From ff59f00356fdc1f02bc83c5acd3f4220ce68f60e Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Fri, 26 May 2017 17:55:51 -0400
Subject: [PATCH] Feed tests: deduplication

---
 tests/Feed/TestFeed.php                       | 14 +++++++
 .../Feed/Deduplication/Hashes-Dates1.php      | 41 +++++++++++++++++++
 .../Feed/Deduplication/Hashes-Dates2.php      | 41 +++++++++++++++++++
 .../Feed/Deduplication/Hashes-Dates3.php      | 41 +++++++++++++++++++
 tests/docroot/Feed/Deduplication/Hashes.php   | 33 +++++++++++++++
 .../Feed/Deduplication/IdenticalHashes.php    | 22 +++++-----
 6 files changed, 181 insertions(+), 11 deletions(-)
 create mode 100644 tests/docroot/Feed/Deduplication/Hashes-Dates1.php
 create mode 100644 tests/docroot/Feed/Deduplication/Hashes-Dates2.php
 create mode 100644 tests/docroot/Feed/Deduplication/Hashes-Dates3.php
 create mode 100644 tests/docroot/Feed/Deduplication/Hashes.php

diff --git a/tests/Feed/TestFeed.php b/tests/Feed/TestFeed.php
index 0bd9bef..6fe949b 100644
--- a/tests/Feed/TestFeed.php
+++ b/tests/Feed/TestFeed.php
@@ -20,6 +20,7 @@ class TestFeed extends \PHPUnit\Framework\TestCase {
     }
 
     function testDeduplicateFeedItems() {
+        // duplicates with dates lead to the newest match being kept
         $t = strtotime("2002-05-19T15:21:36Z");
         $f = new Feed(null, $this->base."Deduplication/Permalink-Dates");
         $this->assertCount(2, $f->newItems);
@@ -30,6 +31,19 @@ class TestFeed extends \PHPUnit\Framework\TestCase {
         $f = new Feed(null, $this->base."Deduplication/IdenticalHashes");
         $this->assertCount(2, $f->newItems);
         $this->assertTime($t, $f->newItems[0]->updatedDate);
+        $f = new Feed(null, $this->base."Deduplication/Hashes-Dates1"); // content differs
+        $this->assertCount(2, $f->newItems);
+        $this->assertTime($t, $f->newItems[0]->updatedDate);
+        $f = new Feed(null, $this->base."Deduplication/Hashes-Dates2"); // title differs
+        $this->assertCount(2, $f->newItems);
+        $this->assertTime($t, $f->newItems[0]->updatedDate);
+        $f = new Feed(null, $this->base."Deduplication/Hashes-Dates3"); // URL differs
+        $this->assertCount(2, $f->newItems);
+        $this->assertTime($t, $f->newItems[0]->updatedDate);
+        // duplicates without dates lead to the topmost entry being kept
+        $f = new Feed(null, $this->base."Deduplication/Hashes");
+        $this->assertCount(2, $f->newItems);
+        $this->assertSame("http://example.com/1", $f->newItems[0]->url);
     }
 
     function testHandleCacheHeadersOn304() {
diff --git a/tests/docroot/Feed/Deduplication/Hashes-Dates1.php b/tests/docroot/Feed/Deduplication/Hashes-Dates1.php
new file mode 100644
index 0000000..364133c
--- /dev/null
+++ b/tests/docroot/Feed/Deduplication/Hashes-Dates1.php
@@ -0,0 +1,41 @@
+<?php return [
+    'mime'    => "application/rss+xml",
+    'content' => <<<MESSAGE_BODY
+<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
+<channel>
+    <title>Test feed</title>
+    <link>http://example.com/</link>
+    <description>A basic feed for testing</description>
+
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 1995 15:21:36 GMT</pubDate>
+        <atom:updated>2002-02-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1</link>
+        <description>Sample article 2</description>
+        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate> <!-- This is the correct item and date -->
+        <atom:updated>2002-04-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1</link>
+        <description>Sample article 3</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The Second</title>
+        <link>http://example.com/2</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
+    </item>
+</channel>
+</rss>
+MESSAGE_BODY
+];
\ No newline at end of file
diff --git a/tests/docroot/Feed/Deduplication/Hashes-Dates2.php b/tests/docroot/Feed/Deduplication/Hashes-Dates2.php
new file mode 100644
index 0000000..3f2f2a2
--- /dev/null
+++ b/tests/docroot/Feed/Deduplication/Hashes-Dates2.php
@@ -0,0 +1,41 @@
+<?php return [
+    'mime'    => "application/rss+xml",
+    'content' => <<<MESSAGE_BODY
+<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
+<channel>
+    <title>Test feed</title>
+    <link>http://example.com/</link>
+    <description>A basic feed for testing</description>
+
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 1995 15:21:36 GMT</pubDate>
+        <atom:updated>2002-02-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The First!</title>
+        <link>http://example.com/1</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate> <!-- This is the correct item and date -->
+        <atom:updated>2002-04-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The First?</title>
+        <link>http://example.com/1</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The Second</title>
+        <link>http://example.com/2</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
+    </item>
+</channel>
+</rss>
+MESSAGE_BODY
+];
\ No newline at end of file
diff --git a/tests/docroot/Feed/Deduplication/Hashes-Dates3.php b/tests/docroot/Feed/Deduplication/Hashes-Dates3.php
new file mode 100644
index 0000000..5eec075
--- /dev/null
+++ b/tests/docroot/Feed/Deduplication/Hashes-Dates3.php
@@ -0,0 +1,41 @@
+<?php return [
+    'mime'    => "application/rss+xml",
+    'content' => <<<MESSAGE_BODY
+<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
+<channel>
+    <title>Test feed</title>
+    <link>http://example.com/</link>
+    <description>A basic feed for testing</description>
+
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1?</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 1995 15:21:36 GMT</pubDate>
+        <atom:updated>2002-02-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate> <!-- This is the correct item and date -->
+        <atom:updated>2002-04-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1?ook=</link>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
+    </item>
+    <item>
+        <title>Article The Second</title>
+        <link>http://example.com/2</link>
+        <description>Sample article 4</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
+    </item>
+</channel>
+</rss>
+MESSAGE_BODY
+];
\ No newline at end of file
diff --git a/tests/docroot/Feed/Deduplication/Hashes.php b/tests/docroot/Feed/Deduplication/Hashes.php
new file mode 100644
index 0000000..f0dbb11
--- /dev/null
+++ b/tests/docroot/Feed/Deduplication/Hashes.php
@@ -0,0 +1,33 @@
+<?php return [
+    'mime'    => "application/rss+xml",
+    'content' => <<<MESSAGE_BODY
+<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
+<channel>
+    <title>Test feed</title>
+    <link>http://example.com/</link>
+    <description>A basic feed for testing</description>
+
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1</link> <!-- This is the correct item -->
+        <description>Sample article 1</description>
+    </item>
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1?</link>
+        <description>Sample article 1</description>
+    </item>
+    <item>
+        <title>Article The First</title>
+        <link>http://example.com/1?ook=</link>
+        <description>Sample article 1</description>
+    </item>
+    <item>
+        <title>Article The Second</title>
+        <link>http://example.com/2</link>
+        <description>Sample article 4</description>
+    </item>
+</channel>
+</rss>
+MESSAGE_BODY
+];
\ No newline at end of file
diff --git a/tests/docroot/Feed/Deduplication/IdenticalHashes.php b/tests/docroot/Feed/Deduplication/IdenticalHashes.php
index a512631..ced4fbd 100644
--- a/tests/docroot/Feed/Deduplication/IdenticalHashes.php
+++ b/tests/docroot/Feed/Deduplication/IdenticalHashes.php
@@ -9,27 +9,27 @@
 
     <item>
         <guid>1</guid>
-        <description>Sample article 2</description>
-        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate> <!-- This is the correct item and date -->
-        <atom:updated>2002-04-19T15:21:36Z</atom:updated>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 1995 15:21:36 GMT</pubDate>
+        <atom:updated>2002-02-19T15:21:36Z</atom:updated>
     </item>
     <item>
         <guid>1</guid>
-        <description>Sample article 2</description>
-        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate> <!-- This is the correct item and date -->
         <atom:updated>2002-04-19T15:21:36Z</atom:updated>
     </item>
     <item>
         <guid>1</guid>
-        <description>Sample article 2</description>
-        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate>
-        <atom:updated>2002-04-19T15:21:36Z</atom:updated>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
     </item>
     <item>
         <guid>2</guid>
-        <description>Sample article 2</description>
-        <pubDate>Sun, 19 May 2002 15:21:36 GMT</pubDate>
-        <atom:updated>2002-04-19T15:21:36Z</atom:updated>
+        <description>Sample article 1</description>
+        <pubDate>Sun, 18 May 2000 15:21:36 GMT</pubDate>
+        <atom:updated>1999-05-19T15:21:36Z</atom:updated>
     </item>
 </channel>
 </rss>