From 2dfe5429047a3ba8bacc5410c6b913e655385f9e Mon Sep 17 00:00:00 2001
From: Brett Spaulding <brett.spaulding@stonesoft.io>
Date: Mon, 5 Aug 2024 22:02:23 -0400
Subject: [PATCH] [WIP] Artist scraping working

---
 .../app/Http/Controllers/SearchController.php | 61 ++++++++++++++++---
 ...9_add_artist_fields_thumbnail_selected.php | 29 +++++++++
 php/src/public/js/app.js                      |  8 +--
 3 files changed, 85 insertions(+), 13 deletions(-)
 create mode 100644 php/src/database/migrations/2024_08_06_015329_add_artist_fields_thumbnail_selected.php

diff --git a/php/src/app/Http/Controllers/SearchController.php b/php/src/app/Http/Controllers/SearchController.php
index 32f530d..d6fd948 100644
--- a/php/src/app/Http/Controllers/SearchController.php
+++ b/php/src/app/Http/Controllers/SearchController.php
@@ -2,6 +2,7 @@
 
 namespace App\Http\Controllers;
 
+use Facebook\WebDriver\WebDriverExpectedCondition;
 use Illuminate\Http\Request;
 
 use Facebook\WebDriver\Remote\DesiredCapabilities;
@@ -12,10 +13,12 @@ use Facebook\WebDriver\WebDriverBy;
 class SearchController extends Controller
 {
 
-    protected function setUp() {
+    protected function setUp()
+    {
         $host = 'http://selenium-hub:4444';
         $capabilities = DesiredCapabilities::chrome();
         $chromeOptions = new ChromeOptions();
+        // TODO: Add '--headless' back in to arguments
         $chromeOptions->addArguments(['--no-sandbox', '--disable-dev-shm-usage']);
         $capabilities->setCapability(ChromeOptions::CAPABILITY_W3C, $chromeOptions);
         $driver = RemoteWebDriver::create($host, $capabilities);
@@ -25,8 +28,6 @@ class SearchController extends Controller
 
     public function search_artist(Request $request, string $artist)
     {
-        \Log::info($artist);
-        \Log::info($request);
         \Log::info('Getting Artist: ' . $artist);
 //        $url = 'https://example.com';
         $url = 'https://music.youtube.com/search?q=' . str_replace(' ', '+', $artist);
@@ -36,12 +37,58 @@ class SearchController extends Controller
         // the URL to the local Selenium Server
         $driver = $this->setUp();
         $driver->get($url);
-        $html = $driver->getPageSource();
 
-        \Log::info($html);
-        \Log::info('=========================================');
+        // Click the artist button to force a "structure" of results
+        $artistBtnXpath = '//a[@title="Show artist results"]';
+        $driver->wait(10, 500)->until(
+            WebDriverExpectedCondition::visibilityOfElementLocated(WebDriverBy::xpath($artistBtnXpath))
+        );
+        $driver->findElement(WebDriverBy::xpath($artistBtnXpath))->click();
+        // Youtube has multiple elements with the same ID (Naughty!).  We will give a reasonable analog time to render.
+        sleep(5);
 
-        $driver->quit();
+        $contentDivs = $driver->findElements(WebDriverBy::cssSelector('#contents'));
+        $divCount = 0;
+        foreach ($contentDivs as $content) {
+            $divCount += 1;
+
+            $artists = $content->findElements(WebDriverBy::xpath('//ytmusic-responsive-list-item-renderer'));
+
+            if ($artists) {
+                $resultCap = 6;
+                $resultIndex = 0;
+                foreach ($artists as $artist) {
+                    // There are a bunch of elements with no text in them; just a quick and dirty filter
+                    $hasText = $artist->getText();
+                    if ($hasText) {
+                        $resultIndex += 1;
+                        \Log::info('===================================================================================================================================');
+                        \Log::info('===================================================================================================================================');
+//                        \Log::info($artist->getDomProperty('innerHTML'));
+
+                        // Artist Data Targeting
+                        $artistThumbnail = $artist->findElement(WebDriverBy::cssSelector('img'))->getAttribute('src');
+                        $artistLink = $artist->findElements(WebDriverBy::cssSelector('a'));
+                        $artistHref = $artistLink[0]->getAttribute('href');
+                        $artistName = $artistLink[0]->getAttribute('aria-label');
+
+                        \Log::info($artistName . ': ' . $artistHref);
+                        \Log::info($artistThumbnail);
 
+                        if($resultCap <= $resultIndex) {
+                            break;
+                        }
+
+                    }
+                }
+
+                if ($divCount === 1) {
+                    break;
+                }
+
+            }
+        }
+
+        $driver->quit();
     }
 }
diff --git a/php/src/database/migrations/2024_08_06_015329_add_artist_fields_thumbnail_selected.php b/php/src/database/migrations/2024_08_06_015329_add_artist_fields_thumbnail_selected.php
new file mode 100644
index 0000000..6043c5b
--- /dev/null
+++ b/php/src/database/migrations/2024_08_06_015329_add_artist_fields_thumbnail_selected.php
@@ -0,0 +1,29 @@
+<?php
+
+use Illuminate\Database\Migrations\Migration;
+use Illuminate\Database\Schema\Blueprint;
+use Illuminate\Support\Facades\Schema;
+
+return new class extends Migration
+{
+    /**
+     * Run the migrations.
+     */
+    public function up(): void
+    {
+        Schema::table('artists', function (Blueprint $table) {
+            $table->string('thumbnail')->nullable()->after('name');
+            $table->boolean('selected')->default(false)->after('thumbnail');
+        });
+    }
+
+    /**
+     * Reverse the migrations.
+     */
+    public function down(): void
+    {
+        Schema::table('artists', function (Blueprint $table) {
+            //
+        });
+    }
+};
diff --git a/php/src/public/js/app.js b/php/src/public/js/app.js
index 8d06134..f8d72d6 100644
--- a/php/src/public/js/app.js
+++ b/php/src/public/js/app.js
@@ -19,15 +19,11 @@ $('#queue_btn').on('click', () => {
 })
 
 $('#download_btn').on('click', () => {
-    console.log('Blocking UI');
     loader.fadeIn(300);
 
     let artist = $('#search_bar').val();
-    // Prevent
-    $('#search_bar').val('');
     let icon = 'error';
     let title = 'What the flip?!';
-    let text = 'You need to add an artist bro..';
 
     setTimeout(() => {
         if (artist) {
@@ -40,7 +36,7 @@ $('#download_btn').on('click', () => {
                     console.log('===========');
                     icon = 'success';
                     title = 'Shazam!';
-                    proc_notification(icon, title, text);
+                    proc_notification(icon, title, 'Artist found');
                     loader.fadeOut(700);
                 },
                 error: (response) => {
@@ -53,7 +49,7 @@ $('#download_btn').on('click', () => {
             });
 
         } else {
-            proc_notification(icon, title, text);
+            proc_notification(icon, title, 'You need to add an artist, c\'mon man!');
             loader.fadeOut(700);
         }
     }, 100);