Expose HttpStatusCode for urls when searching canonically

willwashburn · Nov 19, 2019 · 5f372b0 · 5f372b0
1 parent f7960d6
commit 5f372b0
Show file tree

Hide file tree

Showing 3 changed files with 52 additions and 17 deletions.
diff --git a/README.md b/README.md
@@ -28,9 +28,10 @@ Use composer
 
 ```composer require willwashburn/mushroom```
 
-Alternatively, add ```"willwashburn/mushroom": "~2.10"``` to your composer.json
+Alternatively, add ```"willwashburn/mushroom": "~2.11"``` to your composer.json
 
 ## Change Log
+- v2.11.0 - Expose HttpStatusCode for urls when searching canonically
 - v2.10.0 - Add method to add a domain to JS Redirect Domains array
 - v2.9.1 - Fix bug where some html sources were not cached
 - v2.9.0 - Expose HTML for urls when searching canonically

diff --git a/src/mushroom/Mushroom.php b/src/mushroom/Mushroom.php
@@ -45,6 +45,15 @@ class Mushroom
      */
     private $html = [];
 
+    /**
+     * Some user-land use cases popped up that required the response_code of the
+     * eventual curl request we made when finding canonical urls. We expose a
+     * method to get this content in the event someone needs it.
+     *
+     * @var int[]
+     */
+    private $httpStatusCode = [];
+
     /**
      * Mushroom constructor.
      *
@@ -147,6 +156,22 @@ public function getCachedHtml($url)
         return false;
     }
 
+
+    /**
+     * We occasionally can find use in knowing the http status code
+     *
+     * @param $url
+     *
+     * @return bool
+     */
+    public function getCachedHttpStatusCode($url)
+    {
+        if ($this->httpStatusCode[$url]) {
+            return $this->httpStatusCode[$url];
+        }
+        return false;
+    }
+
     /**
      * @param array $urls              The urls to unfurl
      * @param array $curlOptions       Custom curl options that can be passed in
@@ -202,6 +227,10 @@ private function batchFollow($urls, $curlOptions, $followHttpRefresh, $findCanon
             if ($url['html']) {
                 $this->html[$urls[$key]] = $url['html'];
             }
+
+            if ($url['http_status_code']) {
+                $this->httpStatusCode[$urls[$key]] = $url['http_status_code'];
+            }
         }
 
         if ($followHttpRefresh && count($retries) > 0) {
@@ -342,9 +371,10 @@ private function getUrlFromHandle($ch, $followHttpRefresh, $findCanonical)
                     $url = $this->ensureSchemeAndHost($ch, $url);
 
                     return [
-                        'refresh' => true,
-                        'url'     => $url,
-                        'html'    => $html,
+                        'refresh'          => true,
+                        'url'              => $url,
+                        'html'             => $html,
+                        'http_status_code' => $httpStatusCode,
                     ];
                 }
             }
@@ -366,17 +396,19 @@ private function getUrlFromHandle($ch, $followHttpRefresh, $findCanonical)
                 $url = $this->ensureSchemeAndHost($ch, $url);
 
                 return [
-                    'refresh' => false,
-                    'url'     => $url,
-                    'html'    => $html
+                    'refresh'          => false,
+                    'url'              => $url,
+                    'html'             => $html,
+                    'http_status_code' => $httpStatusCode,
                 ];
             }
         }
 
         return [
-            'refresh' => false,
-            'url'     => $effectiveUrl,
-            'html'    => $html,
+            'refresh'          => false,
+            'url'              => $effectiveUrl,
+            'html'             => $html,
+            'http_status_code' => $httpStatusCode,
         ];
     }
 

diff --git a/tests/ExpandLinksTest.php b/tests/ExpandLinksTest.php
@@ -127,11 +127,11 @@ public function test_setting_curl_options_works()
                ->shouldReceive('curl_getinfo')->getMock()
                ->shouldReceive('curl_setopt_array')
                ->with(M::any(), M::on(function ($arg) use ($expected_curl_opts) {
-                   foreach ( array_keys($expected_curl_opts) as $key ) {
-                       if ( $arg[$key] != $expected_curl_opts[$key] ) {
-                           return false;
-                       }
-                   }
+                foreach (array_keys($expected_curl_opts) as $key) {
+                    if ($arg[$key] != $expected_curl_opts[$key]) {
+                        return false;
+                    }
+                }
 
                    return true;
                }))
@@ -151,7 +151,7 @@ public function test_without_follow_http_redirects()
         ];
 
         $mushroom = new Mushroom();
-        foreach ( $links as list( $link, $expected_result ) ) {
+        foreach ($links as list( $link, $expected_result )) {
             $result = $mushroom->expand($link, [], false);
 
             $this->assertEquals($expected_result, $result);
@@ -164,15 +164,17 @@ public function test_get_html_works()
             ['http://bit.ly/1bdDlXc', 'https://www.google.com/'],
             ['http://www.tailwindapp.com', 'https://www.tailwindapp.com/'],
             ['https://diply.com/article/auntyacid/pinterest-diy-easy-solutions', 'https://diply.com/article/auntyacid/pinterest-diy-easy-solutions'],
+            ['https://www.zazzle.com/cookie_monster_cookies_for_santa_dinner_plate-115773106232089655', 'https://www.zazzle.com/cookie_monster_cookies_for_santa_dinner_plate-115773106232089655'],
         ];
 
         $mushroom = new Mushroom();
-        foreach ( $links as list( $link, $expected_result ) ) {
+        foreach ($links as list( $link, $expected_result )) {
             $result = $mushroom->canonical($link);
 
             $this->assertEquals($expected_result, $result);
 
             $this->assertNotFalse($mushroom->getCachedHtml($link), $link);
+            $this->assertNotEmpty($mushroom->getCachedHttpStatusCode($link), $link);
         }
 
         $this->assertFalse($mushroom->getCachedHtml('https://www.tailwindapp.com'));