refactor for parsing Instagram and GitHub complete

controllers/Parse.php

@@ -98,6 +98,23 @@ class Parse {
       }
     }
 
+    $parser = new p3k\XRay\Parser($this->http);
+    $parsed = $parser->parse($result['body'], $result['url'], $opts);
+
+    // Allow the parser to override the HTTP response code, e.g. a meta-equiv tag
+    if(isset($parsed['code']))
+      $result['code'] = $parsed['code'];
+
+    $data = [
+      'data' => $parsed['data'],
+      'url' => $result['url'],
+      'code' => $result['code']
+    ];
+    if($request->get('include_original') && isset($parsed['original']))
+      $data['original'] = $parsed['original'];
+
+    return $this->respond($response, 200, $data);
+
 
 
     // Check for known services

lib/XRay.php

@@ -13,15 +13,29 @@ class XRay {
     return $rels->parse($url, $opts);
   }
 
-  public function parse($url, $opts=[]) {
-    $fetch = new XRay\Fetch($this->http);
-    $response = $fetch->fetch($url, $opts);
-    return $this->parse_doc($response, $url, $opts);
-  }
+  public function parse($url, $opts_or_body=false, $opts_for_body=[]) {
+    if(!$opts_or_body || is_array($opts_or_body)) {
+      $fetch = new XRay\Fetcher($this->http);
+      $response = $fetch->fetch($url, $opts_or_body);
+      if(!empty($response['error']))
+        return $response;
+      $body = $response['body'];
+      $url = $response['url'];
+      $code = $response['code'];
+      $opts = is_array($opts_or_body) ? $opts_or_body : $opts_for_body;
+    } else {
+      $body = $opts_or_body;
+      $opts = $opts_for_body;
+      $code = null;
+    }
+    $parser = new XRay\Parser($this->http);
 
-  public function parse_doc($response, $url=false, $opts=[]) {
-    
-    
+    $result = $parser->parse($body, $url, $opts);
+    if(!isset($opts['include_original']) || !$opts['include_original'])
+      unset($result['original']);
+    $result['url'] = $url;
+    $result['code'] = isset($result['code']) ? $result['code'] : $code;
+    return $result;
   }
 
 }

lib/XRay/Fetcher.php

@@ -9,6 +9,8 @@ class Fetcher {
   }
 
   public function fetch($url, $opts=[]) {
+    if($opts == false) $opts = [];
+    
     if(isset($opts['timeout']))
       $this->http->set_timeout($opts['timeout']);
     if(isset($opts['max_redirects']))

lib/XRay/Formats/GitHub.php

@@ -19,37 +19,50 @@ class GitHub extends Format {
       || preg_match('~https://github.com/([^/]+)/([^/]+)/issues/(\d+)#issuecomment-(\d+)~', $url, $match);
   }
 
-  public static function fetch($http, $url, $creds) {
-    // Transform the GitHub URL to an API request
+  private static function extract_url_parts($url) {
+    $response = false;
+
     if(preg_match('~https://github.com/([^/]+)/([^/]+)/pull/(\d+)$~', $url, $match)) {
-      $type = 'pull';
-      $org = $match[1];
-      $repo = $match[2];
-      $pull = $match[3];
-      $apiurl = 'https://api.github.com/repos/'.$org.'/'.$repo.'/pulls/'.$pull;
+      $response = [];
+      $response['type'] = 'pull';
+      $response['org'] = $match[1];
+      $response['repo'] = $match[2];
+      $response['pull'] = $match[3];
+      $response['apiurl'] = 'https://api.github.com/repos/'.$response['org'].'/'.$response['repo'].'/pulls/'.$response['pull'];
 
     } elseif(preg_match('~https://github.com/([^/]+)/([^/]+)/issues/(\d+)$~', $url, $match)) {
-      $type = 'issue';
-      $org = $match[1];
-      $repo = $match[2];
-      $issue = $match[3];
-      $apiurl = 'https://api.github.com/repos/'.$org.'/'.$repo.'/issues/'.$issue;
+      $response = [];
+      $response['type'] = 'issue';
+      $response['org'] = $match[1];
+      $response['repo'] = $match[2];
+      $response['issue'] = $match[3];
+      $response['apiurl'] = 'https://api.github.com/repos/'.$response['org'].'/'.$response['repo'].'/issues/'.$response['issue'];
 
     } elseif(preg_match('~https://github.com/([^/]+)/([^/]+)$~', $url, $match)) {
-      $type = 'repo';
-      $org = $match[1];
-      $repo = $match[2];
-      $apiurl = 'https://api.github.com/repos/'.$org.'/'.$repo;
+      $response = [];
+      $response['type'] = 'repo';
+      $response['org'] = $match[1];
+      $response['repo'] = $match[2];
+      $response['apiurl'] = 'https://api.github.com/repos/'.$response['org'].'/'.$response['repo'];
 
     } elseif(preg_match('~https://github.com/([^/]+)/([^/]+)/issues/(\d+)#issuecomment-(\d+)~', $url, $match)) {
-      $type = 'comment';
-      $org = $match[1];
-      $repo = $match[2];
-      $issue = $match[3];
-      $comment = $match[4];
-      $apiurl = 'https://api.github.com/repos/'.$org.'/'.$repo.'/issues/comments/'.$comment;
-
-    } else {
+      $response = [];
+      $response['type'] = 'comment';
+      $response['org'] = $match[1];
+      $response['repo'] = $match[2];
+      $response['issue'] = $match[3];
+      $response['comment'] = $match[4];
+      $response['apiurl'] = 'https://api.github.com/repos/'.$response['org'].'/'.$response['repo'].'/issues/comments/'.$response['comment'];
+
+    }
+
+    return $response;
+  }
+
+  public static function fetch($http, $url, $creds) {
+    $parts = self::extract_url_parts($url);
+
+    if(!$parts) {
       return [
         'error' => 'unsupported_url',
         'error_description' => 'This GitHub URL is not supported',
@@ -62,7 +75,7 @@ class GitHub extends Format {
       $headers[] = 'Authorization: Bearer ' . $creds['github_access_token'];
     }
 
-    $response = $http->get($apiurl, $headers);
+    $response = $http->get($parts['apiurl'], $headers);
     if($response['code'] != 200) {
       return [
         'error' => 'github_error',
@@ -78,20 +91,20 @@ class GitHub extends Format {
     ];
   }
 
-  public static function parse($http, $url, $creds, $json=null) {
+  public static function parse($json, $url) {
+    $data = @json_decode($json, true);
 
-    if(false) {
-    } else {
-      $data = json_decode($json, true);
-    }
+    if(!$data)
+      return self::_unknown();
 
-    if(!$data) {
-      return [null, null, 0];
-    }
+    $parts = self::extract_url_parts($url);
+
+    if(!$parts)
+      return self::_unknown();
 
     // Start building the h-entry
     $entry = array(
-      'type' => ($type == 'repo' ? 'repo' : 'entry'),
+      'type' => ($parts['type'] == 'repo' ? 'repo' : 'entry'),
       'url' => $url,
       'author' => [
         'type' => 'card',
@@ -101,7 +114,7 @@ class GitHub extends Format {
       ]
     );
 
-    if($type == 'repo')
+    if($parts['type'] == 'repo')
       $authorkey = 'owner';
     else
       $authorkey = 'user';
@@ -110,20 +123,20 @@ class GitHub extends Format {
     $entry['author']['photo'] = $data[$authorkey]['avatar_url'];
     $entry['author']['url'] = $data[$authorkey]['html_url'];
 
-    if($type == 'pull') {
-      $entry['name'] = '#' . $pull . ' ' . $data['title'];
-    } elseif($type == 'issue') {
-      $entry['name'] = '#' . $issue . ' ' . $data['title'];
-    } elseif($type == 'repo') {
+    if($parts['type'] == 'pull') {
+      $entry['name'] = '#' . $parts['pull'] . ' ' . $data['title'];
+    } elseif($parts['type'] == 'issue') {
+      $entry['name'] = '#' . $parts['issue'] . ' ' . $data['title'];
+    } elseif($parts['type'] == 'repo') {
       $entry['name'] = $data['name'];
     }
 
-    if($type == 'repo') {
+    if($parts['type'] == 'repo') {
       if(!empty($data['description']))
         $entry['summary'] = $data['description'];
     }
 
-    if($type != 'repo' && !empty($data['body'])) {
+    if($parts['type'] != 'repo' && !empty($data['body'])) {
       $parser = new GithubMarkdown();
 
       $entry['content'] = [
@@ -132,8 +145,8 @@ class GitHub extends Format {
       ];
     }
 
-    if($type == 'comment') {
-      $entry['in-reply-to'] = ['https://github.com/'.$org.'/'.$repo.'/issues/'.$issue];
+    if($parts['type'] == 'comment') {
+      $entry['in-reply-to'] = ['https://github.com/'.$parts['org'].'/'.$parts['repo'].'/issues/'.$parts['issue']];
     }
 
     if(!empty($data['labels'])) {
@@ -144,11 +157,10 @@ class GitHub extends Format {
 
     $entry['published'] = $data['created_at'];
 
-    $r = [
-      'data' => $entry
+    return [
+      'data' => $entry,
+      'original' => $json
     ];
-
-    return [$r, $json, $response['code']];
   }
 
 }

lib/XRay/Formats/Instagram.php

@@ -3,9 +3,8 @@ namespace p3k\XRay\Formats;
 
 use DOMDocument, DOMXPath;
 use DateTime, DateTimeZone;
-use Parse;
 
-class Instagram {
+class Instagram extends Format {
 
   public static function matches_host($url) {
     $host = parse_url($url, PHP_URL_HOST);
@@ -16,12 +15,12 @@ class Instagram {
     return self::matches_host($url);
   }
 
-  public static function parse($html, $url, $http) {
+  public static function parse($http, $html, $url) {
 
     $photoData = self::_extractPhotoDataFromPhotoPage($html);
 
     if(!$photoData)
-      return false;
+      return self::_unknown();
 
     // Start building the h-entry
     $entry = array(
@@ -140,19 +139,18 @@ class Instagram {
 
     $entry['published'] = $published->format('c');
 
-    $response = [
-      'data' => $entry
-    ];
-
     if(count($refs)) {
-      $response['refs'] = $refs;
+      $entry['refs'] = $refs;
     }
 
-    return [$response, [
-      'photo' => $photoData,
-      'profiles' => $profiles,
-      'locations' => $locations
-    ]];
+    return [
+      'data' => $entry,
+      'original' => json_encode([
+        'photo' => $photoData,
+        'profiles' => $profiles,
+        'locations' => $locations
+      ])
+    ];
   }
 
   private static function _buildHCardFromInstagramProfile($profile) {

lib/XRay/Parser.php

@@ -1,12 +1,38 @@
 <?php
 namespace p3k\XRay;
 
+use p3k\XRay\Formats;
+
 class Parser {
+  private $http;
+
+  public function __construct($http) {
+    $this->http = $http;
+  }
 
-  public function parse($url, $body) {
+  public function parse($body, $url, $opts=[]) {
+    if(isset($opts['timeout']))
+      $this->http->set_timeout($opts['timeout']);
+    if(isset($opts['max_redirects']))
+      $this->http->set_max_redirects($opts['max_redirects']);
+
+    // Check if the URL matches a special parser
+
+    if(Formats\Instagram::matches($url)) {
+      return Formats\Instagram::parse($this->http, $body, $url);
+    }
+
+    if(Formats\GitHub::matches($url)) {
+      return Formats\GitHub::parse($body, $url);
+    }
 
     
-    
+
+    return [
+      'data' => [
+        'type' => 'unknown'
+      ]
+    ];
   }
 
 }

tests/InstagramTest.php

@@ -71,8 +71,8 @@ class InstagramTest extends PHPUnit_Framework_TestCase {
 
     $this->assertEquals(2, count($data['data']['category']));
     $this->assertContains('http://tinyletter.com/kmikeym', $data['data']['category']);
-    $this->assertArrayHasKey('http://tinyletter.com/kmikeym', $data['refs']);
-    $this->assertEquals(['type'=>'card','name'=>'Mike Merrill','url'=>'http://tinyletter.com/kmikeym','photo'=>'https://instagram.fsjc1-3.fna.fbcdn.net/t51.2885-19/s320x320/12627953_686238411518831_1544976311_a.jpg'], $data['refs']['http://tinyletter.com/kmikeym']);
+    $this->assertArrayHasKey('http://tinyletter.com/kmikeym', $data['data']['refs']);
+    $this->assertEquals(['type'=>'card','name'=>'Mike Merrill','url'=>'http://tinyletter.com/kmikeym','photo'=>'https://instagram.fsjc1-3.fna.fbcdn.net/t51.2885-19/s320x320/12627953_686238411518831_1544976311_a.jpg'], $data['data']['refs']['http://tinyletter.com/kmikeym']);
   }
 
   public function testInstagramPhotoWithVenue() {
@@ -86,8 +86,8 @@ class InstagramTest extends PHPUnit_Framework_TestCase {
 
     $this->assertEquals(1, count($data['data']['location']));
     $this->assertContains('https://www.instagram.com/explore/locations/109284789535230/', $data['data']['location']);
-    $this->assertArrayHasKey('https://www.instagram.com/explore/locations/109284789535230/', $data['refs']);
-    $venue = $data['refs']['https://www.instagram.com/explore/locations/109284789535230/'];
+    $this->assertArrayHasKey('https://www.instagram.com/explore/locations/109284789535230/', $data['data']['refs']);
+    $venue = $data['data']['refs']['https://www.instagram.com/explore/locations/109284789535230/'];
     $this->assertEquals('XOXO Outpost', $venue['name']);
     $this->assertEquals('45.5261002', $venue['latitude']);
     $this->assertEquals('-122.6558081', $venue['longitude']);