jobifygarden/app/Console/Commands/CurateJobs.php

<?php

namespace App\Console\Commands;

use App\Models\Job;
use App\Models\Tag;
use Illuminate\Console\Attributes\Description;
use Illuminate\Console\Attributes\Signature;
use Illuminate\Console\Command;
use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Log;

#[Signature('jobs:curate {--batch=50}')]
#[Description('Curate jobs by fetching tags from an external API')]
class CurateJobs extends Command
{
    protected $signature = 'jobs:curate {--batch=50}';
    protected $description = 'Curate jobs by fetching tags from an external API';

    /**
     * Execute the console command.
     */
    public function handle(): int
    {
        $batchSize = (int) $this->option('batch');
        //$apiUrl = $this->option('api-url');
        $apiUrl = 'https://olm.c0rrup7.net/api/generate';

        if (empty($apiUrl)) {
            $this->error('API URL is required. Use --api-url option to provide the endpoint.');
            return self::FAILURE;
        }

        $log = Log::channel('jobs_curate');

        $this->info('Starting job curation...');
        $log->info('Starting job curation batch', ['batch_size' => $batchSize]);
        $this->newLine();

        // Fetch uncurated jobs
        $uncuratedJobs = Job::uncurated()
            ->limit($batchSize)
            ->get();

        if ($uncuratedJobs->isEmpty()) {
            $this->info('No uncurated jobs found.');
            return self::SUCCESS;
        }

        $this->info("Found {$uncuratedJobs->count()} uncurated jobs.");
        $this->newLine();

        $curatedCount = 0;
        $failedCount = 0;

        $progressBar = $this->output->createProgressBar($uncuratedJobs->count());
        $progressBar->start();

        foreach ($uncuratedJobs as $job) {
            try {
                $body = 'This is the content of a job ad, you need to read it and extract the relevant tags from it. Respond only with the tags in JSON format, NO OTHER TEXT BEFORE OR AFTER THE JSON.';
                $body .= 'Title: '. htmlentities($job->title);
                $body .= 'Description: '. htmlentities($job->description);
                $body .= 'RETURN ONLY VALID JSON, NO TEXT BEFORE OR AFTER THE JSON!';

                // Make POST request to API with job name and description
                $response = Http::timeout(300)
                    ->withHeaders([
                        'Accept' => 'application/json',
                        'CF-Access-Client-Id' => '8c4cc272c7f08f8d7911a28eb5d7db44.access',
                        'CF-Access-Client-Secret' => 'adeeed2f75236053953cc9dd898c7756bab5e00f7d0901bf7d488b6db8aba336',
                    ])
                    ->post($apiUrl, [
                    'model' => 'granite4.1:3b',
                    'stream' => false,
                    'prompt' => $body,
                ]);

                if ($response->successful()) {

                    $responseArray = $response->json();
                    $responseTagsJson = $responseArray['response'];
                    $log->info("Curated job id: {$job->id} Result: {$responseTagsJson}");
                    $tags = json_decode($responseTagsJson, true);

                    if (empty($tags) || $tags === null) {
                        continue;
                    }

                    //$tags = array_flip($tags);

                    if (is_array($tags) && !empty($tags)) {
                        $tagIds = [];

                        if (isset($tags['tags'])) {
                            $tags = $tags['tags'];
                        }

                        foreach ($tags as $tagName) {
                            // Find or create tag
                            $tag = Tag::firstOrCreate(
                                ['name' => $tagName]
                            );

                            $tagIds[] = $tag->id;
                        }

                        // Sync tags with the job
                        $job->tags()->sync($tagIds);

                        // Mark job as curated
                        $job->update(['is_curated' => true]);

                        $curatedCount++;
                    } else {
                        // Mark as curated even if no tags returned
                        $job->update(['is_curated' => true]);
                        $curatedCount++;
                    }
                } else {
                    $failedCount++;
                    $log->warning("Failed to curate job #{$job->id}: API returned {$response->status()} -> {$response->body()}");
                }
            } catch (\Exception $e) {
                $failedCount++;
                $log->error("Error curating job #{$job->id}: {$e->getMessage()} {$e->getTraceAsString()}");
            }

            $progressBar->advance();
        }

        $progressBar->finish();
        $this->newLine(2);

        $this->info("Successfully curated {$curatedCount} jobs.");
        $log->info("Curation complete", ['curated' => $curatedCount, 'failed' => $failedCount]);

        if ($failedCount > 0) {
            $this->warn("Failed to curate {$failedCount} jobs. Check logs for details.");
        }

        return self::SUCCESS;
    }
}