slake-school-service/tests/Unit/CrawlSourceResolverTest.php

<?php

namespace Tests\Unit;

use App\Models\CrawlAddress;
use App\Models\CrawlSource;
use App\Services\Crawl\CrawlSourceResolver;
use Illuminate\Foundation\Testing\RefreshDatabase;
use Tests\TestCase;

class CrawlSourceResolverTest extends TestCase
{
    use RefreshDatabase;

    public function test_resolves_ai_sjtu_center_to_research_center_api(): void
    {
        CrawlSource::query()->create([
            'name' => '师资列表页（通用 HTML）',
            'target_type' => 'teacher',
            'adapter_code' => 'faculty_list_html',
            'entry_url' => 'https://',
            'match_domains' => ['*'],
            'status' => 1,
            'sort' => 30,
        ]);

        CrawlSource::query()->create([
            'name' => '交大人工智能研究院研究中心',
            'target_type' => 'teacher',
            'adapter_code' => 'ai_sjtu_research_center_api',
            'entry_url' => 'https://ai.sjtu.edu.cn/center',
            'match_domains' => ['ai.sjtu.edu.cn'],
            'status' => 1,
            'sort' => 25,
        ]);

        $resolver = app(CrawlSourceResolver::class);
        $source = $resolver->resolve('https://ai.sjtu.edu.cn/center', 'teacher');

        $this->assertNotNull($source);
        $this->assertSame('ai_sjtu_research_center_api', $source->adapter_code);
    }

    public function test_prefers_crawl_address_bound_source_over_wildcard(): void
    {
        $wildcard = CrawlSource::query()->create([
            'name' => '师资列表页（通用 HTML）',
            'target_type' => 'teacher',
            'adapter_code' => 'faculty_list_html',
            'entry_url' => 'https://',
            'match_domains' => ['*'],
            'status' => 1,
            'sort' => 30,
        ]);

        $aiSource = CrawlSource::query()->create([
            'name' => '交大人工智能研究院研究中心',
            'target_type' => 'teacher',
            'adapter_code' => 'ai_sjtu_research_center_api',
            'entry_url' => 'https://ai.sjtu.edu.cn/center',
            'match_domains' => ['ai.sjtu.edu.cn'],
            'status' => 1,
            'sort' => 25,
        ]);

        CrawlAddress::query()->create([
            'target_type' => 'teacher',
            'name' => '交大 AI 中心',
            'request_url' => 'https://ai.sjtu.edu.cn/center',
            'crawl_source_id' => $aiSource->id,
            'status' => 1,
        ]);

        $resolver = app(CrawlSourceResolver::class);
        $source = $resolver->resolve('https://ai.sjtu.edu.cn/center', 'teacher');

        $this->assertSame('ai_sjtu_research_center_api', $source?->adapter_code);
        $this->assertNotSame($wildcard->id, $source?->id);
    }
}