Update of picoFeed

This commit is contained in:
Frederic Guillot 2013-04-07 10:58:46 -04:00
parent 5f7480c52c
commit 6aa0aaa724
3 changed files with 83 additions and 81 deletions

View File

@ -73,7 +73,16 @@ class Filter
'rss.nytimes.com', 'rss.nytimes.com',
'feeds.wordpress.com', 'feeds.wordpress.com',
'stats.wordpress.com', 'stats.wordpress.com',
'rss.cnn.com' 'rss.cnn.com',
'twitter.com/home?status=',
'twitter.com/share',
'twitter_icon_large.png',
'www.facebook.com/sharer.php',
'facebook_icon_large.png',
'plus.google.com/share',
'www.gstatic.com/images/icons/gplus-16.png',
'www.gstatic.com/images/icons/gplus-32.png',
'www.gstatic.com/images/icons/gplus-64.png'
); );
public $required_attributes = array( public $required_attributes = array(

View File

@ -18,15 +18,15 @@ class Rss10 extends Parser
return false; return false;
} }
$ns = $xml->getNamespaces(true); $namespaces = $xml->getNamespaces(true);
$this->title = (string) $xml->channel->title; $this->title = (string) $xml->channel->title;
$this->url = (string) $xml->channel->link; $this->url = (string) $xml->channel->link;
$this->id = $this->url; $this->id = $this->url;
if (isset($ns['dc'])) { if (isset($namespaces['dc'])) {
$ns_dc = $xml->channel->children($ns['dc']); $ns_dc = $xml->channel->children($namespaces['dc']);
$this->updated = isset($ns_dc->date) ? strtotime($ns_dc->date) : time(); $this->updated = isset($ns_dc->date) ? strtotime($ns_dc->date) : time();
} }
else { else {
@ -36,43 +36,46 @@ class Rss10 extends Parser
foreach ($xml->item as $entry) { foreach ($xml->item as $entry) {
$author = '';
$content = '';
$pubdate = '';
$link = '';
if (isset($ns['feedburner'])) {
$ns_fb = $entry->children($ns['feedburner']);
$link = $ns_fb->origLink;
}
if (isset($ns['dc'])) {
$ns_dc = $entry->children($ns['dc']);
$author = (string) $ns_dc->creator;
$pubdate = (string) $ns_dc->date;
}
if (isset($ns['content'])) {
$ns_content = $entry->children($ns['content']);
$content = (string) $ns_content->encoded;
}
if ($content === '' && isset($entry->description)) {
$content = (string) $entry->description;
}
$item = new \StdClass; $item = new \StdClass;
$item->title = (string) $entry->title; $item->title = (string) $entry->title;
$item->url = $link ?: (string) $entry->link; $item->url = '';
$item->id = $item->url; $item->author= '';
$item->updated = $pubdate ? strtotime($pubdate) : time(); $item->updated = '';
$item->content = $this->filterHtml($content, $item->url); $item->content = '';
$item->author = $author ?: (string) $xml->channel->webMaster;
foreach ($namespaces as $name => $url) {
$namespace = $entry->children($namespaces[$name]);
if (! $item->url && ! empty($namespace->origLink)) $item->url = (string) $namespace->origLink;
if (! $item->author && ! empty($namespace->creator)) $item->author = (string) $namespace->creator;
if (! $item->updated && ! empty($namespace->date)) $item->updated = strtotime((string) $namespace->date);
if (! $item->updated && ! empty($namespace->updated)) $item->updated = strtotime((string) $namespace->updated);
if (! $item->content && ! empty($namespace->encoded)) $item->content = (string) $namespace->encoded;
}
if (empty($item->url)) $item->url = (string) $entry->link;
if (empty($item->updated)) $item->updated = $this->updated;
if (empty($item->content)) {
$item->content = isset($entry->description) ? (string) $entry->description : '';
}
if (empty($item->author)) {
if (isset($entry->author)) {
$item->author = (string) $entry->author;
}
else if (isset($xml->channel->webMaster)) {
$item->author = (string) $xml->channel->webMaster;
}
}
$item->id = $item->url;
$item->content = $this->filterHtml($item->content, $item->url);
$this->items[] = $item; $this->items[] = $item;
} }

View File

@ -9,7 +9,6 @@ class Rss20 extends Parser
$this->content = $this->normalizeData($this->content); $this->content = $this->normalizeData($this->content);
\libxml_use_internal_errors(true); \libxml_use_internal_errors(true);
$xml = \simplexml_load_string($this->content); $xml = \simplexml_load_string($this->content);
if ($xml === false) { if ($xml === false) {
@ -18,73 +17,64 @@ class Rss20 extends Parser
return false; return false;
} }
$ns = $xml->getNamespaces(true); $namespaces = $xml->getNamespaces(true);
$this->title = (string) $xml->channel->title; $this->title = (string) $xml->channel->title;
$this->url = (string) $xml->channel->link; $this->url = (string) $xml->channel->link;
$this->id = $this->url; $this->id = $this->url;
$this->updated = isset($xml->channel->pubDate) ? (string) $xml->channel->pubDate : (string) $xml->channel->lastBuildDate; $this->updated = isset($xml->channel->pubDate) ? (string) $xml->channel->pubDate : (string) $xml->channel->lastBuildDate;
$this->updated = $this->updated ? strtotime($this->updated) : time();
if ($this->updated) {
$this->updated = strtotime($this->updated);
}
else {
$this->updated = time();
}
foreach ($xml->channel->item as $entry) { foreach ($xml->channel->item as $entry) {
$author = ''; $item = new \StdClass;
$content = ''; $item->title = (string) $entry->title;
$pubdate = ''; $item->url = '';
$link = ''; $item->author= '';
$item->updated = '';
$item->content = '';
if (isset($ns['feedburner'])) { foreach ($namespaces as $name => $url) {
$ns_fb = $entry->children($ns['feedburner']); $namespace = $entry->children($namespaces[$name]);
$link = $ns_fb->origLink;
if (! $item->url && ! empty($namespace->origLink)) $item->url = (string) $namespace->origLink;
if (! $item->author && ! empty($namespace->creator)) $item->author = (string) $namespace->creator;
if (! $item->updated && ! empty($namespace->date)) $item->updated = strtotime((string) $namespace->date);
if (! $item->updated && ! empty($namespace->updated)) $item->updated = strtotime((string) $namespace->updated);
if (! $item->content && ! empty($namespace->encoded)) $item->content = (string) $namespace->encoded;
} }
if (isset($ns['dc'])) { if (empty($item->url)) $item->url = (string) $entry->link;
if (empty($item->updated)) $item->updated = strtotime((string) $entry->pubDate) ?: $this->updated;
$ns_dc = $entry->children($ns['dc']); if (empty($item->content)) {
$author = (string) $ns_dc->creator;
$pubdate = (string) $ns_dc->date; $item->content = isset($entry->description) ? (string) $entry->description : '';
} }
if (isset($ns['content'])) { if (empty($item->author)) {
$ns_content = $entry->children($ns['content']);
$content = (string) $ns_content->encoded;
}
if ($content === '' && isset($entry->description)) {
$content = (string) $entry->description;
}
if ($author === '') {
if (isset($entry->author)) { if (isset($entry->author)) {
$author = (string) $entry->author; $item->author = (string) $entry->author;
} }
else if (isset($xml->channel->webMaster)) { else if (isset($xml->channel->webMaster)) {
$author = (string) $xml->channel->webMaster; $item->author = (string) $xml->channel->webMaster;
} }
} }
$item = new \StdClass; if (isset($entry->guid) && isset($entry->guid['isPermaLink']) && (string) $entry->guid['isPermaLink'] != 'false') {
$item->title = (string) $entry->title;
$item->url = $link ?: (string) $entry->link;
$item->id = isset($entry->guid) ? (string) $entry->guid : $item->url;
$item->updated = strtotime($pubdate ?: (string) $entry->pubDate) ?: $this->updated;
$item->content = $this->filterHtml($content, $item->url);
$item->author = $author;
$item->id = (string) $entry->guid;
}
else {
$item->id = $item->url;
}
$item->content = $this->filterHtml($item->content, $item->url);
$this->items[] = $item; $this->items[] = $item;
} }