From 7791847ead28063ed0e304c74dd8288a19f5bd72 Mon Sep 17 00:00:00 2001 From: Charles Teh Date: Wed, 22 Nov 2023 07:27:39 +0800 Subject: [PATCH] Update (fix) --- app/Http/Controllers/Tests/TestController.php | 16 ++++++++++++++++ app/Jobs/Tasks/CrawlRssPostTask.php | 5 +++-- routes/tests.php | 2 ++ 3 files changed, 21 insertions(+), 2 deletions(-) diff --git a/app/Http/Controllers/Tests/TestController.php b/app/Http/Controllers/Tests/TestController.php index eab0050..8fbaed8 100644 --- a/app/Http/Controllers/Tests/TestController.php +++ b/app/Http/Controllers/Tests/TestController.php @@ -5,6 +5,7 @@ use App\Helpers\FirstParty\ImageGen\ImageGen; use App\Http\Controllers\Controller; use App\Jobs\Tasks\BrowseRSSLatestNewsTask; +use App\Jobs\Tasks\CrawlRssPostTask; use App\Models\Post; use App\Notifications\PostWasPublished; use Illuminate\Http\Request; @@ -14,6 +15,21 @@ class TestController extends Controller { + public function crawlTask(Request $request) + { + $id = $request->input('id'); + + if (is_empty($id)) { + return "Missing 'id'."; + } + + CrawlRssPostTask::handle($id); + + return 'ok'; + + } + + public function opml(Request $request) { $raw_posts = BrowseRSSLatestNewsTask::handleSingle('https://hnrss.org/newest?q=ai', 240); diff --git a/app/Jobs/Tasks/CrawlRssPostTask.php b/app/Jobs/Tasks/CrawlRssPostTask.php index 95d38c1..14d7a75 100644 --- a/app/Jobs/Tasks/CrawlRssPostTask.php +++ b/app/Jobs/Tasks/CrawlRssPostTask.php @@ -42,11 +42,10 @@ public static function handle(int $rss_post_id) $service_cost_usage->name = 'smartproxy-CrawlRssPostTask'; $service_cost_usage->reference_1 = 'rss_post'; $service_cost_usage->reference_2 = strval($rss_post_id); - $service_cost_usage->output = null; + $service_cost_usage->output = self::getMarkdownFromHtml($raw_html); $service_cost_usage->save(); } else { - $raw_html = null; $response->throw(); } @@ -54,6 +53,8 @@ public static function handle(int $rss_post_id) $raw_html = null; } + //dd($raw_html); + if (! is_empty($raw_html)) { $rss_post->body = self::getMarkdownFromHtml($raw_html); } else { diff --git a/routes/tests.php b/routes/tests.php index f70bf9c..7b5208d 100644 --- a/routes/tests.php +++ b/routes/tests.php @@ -37,6 +37,8 @@ | */ +Route::get('/crawlTask', [App\Http\Controllers\Tests\TestController::class, 'crawlTask']); + Route::get('/recrawl_from', function (Request $request) { $id = $request->input('id');