From 3bdafad98e9d444dbe11f419f114599c545df5d9 Mon Sep 17 00:00:00 2001
From: Aaron Parecki <aaron@parecki.com>
Date: Thu, 22 Dec 2016 08:06:00 -0800
Subject: [PATCH] parse URLs with fragment IDs

If the input URL contains a fragment, finds the DOM tree at that ID and runs the subtree through the mf2 parser.

closes #15
---
 controllers/Parse.php                     | 34 ++++++++++++++++++++---
 lib/HTTPTest.php                          |  3 ++
 tests/ParseTest.php                       | 24 ++++++++++++++++
 tests/data/source.example.com/fragment-id | 22 +++++++++++++++
 4 files changed, 79 insertions(+), 4 deletions(-)
 create mode 100644 tests/data/source.example.com/fragment-id

diff --git a/controllers/Parse.php b/controllers/Parse.php
index 586181d..ee8691a 100644
--- a/controllers/Parse.php
+++ b/controllers/Parse.php
@@ -154,10 +154,10 @@ class Parse {
       ]);
     }
 
+    $xpath = new DOMXPath($doc);
+
     // If a target parameter was provided, make sure a link to it exists on the page
     if($target=$request->get('target')) {
-      $xpath = new DOMXPath($doc);
-
       $found = [];
       if($target) {
         self::xPathFindNodeWithAttribute($xpath, 'a', 'href', function($u) use($target, &$found){
@@ -190,19 +190,37 @@ class Parse {
       }
     }
 
+    // If the URL has a fragment ID, find the DOM starting at that node and parse it instead
+    $html = $result['body'];
+
+    $fragment = parse_url($url, PHP_URL_FRAGMENT);
+    if($fragment) {
+      $fragElement = self::xPathGetElementById($xpath, $fragment);
+      if($fragElement) {
+        $html = $doc->saveHTML($fragElement);
+        $foundFragment = true;
+      } else {
+        $foundFragment = false;
+      }
+    }
+
     // Now start pulling in the data from the page. Start by looking for microformats2
-    $mf2 = mf2\Parse($result['body'], $result['url']);
+    $mf2 = mf2\Parse($html, $result['url']);
 
     if($mf2 && count($mf2['items']) > 0) {
       $data = Formats\Mf2::parse($mf2, $result['url'], $this->http);
       if($data) {
+        if($fragment) {
+          $data['info'] = [
+            'found_fragment' => $foundFragment
+          ];
+        }
         return $this->respond($response, 200, $data);
       }
     }
 
     // TODO: look for other content like OEmbed or other known services later
 
-
     return $this->respond($response, 200, [
       'data' => [
         'type' => 'unknown',
@@ -217,4 +235,12 @@ class Parse {
     }
   }
 
+  private static function xPathGetElementById($xpath, $id) {
+    $element = null;
+    foreach($xpath->query("//*[@id='$id']") as $el) {
+      $element = $el;
+    }
+    return $element;
+  }
+
 }
diff --git a/lib/HTTPTest.php b/lib/HTTPTest.php
index a0858e3..ca3880e 100644
--- a/lib/HTTPTest.php
+++ b/lib/HTTPTest.php
@@ -10,6 +10,9 @@ class HTTPTest extends HTTPCurl {
   }
 
   public function get($url, $headers=[]) {
+    $parts = parse_url($url);
+    unset($parts['fragment']);
+    $url = \build_url($parts);
     return $this->_read_file($url);
   }
 
diff --git a/tests/ParseTest.php b/tests/ParseTest.php
index c6b2e8f..5475d61 100644
--- a/tests/ParseTest.php
+++ b/tests/ParseTest.php
@@ -358,4 +358,28 @@ class ParseTest extends PHPUnit_Framework_TestCase {
     $this->assertEquals('Tantek Çelik', $data['refs']['https://www.facebook.com/tantek.celik']['name']);
   }
 
+  public function testEntryAtFragmentID() {
+    $url = 'http://source.example.com/fragment-id#comment-1000';
+    $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('http://source.example.com/fragment-id#comment-1000', $data['data']['url']);
+    $this->assertTrue($data['info']['found_fragment']);
+  }
+
+  public function testEntryAtNonExistentFragmentID() {
+    $url = 'http://source.example.com/fragment-id#comment-404';
+    $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('http://source.example.com/fragment-id', $data['data']['url']);
+    $this->assertFalse($data['info']['found_fragment']);
+  }
+
 }
diff --git a/tests/data/source.example.com/fragment-id b/tests/data/source.example.com/fragment-id
new file mode 100644
index 0000000..7a6c566
--- /dev/null
+++ b/tests/data/source.example.com/fragment-id
@@ -0,0 +1,22 @@
+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 class="h-entry">
+    <p class="e-content p-name">This page has comments.</p>
+    <ul>
+      <li class="p-comment h-entry" id="comment-1000">
+        <span class="p-content p-name">Comment text</span>
+        <a href="http://user.example.com/" class="p-author h-card">A. Commenter</a>
+        <a href="#comment-1000" class="u-url">comment permalink</a>
+      </li>
+    </ul>
+    <a href="" class="u-url">permalink</a>
+  </body>
+</html>