From 3669d9c432a5b7a10bc7e8c0ad8091d8999e9aa8 Mon Sep 17 00:00:00 2001
From: j-h-s <j.harrissmall@gmail.com>
Date: Wed, 10 Jan 2018 21:10:01 +0100
Subject: [PATCH] [#71] streamlined twitter scraping a bit

---
 src/AppBundle/Service/TwitterService.php | 123 ++++++++++++++---------
 1 file changed, 73 insertions(+), 50 deletions(-)

diff --git a/src/AppBundle/Service/TwitterService.php b/src/AppBundle/Service/TwitterService.php
index 430c9fe..7d3442d 100644
--- a/src/AppBundle/Service/TwitterService.php
+++ b/src/AppBundle/Service/TwitterService.php
@@ -171,14 +171,16 @@ public function getTweets() {
                 $twTime = \DateTime::createFromFormat('D M d H:i:s P Y', $item->created_at);
                 // original string e.g. 'Mon Sep 08 15:19:11 +0000 2014'
 
-                $tweet = $this->getTweetDetails($item, $twTime);
-
-                if ($tweet === 'txt') {
+                if (empty($item->entities->media)) {
+                    $this->getTweetDetails($item, $twTime);
                     $txtCount++;
+
                 } else {
+                    $tweet = $this->getMedia($item, $twTime);
                     $imgCount += $tweet['img'];
                     $vidCount += $tweet['vid'];
                 }
+
             }
 
             $timeCheck = $twTime->getTimestamp(); // check time of last tweet scraped
@@ -210,16 +212,7 @@ public function getTweets() {
      * @return string
      */
     public function getTweetDetails($item, $twTime) {
-        if (!empty($item->full_text)) {
-            $twText = $item->full_text;
-        } else if (!empty($item->text)) {
-            $twText = $item->text;
-        } else $twText = null;
-
-        if (!empty($item->entities->media)) { // if tweet contains media
-            $tweet = $this->getMediaDetails($item, $twTime, $twText);
-            return $tweet;
-        }
+        $twText = $this->getTwText($item);
 
         $allData = [
             'id'       => $item->id,
@@ -241,61 +234,32 @@ public function getTweetDetails($item, $twTime) {
             $item->favorite_count,
             $allData
         );
-
-        return 'txt';
     }
 
 
     /**
-     * Retrieves the details of an image or video
+     * Determines the type of media that a tweet contains
      * @param  object $item
      * @param  object $twTime
-     * @param  string $twText
      * @return array
      */
-    public function getMediaDetails($item, $twTime, $twText) {
-        $media = $item->extended_entities->media;
-
+    public function getMedia($item, $twTime) {
         $imgCount = 0;
         $vidCount = 0;
 
-        foreach ($media as $photo) { // if tweet contains multiple images
+        $media = $item->extended_entities->media;
+
+        foreach ($media as $photo) {
             if ($photo->type == 'video') {
                 $subType = SocialMedia::SUBTYPE_VIDEO;
                 $vidCount++;
-            } else { // if type == 'photo' or 'animated_gif'
+
+            } else { // if type == 'photo' or 'animted_gif'
                 $subType = SocialMedia::SUBTYPE_IMAGE;
                 $imgCount++;
             }
 
-            $imgSrc = $photo->media_url . ":small";
-            $imgId  = $photo->id;
-            $img    = $this->container
-                ->get('ImageService')
-                ->saveImage('tw', $this->partyCode, $imgSrc, $imgId);
-
-            $allData = [
-                'id'         => $imgId,
-                'posted'     => $twTime->format('Y-m-d H:i:s'), // string
-                'text'       => $twText,
-                'image'      => $img,
-                'img_source' => $imgSrc,
-                'url'        => 'https://twitter.com/statuses/' . $item->id,
-                'likes'      => $item->favorite_count,
-                'retweets'   => $item->retweet_count
-                ];
-
-            $this->db->addSocial(
-                $this->partyCode,
-                SocialMedia::TYPE_TWITTER,
-                $subType,
-                $item->id,
-                $twTime, // DateTime
-                $twText,
-                $img,
-                $item->favorite_count,
-                $allData
-            );
+            $this->getMediaDetails($item, $twTime, $photo, $subType);
         }
 
         $count['img'] = $imgCount;
@@ -303,4 +267,63 @@ public function getMediaDetails($item, $twTime, $twText) {
         return $count;
     }
 
+
+    /**
+     * Retrieves the details of an image or video
+     * @param  object $item
+     * @param  object $twTime
+     * @param  string $photo
+     * @return array
+     */
+    public function getMediaDetails($item, $twTime, $photo, $subType) {
+        $twText = $this->getTwText($item);
+
+        $imgSrc = $photo->media_url . ":small";
+        $imgId  = $photo->id;
+        $img    = $this->container
+            ->get('ImageService')
+            ->saveImage('tw', $this->partyCode, $imgSrc, $imgId);
+
+        $allData = [
+            'id'         => $imgId,
+            'posted'     => $twTime->format('Y-m-d H:i:s'), // string
+            'text'       => $twText,
+            'image'      => $img,
+            'img_source' => $imgSrc,
+            'url'        => 'https://twitter.com/statuses/' . $item->id,
+            'likes'      => $item->favorite_count,
+            'retweets'   => $item->retweet_count
+            ];
+
+        $this->db->addSocial(
+            $this->partyCode,
+            SocialMedia::TYPE_TWITTER,
+            $subType,
+            $item->id,
+            $twTime, // DateTime
+            $twText,
+            $img,
+            $item->favorite_count,
+            $allData
+        );
+    }
+
+
+    /**
+    * Returns text field of a tweet
+    * @param  object $item
+    * @return string
+    */
+    public function getTwText($item) {
+        if (!empty($item->full_text)) {
+            return $item->full_text;
+        }
+
+        if (!empty($item->text)) {
+            return $item->text;
+        }
+
+        return null;
+    }
+
 }