Initial category implementation

lib/Parser/XML/Construct.php

@@ -6,6 +6,8 @@
 declare(strict_types=1);
 namespace MensBeam\Lax\Parser\XML;
 
+use MensBeam\Lax\Category\Category;
+use MensBeam\Lax\Category\Collection as CategoryCollection;
 use MensBeam\Lax\Person\Person;
 use MensBeam\Lax\Person\Collection as PersonCollection;
 use MensBeam\Lax\Text;
@@ -77,7 +79,7 @@ abstract class Construct {
      * @param \DOMNode $context The context node for the XPath query
      * @return \MensBeam\Lax\Date|array|null
      */
-    protected function fetchDate(string $query, int $mode = self::DATE_ANY, \DOMNode $context = null) {
+    protected function fetchDate(string $query, int $mode, \DOMNode $context = null) {
         $out = [];
         $tz = new \DateTimeZone("UTC");
         assert(in_array($mode, [self::DATE_ANY, self::DATE_ALL, self::DATE_EARLIEST, self::DATE_LATEST]));
@@ -327,5 +329,56 @@ abstract class Construct {
     protected function getTitlePod(): ?Text {
         return $this->fetchText("apple:title", self::TEXT_PLAIN);
     }
-    
+
+    protected function getCategoriesAtom(): ?CategoryCollection {
+        $out = new CategoryCollection;
+        foreach ($this->xpath->query("atom:category[@term]") as $node) {
+            $c = new Category;
+            $c->domain = $this->trimText($node->getAttribute("scheme"));
+            $c->label = $this->trimText($node->getAttribute("label"));
+            $c->name = $this->trimText($node->getAttribute("term"));
+            if (strlen($c->name)) {
+                $out[] = $c;
+            }
+        }
+        return count($out) ? $out : null;
+    }
+
+    protected function getCategoriesRss2(): ?CategoryCollection {
+        $out = new CategoryCollection;
+        foreach ($this->xpath->query("category") as $node) {
+            $c = new Category;
+            $c->domain = $this->trimText($node->getAttribute("domain"));
+            $c->name = $this->trimText($node->textContent);
+            if (strlen($c->name)) {
+                $out[] = $c;
+            }
+        }
+        return count($out) ? $out : null;
+    }
+
+    /** Dublin Core doesn't have an obvious category type, so we use 'subject' as a nearest approximation */
+    protected function getCategoriesDC(): ?CategoryCollection {
+        $out = new CategoryCollection;
+        foreach ($this->fetchString("dc:subject", null, true) ?? [] as $text) {
+            if (strlen($text)) {
+                $c = new Category;
+                $c->name = $text;
+                $out[] = $c;
+            }
+        }
+        return count($out) ? $out : null;
+    }
+
+    protected function getCategoriesPod(): ?CategoryCollection {
+        $out = new CategoryCollection;
+        foreach ($this->xpath->query("apple:category|gplay:category") ?? [] as $node) {
+            $c = new Category;
+            $c->name = $this->trimText($node->getAttribute("text"));
+            if (strlen($c->name)) {
+                $out[] = $c;
+            }
+        }
+        return count($out) ? $out : null;
+    }    
 }

lib/Parser/XML/Feed.php

@@ -92,7 +92,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
         $feed->icon = $this->getIcon();
         $feed->image = $this->getImage();
         //$feed->people = $this->getPeople();
-        //$feed->categories = $this->getCategories();
+        $feed->categories = $this->getCategories();
         //$feed->entries = $this->getEntries($feed);
         return $feed;
     }
@@ -105,14 +105,14 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
 
     public function getSchedule(): Schedule {
         $sched = new Schedule;
-        $sched->interval = $this->getSchedIntervalRss1() ?? $this->getSchedIntervalRss2();
+        $sched->interval =  $this->getSchedIntervalRss1() ?? $this->getSchedIntervalRss2();
         $sched->skip = $this->getSchedSkipRss2();
         $sched->expired = $this->getExpiredPod();
         if (is_null($sched->expired) && (($sched->skip & Schedule::DAY_ALL) == Schedule::DAY_ALL || ($sched->skip & Schedule::HOUR_ALL) == Schedule::HOUR_ALL)) {
             $sched->expired = true;
         }
         if ($sched->interval) {
-            $sched->base = $this->getSchedBaseRss1();
+            $sched->base = $this->fetchDate("sched:updateBase", self::DATE_ANY);
         }
         return $sched;
     }
@@ -178,7 +178,11 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
     }
 
     public function getCategories(): CategoryCollection {
-        return $this->getCategoriesAtom() ?? $this->getCategoriesRss2() ?? $this->getCategoriesDC() ?? $this->getCategoriesPod() ?? new CategoryCollection;
+        return $this->getCategoriesAtom() 
+            ?? $this->getCategoriesRss2() 
+            ?? $this->getCategoriesPod() 
+            ?? $this->getCategoriesDC() 
+            ?? new CategoryCollection;
     }
 
     public function getPeople(): PersonCollection {
@@ -233,11 +237,6 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
         return null;
     }
 
-    protected function getSchedBaseRss1(): ?Date {
-        return $this->fetchDate("sched:updateBase");
-    }
-
-
     /** Computes the "skip-schedule" of an RSS feed, the set of days and hours during which a feed should not be fetched */
     protected function getSchedSkipRss2(): ?int {
         $out = 0;

lib/Parser/XML/OldFeed.php

@@ -1,74 +0,0 @@
-<?php
-/** @license MIT
- * Copyright 2018 J. King et al.
- * See LICENSE and AUTHORS files for details */
-
-declare(strict_types=1);
-namespace MensBeam\Lax\Parser\XML\Primitives;
-
-use MensBeam\Lax\Parser\XML\XPath;
-use MensBeam\Lax\Person\Collection as PersonCollection;
-use MensBeam\Lax\Date;
-use MensBeam\Lax\Schedule;
-use MensBeam\Lax\Text;
-use MensBeam\Lax\Url;
-
-trait Feed {
-    /** Primitive to fetch an Atom feed summary
-     *
-     * Atom does not have a 'description' element like the RSSes, but it does have 'subtitle', which fills roughly the same function
-     */
-    protected function getSummaryAtom(): ?Text {
-        return $this->fetchStringAtom("atom:subtitle");
-    }
-
-    /** Primitive to fetch an RSS feed summary */
-    protected function getSummaryRss2(): ?Text {
-        return $this->fetchString("description");
-    }
-
-    /** Primitive to fetch an RDF feed summary */
-    protected function getSummaryRss1(): ?Text {
-        return $this->fetchString("rss1:description|rss0:description");
-    }
-
-    /** Primitive to fetch a Dublin Core feed summary */
-    protected function getSummaryDC(): ?Text {
-        return $this->fetchString("dc:description");
-    }
-
-    /** Primitive to fetch a podcast summary */
-    protected function getSummaryPod(): ?Text {
-        return $this->fetchString("apple:summary|gplay:description") ?? $this->fetchString("apple:subtitle");
-    }
-
-    /** Primitive to fetch a collection of authors associated with an Atom feed */
-    protected function getAuthorsAtom(): ?PersonCollection {
-        return $this->fetchPeopleAtom("atom:author", "author");
-    }
-
-    /** Primitive to fetch an RDF feed's canonical URL */
-    protected function getUrlRss1(): ?Url {
-        // 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 a podcast's canonical URL */
-    protected function getUrlPod(): ?Url {
-        return $this->fetchUrl("apple:new-feed-url");
-    }
-
-    /** Primitive to fetch the modification date of an RSS feed */
-    protected function getDateModifiedRss2(): ?Date {
-        return $this->fetchDate("lastBuildDate") ?? $this->fetchDate("pubDate");
-    }
-}

tests/cases/AbstractParserTestCase.php

@@ -84,6 +84,16 @@ class AbstractParserTestCase extends \PHPUnit\Framework\TestCase {
                     $c[] = $this->makePerson($m);
                 }
                 $f->$k = $c;
+            } elseif ($k === "categories") {
+                $c = new CategoryCollection;
+                foreach ($v as $m) {
+                    $o = new Category;
+                    foreach ($m as $kk => $vv) {
+                        $o->$kk = $vv;
+                    }
+                    $c[] = $o;
+                }
+                $f->$k = $c;
             } elseif ($k === "entries") {
                 $c = [];
                 foreach ($v as $m) {

tests/cases/XML/feed-atom.yaml

@@ -445,3 +445,22 @@ Logo URL:
         format: atom
         version: '1.0'
         image: 'http://example.com/'
+
+Categories:
+    input: >
+        <feed xmlns="http://www.w3.org/2005/Atom">
+            <category term="ook"/>
+            <category term="eek" label="Eek!"/>
+            <category term="ack" label="Ack!" scheme="4:3"/>
+            <category term=""    label="Bogus"/>
+        </feed>
+    output:
+        format: atom
+        version: '1.0'
+        categories:
+            - name: ook
+            - name: eek
+              label: 'Eek!'
+            - name: ack
+              label: 'Ack!'
+              domain: '4:3' # Not treated as a URI