Entry categories

lib/Parser/XML/Construct.php

@@ -389,9 +389,17 @@ abstract class Construct {
         return $this->fetchText("apple:title", self::TEXT_PLAIN);
     }
 
-    protected function getCategoriesAtom(): ?CategoryCollection {
+    protected function getCategoriesFromNode(\DOMNode $context): ?CategoryCollection {
+        return $this->getCategoriesAtom($context)
+            ?? $this->getCategoriesRss2($context)
+            ?? $this->getCategoriesGPlay($context)
+            ?? $this->getCategoriesTunes($context)
+            ?? $this->getCategoriesDC($context);
+    }
+
+    protected function getCategoriesAtom(\DOMNode $context): ?CategoryCollection {
         $out = new CategoryCollection;
-        foreach ($this->xpath->query("atom:category[@term]") as $node) {
+        foreach ($this->xpath->query("atom:category[@term]", $context) as $node) {
             $c = new Category;
             $c->domain = $this->trimText($node->getAttribute("scheme"));
             $c->label = $this->trimText($node->getAttribute("label"));
@@ -403,9 +411,9 @@ abstract class Construct {
         return count($out) ? $out : null;
     }
 
-    protected function getCategoriesRss2(): ?CategoryCollection {
+    protected function getCategoriesRss2(\DOMNode $context): ?CategoryCollection {
         $out = new CategoryCollection;
-        foreach ($this->xpath->query("rss2:category", $this->subject) as $node) {
+        foreach ($this->xpath->query("rss2:category", $context) as $node) {
             $c = new Category;
             $c->domain = $this->trimText($node->getAttribute("domain"));
             $c->name = $this->trimText($node->textContent);
@@ -417,9 +425,9 @@ abstract class Construct {
     }
 
     /** Dublin Core doesn't have an obvious category type, so we use 'subject' as a nearest approximation */
-    protected function getCategoriesDC(): ?CategoryCollection {
+    protected function getCategoriesDC(\DOMNode $context): ?CategoryCollection {
         $out = new CategoryCollection;
-        foreach ($this->fetchString("dc:subject|dct:subject", null, true) ?? [] as $text) {
+        foreach ($this->fetchString("dc:subject|dct:subject", null, true, $context) ?? [] as $text) {
             if (strlen($text)) {
                 $c = new Category;
                 $c->name = $text;
@@ -429,9 +437,9 @@ abstract class Construct {
         return count($out) ? $out : null;
     }
 
-    protected function getCategoriesTunes(): ?CategoryCollection {
+    protected function getCategoriesTunes(\DOMNode $context): ?CategoryCollection {
         $out = new CategoryCollection;
-        foreach ($this->xpath->query("apple:category", $this->subject) as $node) {
+        foreach ($this->xpath->query("apple:category", $context) as $node) {
             $c = new Category;
             $c->name = $this->trimText($node->getAttribute("text"));
             if (strlen($c->name)) {
@@ -448,9 +456,9 @@ abstract class Construct {
         return count($out) ? $out : null;
     }
 
-    protected function getCategoriesGPlay(): ?CategoryCollection {
+    protected function getCategoriesGPlay(\DOMNode $context): ?CategoryCollection {
         $out = new CategoryCollection;
-        foreach ($this->xpath->query("gplay:category", $this->subject) as $node) {
+        foreach ($this->xpath->query("gplay:category", $context) as $node) {
             $c = new Category;
             $c->name = $this->trimText($node->getAttribute("text"));
             if (strlen($c->name)) {

lib/Parser/XML/Entry.php

@@ -141,7 +141,16 @@ class Entry extends Construct implements \MensBeam\Lax\Parser\Entry {
     }
 
     public function getCategories(): CategoryCollection {
-        return new CategoryCollection;
+        // first try to get categories from the entry itself
+        $list = $this->getCategoriesFromNode($this->subject);
+        if (!$list) {
+            // if there are none, try to get some from the entry's Atom <source> element, if any
+            $src = $this->fetchElement("atom:source");
+            if ($src) {
+                $list = $this->getCategoriesFromNode($src);
+            }
+        }
+        return $list ?? new CategoryCollection;
     }
 
     public function getEnclosures(): EnclosureCollection {

lib/Parser/XML/Feed.php

@@ -191,12 +191,7 @@ class Feed extends Construct implements \MensBeam\Lax\Parser\Feed {
     }
 
     public function getCategories(): CategoryCollection {
-        return $this->getCategoriesAtom()
-            ?? $this->getCategoriesRss2()
-            ?? $this->getCategoriesGPlay()
-            ?? $this->getCategoriesTunes()
-            ?? $this->getCategoriesDC()
-            ?? new CategoryCollection;
+            return $this->getCategoriesFromNode($this->subject) ?? new CategoryCollection;
     }
 
     public function getPeople(): PersonCollection {

lib/Parser/XML/OldEntry.php

@@ -1,35 +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;
-
-trait Entry {
-    /** Primitive to fetch a collection of authors associated with an Atom entry
-     *
-     * This differs from feeds in that an entry's <source> element (which possibly contains metadata for the source feed) is checked for authors if the entry itself has none
-     */
-    protected function getAuthorsAtom() {
-        return $this->fetchPeopleAtom("atom:author", "author") ?? $this->fetchPeopleAtom("atom:source[1]/atom:author", "author");
-    }
-
-    /** 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;
-        }
-    }
-
-    /** Primitive to fetch the modification date of an RSS feed */
-    protected function getDateModifiedRss2() {
-        return $this->fetchDate("pubDate");
-    }
-}

tests/cases/XML/entry-atom.yaml

@@ -249,3 +249,37 @@ Authors and contributors:
                       role: contributor
                     - name: Curly
                       role: contributor
+
+Categories:
+    input: >
+        <feed xmlns="http://www.w3.org/2005/Atom">
+            <entry>
+                <category term="ook"/>
+                <category term="eek"/>
+            </entry>
+            <entry>
+                <source>
+                    <category term="ook"/>
+                    <category term="eek"/>
+                </source>
+            </entry>
+            <entry>
+                <category term="ack"/>
+                <source>
+                    <category term="ook"/>
+                    <category term="eek"/>
+                </source>
+            </entry>
+        </feed>
+    output:
+        format: atom
+        version: '1.0'
+        entries:
+            - categories:
+                - name: ook
+                - name: eek
+            - categories:
+                - name: ook
+                - name: eek
+            - categories:
+                - name: ack

tests/cases/XML/entry-mixed.yaml

@@ -93,3 +93,31 @@ Entry author:
             - people:
                     - name: 'Jane Doe'
                       role: author
+
+Categories:
+    input: >
+        <rss><channel xmlns:atom="http://www.w3.org/2005/Atom">
+            <item>
+                <category>ook</category>
+            </item>
+            <item>
+                <atom:source>
+                    <category>ook</category>
+                </atom:source>
+            </item>
+            <item>
+                <category>eek</category>
+                <atom:source>
+                    <category>ook</category>
+                </atom:source>
+            </item>
+        </channel></rss>
+    output:
+        format: rss
+        entries:
+            - categories:
+                - name: ook
+            - categories:
+                - name: ook
+            - categories:
+                - name: eek

tests/cases/XML/entry-other.yaml

@@ -102,3 +102,47 @@ Mixed podcast authors: # Google Play is arbitrarily preferred
                     - name: 'John Doe'
                       mail: 'john.doe@example.com'
                       role: author
+
+iTunes categories:
+    input: >
+        <rss><channel xmlns:itunes="http://www.itunes.com/dtds/podcast-1.0.dtd">
+            <item>
+                <itunes:category text="Arts"/>
+                <itunes:category text="Arts  ">
+                    <itunes:category text="  Fashion    &amp; Beauty"/>
+                </itunes:category>
+                <itunes:category text="Arts  ">
+                    <itunes:category text="   "/>
+                    <itunes:category text="Books"/>
+                    <itunes:category text="Design"/>
+                </itunes:category>
+                <itunes:category text="   "/>
+                <itunes:category>Bogus</itunes:category>
+            </item>
+        </channel></rss>
+    output:
+        format: rss
+        entries:
+            - categories:
+                - name: Arts
+                - name: Arts
+                  subcategory: 'Fashion & Beauty'
+                - name: Arts
+                  subcategory: Books
+
+Google Play categories:
+    input: >
+        <rss><channel xmlns:play="http://www.google.com/schemas/play-podcasts/1.0">
+            <item>
+                <play:category text="Arts"/>
+                <play:category text="   "/>
+                <play:category>Bogus</play:category>
+                <play:category text="Music"/>
+            </item>
+        </channel></rss>
+    output:
+        format: rss
+        entries:
+            - categories:
+                - name: Arts
+                - name: Music

tests/cases/XML/entry-rss0.yaml

@@ -1,54 +0,0 @@
-Empty entry:
-    input: >
-        <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns="http://channel.netscape.com/rdf/simple/0.9/">
-            <channel>
-                <item/>
-            </channel>
-            <item/>
-        </rdf:RDF>
-    output:
-        format: rdf
-        version: '0.90'
-
-Entry link:
-    input: >
-        <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns="http://channel.netscape.com/rdf/simple/0.9/">
-            <channel>
-                <item>
-                    <link>http://example.com/</link>
-                </item>
-            </channel>
-            <item>
-                <link>http://example.com/</link>
-            </item>
-        </rdf:RDF>
-    output:
-        format: rdf
-        version: '0.90'
-        entries:
-            - link: 'http://example.com/'
-            - link: 'http://example.com/'
-
-Entry title:
-    input: >
-        <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns="http://channel.netscape.com/rdf/simple/0.9/">
-            <item>
-                <title>Loose  title</title>
-            </item>
-        </rdf:RDF>
-    output:
-        format: rdf
-        version: '0.90'
-        entries:
-            - title: {loose: 'Loose  title'}
-
-Entry content: # RSS 0.90 didn't have entry descriptions
-    input: >
-        <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns="http://channel.netscape.com/rdf/simple/0.9/">
-            <item>
-                <description>Bogus description</description>
-            </item>
-        </rdf:RDF>
-    output:
-        format: rdf
-        version: '0.90'

tests/cases/XML/entry-rss1.yaml

@@ -260,3 +260,19 @@ Dublin Core creators and contributors:
                       role: contributor
                     - name: Curly
                       role: contributor
+
+Categories by way of Dublin Core subjects:
+    input: >
+        <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns="http://purl.org/rss/1.0/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:term="http://purl.org/dc/terms/">
+            <item>
+                <dc:subject>Category, sort  of</dc:subject>
+                <term:subject>blah</term:subject>
+            </item>
+        </rdf:RDF>
+    output:
+        format: rdf
+        version: '1.0'
+        entries:
+            - categories:
+                - name: 'Category, sort of'
+                - name: 'blah'

tests/cases/XML/entry-rss2.yaml

@@ -159,3 +159,20 @@ Entry author:
                     - name: 'John Doe'
                       mail: 'john.doe@example.com'
                       role: author
+
+Categories:
+    input: >
+        <rss><channel>
+            <item>
+                <category>Category the  first  </category>
+                <category domain="ook  eek">Category the  second  </category>
+                <category/>
+            </item>
+        </channel></rss>
+    output:
+        format: rss
+        entries:
+            - categories:
+                - name: 'Category the first'
+                - name: 'Category the second'
+                  domain: 'ook eek'