爬虫开发

app/Console/Commands/CrawlerRunCommand.php

@@ -0,0 +1,64 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Console\Commands;
+
+use App\Enums\CrawlTriggerType;
+use App\Jobs\RunCrawlRuleJob;
+use App\Models\CrawlRule;
+use App\Services\Crawler\CrawlRuleScheduleService;
+use Illuminate\Console\Command;
+
+class CrawlerRunCommand extends Command
+{
+    protected $signature = 'crawler:run {ruleId? : 指定规则ID} {--all : 忽略cron，执行全部启用规则} {--sync : 同步执行，不走队列}';
+
+    protected $description = '执行采集规则';
+
+    public function handle(CrawlRuleScheduleService $scheduleService): int
+    {
+        $ruleId = $this->argument('ruleId');
+
+        $query = CrawlRule::query()->where('enabled', true);
+
+        if ($ruleId !== null) {
+            $query->whereKey((int) $ruleId);
+        }
+
+        $rules = $query->orderBy('id')->get();
+
+        if ($rules->isEmpty()) {
+            $this->warn('没有可执行的采集规则');
+
+            return self::SUCCESS;
+        }
+
+        $shouldRunAll = (bool) $this->option('all') || $ruleId !== null;
+        $sync = (bool) $this->option('sync');
+
+        $dispatched = 0;
+
+        foreach ($rules as $rule) {
+            if (! $shouldRunAll && ! $scheduleService->isDue($rule)) {
+                continue;
+            }
+
+            if ($sync) {
+                RunCrawlRuleJob::dispatchSync($rule->id, CrawlTriggerType::Schedule->value);
+            } else {
+                RunCrawlRuleJob::dispatch($rule->id, CrawlTriggerType::Schedule->value);
+            }
+
+            $dispatched++;
+            $this->info(sprintf('已提交规则 #%d %s', $rule->id, $rule->name));
+        }
+
+        if ($dispatched === 0) {
+            $this->line('当前无到期规则');
+        }
+
+        return self::SUCCESS;
+    }
+}
+