add tests for feeds, catch case when a permalink has other h-entrys

lib/Formats/Mf2.php

@@ -6,56 +6,81 @@ use HTMLPurifier, HTMLPurifier_Config;
 class Mf2 {
 
   public static function parse($mf2, $url, $http) {
+    if(count($mf2['items']) == 0)
+      return false;
 
-    // TODO: Check if the list of items is a bunch of h-entrys and return as a feed
+    // Check if the list of items is a bunch of h-entrys and return as a feed
+    // Unless this page's URL matches one of the entries, then treat it as a permalink
+    $hentrys = 0;
+    $lastSeenEntry = false;
+    foreach($mf2['items'] as $item) {
+      if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
+        if(array_key_exists('url', $item['properties'])) {
+          $urls = $item['properties']['url'];
+          $urls = array_map('\normalize_url', $urls);
+          if(in_array($url, $urls)) {
+            return self::parseAsHEntry($mf2, $item, $http, $url);
+          }
+          $lastSeenEntry = $item;
+        }
+        $hentrys++;
+      }
+    }
 
+    // If there was more than one h-entry on the page, treat the whole page as a feed
+    if($hentrys > 1) {
+      return self::parseAsHFeed($mf2, $http);
+    }
+
+    // If the first item is an h-feed, parse as a feed
+    $first = $mf2['items'][0];
+    if(in_array('h-feed', $first['type'])) {
+      return self::parseAsHFeed($mf2, $http);
+    }
 
-    if($item = $mf2['items'][0]) {
-      // If the first item is a feed, the page is a feed
-      if(in_array('h-feed', $item['type'])) {
-        return self::parseAsHFeed($mf2, $http);
+    // Check each top-level h-card, and if there is one that matches this URL, the page is an h-card
+    foreach($mf2['items'] as $item) {
+      if(in_array('h-card', $item['type'])
+        and array_key_exists('url', $item['properties'])
+      ) {
+        $urls = $item['properties']['url'];
+        $urls = array_map('\normalize_url', $urls);
+        if(in_array($url, $urls)) {
+          // TODO: check for children h-entrys (like tantek.com), or sibling h-entries (like aaronparecki.com)
+          // and return the result as a feed instead
+          return self::parseAsHCard($item, $http, $url);
+        }
       }
+    }
 
-      // Check each top-level h-card, and if there is one that matches this URL, the page is an h-card
-      foreach($mf2['items'] as $i) {
-        if(in_array('h-card', $i['type'])
-          and array_key_exists('url', $i['properties'])
-        ) {
-          $urls = $i['properties']['url'];
-          $urls = array_map('\normalize_url', $urls);
-          if(in_array($url, $urls)) {
-            // TODO: check for children h-entrys (like tantek.com), or sibling h-entries (like aaronparecki.com)
-            // and return the result as a feed instead
-            return self::parseAsHCard($i, $http, $url);
-          }
+    // If there was only one h-entry, but the URL for it is not the same as this page, then treat as a feed
+    if($hentrys == 1) {
+      if($lastSeenEntry) {
+        $urls = $lastSeenEntry['properties']['url'];
+        $urls = array_map('\normalize_url', $urls);
+        if(count($urls) && !in_array($url, $urls)) {
+          return self::parseAsHFeed($mf2, $http);
         }
       }
     }
 
+    // Fallback case, but hopefully we have found something before this point
     foreach($mf2['items'] as $item) {
       // Otherwise check for an h-entry
       if(in_array('h-entry', $item['type']) || in_array('h-cite', $item['type'])) {
-        return self::parseAsHEntry($mf2, $http);
+        return self::parseAsHEntry($mf2, $item, $http);
       }
     }
 
     return false;
   }
 
-  private static function parseAsHEntry($mf2, $http) {
+  private static function parseAsHEntry($mf2, $item, $http) {
     $data = [
       'type' => 'entry'
     ];
     $refs = [];
 
-    // Find the first h-entry
-    foreach($mf2['items'] as $i) {
-      if(in_array('h-entry', $i['type']) || in_array('h-cite', $i['type'])) {
-        $item = $i;
-        continue;
-      }
-    }
-
     // Single plaintext values
     $properties = ['url','published','summary','rsvp'];
     foreach($properties as $p) {
@@ -166,12 +191,12 @@ class Mf2 {
         'url' => null,
         'photo' => null
       ],
-      'items' => [],
       'todo' => 'Not yet implemented. Please see https://github.com/aaronpk/XRay/issues/1'
     ];
 
     return [
-      'data' => $data
+      'data' => $data,
+      'entries' => []
     ];
   }
 

tests/FeedTest.php

@@ -0,0 +1,86 @@
+<?php
+use Symfony\Component\HttpFoundation\Request;
+use Symfony\Component\HttpFoundation\Response;
+
+class FeedTest extends PHPUnit_Framework_TestCase {
+
+  private $http;
+
+  public function setUp() {
+    $this->client = new Parse();
+    $this->client->http = new p3k\HTTPTest(dirname(__FILE__).'/data/');
+  }
+
+  private function parse($params) {
+    $request = new Request($params);
+    $response = new Response();
+    return $this->client->parse($request, $response);
+  }
+
+  public function testListOfHEntrys() {
+    $url = 'http://feed.example.com/list-of-hentrys';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertEquals('feed', $data->data->type);
+  }
+
+  public function testListOfHEntrysWithHCard() {
+    $url = 'http://feed.example.com/list-of-hentrys-with-h-card';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertEquals('feed', $data->data->type);
+  }
+
+  public function testShortListOfHEntrysWithHCard() {
+    $url = 'http://feed.example.com/short-list-of-hentrys-with-h-card';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertEquals('feed', $data->data->type);
+  }
+
+  public function testTopLevelHFeed() {
+    $url = 'http://feed.example.com/top-level-h-feed';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertEquals('feed', $data->data->type);
+  }
+
+  public function testHCardWithChildHEntrys() {
+    $url = 'http://feed.example.com/h-card-with-child-h-entrys';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertEquals('card', $data->data->type);
+  }
+
+  public function testHCardWithChildHFeed() {
+    $url = 'http://feed.example.com/h-card-with-child-h-feed';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertEquals('card', $data->data->type);
+  }
+
+}

tests/ParseTest.php

@@ -202,4 +202,15 @@ class ParseTest extends PHPUnit_Framework_TestCase {
     $this->assertEquals('yes', $data['data']['rsvp']);
   }
 
+  public function testMultipleHEntryOnPermalink() {
+    $url = 'http://source.example.com/multiple-h-entry-on-permalink';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body, true);
+    $this->assertEquals('entry', $data['data']['type']);
+    $this->assertEquals('Primary Post', $data['data']['name']);
+  }
+
 }

tests/data/feed.example.com/h-card-with-child-h-entrys

@@ -11,7 +11,7 @@ Connection: keep-alive
   <body>
 
     <div class="h-card">
-      <a href="/author" class="u-url p-name">Author Name</a>
+      <a href="/h-card-with-child-h-entrys" class="u-url p-name">Author Name</a>
 
       <ul>
         <li class="h-entry">

tests/data/feed.example.com/h-card-with-child-h-feed

@@ -11,7 +11,7 @@ Connection: keep-alive
   <body>
 
     <div class="h-card">
-      <a href="/author" class="u-url p-name">Author Name</a>
+      <a href="/h-card-with-child-h-feed" class="u-url p-name">Author Name</a>
 
       <ul class="h-feed">
         <li class="h-entry">

tests/data/feed.example.com/short-list-of-hentrys-with-h-card

@@ -13,6 +13,7 @@ Connection: keep-alive
     <ul>
       <li class="h-entry">
         <a href="/1" class="u-url p-name">One</a>
+        <a href="/author" class="p-author h-card">Author</a>
       </li>
     </ul>
 

tests/data/source.example.com/multiple-h-entry-on-permalink

@@ -0,0 +1,21 @@
+HTTP/1.1 200 OK
+Server: Apache
+Date: Wed, 09 Dec 2015 03:29:14 GMT
+Content-Type: text/html; charset=utf-8
+Connection: keep-alive
+
+<html>
+  <head>
+    <title>Test</title>
+  </head>
+  <body>
+    <div class="h-entry">
+      <h2 class="p-name">Primary Post</h2>
+      <a href="/multiple-h-entry-on-permalink">permalink</a>
+    </div>
+    <div class="h-entry">
+      <h3 class="p-name">Next Post</h3>
+      <a href="/another-post">read more</a>
+    </div>
+  </body>
+</html>

tests/data/source.example.com/person-tag-is-url

@@ -11,5 +11,6 @@ Connection: keep-alive
   <body class="h-entry">
     <h2 class="p-name">Hello World</h2>
     <a href="http://alice.example.com/" class="u-category">Alice</a>
+    <a href="/person-tag-is-url" class="u-url">permalink</a>
   </body>
 </html>

tests/data/source.example.com/reply-is-url

@@ -11,5 +11,6 @@ Connection: keep-alive
   <body class="h-entry">
     <a href="http://example.com/100" class="u-in-reply-to">in reply to</a>
     <p class="e-content">This page has a link to <a href="http://target.example.com">target.example.com</a> and some <b>formatted text</b>.</p>
+    <a href="/reply-is-url" class="u-url">permalink</a>
   </body>
 </html>