better handling of troublesome feeds.

This commit is contained in:
Friendika 2011-04-04 19:36:18 -07:00
parent 178362e50b
commit 793967a1d3
7 changed files with 62 additions and 27 deletions

View File

@ -1478,7 +1478,9 @@ function lrdd($uri) {
return array(); return array();
logger('lrdd: host_meta: ' . $xml, LOGGER_DATA); logger('lrdd: host_meta: ' . $xml, LOGGER_DATA);
$h = simplexml_load_string($xml);
$h = parse_xml_string($xml);
$arr = convert_xml_element_to_array($h); $arr = convert_xml_element_to_array($h);
if(isset($arr['xrd']['property'])) { if(isset($arr['xrd']['property'])) {
@ -1550,6 +1552,8 @@ function lrdd($uri) {
$headers = $a->get_curl_headers(); $headers = $a->get_curl_headers();
logger('lrdd: headers=' . $headers, LOGGER_DEBUG); logger('lrdd: headers=' . $headers, LOGGER_DEBUG);
// don't try and parse raw xml as html
if(! strstr($html,'<?xml')) {
require_once('library/HTML5/Parser.php'); require_once('library/HTML5/Parser.php');
$dom = @HTML5_Parser::parse($html); $dom = @HTML5_Parser::parse($html);
@ -1563,6 +1567,7 @@ function lrdd($uri) {
} }
} }
} }
}
if(isset($pagelink)) if(isset($pagelink))
return(fetch_xrd_links($pagelink)); return(fetch_xrd_links($pagelink));
@ -1638,7 +1643,7 @@ function fetch_xrd_links($url) {
return array(); return array();
logger('fetch_xrd_links: ' . $xml, LOGGER_DATA); logger('fetch_xrd_links: ' . $xml, LOGGER_DATA);
$h = simplexml_load_string($xml); $h = parse_xml_string($xml);
$arr = convert_xml_element_to_array($h); $arr = convert_xml_element_to_array($h);
$links = array(); $links = array();
@ -2759,3 +2764,18 @@ function lang_selector() {
$o .= '</select></form></div>'; $o .= '</select></form></div>';
return $o; return $o;
}} }}
if(! function_exists('parse_xml_string')) {
function parse_xml_string($s) {
if(! strstr($s,'<?xml'))
return false;
$s2 = substr($s,strpos($s,'<?xml'));
libxml_use_internal_errors(true);
$x = @simplexml_load_string($s2);
if(count(libxml_get_errors()))
foreach(libxml_get_errors() as $err)
logger('libxml: parse: ' . $err, LOGGER_DATA);
libxml_clear_errors();
return $x;
}}

View File

@ -216,7 +216,7 @@ function scrape_feed($url) {
} }
if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) { if(stristr($line,'application/rss+xml') || stristr($s,'<rss')) {
$ret['feed_rss'] = $url; $ret['feed_rss'] = $url;
return ret; return $ret;
} }
} }
} }

View File

@ -180,7 +180,7 @@ function construct_activity_object($item) {
if($item['object']) { if($item['object']) {
$o = '<as:object>' . "\r\n"; $o = '<as:object>' . "\r\n";
$r = @simplexml_load_string($item['object']); $r = parse_xml_string($item['object']);
if($r->type) if($r->type)
$o .= '<as:object-type>' . xmlify($r->type) . '</as:object-type>' . "\r\n"; $o .= '<as:object-type>' . xmlify($r->type) . '</as:object-type>' . "\r\n";
if($r->id) if($r->id)
@ -206,7 +206,7 @@ function construct_activity_target($item) {
if($item['target']) { if($item['target']) {
$o = '<as:target>' . "\r\n"; $o = '<as:target>' . "\r\n";
$r = @simplexml_load_string($item['target']); $r = parse_xml_string($item['target']);
if($r->type) if($r->type)
$o .= '<as:object-type>' . xmlify($r->type) . '</as:object-type>' . "\r\n"; $o .= '<as:object-type>' . xmlify($r->type) . '</as:object-type>' . "\r\n";
if($r->id) if($r->id)
@ -241,8 +241,14 @@ function get_atom_elements($feed,$item) {
$res = array(); $res = array();
$author = $item->get_author(); $author = $item->get_author();
if($author) {
$res['author-name'] = unxmlify($author->get_name()); $res['author-name'] = unxmlify($author->get_name());
$res['author-link'] = unxmlify($author->get_link()); $res['author-link'] = unxmlify($author->get_link());
}
else {
$res['author-name'] = unxmlify($feed->get_title());
$res['author-link'] = unxmlify($feed->get_permalink());
}
$res['uri'] = unxmlify($item->get_id()); $res['uri'] = unxmlify($item->get_id());
$res['title'] = unxmlify($item->get_title()); $res['title'] = unxmlify($item->get_title());
$res['body'] = unxmlify($item->get_content()); $res['body'] = unxmlify($item->get_content());
@ -343,7 +349,6 @@ function get_atom_elements($feed,$item) {
// the wild, by sanitising it and converting supported tags to bbcode before we rip out any remaining // the wild, by sanitising it and converting supported tags to bbcode before we rip out any remaining
// html. // html.
if((strpos($res['body'],'<') !== false) || (strpos($res['body'],'>') !== false)) { if((strpos($res['body'],'<') !== false) || (strpos($res['body'],'>') !== false)) {
$res['body'] = preg_replace('#<object[^>]+>.+?' . 'http://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+).+?</object>#s', $res['body'] = preg_replace('#<object[^>]+>.+?' . 'http://www.youtube.com/((?:v|cp)/[A-Za-z0-9\-_=]+).+?</object>#s',
@ -783,7 +788,7 @@ function dfrn_deliver($owner,$contact,$atom, $dissolve = false) {
return 3; return 3;
} }
$res = simplexml_load_string($xml); $res = parse_xml_string($xml);
if((intval($res->status) != 0) || (! strlen($res->challenge)) || (! strlen($res->dfrn_id))) if((intval($res->status) != 0) || (! strlen($res->challenge)) || (! strlen($res->dfrn_id)))
return (($res->status) ? $res->status : 3); return (($res->status) ? $res->status : 3);
@ -878,7 +883,7 @@ function dfrn_deliver($owner,$contact,$atom, $dissolve = false) {
return 3; return 3;
} }
$res = simplexml_load_string($xml); $res = parse_xml_string($xml);
return $res->status; return $res->status;
@ -916,6 +921,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $secure_fee
if($feed->error()) if($feed->error())
logger('consume_feed: Error parsing XML: ' . $feed->error()); logger('consume_feed: Error parsing XML: ' . $feed->error());
$permalink = $feed->get_permalink();
// Check at the feed level for updated contact name and/or photo // Check at the feed level for updated contact name and/or photo
@ -1230,6 +1236,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $secure_fee
// Head post of a conversation. Have we seen it? If not, import it. // Head post of a conversation. Have we seen it? If not, import it.
$item_id = $item->get_id(); $item_id = $item->get_id();
$datarray = get_atom_elements($feed,$item); $datarray = get_atom_elements($feed,$item);
$r = q("SELECT `uid`, `last-child`, `edited`, `body` FROM `item` WHERE `uri` = '%s' AND `uid` = %d LIMIT 1", $r = q("SELECT `uid`, `last-child`, `edited`, `body` FROM `item` WHERE `uri` = '%s' AND `uid` = %d LIMIT 1",
@ -1275,7 +1282,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $secure_fee
if(! is_array($contact)) if(! is_array($contact))
return; return;
if($contact['network'] === 'stat') { if($contact['network'] === 'stat' || stristr($permalink,'twitter.com')) {
if(strlen($datarray['title'])) if(strlen($datarray['title']))
unset($datarray['title']); unset($datarray['title']);
$datarray['last-child'] = 1; $datarray['last-child'] = 1;

View File

@ -203,7 +203,7 @@ function poller_run($argv, $argc){
} }
$res = simplexml_load_string($xml); $res = parse_xml_string($xml);
if(intval($res->status) == 1) { if(intval($res->status) == 1) {
logger("poller: $url replied status 1 - marking for death "); logger("poller: $url replied status 1 - marking for death ");

View File

@ -240,7 +240,7 @@ function dfrn_confirm_post(&$a,$handsfree = null) {
notice( t('Unexpected response from remote site: ') . EOL . $leading_junk . EOL ); notice( t('Unexpected response from remote site: ') . EOL . $leading_junk . EOL );
} }
$xml = simplexml_load_string($res); $xml = parse_xml_string($res);
$status = (int) $xml->status; $status = (int) $xml->status;
$message = unxmlify($xml->message); // human readable text of what may have gone wrong. $message = unxmlify($xml->message); // human readable text of what may have gone wrong.
switch($status) { switch($status) {

View File

@ -69,7 +69,7 @@ function dfrn_poll_init(&$a) {
if(strlen($s)) { if(strlen($s)) {
$xml = simplexml_load_string($s); $xml = parse_xml_string($s);
if((int) $xml->status == 1) { if((int) $xml->status == 1) {
$_SESSION['authenticated'] = 1; $_SESSION['authenticated'] = 1;
@ -468,7 +468,7 @@ function dfrn_poll_content(&$a) {
if(strlen($s) && strstr($s,'<?xml')) { if(strlen($s) && strstr($s,'<?xml')) {
$xml = simplexml_load_string($s); $xml = parse_xml_string($s);
logger('dfrn_poll: profile: parsed xml: ' . print_r($xml,true), LOGGER_DATA); logger('dfrn_poll: profile: parsed xml: ' . print_r($xml,true), LOGGER_DATA);

View File

@ -15,7 +15,8 @@ function follow_post(&$a) {
$email_conversant = false; $email_conversant = false;
if($url) { if($url) {
$links = @lrdd($url); $links = lrdd($url);
if(count($links)) { if(count($links)) {
foreach($links as $link) { foreach($links as $link) {
if($link['@attributes']['rel'] === NAMESPACE_DFRN) if($link['@attributes']['rel'] === NAMESPACE_DFRN)
@ -107,7 +108,7 @@ function follow_post(&$a) {
if((! isset($vcard)) && (! $poll)) { if((! isset($vcard)) && (! $poll)) {
$ret = scrape_feed($url); $ret = scrape_feed($url);
logger('mod_follow: scrape_feed returns: ' . print_r($ret,true), LOGGER_DATA);
if(count($ret) && ($ret['feed_atom'] || $ret['feed_rss'])) { if(count($ret) && ($ret['feed_atom'] || $ret['feed_rss'])) {
$poll = ((x($ret,'feed_atom')) ? unamp($ret['feed_atom']) : unamp($ret['feed_rss'])); $poll = ((x($ret,'feed_atom')) ? unamp($ret['feed_atom']) : unamp($ret['feed_rss']));
$vcard = array(); $vcard = array();
@ -156,7 +157,14 @@ function follow_post(&$a) {
} }
if((! $vcard['photo']) && strlen($email)) if((! $vcard['photo']) && strlen($email))
$vcard['photo'] = gravatar_img($email); $vcard['photo'] = gravatar_img($email);
if($poll === $profile)
$lnk = $feed->get_permalink();
if(isset($lnk) && strlen($lnk))
$profile = $lnk;
if(! (x($vcard,'fn')))
$vcard['fn'] = notags($feed->get_title());
if(! (x($vcard,'fn')))
$vcard['fn'] = notags($feed->get_description());
$network = 'feed'; $network = 'feed';
$priority = 2; $priority = 2;
} }