diff options
Diffstat (limited to 'classes/feeditem')
| -rw-r--r-- | classes/feeditem/atom.php | 76 | ||||
| -rw-r--r-- | classes/feeditem/common.php | 28 | ||||
| -rw-r--r-- | classes/feeditem/rss.php | 76 |
3 files changed, 160 insertions, 20 deletions
diff --git a/classes/feeditem/atom.php b/classes/feeditem/atom.php index 9680748f9..dfac7149f 100644 --- a/classes/feeditem/atom.php +++ b/classes/feeditem/atom.php @@ -1,5 +1,6 @@ <?php class FeedItem_Atom extends FeedItem_Common { + function get_id() { $id = $this->elem->getElementsByTagName("id")->item(0); @@ -30,6 +31,7 @@ class FeedItem_Atom extends FeedItem_Common { } } + function get_link() { $links = $this->elem->getElementsByTagName("link"); @@ -38,8 +40,13 @@ class FeedItem_Atom extends FeedItem_Common { (!$link->hasAttribute("rel") || $link->getAttribute("rel") == "alternate" || $link->getAttribute("rel") == "standout")) { + $base = $this->xpath->evaluate("string(ancestor-or-self::*[@xml:base][1]/@xml:base)", $link); + + if ($base) + return rewrite_relative_url($base, trim($link->getAttribute("href"))); + else + return trim($link->getAttribute("href")); - return $link->getAttribute("href"); } } } @@ -48,7 +55,7 @@ class FeedItem_Atom extends FeedItem_Common { $title = $this->elem->getElementsByTagName("title")->item(0); if ($title) { - return $title->nodeValue; + return trim($title->nodeValue); } } @@ -58,7 +65,13 @@ class FeedItem_Atom extends FeedItem_Common { if ($content) { if ($content->hasAttribute('type')) { if ($content->getAttribute('type') == 'xhtml') { - return $this->doc->saveXML($content->firstChild->nextSibling); + for ($i = 0; $i < $content->childNodes->length; $i++) { + $child = $content->childNodes->item($i); + + if ($child->hasChildNodes()) { + return $this->doc->saveXML($child); + } + } } } @@ -72,7 +85,13 @@ class FeedItem_Atom extends FeedItem_Common { if ($content) { if ($content->hasAttribute('type')) { if ($content->getAttribute('type') == 'xhtml') { - return $this->doc->saveXML($content->firstChild->nextSibling); + for ($i = 0; $i < $content->childNodes->length; $i++) { + $child = $content->childNodes->item($i); + + if ($child->hasChildNodes()) { + return $this->doc->saveXML($child); + } + } } } @@ -87,13 +106,13 @@ class FeedItem_Atom extends FeedItem_Common { foreach ($categories as $cat) { if ($cat->hasAttribute("term")) - array_push($cats, $cat->getAttribute("term")); + array_push($cats, trim($cat->getAttribute("term"))); } $categories = $this->xpath->query("dc:subject", $this->elem); foreach ($categories as $cat) { - array_push($cats, $cat->nodeValue); + array_push($cats, trim($cat->nodeValue)); } return $cats; @@ -126,6 +145,51 @@ class FeedItem_Atom extends FeedItem_Common { $enc->type = $enclosure->getAttribute("type"); $enc->link = $enclosure->getAttribute("url"); $enc->length = $enclosure->getAttribute("length"); + $enc->height = $enclosure->getAttribute("height"); + $enc->width = $enclosure->getAttribute("width"); + + $desc = $this->xpath->query("media:description", $enclosure)->item(0); + if ($desc) $enc->title = strip_tags($desc->nodeValue); + + array_push($encs, $enc); + } + + + $enclosures = $this->xpath->query("media:group", $this->elem); + + foreach ($enclosures as $enclosure) { + $enc = new FeedEnclosure(); + + $content = $this->xpath->query("media:content", $enclosure)->item(0); + + if ($content) { + $enc->type = $content->getAttribute("type"); + $enc->link = $content->getAttribute("url"); + $enc->length = $content->getAttribute("length"); + $enc->height = $content->getAttribute("height"); + $enc->width = $content->getAttribute("width"); + + $desc = $this->xpath->query("media:description", $content)->item(0); + if ($desc) { + $enc->title = strip_tags($desc->nodeValue); + } else { + $desc = $this->xpath->query("media:description", $enclosure)->item(0); + if ($desc) $enc->title = strip_tags($desc->nodeValue); + } + + array_push($encs, $enc); + } + } + + $enclosures = $this->xpath->query("media:thumbnail", $this->elem); + + foreach ($enclosures as $enclosure) { + $enc = new FeedEnclosure(); + + $enc->type = "image/generic"; + $enc->link = $enclosure->getAttribute("url"); + $enc->height = $enclosure->getAttribute("height"); + $enc->width = $enclosure->getAttribute("width"); array_push($encs, $enc); } diff --git a/classes/feeditem/common.php b/classes/feeditem/common.php index 0787a42cb..80bebf8fb 100644 --- a/classes/feeditem/common.php +++ b/classes/feeditem/common.php @@ -8,6 +8,17 @@ abstract class FeedItem_Common extends FeedItem { $this->elem = $elem; $this->xpath = $xpath; $this->doc = $doc; + + try { + + $source = $elem->getElementsByTagName("source")->item(0); + + // we don't need <source> element + if ($source) + $elem->removeChild($source); + } catch (DOMException $e) { + // + } } function get_author() { @@ -33,13 +44,26 @@ abstract class FeedItem_Common extends FeedItem { } } - // todo function get_comments_url() { + //RSS only. Use a query here to avoid namespace clashes (e.g. with slash). + //might give a wrong result if a default namespace was declared (possible with XPath 2.0) + $com_url = $this->xpath->query("comments", $this->elem)->item(0); + + if($com_url) + return $com_url->nodeValue; + + //Atom Threading Extension (RFC 4685) stuff. Could be used in RSS feeds, so it's in common. + //'text/html' for type is too restrictive? + $com_url = $this->xpath->query("atom:link[@rel='replies' and contains(@type,'text/html')]/@href", $this->elem)->item(0); + if($com_url) + return $com_url->nodeValue; } function get_comments_count() { - $comments = $this->xpath->query("slash:comments", $this->elem)->item(0); + //also query for ATE stuff here + $query = "slash:comments|thread:total|atom:link[@rel='replies']/@thread:count"; + $comments = $this->xpath->query($query, $this->elem)->item(0); if ($comments) { return $comments->nodeValue; diff --git a/classes/feeditem/rss.php b/classes/feeditem/rss.php index e5960243c..c9a7467cd 100644 --- a/classes/feeditem/rss.php +++ b/classes/feeditem/rss.php @@ -33,20 +33,20 @@ class FeedItem_RSS extends FeedItem_Common { || $link->getAttribute("rel") == "alternate" || $link->getAttribute("rel") == "standout")) { - return $link->getAttribute("href"); + return trim($link->getAttribute("href")); } } $link = $this->elem->getElementsByTagName("guid")->item(0); if ($link && $link->hasAttributes() && $link->getAttribute("isPermaLink") == "true") { - return $link->nodeValue; + return trim($link->nodeValue); } $link = $this->elem->getElementsByTagName("link")->item(0); if ($link) { - return $link->nodeValue; + return trim($link->nodeValue); } } @@ -54,21 +54,26 @@ class FeedItem_RSS extends FeedItem_Common { $title = $this->elem->getElementsByTagName("title")->item(0); if ($title) { - return $title->nodeValue; + return trim($title->nodeValue); } } function get_content() { - $content = $this->xpath->query("content:encoded", $this->elem)->item(0); + $contentA = $this->xpath->query("content:encoded", $this->elem)->item(0); + $contentB = $this->elem->getElementsByTagName("description")->item(0); - if ($content) { - return $content->nodeValue; + if ($contentA && !$contentB) { + return $contentA->nodeValue; } - $content = $this->elem->getElementsByTagName("description")->item(0); - if ($content) { - return $content->nodeValue; + if ($contentB && !$contentA) { + return $contentB->nodeValue; + } + + if ($contentA && $contentB) { + return mb_strlen($contentA->nodeValue) > mb_strlen($contentB->nodeValue) ? + $contentA->nodeValue : $contentB->nodeValue; } } @@ -85,13 +90,13 @@ class FeedItem_RSS extends FeedItem_Common { $cats = array(); foreach ($categories as $cat) { - array_push($cats, $cat->nodeValue); + array_push($cats, trim($cat->nodeValue)); } $categories = $this->xpath->query("dc:subject", $this->elem); foreach ($categories as $cat) { - array_push($cats, $cat->nodeValue); + array_push($cats, trim($cat->nodeValue)); } return $cats; @@ -108,6 +113,8 @@ class FeedItem_RSS extends FeedItem_Common { $enc->type = $enclosure->getAttribute("type"); $enc->link = $enclosure->getAttribute("url"); $enc->length = $enclosure->getAttribute("length"); + $enc->height = $enclosure->getAttribute("height"); + $enc->width = $enclosure->getAttribute("width"); array_push($encs, $enc); } @@ -120,6 +127,51 @@ class FeedItem_RSS extends FeedItem_Common { $enc->type = $enclosure->getAttribute("type"); $enc->link = $enclosure->getAttribute("url"); $enc->length = $enclosure->getAttribute("length"); + $enc->height = $enclosure->getAttribute("height"); + $enc->width = $enclosure->getAttribute("width"); + + $desc = $this->xpath->query("media:description", $enclosure)->item(0); + if ($desc) $enc->title = strip_tags($desc->nodeValue); + + array_push($encs, $enc); + } + + + $enclosures = $this->xpath->query("media:group", $this->elem); + + foreach ($enclosures as $enclosure) { + $enc = new FeedEnclosure(); + + $content = $this->xpath->query("media:content", $enclosure)->item(0); + + if ($content) { + $enc->type = $content->getAttribute("type"); + $enc->link = $content->getAttribute("url"); + $enc->length = $content->getAttribute("length"); + $enc->height = $content->getAttribute("height"); + $enc->width = $content->getAttribute("width"); + + $desc = $this->xpath->query("media:description", $content)->item(0); + if ($desc) { + $enc->title = strip_tags($desc->nodeValue); + } else { + $desc = $this->xpath->query("media:description", $enclosure)->item(0); + if ($desc) $enc->title = strip_tags($desc->nodeValue); + } + + array_push($encs, $enc); + } + } + + $enclosures = $this->xpath->query("media:thumbnail", $this->elem); + + foreach ($enclosures as $enclosure) { + $enc = new FeedEnclosure(); + + $enc->type = "image/generic"; + $enc->link = $enclosure->getAttribute("url"); + $enc->height = $enclosure->getAttribute("height"); + $enc->width = $enclosure->getAttribute("width"); array_push($encs, $enc); } |