# Tidy for PHP version 4
if (substr(phpversion(), 0, 1) == 4) {
tidy_setopt('uppercase-attributes', TRUE);
tidy_setopt('wrap', 800);
tidy_parse_string($input_string);
$cleaned_html = tidy_get_output();
}
# Tidy for PHP version 5
if (substr(phpversion(), 0, 1) == 5) {
$config = array('uppercase-attributes' => true, 'wrap' => 800);
$tidy = new tidy();
$tidy->parseString($input_string, $config, 'utf8');
$tidy->cleanRepair();
$cleaned_html = tidy_get_output($tidy);
}
} else {
# Tidy not configured for this computer
$cleaned_html = $input_string;
}
return $cleaned_html;
}
/**
* Convert a print output to XSL-FO
*
* @param string $print_output The print output
* @return string XSL-FO code
* @access public
*/
function processPrintoutput2FO($print_output)
{
global $ilLog;
if (extension_loaded("tidy")) {
$config = array("indent" => false, "output-xml" => true, "numeric-entities" => true);
$tidy = new tidy();
$tidy->parseString($print_output, $config, 'utf8');
$tidy->cleanRepair();
$print_output = tidy_get_output($tidy);
$print_output = preg_replace("/^.*?(<html)/", "\\1", $print_output);
} else {
$print_output = str_replace(" ", " ", $print_output);
$print_output = str_replace("⊗", "X", $print_output);
}
$xsl = file_get_contents("./Modules/Survey/xml/question2fo.xsl");
// additional font support
$xsl = str_replace('font-family="Helvetica, unifont"', 'font-family="' . $GLOBALS['ilSetting']->get('rpc_pdf_font', 'Helvetica, unifont') . '"', $xsl);
$args = array('/_xml' => $print_output, '/_xsl' => $xsl);
$xh = xslt_create();
$params = array();
$output = xslt_process($xh, "arg:/_xml", "arg:/_xsl", NULL, $args, $params);
xslt_error($xh);
xslt_free($xh);
$ilLog->write($output);
return $output;
}
/**
* Use the HTML tidy PECL extension to use the tidy library in-process,
* saving the overhead of spawning a new process.
*
* 'pear install tidy' should be able to compile the extension module.
*
* @private
* @static
*/
function internalTidy($text)
{
global $wgTidyConf, $IP;
$fname = 'Parser::internalTidy';
wfProfileIn($fname);
$tidy = new tidy();
$tidy->parseString($text, $wgTidyConf, 'utf8');
$tidy->cleanRepair();
if ($tidy->getStatus() == 2) {
// 2 is magic number for fatal error
// http://www.php.net/manual/en/function.tidy-get-status.php
$cleansource = null;
} else {
$cleansource = tidy_get_output($tidy);
}
wfProfileOut($fname);
return $cleansource;
}
/**
* Use the HTML tidy PECL extension to use the tidy library in-process,
* saving the overhead of spawning a new process.
*
* 'pear install tidy' should be able to compile the extension module.
*
* @private
* @static
*/
function internalTidy($text)
{
global $wgTidyConf, $IP, $wgDebugTidy;
$fname = 'Parser::internalTidy';
wfProfileIn($fname);
$tidy = new tidy();
$tidy->parseString($text, $wgTidyConf, 'utf8');
$tidy->cleanRepair();
if ($tidy->getStatus() == 2) {
// 2 is magic number for fatal error
// http://www.php.net/manual/en/function.tidy-get-status.php
$cleansource = null;
} else {
$cleansource = tidy_get_output($tidy);
}
if ($wgDebugTidy && $tidy->getStatus() > 0) {
$cleansource .= "<!--\nTidy reports:\n" . str_replace('-->', '-->', $tidy->errorBuffer) . "\n-->";
}
wfProfileOut($fname);
return $cleansource;
}
/**
* Use HTML Tidy to validate the $text
* Only runs when $config['HTML_Tidy'] is off
*
* @param string $text The html content to be checked. Passed by reference
*/
static function tidyFix(&$text, $ignore_config = false)
{
global $config;
if (!$ignore_config) {
if (empty($config['HTML_Tidy']) || $config['HTML_Tidy'] == 'off') {
return true;
}
}
if (!function_exists('tidy_parse_string')) {
return false;
}
$options = array();
$options['wrap'] = 0;
//keeps tidy from wrapping... want the least amount of space changing as possible.. could get rid of spaces between words with the str_replaces below
$options['doctype'] = 'omit';
//omit, auto, strict, transitional, user
$options['drop-empty-paras'] = true;
//drop empty paragraphs
$options['output-xhtml'] = true;
//need this so that <br> will be <br/> .. etc
$options['show-body-only'] = true;
$options['hide-comments'] = false;
//$options['anchor-as-name'] = true; //default is true, but not alwasy availabel. When true, adds an id attribute to anchor; when false, removes the name attribute... poorly designed, but we need it to be true
//
// php4
//
if (function_exists('tidy_setopt')) {
$options['char-encoding'] = 'utf8';
gp_edit::tidyOptions($options);
$tidy = tidy_parse_string($text);
tidy_clean_repair();
if (tidy_get_status() === 2) {
// 2 is magic number for fatal error
// http://www.php.net/manual/en/function.tidy-get-status.php
$tidyErrors[] = 'Tidy found serious XHTML errors: <br/>' . nl2br(htmlspecialchars(tidy_get_error_buffer($tidy)));
return false;
}
$text = tidy_get_output();
//
// php5
//
} else {
$tidy = tidy_parse_string($text, $options, 'utf8');
tidy_clean_repair($tidy);
if (tidy_get_status($tidy) === 2) {
// 2 is magic number for fatal error
// http://www.php.net/manual/en/function.tidy-get-status.php
$tidyErrors[] = 'Tidy found serious XHTML errors: <br/>' . nl2br(htmlspecialchars(tidy_get_error_buffer($tidy)));
return false;
}
$text = tidy_get_output($tidy);
}
return true;
}
<?php
/*
* cleanhtml.php
*
* A simple script to clean and repair HTML,XHTML,PHP,ASP,etc. documents
* if no file is provided, it reads from standard input.
*
* NOTE: Works only with tidy for PHP 4.3.x, for tidy in PHP 5 see cleanhtml5.php
*
* By: John Coggeshall <*****@*****.**>
*
* Usage: php cleanhtml.php [filename]
*
*/
if (!isset($_SERVER['argv'][1])) {
$data = file_get_contents("php://stdin");
tidy_parse_string($data);
} else {
tidy_parse_file($_SERVER['argv'][1]);
}
tidy_clean_repair();
if (tidy_warning_count() || tidy_error_count()) {
echo "\n\nThe following errors or warnings occurred:\n";
echo tidy_get_error_buffer();
echo "\n";
}
echo tidy_get_output();
/**
* Use the HTML tidy extension to use the tidy library in-process,
* saving the overhead of spawning a new process.
*
* @param string $text HTML to check
* @param bool $stderr Whether to read result from error status instead of output
* @param int &$retval Exit code (-1 on internal error)
* @return string|null
*/
private static function phpClean($text, $stderr = false, &$retval = null)
{
global $wgTidyConf, $wgDebugTidy;
if (!wfIsHHVM() && !class_exists('tidy') || wfIsHHVM() && !function_exists('tidy_repair_string')) {
wfWarn("Unable to load internal tidy class.");
$retval = -1;
return null;
}
$tidy = new tidy();
$tidy->parseString($text, $wgTidyConf, 'utf8');
if ($stderr) {
$retval = $tidy->getStatus();
return $tidy->errorBuffer;
}
$tidy->cleanRepair();
$retval = $tidy->getStatus();
if ($retval == 2) {
// 2 is magic number for fatal error
// http://www.php.net/manual/en/function.tidy-get-status.php
$cleansource = null;
} else {
$cleansource = tidy_get_output($tidy);
if ($wgDebugTidy && $retval > 0) {
$cleansource .= "<!--\nTidy reports:\n" . str_replace('-->', '-->', $tidy->errorBuffer) . "\n-->";
}
}
return $cleansource;
}
}
if (check_for_add_field($line)) {
continue;
}
if (check_for_set_map($line)) {
continue;
}
}
}
$text = $config->saveXML($topNode);
$tidy = new tidy();
$config = array('input-xml' => true, 'output-xml' => true, 'indent' => true, 'wrap' => 0);
$tidy->isXML();
$tidy->parseString($text, $config, 'UTF8');
$tidy->cleanRepair();
file_put_contents("config_formClass.xml", tidy_get_output($tidy) . "\n");
function check_for_add_field($line)
{
global $config;
global $red;
global $black;
global $classNode;
global $fieldsNode;
if (!preg_match('/^\\s*\\$this->addField\\(\\s*(.*)\\s*\\);\\s*$/', $line, $matches)) {
if (preg_match('/addField/', $line)) {
echo "{$red}Found addField() but don't know how to deal with it at:{$black}\n\t{$line}\n";
}
return false;
}
$fields = explode(',', $matches[1]);
//we hope that there are no commas in the names
private function repairHtml($node)
{
if (class_exists('tidy')) {
$c = array('clean' => true, 'output-xhtml' => true, 'show-body-only' => true, 'input-xml' => true);
$tidy = new tidy();
$tidy->parseString($node, $c, 'utf8');
$tidy->cleanRepair();
$node = tidy_get_output($tidy);
$tidy->diagnose();
// SPConfig::debOut( $tidy->errorBuffer );
return $node;
}
return "Error - cannot repair";
}
public function htmlIndentation($html)
{
if (class_exists('tidy')) {
$config = array('char-encoding' => 'utf8', 'vertical-space' => false, 'indent' => true, 'wrap' => 0, 'word-2000' => 1, 'break-before-br' => true, 'indent-cdata' => true);
$tidy = new \Tidy();
$tidy->parseString($html, $config);
return str_replace('>' . PHP_EOL . '</', '></', tidy_get_output($tidy));
} else {
return $html;
}
}
/**
* Use the HTML tidy PECL extension to use the tidy library in-process,
* saving the overhead of spawning a new process.
*
* 'pear install tidy' should be able to compile the extension module.
*/
private static function execInternalTidy($text, $stderr = false, &$retval = null)
{
global $wgTidyConf, $IP, $wgDebugTidy;
wfProfileIn(__METHOD__);
$tidy = new tidy();
$tidy->parseString($text, $wgTidyConf, 'utf8');
if ($stderr) {
$retval = $tidy->getStatus();
return $tidy->errorBuffer;
} else {
$tidy->cleanRepair();
$retval = $tidy->getStatus();
if ($retval == 2) {
// 2 is magic number for fatal error
// http://www.php.net/manual/en/function.tidy-get-status.php
$cleansource = null;
} else {
$cleansource = tidy_get_output($tidy);
}
if ($wgDebugTidy && $retval > 0) {
$cleansource .= "<!--\nTidy reports:\n" . str_replace('-->', '-->', $tidy->errorBuffer) . "\n-->";
}
wfProfileOut(__METHOD__);
return $cleansource;
}
}
$datetext = str_replace('juliol', 'July', $datetext);
$datetext = str_replace('agost', 'August', $datetext);
$datetext = str_replace('setembre', 'September', $datetext);
$datetext = str_replace('octubre', 'October', $datetext);
$datetext = str_replace('novembre', 'November', $datetext);
$datetext = str_replace('desembre', 'December', $datetext);
$date = date_create_from_format('F d, Y H:i:s', $datetext . ' 00:00:00');
$newItem->setDate($date->format('Y-m-d H:i:s'));
//Now add the feed item
$TestFeed->addItem($newItem);
$feed_count++;
}
$texts = $html->find('text');
$go_on = FALSE;
foreach ($texts as $text) {
if ($text->plaintext == '« Older Entries') {
//Not sleeping, Wordpress.com does not appear to be rate-limited
$html_text = file_get_contents($text->parent->href) or exit(1);
$tidy = tidy_parse_string($html_text, $tidy_config, 'UTF8');
tidy_clean_repair($tidy);
$html = str_get_html(tidy_get_output($tidy));
$go_on = TRUE;
break;
}
}
}
if ($feed_count == 0) {
//No error but no feeds, this is wrong
exit(1);
}
$TestFeed->generateFeed();
<?php
$tidy = tidy_parse_file("intro2_ex1.html");
tidy_clean_repair($tidy);
echo tidy_get_output($tidy);
