math_cms/app/Jobs/ProcessMarkdownCandidateBatch.php

<?php

namespace App\Jobs;

use App\Models\MarkdownImport;
use App\Models\PreQuestionCandidate;
use App\Services\MarkdownQuestionParser;
use App\Services\PdfStorageService;
use Illuminate\Bus\Queueable;
use Illuminate\Contracts\Queue\ShouldQueue;
use Illuminate\Foundation\Bus\Dispatchable;
use Illuminate\Queue\InteractsWithQueue;
use Illuminate\Queue\SerializesModels;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\Log;

class ProcessMarkdownCandidateBatch implements ShouldQueue
{
    use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;

    public int $timeout = 300; // 5分钟超时
    public int $tries = 3;

    public function __construct(
        public int $markdownImportId,
        public int $sequenceStart,
        public int $sequenceEnd
    ) {
        //
    }

    public function handle(MarkdownQuestionParser $parser, PdfStorageService $uploader): void
    {
        $import = MarkdownImport::find($this->markdownImportId);
        if (!$import) {
            Log::error('MarkdownImport not found for batch', [
                'id' => $this->markdownImportId,
                'sequence_start' => $this->sequenceStart,
                'sequence_end' => $this->sequenceEnd,
            ]);
            return;
        }

        Log::info('Markdown batch started', [
            'import_id' => $this->markdownImportId,
            'sequence_start' => $this->sequenceStart,
            'sequence_end' => $this->sequenceEnd,
        ]);

        $records = PreQuestionCandidate::query()
            ->where('import_id', $this->markdownImportId)
            ->whereBetween('sequence', [$this->sequenceStart, $this->sequenceEnd])
            ->orderBy('sequence')
            ->get();

        $processed = 0;
        $failed = 0;

        foreach ($records as $record) {
            try {
                $meta = $record->meta ?? [];
                if (!empty($meta['ai_parsed'])) {
                    $processed++;
                    continue;
                }

                $existingConfidence = $record->confidence ?? $record->ai_confidence;

                // 置信度高（>=0.85）时跳过再次 AI 解析，直接计入进度
                if ($existingConfidence !== null && (float) $existingConfidence >= 0.85) {
                    $this->markParsed($record);
                    $processed++;
                    continue;
                }

                // 快速过滤卷子/区块标题，避免误判为题目再次走 AI
                if (!$this->isLikelyQuestion((string) $record->raw_markdown)) {
                    $meta['ai_parsed'] = true;
                    $meta['ai_parsed_at'] = now()->toDateTimeString();

                    $record->update([
                        'is_question_candidate' => false,
                        'ai_confidence' => 0.0,
                        'confidence' => 0.0,
                        'is_valid_question' => false,
                        'status' => 'rejected',
                        'meta' => $meta,
                    ]);
                    $processed++;
                    continue;
                }

                // 已经处理过的不重复处理
                if (in_array($record->status, ['pending', 'reviewed', 'accepted', 'rejected'], true) && $record->stem !== null) {
                    $this->markParsed($record);
                    continue;
                }

                $parsed = $parser->parseRawMarkdown((string) $record->raw_markdown, (int) $record->index);

                $meta = $record->meta ?? [];
                $meta['ai_parsed'] = true;
                $meta['ai_parsed_at'] = now()->toDateTimeString();

                // 结构化后立即回传图片
                $uploadedImages = [];
                if (!empty($parsed['images'])) {
                    foreach ($parsed['images'] as $idx => $imgUrl) {
                        $path = "imports/images/{$record->id}_{$idx}.jpg";
                        $uploadedImages[] = $uploader->put($path, (string)@file_get_contents($imgUrl)) ?: $imgUrl;
                    }
                }
                $meta['images_uploaded'] = !empty($uploadedImages);

                $record->update([
                    'stem' => $parsed['stem'] ?? null,
                    'options' => $parsed['options'] ?? null,
                    'images' => !empty($uploadedImages) ? $uploadedImages : $parsedImages,
                    'tables' => $parsed['tables'] ?? [],
                    'is_question_candidate' => (bool) ($parsed['is_question_candidate'] ?? false),
                    'ai_confidence' => $parsed['ai_confidence'] ?? null,
                    'status' => 'pending',
                    'meta' => $meta,
                ]);

                $processed++;
            } catch (\Throwable $e) {
                $failed++;

                Log::warning('Markdown batch item failed', [
                    'import_id' => $this->markdownImportId,
                    'candidate_id' => $record->id,
                    'sequence' => $record->sequence,
                    'index' => $record->index,
                    'error' => $e->getMessage(),
                ]);
            }
        }

        $this->refreshProgress();

        Log::info('Markdown batch finished', [
            'import_id' => $this->markdownImportId,
            'sequence_start' => $this->sequenceStart,
            'sequence_end' => $this->sequenceEnd,
            'processed' => $processed,
            'failed' => $failed,
        ]);

        $this->finalizeIfDone();
    }

    private function finalizeIfDone(): void
    {
        $import = MarkdownImport::find($this->markdownImportId);
        if (!$import) {
            return;
        }

        $total = (int) ($import->progress_total ?? 0);
        $current = (int) ($import->progress_current ?? 0);

        if ($total <= 0 || $current < $total) {
            return;
        }

        // 只要有一个 batch 到达“完成条件”，就尝试做一次幂等的最终状态更新
        $updated = DB::table('markdown_imports')
            ->where('id', $this->markdownImportId)
            ->where('status', 'processing')
            ->update([
                'status' => 'parsed',
                'progress_stage' => MarkdownImport::STAGE_PARSED,
                'progress_message' => '解析完成，等待人工校对',
                'progress_updated_at' => now(),
                'processing_finished_at' => now(),
            ]);

        if ($updated) {
            Log::info('Markdown import finalized', [
                'import_id' => $this->markdownImportId,
                'progress_total' => $total,
                'progress_current' => $current,
            ]);
        }
    }

    private function refreshProgress(): void
    {
        $total = PreQuestionCandidate::query()
            ->where('import_id', $this->markdownImportId)
            ->where('status', '!=', PreQuestionCandidate::STATUS_SUPERSEDED)
            ->count();

        $parsed = PreQuestionCandidate::query()
            ->where('import_id', $this->markdownImportId)
            ->where('status', '!=', PreQuestionCandidate::STATUS_SUPERSEDED)
            ->where(function ($query) {
                $query->whereRaw("JSON_UNQUOTE(JSON_EXTRACT(meta, '$.ai_parsed')) = 'true'")
                    ->orWhereNotNull('stem')
                    ->orWhereNotNull('ai_confidence')
                    ->orWhereNotNull('confidence');
            })
            ->count();

        DB::table('markdown_imports')
            ->where('id', $this->markdownImportId)
            ->update([
                'progress_total' => $total,
                'progress_current' => min($parsed, $total),
                'progress_updated_at' => now(),
                'progress_stage' => MarkdownImport::STAGE_AI_PARSING,
                'progress_message' => 'AI 解析中…',
            ]);
    }

    private function markParsed(PreQuestionCandidate $record): void
    {
        $meta = $record->meta ?? [];
        if (!empty($meta['ai_parsed'])) {
            return;
        }

        $meta['ai_parsed'] = true;
        $meta['ai_parsed_at'] = now()->toDateTimeString();
        $record->update(['meta' => $meta]);
    }

    /**
     * 轻量启发式判断是否像一道题目，过滤卷子/部分标题和说明文字。
     */
    private function isLikelyQuestion(string $raw): bool
    {
        $text = trim(strip_tags($raw));
        $length = mb_strlen($text);

        // Markdown 标题或“卷/部分/说明”且文本很短，视为非题
        if (preg_match('/^#+\\s+/m', $raw)) {
            return false;
        }

        if (preg_match('/(第[一二三四五六七八九十IVX]+[卷部分]|题型|说明|试卷)/u', $text) && $length <= 80) {
            return false;
        }

        // 过短且无问句/命令词/选项特征
        if ($length < 25 && !preg_match('/[\\?？求解求证计算]|[A-D]\\.|（本小题满分/u', $text)) {
            return false;
        }

        return true;
    }
}