Keep track of articles to post relationships

Themroc · Themroc · commit aa75cb590001 · 2022-02-02T20:26:31.000+01:00
diff --git a/jobs/GetFeedUpdates.php b/jobs/GetFeedUpdates.php
@@ -11,29 +11,30 @@
 
 use sij\humhub\modules\rss\components\MarkdownHelper;
 use sij\humhub\modules\rss\components\RssElement;
-use sij\humhub\modules\rss\controllers\RssController;
+use sij\humhub\modules\rss\models\RssPosts;
 
 /**
  * Reads the RSS Feed URL for this space
  * and adds any new posts into the stream.
  */
 class GetFeedUpdates extends ActiveJob
 {
+    /**
+     * @var string mutex to acquire
+     */
+    const MUTEX_ID = 'rss-queue';
 
     public $space; # humhub\modules\space\models\Space
     public $force; # when true, ignore any cached values (used when config changed)
 
     private $logFileHandle;
 
     # configuration settings
-    private $cfg_url; # the URL of the RSS feed
-    private $cfg_article; # 'summary' or 'full'
-    private $cfg_pictures; # 'yes' or 'no'
-    private $cfg_maxwidth; # maximum width of pictures
+    private $cfg_url;       # the URL of the RSS feed
+    private $cfg_article;   # 'summary' or 'full'
+    private $cfg_pictures;  # 'yes' or 'no'
+    private $cfg_maxwidth;  # maximum width of pictures
     private $cfg_maxheight; # maximum height of pictures
-    private $cfg_owner; # user id of the owner of the posts
-    private $cfg_dayshistory; # maximum number of days history
-    private $cfg_daysfuture; # maximum number of days into the future
 
     private $created_by; # int: user id number
     private $rss_folder; # string: full name of the folder to store the RSS feed
@@ -46,25 +47,18 @@ class GetFeedUpdates extends ActiveJob
     private $feed_copyright; # string: copyright information for this RSS feed
     private $article; # string: markdown document representing HTML feed item body
     private $now; # DateTime: the current time
-    private $oldest; # oldest date we are accepting
-    private $newest; # newest date we are accepting
     private $items; # array of sij\humhub\modules\rss\components\RssElement keyed by pubDate
 
-    /**
-     * @var string mutex to acquire
-     */
-    const MUTEX_ID = 'rss-queue';
-
     private function log($message) {
         if ( $this->logFileHandle ) {
             fwrite($this->logFileHandle, $message);
             fflush($this->logFileHandle);
         }
     }
 
-/**
- * Creates a new Post in the Space
- */
+    /**
+     * Creates a new Post in the Space
+     */
     private function postError($title, $info)
     {
 
@@ -89,26 +83,31 @@ private function postError($title, $info)
 
     }
 
-/**
- * Creates a new Post in the Space
- */
-    private function postMessage($message, $datePublished = false)
+    /**
+     * Creates a new Post in the Space
+     */
+    private function postMessage($message, $link = false, $datePublished = false)
     {
-
         $post = null;
 
+        // find previous version of the post via db
+        if ( $link ) {
+            $url2id = RssPosts::findOne(['rss_link' => $link]);
+            if ( $url2id !== null )
+                $post = Post::findOne($url2id->post_id);
+        }
+
         // attempt to locate a previous version of the post
-        if ( $datePublished ) {
+        // guess this should go some day - themroc
+        if ( $post === null && $datePublished ) {
             $stamp = $datePublished->format("Y-m-d H:i:s");
             $oldContent = Content::findAll([
                 'contentcontainer_id' => $this->space->contentcontainer_id,
-//              'created_by' => $this->created_by, // cant rely on this field if config changed
+                'created_by' => $this->created_by,
                 'created_at' => $stamp,
             ]);
             if ( count($oldContent) == 1 ) {
                 $post = Post::findOne($oldContent[0]->object_id);
-                $this->log("\n\n### update Post\n");
-                Console::stdout("RSS queue: updating post... ");
             }
         }
 
@@ -117,6 +116,19 @@ private function postMessage($message, $datePublished = false)
             $post = new Post($this->space);
             $this->log("\n\n### new Post\n");
             Console::stdout("RSS queue: creating new post... ");
+        } else {
+            if ( $datePublished ) {
+                if ($stamp > $post->created_at) {
+                    $this->log("\n\n### update Post\n");
+                    Console::stdout("RSS queue: updating post... ");
+                } else {
+                    return; // not changed
+                }
+            } else {
+                if ( $post !== null ) {
+                    return; // we assume it hasn't changed - better miss an update than rewrite the post every time.
+                }
+            }
         }
 
         $post->created_by =
@@ -134,6 +146,13 @@ private function postMessage($message, $datePublished = false)
         $post->save();
         $this->log(print_r($post, true));
 
+        if (! $url2id ) {
+            $url2id = new RssPosts();
+            $url2id->rss_link = $link;
+            $url2id->post_id = $post->id;
+            $url2id->save();
+        }
+
         // make it look like the space post was created at the same time as the RSS article
         // note $post->save() always sets the time stamps to "now"
         if ( $datePublished ) {
@@ -189,10 +208,11 @@ private function parseNewsItem($item)
             $datePublished = false;
         }
 
-        // if the published date is known, skip the item if it is out of range
+        // check if published date is known and in the future
         if ( $datePublished ) {
-            if ( $datePublished < $this->oldest ) return;
-            if ( $datePublished > $this->newest ) return;
+            if ( $datePublished > $this->now ) {
+                return; // don't publish it here if it's not time yet
+            }
         }
 
         // choose which version of the article to use
@@ -259,7 +279,7 @@ private function parseNewsItem($item)
         }
 
         // post the message in the stream
-        $this->postMessage($message, $datePublished);
+        $this->postMessage($message, $link, $datePublished);
 
     }
 
@@ -453,35 +473,33 @@ private function downloadNewsFeed()
  */
     public function run()
     {
-        if (! Yii::$app->mutex->acquire(static::MUTEX_ID)) {
+        if (!Yii::$app->mutex->acquire(static::MUTEX_ID)) {
             Console::stdout("RSS queue execution skipped - already running!\n");
             return;
         }
 
+#        Yii::warning('Running RSS queue');
+        Yii::error('Running RSS queue');
+
 ####### $this->logFileHandle = fopen(dirname(__FILE__) . '/log.txt', 'w');
 
         $this->cfg_url = $this->space->getSetting('url', 'rss');
         $this->cfg_article = $this->space->getSetting('article', 'rss', 'summary');
         $this->cfg_pictures = $this->space->getSetting('pictures', 'rss', 'yes');
         $this->cfg_maxwidth = (int)$this->space->getSetting('maxwidth', 'rss', '500');
         $this->cfg_maxheight = (int)$this->space->getSetting('maxheight', 'rss', '500');
-        $this->cfg_owner = $this->space->getSetting('owner', 'rss', '');
-        $this->cfg_dayshistory = (int)$this->space->getSetting('dayshistory', 'rss', '31');
-        $this->cfg_daysfuture = (int)$this->space->getSetting('daysfuture', 'rss', '1');
 
         MarkdownHelper::$cfg_pictures = $this->cfg_pictures;
         MarkdownHelper::$cfg_maxwidth = $this->cfg_maxwidth;
         MarkdownHelper::$cfg_maxheight = $this->cfg_maxheight;
 
-        $this->created_by = RssController::vetOwner($this->cfg_owner, $this->space)->id;
+        $this->created_by = $this->space->created_by;
 
         $this->rss_folder = Yii::getAlias('@runtime/rss');
         $this->rss_file = $this->rss_folder . '/' . $this->space->guid . '.xml';
         $this->new_file = $this->rss_folder . '/' . $this->space->guid . '.new';
 
-        $this->now = new \DateTime('now');
-        $this->oldest = (new \DateTime('now'))->sub(new \DateInterval('P' . $this->cfg_dayshistory . 'D'));
-        $this->newest = (new \DateTime('now'))->add(new \DateInterval('P' . $this->cfg_daysfuture . 'D'));
+        $this->now = new \DateTime("now");
 
         $this->log("\n\n### run at " . print_r($this->now, true));
 
diff --git a/migrations/m220202_180401_initial.php b/migrations/m220202_180401_initial.php
@@ -0,0 +1,31 @@
+<?php
+
+use yii\db\Migration;
+
+class m220202_180401_initial extends Migration
+{
+
+    public function up()
+    {
+        $this->createTable('rss_posts', [
+            'id' => 'pk',
+            'rss_link' => 'varchar(1024) NOT NULL',
+            'post_id' => 'int(11) NOT NULL',
+        ], '');
+
+        // creates index for column `rss_link`
+        // hopefully mysql is smart enaugh to handle varchar(1024) decently
+        $this->createIndex(
+            'idx-rss_posts-rss_link',
+            'rss_posts',
+            'rss_link',
+        );
+    }
+
+    public function down()
+    {
+        echo "m220202_180401_initial does not support migration down.\n";
+        return false;
+    }
+
+}
diff --git a/migrations/uninstall.php b/migrations/uninstall.php
@@ -0,0 +1,18 @@
+<?php
+
+use yii\db\Migration;
+
+class uninstall extends Migration
+{
+    public function up()
+    {
+        $this->dropTable('rss_posts');
+    }
+
+    public function down()
+    {
+        echo "uninstall does not support migration down.\n";
+        return false;
+    }
+
+}
diff --git a/models/RssPosts.php b/models/RssPosts.php
@@ -0,0 +1,66 @@
+<?php
+
+namespace sij\humhub\modules\rss\models;
+
+use Yii;
+use humhub\components\ActiveRecord;
+#use humhub\modules\user\models\User;
+#use humhub\modules\tasks\models\Task;
+
+/**
+ * This is the model class for table "rss_posts".
+ *
+ * The followings are the available columns in table 'rss_posts':
+ * @property integer $id
+ * @property string $rss_link
+ * @property integer $post_id
+ */
+class RssPosts extends ActiveRecord
+{
+
+    /**
+     * @return string the associated database table name
+     */
+    public static function tableName()
+    {
+        return 'rss_posts';
+    }
+
+    /**
+     * @return array validation rules for model attributes.
+     */
+    public function rules()
+    {
+        return [
+            [ ['rss_link', 'post_id'], 'required' ],
+            [ ['rss_link'], 'string' ],
+            [ ['post_id'], 'integer' ],
+        ];
+    }
+
+/*
+    public function getTask()
+    {
+        return $this->hasOne(Task::className(), ['id' => 'task_id']);
+    }
+*/
+
+/*
+    public function afterSave($insert, $changedAttributes)
+    {
+        if ($insert) {
+            if(version_compare(Yii::$app->version, '1.2', '<')) {
+                $notification = new \humhub\modules\tasks\notifications\Assigned();
+                $notification->source = $this->task;
+                $notification->originator = Yii::$app->user->getIdentity();
+                $notification->send($this->user);
+            } else {
+                \humhub\modules\tasks\notifications\Assigned::instance()->from(Yii::$app->user->getIdentity())->about($this->task)->send($this->user);
+            }
+        }
+
+        return parent::afterSave($insert, $changedAttributes);
+    }
+*/
+
+}
