Merge branch 'feature/parse-webmentions-command' into develop

app/Console/Commands/ParseCachedWebMentions.php

@@ -0,0 +1,70 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\WebMention;
+use Illuminate\Console\Command;
+use Illuminate\FileSystem\FileSystem;
+
+class ParseCachedWebMentions extends Command
+{
+    /**
+     * The name and signature of the console command.
+     *
+     * @var string
+     */
+    protected $signature = 'webmentions:parsecached';
+
+    /**
+     * The console command description.
+     *
+     * @var string
+     */
+    protected $description = 'Re-parse the webmention’s cached HTML';
+
+    /**
+     * Create a new command instance.
+     *
+     * @return void
+     */
+    public function __construct()
+    {
+        parent::__construct();
+    }
+
+    /**
+     * Execute the console command.
+     *
+     * @return mixed
+     */
+    public function handle(FileSystem $filesystem)
+    {
+        $HTMLfiles = $filesystem->allFiles(storage_path() . '/HTML');
+        foreach($HTMLfiles as $file) {
+            $filepath = $file->getPathname();
+            $html = $filesystem->get($filepath);
+            $url = $this->URLFromFilename($filepath);
+            $microformats = \Mf2\parse($html, $url);
+            $webmention = WebMention::where('source', $url)->firstOrFail();
+            $webmention->mf2 = json_encode($microformats);
+            $webmention->save();
+        }
+    }
+
+    /**
+     * Determine the source URL from a filename.
+     *
+     * @param  string
+     * @return string
+     */
+    private function URLFromFilename($filepath)
+    {
+        $dir = mb_substr($filepath, mb_strlen(storage_path() . '/HTML/'));
+        $url = str_replace(['http/', 'https/'], ['http://', 'https://'], $dir);
+        if (mb_substr($url, -1) == 'index.html') {
+            $url = mb_substr($url, 0, mb_strlen($url) - 10);
+        }
+
+        return $url;
+    }
+}

app/Console/Commands/ReDownloadWebMentions.php

@@ -0,0 +1,47 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\WebMention;
+use Illuminate\Console\Command;
+use App\Jobs\DownloadWebMention;
+
+class ReDownloadWebMentions extends Command
+{
+    /**
+     * The name and signature of the console command.
+     *
+     * @var string
+     */
+    protected $signature = 'webmentions:redownload';
+
+    /**
+     * The console command description.
+     *
+     * @var string
+     */
+    protected $description = 'Redownload the HTML content of webmentions';
+
+    /**
+     * Create a new command instance.
+     *
+     * @return void
+     */
+    public function __construct()
+    {
+        parent::__construct();
+    }
+
+    /**
+     * Execute the console command.
+     *
+     * @return mixed
+     */
+    public function handle()
+    {
+        $webmentions = WebMention::all();
+        foreach ($webmentions as $webmention) {
+            $this->dispatch(new DownloadWebMention($webmention->source));
+        }
+    }
+}

app/Console/Kernel.php

@@ -14,6 +14,8 @@ class Kernel extends ConsoleKernel
      */
     protected $commands = [
         Commands\SecurityCheck::class,
+        Commands\ParseCachedWebMentions::class,
+        Commands\ReDownloadWebMentions::class,
     ];
 
     /**

app/Jobs/DownloadWebMention.php

@@ -0,0 +1,67 @@
+<?php
+
+namespace App\Jobs;
+
+use GuzzleHttp\Client;
+use Illuminate\Bus\Queueable;
+use Illuminate\Queue\SerializesModels;
+use Illuminate\Queue\InteractsWithQueue;
+use Illuminate\Contracts\Queue\ShouldQueue;
+
+class DownloadWebMention implements ShouldQueue
+{
+    use InteractsWithQueue, Queueable, SerializesModels;
+
+    /**
+     * The webmention source URL.
+     *
+     * @var
+     */
+    protected $source;
+
+    /**
+     * Create a new job instance.
+     *
+     * @return void
+     */
+    public function __construct(string $source)
+    {
+        $this->source = $source;
+    }
+
+    /**
+     * Execute the job.
+     *
+     * @return void
+     */
+    public function handle(Client $guzzle)
+    {
+        $response = $guzzle->request('GET', $source);
+        //4XX and 5XX responses should get Guzzle to throw an exception,
+        //Laravel should catch and retry these automatically.
+        if ($response->getStatusCode() == '200') {
+            $filesystem = \Illuminate\FileSystem\FileSystem();
+            $filesystem->put(
+                $this->createFilenameFromURL($source),
+                (string) $response->getBody())
+            }
+        }
+    }
+
+    /**
+     * Create a file path from a URL. This is used when caching the HTML
+     * response.
+     *
+     * @param  string  The URL
+     * @return string  The path name
+     */
+    private function createFilenameFromURL($url)
+    {
+        $url = str_replace(['https://', 'http://'], ['https/', 'http/'], $url);
+        if (substr($url, -1) == '/') {
+            $url = $url . 'index.html';
+        }
+
+        return $url;
+    }
+}

app/Jobs/ProcessWebMention.php

@@ -4,10 +4,8 @@ namespace App\Jobs;
 
 use Mf2;
 use App\Note;
-use HTMLPurifier;
 use App\WebMention;
 use GuzzleHttp\Client;
-use HTMLPurifier_Config;
 use Illuminate\Queue\SerializesModels;
 use Illuminate\Queue\InteractsWithQueue;
 use Jonnybarnes\WebmentionsParser\Parser;
@@ -22,7 +20,6 @@ class ProcessWebMention extends Job implements ShouldQueue
 
     protected $note;
     protected $source;
-    protected $guzzle;
 
     /**
      * Create a new job instance.
@@ -31,28 +28,26 @@ class ProcessWebMention extends Job implements ShouldQueue
      * @param  string $source
      * @return void
      */
-    public function __construct(Note $note, $source, Client $guzzle = null)
+    public function __construct(Note $note, $source)
     {
         $this->note = $note;
         $this->source = $source;
-        $this->guzzle = $guzzle ?? new Client();
     }
 
     /**
      * Execute the job.
      *
-     * @param  \Jonnybarnes\WebmentionsParser\Parser $parser
+     * @param  \Jonnybarnes\WebmentionsParser\Parser  $parser
+     * @param  \GuzzleHttp\Client  $guzzle
      * @return void
      */
-    public function handle(Parser $parser)
+    public function handle(Parser $parser, Client $guzzle)
     {
-        $sourceURL = parse_url($this->source);
-        $baseURL = $sourceURL['scheme'] . '://' . $sourceURL['host'];
-        $remoteContent = $this->getRemoteContent($this->source);
+        $remoteContent = $this->getRemoteContent($this->source, $guzzle);
         if ($remoteContent === null) {
             throw new RemoteContentNotFoundException;
         }
-        $microformats = Mf2\parse($remoteContent, $baseURL);
+        $microformats = Mf2\parse($remoteContent, $this->source);
         $webmentions = WebMention::where('source', $this->source)->get();
         foreach ($webmentions as $webmention) {
             //check webmention still references target
@@ -65,7 +60,6 @@ class ProcessWebMention extends Job implements ShouldQueue
                     return;
                 }
                 //webmenion is still a reply, so update content
-                $microformats = $this->filterHTML($microformats);
                 $this->dispatch(new SaveProfileImage($microformats));
                 $webmention->mf2 = json_encode($microformats);
                 $webmention->save();
@@ -94,7 +88,6 @@ class ProcessWebMention extends Job implements ShouldQueue
         $webmention = new WebMention();
         $type = $parser->getMentionType($microformats); //throw error here?
         $this->dispatch(new SaveProfileImage($microformats));
-        $microformats = $this->filterHTML($microformats);
         $webmention->source = $this->source;
         $webmention->target = $this->note->longurl;
         $webmention->commentable_id = $this->note->id;
@@ -107,13 +100,14 @@ class ProcessWebMention extends Job implements ShouldQueue
     /**
      * Retreive the remote content from a URL, and caches the result.
      *
-     * @param  string       The URL to retreive content from
-     * @return string|null  The HTML from the URL (or null if error)
+     * @param  string  $url
+     * @param  GuzzleHttp\client  $guzzle
+     * @return string|null
      */
-    private function getRemoteContent($url)
+    private function getRemoteContent($url, Client $guzzle)
     {
         try {
-            $response = $this->guzzle->request('GET', $url);
+            $response = $guzzle->request('GET', $url);
         } catch (RequestException $e) {
             return;
         }
@@ -139,43 +133,11 @@ class ProcessWebMention extends Job implements ShouldQueue
      */
     private function createFilenameFromURL($url)
     {
-        $url = str_replace(['https://', 'http://'], ['', ''], $url);
+        $url = str_replace(['https://', 'http://'], ['https/', 'http/'], $url);
         if (substr($url, -1) == '/') {
             $url = $url . 'index.html';
         }
 
         return $url;
     }
-
-    /**
-     * Filter the HTML in a reply webmention.
-     *
-     * @param  array  The unfiltered microformats
-     * @return array  The filtered microformats
-     */
-    private function filterHTML($microformats)
-    {
-        if (isset($microformats['items'][0]['properties']['content'][0]['html'])) {
-            $microformats['items'][0]['properties']['content'][0]['html_purified'] = $this->useHTMLPurifier(
-                $microformats['items'][0]['properties']['content'][0]['html']
-            );
-        }
-
-        return $microformats;
-    }
-
-    /**
-     * Set up and use HTMLPurifer on some HTML.
-     *
-     * @param  string  The HTML to be processed
-     * @return string  The processed HTML
-     */
-    private function useHTMLPurifier($html)
-    {
-        $config = HTMLPurifier_Config::createDefault();
-        $config->set('Cache.SerializerPath', storage_path() . '/HTMLPurifier');
-        $purifier = new HTMLPurifier($config);
-
-        return $purifier->purify($html);
-    }
 }

app/Observers/WebMentionObserver.php

@@ -0,0 +1,65 @@
+<?php
+
+namespace App\Observers;
+
+use HTMLPurifier;
+use App\WebMention;
+use HTMLPurifier_Config;
+
+class WebMentionObserver
+{
+    /**
+     * Listen for the created event.
+     *
+     * @param  WebMention  $webmention
+     * @return void
+     */
+    public function created(WebMention $webmention)
+    {
+        $this->addFilteredHTML($webmention);
+    }
+
+    /**
+     * Listen for the updated event.
+     *
+     * @param  WebMention  $webmention
+     * @return void
+     */
+    public function updated(WebMention $webmention)
+    {
+        $this->addFilteredHTML($webmention);
+    }
+
+    /**
+     * Filter the HTML in a reply webmention.
+     *
+     * @param  WebMention  The WebMention model
+     * @return void
+     */
+    private function addFilteredHTML(WebMention $webmention)
+    {
+        $mf2 = json_decode($webmention->mf2);
+        if (isset($mf2['items'][0]['properties']['content'][0]['html'])) {
+            $mf2['items'][0]['properties']['content'][0]['html_purified'] = $this->useHTMLPurifier(
+                $mf2['items'][0]['properties']['content'][0]['html']
+            );
+        }
+        $webmention->mf2 = json_encode($mf2);
+        $webmetion->save();
+    }
+
+    /**
+     * Set up and use HTMLPurifer on some HTML.
+     *
+     * @param  string  The HTML to be processed
+     * @return string  The processed HTML
+     */
+    private function useHTMLPurifier($html)
+    {
+        $config = HTMLPurifier_Config::createDefault();
+        $config->set('Cache.SerializerPath', storage_path() . '/HTMLPurifier');
+        $purifier = new HTMLPurifier($config);
+
+        return $purifier->purify($html);
+    }
+}

app/Providers/AppServiceProvider.php

@@ -5,6 +5,8 @@ namespace App\Providers;
 use App\Tag;
 use App\Note;
 use Validator;
+use App\WebMention;
+use App\Observers\WebMentionObserver;
 use Illuminate\Support\ServiceProvider;
 
 class AppServiceProvider extends ServiceProvider
@@ -45,6 +47,9 @@ class AppServiceProvider extends ServiceProvider
                 $note->tags()->attach($tagsToAdd);
             }
         });
+
+        //observer the webmention model
+        WebMention::observe(WebMentionObserver::class);
     }
 
     /**

changelog.md

@@ -5,6 +5,7 @@
   - Added `integrity` values to external assets (issue#10)
   - Move mapbox links into own sub-view (issue#11)
   - Updated mapbox version (issue#12)
+  - Massive refactor of webmention code, allowing for re-parse command (issue#8)
 
 ## Version 0.0.10 (2016-09-10)
   - Add an artisan command for sensiolab’s security check