From 7dea53055bf188cba40f6527b0997db3b03b3aa0 Mon Sep 17 00:00:00 2001
From: Kaishiyoku <dmp@nfze.de>
Date: Wed, 23 May 2018 15:41:38 +0200
Subject: [PATCH] closed #4 fixed rating crawl logic

---
 app/Console/Commands/CrawlImages.php          | 35 +++++++++++++------
 .../2018_05_13_165545_create_images_table.php |  7 +---
 2 files changed, 25 insertions(+), 17 deletions(-)

diff --git a/app/Console/Commands/CrawlImages.php b/app/Console/Commands/CrawlImages.php
index 36ac460..2a66fff 100644
--- a/app/Console/Commands/CrawlImages.php
+++ b/app/Console/Commands/CrawlImages.php
@@ -97,7 +97,18 @@ public function handle()
 
     private function getContent($uri)
     {
-        return file_get_contents(env('CRAWLER_BASE_URL') . $uri);
+        $ch = curl_init();
+
+        curl_setopt($ch, CURLOPT_HEADER, 0);
+        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
+        curl_setopt($ch, CURLOPT_FRESH_CONNECT, true);
+        curl_setopt($ch, CURLOPT_URL, env('CRAWLER_BASE_URL') . $uri);
+
+        $content = curl_exec($ch);
+
+        curl_close($ch);
+
+        return $content;
     }
 
     private function getListContent($pageNumber = null)
@@ -145,15 +156,17 @@ private function getImages(Collection $uris, $isTest = false)
 
             $crawler = new Crawler($this->getContent($uri));
 
-            $rating = collect(
-                $crawler
-                    ->filterXPath('//table/tr/td')
-                    ->each(function (Crawler $node) {
-                        return $this->replaceNewLines($node->text());
-                    })
-            )->filter(function ($value) {
-                return in_array($value, $this->getRatings());
-            })->first();
+            $rating = collect($crawler
+                ->filterXPath('//table[@class="image_info form"]/tr')
+                ->each(function (Crawler $node) {
+                    $label = $this->replaceNewLines($node->children()->getNode(0)->textContent);
+                    $value = $this->replaceNewLines($node->children()->getNode(1)->textContent);
+
+                    return compact('label', 'value');
+                })
+            )->filter(function ($item) {
+                return $item['label'] == 'Rating' && in_array($item['value'], $this->getRatings());
+            })->first()['value'];
 
             $imageNode = $crawler->filter('img#main_image')->first();
 
@@ -188,7 +201,7 @@ private function getImages(Collection $uris, $isTest = false)
                 $image->save();
             }
 
-            $this->verbose(function () use ($externalId) {$this->line('  #' . $externalId);});
+            $this->verbose(function () use ($externalId, $rating) {$this->line('  #' . $externalId . '|' . $rating);});
         });
     }
 
diff --git a/database/migrations/2018_05_13_165545_create_images_table.php b/database/migrations/2018_05_13_165545_create_images_table.php
index 7330713..985cf5b 100644
--- a/database/migrations/2018_05_13_165545_create_images_table.php
+++ b/database/migrations/2018_05_13_165545_create_images_table.php
@@ -16,12 +16,7 @@ public function up()
         Schema::create('images', function (Blueprint $table) {
             $table->unsignedInteger('external_id')->unique();
             $table->string('url', 1528);
-            $table->enum('rating', [
-                'Unknown',
-                'Safe',
-                'Questionable',
-                'Explicit',
-            ]);
+            $table->enum('rating', explode(',', env('CRAWLER_RATINGS')));
             $table->timestamps();
 
             $table->primary('external_id');