Add Atom parsing for channel image and pub date

prof18 · Jun 11, 2023 · f2cc211 · f2cc211
1 parent 92dec79
commit f2cc211
Show file tree

Hide file tree

Showing 4 changed files with 91 additions and 10 deletions.
diff --git a/rssparser/src/main/java/com/prof/rssparser/core/parser/atom/AtomParser.kt b/rssparser/src/main/java/com/prof/rssparser/core/parser/atom/AtomParser.kt
@@ -24,7 +24,6 @@ import com.prof.rssparser.ItunesArticleData
 import com.prof.rssparser.ItunesChannelData
 import com.prof.rssparser.core.CoreXMLParser.getImageUrl
 import com.prof.rssparser.utils.AtomKeyword
-import com.prof.rssparser.utils.RSSKeyword
 import com.prof.rssparser.utils.attributeValue
 import com.prof.rssparser.utils.contains
 import com.prof.rssparser.utils.nextTrimmedText
@@ -64,6 +63,12 @@ internal fun extractAtomContent(xmlPullParser: XmlPullParser): Channel {
                 }
                 //endregion
 
+                //region Channel tags
+                xmlPullParser.contains(AtomKeyword.Icon) -> {
+                    channelImageBuilder.url(xmlPullParser.nextTrimmedText())
+                }
+                //endregion
+
                 //region Item tags
                 xmlPullParser.contains(AtomKeyword.Entry.Author) -> {
                     if (insideItem) {
@@ -113,14 +118,15 @@ internal fun extractAtomContent(xmlPullParser: XmlPullParser): Channel {
                     }
                 }
 
-                xmlPullParser.contains(AtomKeyword.Entry.PubDate) -> {
+                xmlPullParser.contains(AtomKeyword.Updated) -> {
+                    if (insideChannel) {
+                        channelBuilder.lastBuildDate(xmlPullParser.nextTrimmedText())
+                    }
+                }
+
+                xmlPullParser.contains(AtomKeyword.Entry.Published) -> {
                     if (insideItem) {
-                        val nextTokenType = xmlPullParser.next()
-                        if (nextTokenType == XmlPullParser.TEXT) {
-                            articleBuilder.pubDate(xmlPullParser.text.trim())
-                        }
-                        // Skip to be able to find date inside 'tag' tag
-                        continue@loop
+                        articleBuilder.pubDate(xmlPullParser.nextTrimmedText())
                     }
                 }
 

diff --git a/rssparser/src/main/java/com/prof/rssparser/utils/AtomKeyword.kt b/rssparser/src/main/java/com/prof/rssparser/utils/AtomKeyword.kt
@@ -3,18 +3,19 @@ package com.prof.rssparser.utils
 internal sealed class AtomKeyword(val value: String) {
     object Atom : AtomKeyword("feed")
     object Title : AtomKeyword("title")
+    object Icon: AtomKeyword("icon")
     object Link : AtomKeyword("link") {
         object Href : AtomKeyword("href")
         object Rel : AtomKeyword("rel")
         object Edit : AtomKeyword("edit")
     }
     object Subtitle : AtomKeyword("subtitle")
-
+    object Updated: AtomKeyword("updated")
     object Entry {
         object Item : AtomKeyword("entry")
         object Guid : AtomKeyword("id")
         object Content : AtomKeyword("content")
-        object PubDate : AtomKeyword("updated")
+        object Published : AtomKeyword("published")
         object Category : AtomKeyword("category")
         object Term : AtomKeyword("term")
         object Description : AtomKeyword("summary")

diff --git a/rssparser/src/test/java/com/prof/rssparser/core/CoreXMLParserAtomTest.kt b/rssparser/src/test/java/com/prof/rssparser/core/CoreXMLParserAtomTest.kt
@@ -0,0 +1,39 @@
+package com.prof.rssparser.core
+
+import com.prof.rssparser.Image
+import com.prof.rssparser.testutils.BaseCoreXMLParserTest
+
+class CoreXMLParserAtomTest: BaseCoreXMLParserTest(
+    feedPath = "/feed-atom-test.xml",
+    channelTitle = "The Verge -  All Posts",
+    channelLink = "https://www.theverge.com/",
+    channelImage = Image(
+        title = null,
+        link = null,
+        description = null,
+        url = "https://cdn.vox-cdn.com/community_logos/52801/VER_Logomark_32x32..png"
+    ),
+    channelLastBuildDate = "2023-05-26T17:30:31-04:00",
+    channelUpdatePeriod = null,
+    articleGuid = "https://www.theverge.com/2023/5/26/23739273/google-sonos-smart-speaker-patent-lawsuit-ruling",
+    articleTitle = "Sonos wins \$32.5 million patent infringement victory over Google",
+    articleAuthor = "Chris Welch",
+    articleLink = "https://www.theverge.com/2023/5/26/23739273/google-sonos-smart-speaker-patent-lawsuit-ruling",
+    articlePubDate = "2023-05-26T17:30:31-04:00",
+    articleContent = """
+<figure>
+            <img alt="A photo of the Sonos Era 300 on a kitchen dining table." src="https://cdn.vox-cdn.com/thumbor/oCea2Vc5FYLWqQXGUmA4O-rRrM0=/0x0:2040x1360/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/72316887/DSCF0491.0.jpg" />
+            <figcaption>Photo by Chris Welch / The Verge</figcaption>
+            </figure>
+
+            <p id="b46vcm">Google has been ordered to pay Sonos ${'$'}32.5 million for infringing on the company’s smart speaker patent. A <a href="https://www.documentcloud.org/documents/23826599-google-sonos-trial-verdict?responsive=1&amp;title=1">jury verdict</a> issued in a San Francisco courtroom on Friday found that Google’s smart speakers and media players infringed on one of two Sonos patents at issue.</p>
+            <p id="keHPBL"><a href="https://www.theverge.com/2020/1/7/21055048/sonos-google-lawsuit-sues-speakers-assistant-amazon">The legal battle started in 2020</a> when Sonos accused Google of copying its patented multiroom audio technology after the companies partnered in 2013. <a href="https://www.theverge.com/2022/1/6/22871121/sonos-google-patent-itc-ruling-decision-import-ban">Sonos went on to win its case at the US International Trade Commission</a>, resulting in a limited import ban on some of the Google devices in question. Google has also <a href="https://www.theverge.com/2022/1/6/22871304/google-home-speaker-group-volume-control-changes-sonos-patent-decision">had to pull some features</a> from its lineup of smart speakers and smart displays.</p>
+            <figure class="e-image">
+
+            <cite>Image: United States District Court for the Northern District of...</cite></figure>
+            <p>
+            <a href="https://www.theverge.com/2023/5/26/23739273/google-sonos-smart-speaker-patent-lawsuit-ruling">Continue reading&hellip;</a>
+            </p>
+    """.trimIndent(),
+    articleImage = "https://cdn.vox-cdn.com/thumbor/oCea2Vc5FYLWqQXGUmA4O-rRrM0=/0x0:2040x1360/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/72316887/DSCF0491.0.jpg"
+)
diff --git a/rssparser/src/test/resources/feed-atom-test.xml b/rssparser/src/test/resources/feed-atom-test.xml
@@ -0,0 +1,35 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
+    <title>The Verge -  All Posts</title>
+    <icon>https://cdn.vox-cdn.com/community_logos/52801/VER_Logomark_32x32..png</icon>
+    <updated>2023-05-26T17:30:31-04:00</updated>
+    <id>https://www.theverge.com/rss/full.xml</id>
+    <link type="text/html" href="https://www.theverge.com/" rel="alternate"/>
+    <entry>
+        <published>2023-05-26T17:30:31-04:00</published>
+        <updated>2023-05-26T17:30:31-04:00</updated>
+        <title>Sonos wins $32.5 million patent infringement victory over Google</title>
+        <content type="html">
+
+            &lt;figure&gt;
+            &lt;img alt="A photo of the Sonos Era 300 on a kitchen dining table." src="https://cdn.vox-cdn.com/thumbor/oCea2Vc5FYLWqQXGUmA4O-rRrM0=/0x0:2040x1360/1310x873/cdn.vox-cdn.com/uploads/chorus_image/image/72316887/DSCF0491.0.jpg" /&gt;
+            &lt;figcaption&gt;Photo by Chris Welch / The Verge&lt;/figcaption&gt;
+            &lt;/figure&gt;
+
+            &lt;p id="b46vcm"&gt;Google has been ordered to pay Sonos $32.5 million for infringing on the company’s smart speaker patent. A &lt;a href="https://www.documentcloud.org/documents/23826599-google-sonos-trial-verdict?responsive=1&amp;amp;title=1"&gt;jury verdict&lt;/a&gt; issued in a San Francisco courtroom on Friday found that Google’s smart speakers and media players infringed on one of two Sonos patents at issue.&lt;/p&gt;
+            &lt;p id="keHPBL"&gt;&lt;a href="https://www.theverge.com/2020/1/7/21055048/sonos-google-lawsuit-sues-speakers-assistant-amazon"&gt;The legal battle started in 2020&lt;/a&gt; when Sonos accused Google of copying its patented multiroom audio technology after the companies partnered in 2013. &lt;a href="https://www.theverge.com/2022/1/6/22871121/sonos-google-patent-itc-ruling-decision-import-ban"&gt;Sonos went on to win its case at the US International Trade Commission&lt;/a&gt;, resulting in a limited import ban on some of the Google devices in question. Google has also &lt;a href="https://www.theverge.com/2022/1/6/22871304/google-home-speaker-group-volume-control-changes-sonos-patent-decision"&gt;had to pull some features&lt;/a&gt; from its lineup of smart speakers and smart displays.&lt;/p&gt;
+            &lt;figure class="e-image"&gt;
+
+            &lt;cite&gt;Image: United States District Court for the Northern District of...&lt;/cite&gt;&lt;/figure&gt;
+            &lt;p&gt;
+            &lt;a href="https://www.theverge.com/2023/5/26/23739273/google-sonos-smart-speaker-patent-lawsuit-ruling"&gt;Continue reading&amp;hellip;&lt;/a&gt;
+            &lt;/p&gt;
+
+        </content>
+        <link rel="alternate" type="text/html" href="https://www.theverge.com/2023/5/26/23739273/google-sonos-smart-speaker-patent-lawsuit-ruling"/>
+        <id>https://www.theverge.com/2023/5/26/23739273/google-sonos-smart-speaker-patent-lawsuit-ruling</id>
+        <author>
+            <name>Chris Welch</name>
+        </author>
+    </entry>
+</feed>