registry = $registry;
}
public function parse(&$data, $encoding, $url = '')
{
if (function_exists('Mf2\parse')) {
// Check for both h-feed and h-entry, as both a feed with no entries
// and a list of entries without an h-feed wrapper are both valid.
$position = 0;
while ($position = strpos($data, 'h-feed', $position)) {
$start = $position < 200 ? 0 : $position - 200;
$check = substr($data, $start, 400);
if (preg_match('/class="[^"]*h-feed/', $check)) {
return $this->parse_microformats($data, $url);
}
$position += 7;
}
$position = 0;
while ($position = strpos($data, 'h-entry', $position)) {
$start = $position < 200 ? 0 : $position - 200;
$check = substr($data, $start, 400);
if (preg_match('/class="[^"]*h-entry/', $check)) {
return $this->parse_microformats($data, $url);
}
$position += 7;
}
}
// Use UTF-8 if we get passed US-ASCII, as every US-ASCII character is a UTF-8 character
if (strtoupper($encoding) === 'US-ASCII')
{
$this->encoding = 'UTF-8';
}
else
{
$this->encoding = $encoding;
}
// Strip BOM:
// UTF-32 Big Endian BOM
if (substr($data, 0, 4) === "\x00\x00\xFE\xFF")
{
$data = substr($data, 4);
}
// UTF-32 Little Endian BOM
elseif (substr($data, 0, 4) === "\xFF\xFE\x00\x00")
{
$data = substr($data, 4);
}
// UTF-16 Big Endian BOM
elseif (substr($data, 0, 2) === "\xFE\xFF")
{
$data = substr($data, 2);
}
// UTF-16 Little Endian BOM
elseif (substr($data, 0, 2) === "\xFF\xFE")
{
$data = substr($data, 2);
}
// UTF-8 BOM
elseif (substr($data, 0, 3) === "\xEF\xBB\xBF")
{
$data = substr($data, 3);
}
if (substr($data, 0, 5) === '')) !== false)
{
$declaration = $this->registry->create('XML_Declaration_Parser', array(substr($data, 5, $pos - 5)));
if ($declaration->parse())
{
$data = substr($data, $pos + 2);
$data = 'version . '" encoding="' . $encoding . '" standalone="' . (($declaration->standalone) ? 'yes' : 'no') . '"?>' ."\n". $this->declare_html_entities() . $data;
}
else
{
$this->error_string = 'SimplePie bug! Please report this!';
return false;
}
}
$return = true;
static $xml_is_sane = null;
if ($xml_is_sane === null)
{
$parser_check = xml_parser_create();
xml_parse_into_struct($parser_check, '&', $values);
xml_parser_free($parser_check);
$xml_is_sane = isset($values[0]['value']);
}
// Create the parser
if ($xml_is_sane)
{
$xml = xml_parser_create_ns($this->encoding, $this->separator);
xml_parser_set_option($xml, XML_OPTION_SKIP_WHITE, 1);
xml_parser_set_option($xml, XML_OPTION_CASE_FOLDING, 0);
xml_set_object($xml, $this);
xml_set_character_data_handler($xml, 'cdata');
xml_set_element_handler($xml, 'tag_open', 'tag_close');
// Parse!
if (!xml_parse($xml, $data, true))
{
$this->error_code = xml_get_error_code($xml);
$this->error_string = xml_error_string($this->error_code);
$return = false;
}
$this->current_line = xml_get_current_line_number($xml);
$this->current_column = xml_get_current_column_number($xml);
$this->current_byte = xml_get_current_byte_index($xml);
xml_parser_free($xml);
return $return;
}
else
{
libxml_clear_errors();
$xml = new XMLReader();
$xml->xml($data);
while (@$xml->read())
{
switch ($xml->nodeType)
{
case constant('XMLReader::END_ELEMENT'):
if ($xml->namespaceURI !== '')
{
$tagName = $xml->namespaceURI . $this->separator . $xml->localName;
}
else
{
$tagName = $xml->localName;
}
$this->tag_close(null, $tagName);
break;
case constant('XMLReader::ELEMENT'):
$empty = $xml->isEmptyElement;
if ($xml->namespaceURI !== '')
{
$tagName = $xml->namespaceURI . $this->separator . $xml->localName;
}
else
{
$tagName = $xml->localName;
}
$attributes = array();
while ($xml->moveToNextAttribute())
{
if ($xml->namespaceURI !== '')
{
$attrName = $xml->namespaceURI . $this->separator . $xml->localName;
}
else
{
$attrName = $xml->localName;
}
$attributes[$attrName] = $xml->value;
}
$this->tag_open(null, $tagName, $attributes);
if ($empty)
{
$this->tag_close(null, $tagName);
}
break;
case constant('XMLReader::TEXT'):
case constant('XMLReader::CDATA'):
$this->cdata(null, $xml->value);
break;
}
}
if ($error = libxml_get_last_error())
{
$this->error_code = $error->code;
$this->error_string = $error->message;
$this->current_line = $error->line;
$this->current_column = $error->column;
return false;
}
else
{
return true;
}
}
}
public function get_error_code()
{
return $this->error_code;
}
public function get_error_string()
{
return $this->error_string;
}
public function get_current_line()
{
return $this->current_line;
}
public function get_current_column()
{
return $this->current_column;
}
public function get_current_byte()
{
return $this->current_byte;
}
public function get_data()
{
return $this->data;
}
public function tag_open($parser, $tag, $attributes)
{
list($this->namespace[], $this->element[]) = $this->split_ns($tag);
$attribs = array();
foreach ($attributes as $name => $value)
{
list($attrib_namespace, $attribute) = $this->split_ns($name);
$attribs[$attrib_namespace][$attribute] = $value;
}
if (isset($attribs[SIMPLEPIE_NAMESPACE_XML]['base']))
{
$base = $this->registry->call('Misc', 'absolutize_url', array($attribs[SIMPLEPIE_NAMESPACE_XML]['base'], end($this->xml_base)));
if ($base !== false)
{
$this->xml_base[] = $base;
$this->xml_base_explicit[] = true;
}
}
else
{
$this->xml_base[] = end($this->xml_base);
$this->xml_base_explicit[] = end($this->xml_base_explicit);
}
if (isset($attribs[SIMPLEPIE_NAMESPACE_XML]['lang']))
{
$this->xml_lang[] = $attribs[SIMPLEPIE_NAMESPACE_XML]['lang'];
}
else
{
$this->xml_lang[] = end($this->xml_lang);
}
if ($this->current_xhtml_construct >= 0)
{
$this->current_xhtml_construct++;
if (end($this->namespace) === SIMPLEPIE_NAMESPACE_XHTML)
{
$this->data['data'] .= '<' . end($this->element);
if (isset($attribs['']))
{
foreach ($attribs[''] as $name => $value)
{
$this->data['data'] .= ' ' . $name . '="' . htmlspecialchars($value, ENT_COMPAT, $this->encoding) . '"';
}
}
$this->data['data'] .= '>';
}
}
else
{
$this->datas[] =& $this->data;
$this->data =& $this->data['child'][end($this->namespace)][end($this->element)][];
$this->data = array('data' => '', 'attribs' => $attribs, 'xml_base' => end($this->xml_base), 'xml_base_explicit' => end($this->xml_base_explicit), 'xml_lang' => end($this->xml_lang));
if ((end($this->namespace) === SIMPLEPIE_NAMESPACE_ATOM_03 && in_array(end($this->element), array('title', 'tagline', 'copyright', 'info', 'summary', 'content')) && isset($attribs['']['mode']) && $attribs['']['mode'] === 'xml')
|| (end($this->namespace) === SIMPLEPIE_NAMESPACE_ATOM_10 && in_array(end($this->element), array('rights', 'subtitle', 'summary', 'info', 'title', 'content')) && isset($attribs['']['type']) && $attribs['']['type'] === 'xhtml')
|| (end($this->namespace) === SIMPLEPIE_NAMESPACE_RSS_20 && in_array(end($this->element), array('title')))
|| (end($this->namespace) === SIMPLEPIE_NAMESPACE_RSS_090 && in_array(end($this->element), array('title')))
|| (end($this->namespace) === SIMPLEPIE_NAMESPACE_RSS_10 && in_array(end($this->element), array('title'))))
{
$this->current_xhtml_construct = 0;
}
}
}
public function cdata($parser, $cdata)
{
if ($this->current_xhtml_construct >= 0)
{
$this->data['data'] .= htmlspecialchars($cdata, ENT_QUOTES, $this->encoding);
}
else
{
$this->data['data'] .= $cdata;
}
}
public function tag_close($parser, $tag)
{
if ($this->current_xhtml_construct >= 0)
{
$this->current_xhtml_construct--;
if (end($this->namespace) === SIMPLEPIE_NAMESPACE_XHTML && !in_array(end($this->element), array('area', 'base', 'basefont', 'br', 'col', 'frame', 'hr', 'img', 'input', 'isindex', 'link', 'meta', 'param')))
{
$this->data['data'] .= '' . end($this->element) . '>';
}
}
if ($this->current_xhtml_construct === -1)
{
$this->data =& $this->datas[count($this->datas) - 1];
array_pop($this->datas);
}
array_pop($this->element);
array_pop($this->namespace);
array_pop($this->xml_base);
array_pop($this->xml_base_explicit);
array_pop($this->xml_lang);
}
public function split_ns($string)
{
static $cache = array();
if (!isset($cache[$string]))
{
if ($pos = strpos($string, $this->separator))
{
static $separator_length;
if (!$separator_length)
{
$separator_length = strlen($this->separator);
}
$namespace = substr($string, 0, $pos);
$local_name = substr($string, $pos + $separator_length);
if (strtolower($namespace) === SIMPLEPIE_NAMESPACE_ITUNES)
{
$namespace = SIMPLEPIE_NAMESPACE_ITUNES;
}
// Normalize the Media RSS namespaces
if ($namespace === SIMPLEPIE_NAMESPACE_MEDIARSS_WRONG ||
$namespace === SIMPLEPIE_NAMESPACE_MEDIARSS_WRONG2 ||
$namespace === SIMPLEPIE_NAMESPACE_MEDIARSS_WRONG3 ||
$namespace === SIMPLEPIE_NAMESPACE_MEDIARSS_WRONG4 ||
$namespace === SIMPLEPIE_NAMESPACE_MEDIARSS_WRONG5 )
{
$namespace = SIMPLEPIE_NAMESPACE_MEDIARSS;
}
$cache[$string] = array($namespace, $local_name);
}
else
{
$cache[$string] = array('', $string);
}
}
return $cache[$string];
}
private function parse_hcard($data, $category = false) {
$name = '';
$link = '';
// Check if h-card is set and pass that information on in the link.
if (isset($data['type']) && in_array('h-card', $data['type'])) {
if (isset($data['properties']['name'][0])) {
$name = $data['properties']['name'][0];
}
if (isset($data['properties']['url'][0])) {
$link = $data['properties']['url'][0];
if ($name === '') {
$name = $link;
}
else {
// can't have commas in categories.
$name = str_replace(',', '', $name);
}
$person_tag = $category ? '' : '';
return ''.$person_tag.$name.'';
}
}
return isset($data['value']) ? $data['value'] : '';
}
private function parse_microformats(&$data, $url) {
$feed_title = '';
$feed_author = NULL;
$author_cache = array();
$items = array();
$entries = array();
$mf = Mf2\parse($data, $url);
// First look for an h-feed.
$h_feed = array();
foreach ($mf['items'] as $mf_item) {
if (in_array('h-feed', $mf_item['type'])) {
$h_feed = $mf_item;
break;
}
// Also look for an h-feed in the children of each top level item.
if (!isset($mf_item['children'][0]['type'])) continue;
if (in_array('h-feed', $mf_item['children'][0]['type'])) {
$h_feed = $mf_item['children'][0];
// In this case the parent of the h-feed may be an h-card, so use it as
// the feed_author.
if (in_array('h-card', $mf_item['type'])) $feed_author = $mf_item;
break;
}
}
if (isset($h_feed['children'])) {
$entries = $h_feed['children'];
// Also set the feed title and store author from the h-feed if available.
if (isset($mf['items'][0]['properties']['name'][0])) {
$feed_title = $mf['items'][0]['properties']['name'][0];
}
if (isset($mf['items'][0]['properties']['author'][0])) {
$feed_author = $mf['items'][0]['properties']['author'][0];
}
}
else {
$entries = $mf['items'];
}
for ($i = 0; $i < count($entries); $i++) {
$entry = $entries[$i];
if (in_array('h-entry', $entry['type'])) {
$item = array();
$title = '';
$description = '';
if (isset($entry['properties']['url'][0])) {
$link = $entry['properties']['url'][0];
if (isset($link['value'])) $link = $link['value'];
$item['link'] = array(array('data' => $link));
}
if (isset($entry['properties']['uid'][0])) {
$guid = $entry['properties']['uid'][0];
if (isset($guid['value'])) $guid = $guid['value'];
$item['guid'] = array(array('data' => $guid));
}
if (isset($entry['properties']['name'][0])) {
$title = $entry['properties']['name'][0];
if (isset($title['value'])) $title = $title['value'];
$item['title'] = array(array('data' => $title));
}
if (isset($entry['properties']['author'][0]) || isset($feed_author)) {
// author is a special case, it can be plain text or an h-card array.
// If it's plain text it can also be a url that should be followed to
// get the actual h-card.
$author = isset($entry['properties']['author'][0]) ?
$entry['properties']['author'][0] : $feed_author;
if (!is_string($author)) {
$author = $this->parse_hcard($author);
}
else if (strpos($author, 'http') === 0) {
if (isset($author_cache[$author])) {
$author = $author_cache[$author];
}
else {
$mf = Mf2\fetch($author);
foreach ($mf['items'] as $hcard) {
// Only interested in an h-card by itself in this case.
if (!in_array('h-card', $hcard['type'])) {
continue;
}
// It must have a url property matching what we fetched.
if (!isset($hcard['properties']['url']) ||
!(in_array($author, $hcard['properties']['url']))) {
continue;
}
// Save parse_hcard the trouble of finding the correct url.
$hcard['properties']['url'][0] = $author;
// Cache this h-card for the next h-entry to check.
$author_cache[$author] = $this->parse_hcard($hcard);
$author = $author_cache[$author];
break;
}
}
}
$item['author'] = array(array('data' => $author));
}
if (isset($entry['properties']['photo'][0])) {
// If a photo is also in content, don't need to add it again here.
$content = '';
if (isset($entry['properties']['content'][0]['html'])) {
$content = $entry['properties']['content'][0]['html'];
}
$photo_list = array();
for ($j = 0; $j < count($entry['properties']['photo']); $j++) {
$photo = $entry['properties']['photo'][$j];
if (strpos($content, $photo) === false) {
$photo_list[] = $photo;
}
}
// When there's more than one photo show the first and use a lightbox.
$count = count($photo_list);
if ($count > 1) {
$description = '
';
for ($j = 0; $j < $count; $j++) {
$hidden = $j === 0 ? '' : 'class="hidden" ';
$description .= ''.
'';
}
$description .= '
'.$count.' photos
';
}
else if ($count == 1) {
$description = '';
}
}
if (isset($entry['properties']['content'][0]['html'])) {
// e-content['value'] is the same as p-name when they are on the same
// element. Use this to replace title with a strip_tags version so
// that alt text from images is not included in the title.
if ($entry['properties']['content'][0]['value'] === $title) {
$title = strip_tags($entry['properties']['content'][0]['html']);
$item['title'] = array(array('data' => $title));
}
$description .= $entry['properties']['content'][0]['html'];
if (isset($entry['properties']['in-reply-to'][0]['value'])) {
$in_reply_to = $entry['properties']['in-reply-to'][0]['value'];
$description .= ' '.
''.$in_reply_to.'
';
}
$item['description'] = array(array('data' => $description));
}
if (isset($entry['properties']['category'])) {
$category_csv = '';
// Categories can also contain h-cards.
foreach ($entry['properties']['category'] as $category) {
if ($category_csv !== '') $category_csv .= ', ';
if (is_string($category)) {
// Can't have commas in categories.
$category_csv .= str_replace(',', '', $category);
}
else {
$category_csv .= $this->parse_hcard($category, true);
}
}
$item['category'] = array(array('data' => $category_csv));
}
if (isset($entry['properties']['published'][0])) {
$timestamp = strtotime($entry['properties']['published'][0]);
$pub_date = date('F j Y g:ia', $timestamp).' GMT';
$item['pubDate'] = array(array('data' => $pub_date));
}
// The title and description are set to the empty string to represent
// a deleted item (which also makes it an invalid rss item).
if (isset($entry['properties']['deleted'][0])) {
$item['title'] = array(array('data' => ''));
$item['description'] = array(array('data' => ''));
}
$items[] = array('child' => array('' => $item));
}
}
// Mimic RSS data format when storing microformats.
$link = array(array('data' => $url));
$image = '';
if (!is_string($feed_author) &&
isset($feed_author['properties']['photo'][0])) {
$image = array(array('child' => array('' => array('url' =>
array(array('data' => $feed_author['properties']['photo'][0]))))));
}
// Use the a name given for the h-feed, or get the title from the html.
if ($feed_title !== '') {
$feed_title = array(array('data' => htmlspecialchars($feed_title)));
}
else if ($position = strpos($data, '
')) {
$start = $position < 200 ? 0 : $position - 200;
$check = substr($data, $start, 400);
$matches = array();
if (preg_match('/(.+)<\/title>/', $check, $matches)) {
$feed_title = array(array('data' => htmlspecialchars($matches[1])));
}
}
$channel = array('channel' => array(array('child' => array('' =>
array('link' => $link, 'image' => $image, 'title' => $feed_title,
'item' => $items)))));
$rss = array(array('attribs' => array('' => array('version' => '2.0')),
'child' => array('' => $channel)));
$this->data = array('child' => array('' => array('rss' => $rss)));
return true;
}
private function declare_html_entities() {
// This is required because the RSS specification says that entity-encoded
// html is allowed, but the xml specification says they must be declared.
return ' ]>';
}
}