Add canonical feed URLs

lib/Feed.php

@@ -9,7 +9,9 @@ namespace JKingWeb\Lax;
 use JKingWeb\Lax\Person\Person;
 use JKingWeb\Lax\Person\Collection as PersonCollection;
 
-abstract class Feed {    
+abstract class Feed {
+    protected $reqUrl;
+
     public $type;
     public $version;
     public $url;
@@ -26,6 +28,7 @@ abstract class Feed {
     /** Parses the feed to extract sundry metadata */
     protected function parse() {
         $this->id = $this->getId();
+        $this->url = $this->getUrl();
         $this->link = $this->getLink();
         $this->title = $this->getTitle();
         $this->summary = $this->getSummary();
@@ -36,6 +39,9 @@ abstract class Feed {
         $this->title = strlen($this->title) ? $this->title : $this->link;
     }
     
+    /** General function to fetch the canonical feed URL */
+    abstract public function getUrl(): string;
+    
     /** General function to fetch the feed title */
     abstract public function getTitle(): string;
 

lib/JSON/Feed.php

@@ -12,23 +12,29 @@ use JKingWeb\Lax\Person\Collection as PersonCollection;
 class Feed extends \JKingWeb\Lax\Feed {
     use Construct;
 
-    protected $reqUrl;
-
     /** Constructs a parsed feed */
     public function __construct(string $data, string $contentType = "", string $url = "") {
-        $this->reqUrl = $url;
         $this->init($data, $contentType);
         $this->parse();
     }
 
     /** Performs initialization of the instance */
-    protected function init(string $data, string $contentType = "") {
+    protected function init(string $data, string $contentType = "", string $url = "") {
+        $this->reqUrl = $url;
         $this->json = json_decode($data);
         $this->url = $this->reqUrl;
         $this->type = "json";
         $this->version = $this->fetchMember("version", "str") ?? "";
     }
     
+    /** General function to fetch the canonical feed URL
+     * 
+     * If the feed does not include a canonical URL, the request URL is returned instead
+     */
+    public function getUrl(): string {
+        return $this->fetchUrl("feed_url") ?? $this->reqUrl;
+    }
+    
     /** General function to fetch the feed title */
     public function getTitle(): string {
         return $this->fetchMember("title", "str") ?? "";
@@ -57,8 +63,7 @@ class Feed extends \JKingWeb\Lax\Feed {
      * For JSON feeds this is always the feed URL specified in the feed
     */
     public function getId(): string {
-        //return $this->getUrl();
-        return "";
+        return $this->fetchUrl("feed_url") ?? "";
     }
 
     /** General function to fetch a collection of people associated with a feed */

lib/XML/Construct.php

@@ -21,7 +21,10 @@ trait Construct {
     /** Retrieves an element node based on an XPath query */
     protected function fetchElement(string $query, \DOMNode $context = null) {
         $context = $context ?? $this->subject;
-        $node = $this->xpath->query("(".$query.")[1]", $context ?? $this->subject);
+        $node = @$this->xpath->query("(".$query.")[1]", $context ?? $this->subject);
+        if ($node===false) {
+            throw new \Exception("Invalid XPath query: $query");
+        }
         return ($node->length) ? $node->item(0) : null;
     }
 

lib/XML/Feed.php

@@ -22,6 +22,7 @@ class Feed extends \JKingWeb\Lax\Feed {
 
     /** Performs initialization of the instance */
     protected function init(string $data, string $contentType = "", string $url = "") {
+        $this->reqUrl = $url;
         $this->document = new \DOMDocument();
         $this->document->loadXML($data, \LIBXML_BIGLINES | \LIBXML_COMPACT);
         $this->document->documentURI = $url;
@@ -54,6 +55,14 @@ class Feed extends \JKingWeb\Lax\Feed {
         $this->url = $url;
     }
     
+    /** General function to fetch the canonical feed URL
+     * 
+     * If the feed does not include a canonical URL, the request URL is returned instead
+     */
+    public function getUrl(): string {
+        return $this->getUrlAtom() ?? $this->getUrlRss1() ?? $this->getUrlPod() ?? $this->reqUrl;
+    }
+    
     /** General function to fetch the feed title */
     public function getTitle(): string {
         return $this->getTitleAtom() ?? $this->getTitleRss1() ?? $this->getTitleRss2() ?? $this->getTitleDC() ?? $this->getTitlePod() ?? "";

lib/XML/Primitives/Construct.php

@@ -183,4 +183,10 @@ trait Construct {
         $out[] = $p;
         return $out;
     }
+
+    /** Primitive to fetch an Atom feed or entry's canonical URL */
+    protected function getUrlAtom() {
+        $node = $this->fetchAtomRelations("self");
+        return $node->length ? $this->resolveNodeUrl($node->item(0), "href") : null;
+    }
 }

lib/XML/Primitives/Entry.php

@@ -7,6 +7,7 @@ declare(strict_types=1);
 namespace JKingWeb\Lax\XML\Primitives;
 
 use JKingWeb\Lax\Person\Collection as PersonCollection;
+use JKingWeb\Lax\XML\XPath;
 
 trait Entry {
 
@@ -55,4 +56,15 @@ trait Entry {
         }
         return $out;
     }
+
+    /** Primitive to fetch an RDF entry's canonical URL */
+    protected function getUrlRss1() {
+        // XPath doesn't seem to like the query we'd need for this, so it must be done the hard way.
+        $node = $this->subject;
+        if ($node->localName=="item" && ($node->namespaceURI==XPath::NS['rss1'] || $node->namespaceURI==XPath::NS['rss0']) && $node->hasAttributeNS(XPath::NS['rdf'], "about")) {
+            return $this->resolveNodeUrl($node, "about", XPath::NS['rdf']);
+        } else {
+            return null;
+        }
+    }
 }

lib/XML/Primitives/Feed.php

@@ -7,6 +7,7 @@ declare(strict_types=1);
 namespace JKingWeb\Lax\XML\Primitives;
 
 use JKingWeb\Lax\Person\Collection as PersonCollection;
+use JKingWeb\Lax\XML\XPath;
 
 trait Feed {
 
@@ -76,4 +77,29 @@ trait Feed {
         }
         return $out;
     }
+
+    /** Primitive to fetch an RDF feed's canonical URL */
+    protected function getUrlRss1() {
+        // XPath doesn't seem to like the query we'd need for this, so it must be done the hard way.
+        $node = $this->subject;
+        if ($node->hasAttributeNS(XPath::NS['rdf'], "about")) {
+            if (
+                ($node->localName=="channel" && ($node->namespaceURI==XPath::NS['rss1'] || $node->namespaceURI==XPath::NS['rss0'])) ||
+                ($node==$node->ownerDocument->documentElement && $node->localName=="RDF" && $node->namespaceURI==XPath::NS['rdf'])
+            ) {
+                return $this->resolveNodeUrl($node, "about", XPath::NS['rdf']);
+            }
+        }
+        return null;
+    }
+
+    /** Primitive to fetch an podcast's canonical URL */
+    protected function getUrlPod() {
+        $node =  $this->fetchElement("./apple:new-feed-url");
+        if ($node) {
+            return $this->resolveNodeUrl($node);
+        } else {
+            return null;
+        }
+    }
 }