improve support for feeds with invalid unicode entities

master
Andrew Dolgov 11 years ago
parent 38f439708c
commit f612dbe8a0

@ -18,36 +18,46 @@ class FeedParser {
$this->doc = new DOMDocument();
$this->doc->loadXML($data);
mb_substitute_character("none");
$error = libxml_get_last_error();
// libxml compiled without iconv?
if ($error && $error->code == 32) {
if ($error && ($error->code == 32 || $error->code == 9)) {
if (preg_match('/^(<\?xml[\t\n\r ].*?encoding=["\'])(.+?)(["\'].*?\?>)/s', $data, $matches) === 1) {
libxml_clear_errors();
$enc = $matches[2];
$data = iconv($enc, 'UTF-8//IGNORE', $data);
$data = mb_convert_encoding($data, 'UTF-8', $enc);
$data = preg_replace('/^<\?xml[\t\n\r ].*?\?>/s', $matches[1] . "UTF-8" . $matches[3] , $data);
$this->doc = new DOMDocument();
$this->doc->loadXML($data);
$error = libxml_get_last_error();
// apparently not all UTF-8 characters are valid for XML
$data = preg_replace('/[^\x{0009}\x{000a}\x{000d}\x{0020}-\x{D7FF}\x{E000}-\x{FFFD}]+/u', ' ', $data);
if ($data) {
libxml_clear_errors();
$this->doc = new DOMDocument();
$this->doc->loadXML($data);
$error = libxml_get_last_error();
}
}
}
// some terrible invalid unicode entity?
if ($error && $error->code == 9) {
libxml_clear_errors();
$data = mb_convert_encoding($data, 'UTF-8', 'UTF-8');
// we might want to try guessing input encoding here too
$data = iconv("UTF-8", "UTF-8//IGNORE", $data);
if ($data) {
libxml_clear_errors();
$this->doc = new DOMDocument();
$this->doc->loadXML($data);
$this->doc = new DOMDocument();
$this->doc->loadXML($data);
$error = libxml_get_last_error();
$error = libxml_get_last_error();
}
}
$this->error = $this->format_error($error);

Loading…
Cancel
Save