function convertData($html)
{
// Style according to the Netiquette
$html = preg_replace('#<(?:b|strong)\\b[^>]*>(\\s*)#iu', '$1*', $html);
$html = preg_replace('#(\\s*)</(?:b|strong)\\b[^>]*>#iu', '*$1', $html);
$html = preg_replace('#<u\\b[^>]*>(\\s*)#iu', '$1_', $html);
$html = preg_replace('#(\\s*)</u\\b[^>]*>#iu', '_$1', $html);
// Remove <sub> and <sup> tags
$html = preg_replace('#<(/?)su[bp]\\b([^>]*)>#iu', '<$1span$2>', $html);
// Fill empty alt attributes with whitespace, clear src attributes
$html = preg_replace('#(<[^>]+\\balt=")"#iu', '$1 "', $html);
$html = preg_replace('#(<[^>]+\\bsrc=")(?:[^"]*)"#iu', '$1"', $html);
// Inline URLs
$html = preg_replace_callback('#<a\\b[^>]*\\shref="([^"]*)"[^>]*>(.*?)</a\\b[^>]*>#isu', array(__CLASS__, 'buildTextAnchor'), $html);
// Convert html-entities to UTF-8 for w3m
$html = str_replace(array('"', '<', '>', ''', '"', '<', '>', "'"), array('&quot;', '&lt;', '&gt;', '&#039;', '"', '<', '>', '''), FILTER::get($html, 'text'));
$html = html_entity_decode($html, ENT_COMPAT, 'UTF-8');
$file = tempnam(PATCHWORK_ZCACHE, 'converter');
Patchwork::writeFile($file, $html);
$html = escapeshellarg($file);
$html = `w3m -dump -cols {$this->cols} -T text/html -I UTF-8 -O UTF-8 {$html}`;
$html = str_replace(self::$charMap[0], self::$charMap[1], $html);
$html = strtr($html, self::$textAnchor);
self::$textAnchor = array();
unlink($file);
return $html;
}
protected function init(&$param)
{
if (isset($this->form->rawValues[$this->name])) {
$value =& $this->form->rawValues[$this->name];
$value = FILTER::get($value, 'html');
}
parent::init($param);
if (isset($param['toolbarSet'])) {
$this->toolbarSet = $param['toolbarSet'];
}
if (isset($param['config'])) {
$this->config = $param['config'];
}
}
public final function __construct($args = array())
{
$class = get_class($this);
isset($this->contentType) or $this->contentType = constant($class . '::contentType');
$a = (array) $this->get;
$this->get = (object) array();
$_GET = array();
foreach ($a as $key => &$a) {
if (is_string($key)) {
$default = $a;
$a = $key;
} else {
$default = '';
}
false !== strpos($a, "") && ($a = str_replace("", '', $a));
if (false !== strpos($a, '\\')) {
$a = strtr($a, array('\\\\' => '\\', '\\:' => ""));
$a = explode(':', $a);
$b = count($a);
do {
false !== strpos($a[--$b], "") && ($a[$b] = strtr($a[$b], "", ':'));
} while ($b);
} else {
$a = explode(':', $a);
}
$key = array_shift($a);
$b = isset($args[$key]) ? (string) $args[$key] : $default;
false !== strpos($b, "") && ($b = str_replace("", '', $b));
if ($a) {
$b = FILTER::get($b, array_shift($a), $a);
if (false === $b) {
$b = $default;
}
}
$_GET[$key] = $this->get->{$key} = $b;
}
$this->control();
if (!$this->contentType && '' !== ($a = strtolower(pathinfo(p\Superloader::class2file($class), PATHINFO_EXTENSION)))) {
$this->contentType = isset(p\StaticResource::$contentType['.' . $a]) ? p\StaticResource::$contentType['.' . $a] : 'application/octet-stream';
}
$this->contentType && header('Content-Type: ' . $this->contentType);
}
protected function init(&$param)
{
empty($param['disabled']) || ($this->disabled = true);
if ($this->disabled || !empty($param['readonly'])) {
$this->readonly = true;
}
if (isset($param['valid'])) {
$this->valid = $param['valid'];
} else {
if (!isset($this->valid)) {
$this->valid = 'char';
if (!isset($param[0]) && '' !== $this->validDefaultRx) {
$this->validArgs[] = $this->validDefaultRx;
$this->validmsg = T($this->validDefaultMsg);
}
}
}
if (!empty($param['multiple'])) {
$this->isdata = false;
$this->multiple = true;
}
isset($param['isdata']) && ($this->isdata = (bool) $param['isdata']);
$i = 0;
while (isset($param[$i])) {
$this->validArgs[] =& $param[$i++];
}
isset($param['validmsg']) && ($this->validmsg = $param['validmsg']);
isset($param['validMsg']) && ($this->validmsg = $param['validMsg']);
$this->validmsg || ($this->validmsg = FILTER::getMsg($this->valid, $this->validArgs));
if (!$this->readonly && isset($this->form->rawValues[$this->name])) {
$value = $this->form->rawValues[$this->name];
if (is_string($value) && false !== strpos($value, "")) {
$value = str_replace("", '', $value);
$this->form->rawValues[$this->name] = $value;
}
} else {
if (isset($param['default'])) {
$value = $param['default'];
if ($this->multiple && !is_array($value)) {
$value = explode(',', $value);
$value = array_map('rawurldecode', $value);
}
$this->setValue($value);
$value =& $this->value;
} else {
$value = '';
}
}
if ($this->multiple) {
$this->status = '';
if ($value) {
if (is_array($value)) {
$status = true;
foreach ($value as $i => &$v) {
if ('' === $v) {
unset($value[$i]);
} else {
$a = FILTER::get($v, $this->valid, $this->validArgs);
if (false === $a) {
$status = false;
} else {
$v = $a;
$status = true && $status;
}
}
}
$value && ($this->status = $status);
} else {
$this->status = false;
$value = array();
}
} else {
$value = array();
}
} else {
if ('' === (string) $value) {
$this->status = '';
} else {
$this->status = FILTER::get($value, $this->valid, $this->validArgs);
if ('' !== $this->status && false !== $this->status) {
$value = $this->status;
$this->status = true;
}
}
}
$this->setValue($value);
}
function convertFile($file)
{
$file = escapeshellarg($file);
$file = `antiword -t -w 0 -m UTF-8 {$file}`;
return FILTER::get($file, 'text');
}
function convertFile($file)
{
$file = escapeshellarg($file);
$file = `pdftotext -enc UTF-8 {$file} -`;
return FILTER::get($file, 'text');
}
