math_cms/app/Services/QuestionTemReviewService.php

<?php

namespace App\Services;

use App\Models\Question;
use App\Support\AnswerSolutionStepMarkerInjector;
use Illuminate\Support\Facades\Cache;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\Schema;
use Illuminate\Support\Str;

/**
 * questions_tem 质检入库页：知识点排序、PDF 口径预览、写入 questions
 */
class QuestionTemReviewService
{
    /** Session 键：供「入库题目调难度」页列举的本轮已入库 question.id */
    public const SESSION_TUNING_QUESTION_IDS = 'import_difficulty_tune_question_ids';

    /**
     * 将成功入库的正式题 ID 合并进会话，供调难度页使用。
     *
     * @param  list<int>  $questionIds
     */
    public static function mergeQuestionIdsIntoTuningSession(array $questionIds): void
    {
        $questionIds = array_values(array_unique(array_filter(array_map('intval', $questionIds))));
        if ($questionIds === []) {
            return;
        }
        $existing = session(self::SESSION_TUNING_QUESTION_IDS, []);
        if (! is_array($existing)) {
            $existing = [];
        }
        session([
            self::SESSION_TUNING_QUESTION_IDS => array_values(array_unique(array_merge($existing, $questionIds))),
        ]);
    }

    /**
     * 左侧：按 questions 表中该知识点正式题数量升序（题少的在前），仅包含 questions_tem 中出现过的 kp_code
     *
     * @param  ?int  $limit  为 null 时不截断（质检页需完整列表 + 搜索，否则题量大的 KP 如 B01 会落在 500 条之后而无法检索）
     * @return list<array{kp_code: string, kp_name: string, questions_count: int, tem_count: int, tem_importable_count: int}>
     */
    public function listKnowledgePointsByQuestionsAsc(?int $limit = null, ?int $gradeFilter = null, ?int $semesterFilter = null): array
    {
        if (! Schema::hasTable('questions_tem')) {
            return [];
        }

        $gradeKpWhitelist = null;
        if ($gradeFilter !== null || $semesterFilter !== null) {
            $gradeKpWhitelist = $this->kpCodesFromCatalogFilter($gradeFilter, $semesterFilter);
            if ($gradeKpWhitelist === []) {
                return [];
            }
        }

        $temQuery = DB::table('questions_tem')
            ->whereNotNull('kp_code')
            ->where('kp_code', '!=', '');
        if ($gradeKpWhitelist !== null) {
            $temQuery->whereIn('kp_code', $gradeKpWhitelist);
        }
        $temKps = $temQuery->distinct()->pluck('kp_code')->all();

        if ($temKps === []) {
            return [];
        }

        $kpNames = [];
        if (Schema::hasTable('knowledge_points')) {
            $kpNames = DB::table('knowledge_points')
                ->whereIn('kp_code', $temKps)
                ->pluck('name', 'kp_code')
                ->toArray();
        }

        $counts = [];
        if (Schema::hasTable('questions')) {
            $countQuery = DB::table('questions')->whereIn('kp_code', $temKps);
            $counts = $countQuery
                ->selectRaw('kp_code, COUNT(*) as c')
                ->groupBy('kp_code')
                ->pluck('c', 'kp_code')
                ->toArray();
        }

        $temCountQuery = DB::table('questions_tem')->whereIn('kp_code', $temKps);
        $temCounts = $temCountQuery
            ->selectRaw('kp_code, COUNT(*) as c')
            ->groupBy('kp_code')
            ->pluck('c', 'kp_code')
            ->toArray();

        $importableMap = $this->importableTemCountsForKpCodes($temKps, $gradeFilter);

        $rows = [];
        foreach ($temKps as $kp) {
            $name = isset($kpNames[$kp]) ? trim((string) $kpNames[$kp]) : '';
            $rows[] = [
                'kp_code' => $kp,
                'kp_name' => $name,
                'questions_count' => (int) ($counts[$kp] ?? 0),
                'tem_count' => (int) ($temCounts[$kp] ?? 0),
                'tem_importable_count' => (int) ($importableMap[$kp] ?? 0),
            ];
        }

        usort($rows, function ($a, $b) {
            if ($a['questions_count'] === $b['questions_count']) {
                return strcmp($a['kp_code'], $b['kp_code']);
            }

            return $a['questions_count'] <=> $b['questions_count'];
        });

        if ($limit !== null && $limit > 0) {
            return array_slice($rows, 0, $limit);
        }

        return $rows;
    }

    /**
     * 与中间列表 {@see listTemQuestionsForKp}（excludeFormalDuplicates=true）同口径：本 KP 下、题干未与正式库重复、且题干非空的 questions_tem 行数。
     *
     * @param  list<string>  $kpCodes
     * @return array<string, int>  kp_code => count
     */
    private function importableTemCountsForKpCodes(array $kpCodes, ?int $gradeFilter = null): array
    {
        $kpCodes = array_values(array_unique(array_filter($kpCodes)));
        if ($kpCodes === [] || ! Schema::hasTable('questions_tem')) {
            return [];
        }

        if (! Schema::hasTable('questions')) {
            return collect($kpCodes)
                ->mapWithKeys(function (string $k): array {
                    $q = DB::table('questions_tem')->where('kp_code', $k);
                    return [$k => (int) $q->count()];
                })
                ->all();
        }

        $placeholders = implode(',', array_fill(0, count($kpCodes), '?'));
        $sql = "
            SELECT t.kp_code AS kp_code, COUNT(*) AS c
            FROM questions_tem t
            WHERE t.kp_code IN ($placeholders)
            AND t.stem IS NOT NULL AND t.stem != ''
            AND NOT EXISTS (
                SELECT 1 FROM questions q
                WHERE q.kp_code = t.kp_code
                AND q.stem != ''
                AND q.stem = t.stem
            )
            GROUP BY t.kp_code
        ";

        $out = array_fill_keys($kpCodes, 0);
        foreach (DB::select($sql, $kpCodes) as $row) {
            $k = (string) ($row->kp_code ?? '');
            if ($k !== '') {
                $out[$k] = (int) ($row->c ?? 0);
            }
        }

        return $out;
    }

    /**
     * 与入库、判重一致：questions_tem.stem（当前库表仅此 longtext 题干列，无 content）。
     */
    public function normalizedStemFromTemRow(object|array $row): string
    {
        $arr = is_array($row) ? $row : (array) $row;

        return (string) ($arr['stem'] ?? '');
    }

    /**
     * 中间：某知识点下 questions_tem 题目（限制条数）
     *
     * @param  bool  $excludeFormalDuplicates  为 true 时排除「正式库 questions 已存在同 kp_code + 同 stem」的待审行，与 {@see existsDuplicateInQuestions} 一致，减少无效质检
     * @return list<object>
     */
    public function listTemQuestionsForKp(string $kpCode, int $limit = 300, bool $excludeFormalDuplicates = true, ?int $gradeFilter = null): array
    {
        if (! Schema::hasTable('questions_tem') || $kpCode === '') {
            return [];
        }

        $formalStemSet = [];
        if ($excludeFormalDuplicates && Schema::hasTable('questions')) {
            $qFormal = DB::table('questions')->where('kp_code', $kpCode);
            foreach ($qFormal->pluck('stem') as $stem) {
                if ($stem === null || $stem === '') {
                    continue;
                }
                $formalStemSet[(string) $stem] = true;
            }
        }

        $out = [];
        $q = DB::table('questions_tem')->where('kp_code', $kpCode)->orderBy('id');
        foreach ($q->lazyById(200) as $row) {
            if ($excludeFormalDuplicates && $formalStemSet !== []) {
                $stem = $this->normalizedStemFromTemRow($row);
                if ($stem !== '' && isset($formalStemSet[$stem])) {
                    continue;
                }
            }
            $out[] = $row;
            if (count($out) >= $limit) {
                break;
            }
        }

        return $out;
    }

    /**
     * 与 ExamPdfExportService::renderPreviewHtml 一致：公式预处理 + 解析换行，供页面 KaTeX 渲染
     *
     * @param  array<string, mixed>  $row  questions_tem 一行转数组
     * @return array{stem: string, options: ?array, answer: string, solution: string, question_type: string}
     */
    public function buildPdfStylePreviewFields(array $row): array
    {
        $stem = (string) ($row['stem'] ?? '');
        $answer = (string) ($row['answer'] ?? $row['correct_answer'] ?? '');
        $solution = AnswerSolutionStepMarkerInjector::enrichIfNeeded(
            (string) ($row['solution'] ?? ''),
            $row['question_type'] ?? $row['tags'] ?? 'answer'
        );
        $questionType = strtolower((string) ($row['question_type'] ?? $row['tags'] ?? 'answer'));

        $options = $row['options'] ?? null;
        if (is_string($options) && trim($options) !== '') {
            $decoded = json_decode($options, true);
            $options = is_array($decoded) ? $decoded : null;
        }

        $processedStem = MathFormulaProcessor::processFormulas($stem);
        $processedAnswer = MathFormulaProcessor::processFormulas($answer);
        $processedSolution = MathFormulaProcessor::processFormulas($this->formatNewlinesForPdf($solution));

        $processedOptions = null;
        if (is_array($options)) {
            $processedOptions = [];
            foreach ($options as $key => $value) {
                if (is_array($value)) {
                    $text = (string) ($value['text'] ?? $value['value'] ?? reset($value) ?? '');
                    $processedOptions[$key] = MathFormulaProcessor::processFormulas($text);
                } else {
                    $processedOptions[$key] = MathFormulaProcessor::processFormulas((string) $value);
                }
            }
        }

        return [
            'stem' => $processedStem,
            'options' => $processedOptions,
            'answer' => $processedAnswer,
            'solution' => $processedSolution,
            'question_type' => $questionType,
        ];
    }

    private function formatNewlinesForPdf(?string $text): string
    {
        if ($text === null || $text === '') {
            return '';
        }
        $text = preg_replace('/\\\\n(?![a-zA-Z])/', '<br>', $text);

        return (string) preg_replace('/(<br>\s*){3,}/', '<br><br>', $text);
    }

    /**
     * 是否已在 questions 中存在（同 kp + 题干完全一致则视为重复）
     */
    public function existsDuplicateInQuestions(string $kpCode, string $stem): bool
    {
        if ($stem === '' || ! Schema::hasTable('questions')) {
            return false;
        }

        return Question::query()
            ->where('kp_code', $kpCode)
            ->where('stem', $stem)
            ->exists();
    }

    /**
     * 待审行默认难度：与入库写入规则一致，限制在 [0.00, 0.90] 并保留两位小数
     *
     * @param  object|array<string, mixed>  $row
     */
    public function defaultDifficultyForTemRow(object|array $row): float
    {
        $arr = is_array($row) ? $row : (array) $row;
        $d = 0.5;
        if (array_key_exists('difficulty', $arr) && $arr['difficulty'] !== null && $arr['difficulty'] !== '') {
            $d = (float) $arr['difficulty'];
        }

        return max(0.0, min(0.9, round($d, 2)));
    }

    /**
     * 将 questions_tem 一行写入 questions（入库）
     *
     * @param  ?float  $difficultyOverride  若传入则作为 questions.difficulty（仍限制 0.00–0.90、两位小数）；null 时按表内字段或默认 0.5
     * @return array{ok: bool, message: string, question_id: ?int}
     */
    public function importTemRowToQuestions(int $temId, ?float $difficultyOverride = null): array
    {
        if (! Schema::hasTable('questions_tem')) {
            return ['ok' => false, 'message' => 'questions_tem 表不存在', 'question_id' => null];
        }

        $row = DB::table('questions_tem')->where('id', $temId)->first();
        if (! $row) {
            return ['ok' => false, 'message' => '待入库题目不存在', 'question_id' => null];
        }

        $arr = (array) $row;
        $stem = $this->normalizedStemFromTemRow($arr);
        $kp = (string) ($arr['kp_code'] ?? '');
        if ($stem === '' || $kp === '') {
            return ['ok' => false, 'message' => '题干或知识点为空', 'question_id' => null];
        }

        if ($this->existsDuplicateInQuestions($kp, $stem)) {
            return ['ok' => false, 'message' => '正式库已存在相同知识点且题干一致的题目', 'question_id' => null];
        }

        $options = $arr['options'] ?? null;
        if (is_string($options) && trim($options) !== '') {
            $decoded = json_decode($options, true);
            $options = is_array($decoded) ? $decoded : null;
        }

        $questionType = $this->normalizeQuestionTypeForDb($arr['question_type'] ?? $arr['tags'] ?? 'answer');

        $difficulty = $difficultyOverride !== null
            ? max(0.0, min(0.9, round($difficultyOverride, 2)))
            : $this->defaultDifficultyForTemRow($arr);

        $rawSolution = (string) ($arr['solution'] ?? '');
        $solutionStored = AnswerSolutionStepMarkerInjector::enrichIfNeeded($rawSolution, $arr['question_type'] ?? $arr['tags'] ?? 'answer');

        $payload = [
            'question_code' => 'QT'.strtoupper(Str::random(12)),
            'question_type' => $questionType,
            'kp_code' => $kp,
            'stem' => $stem,
            'options' => $options,
            'answer' => (string) ($arr['answer'] ?? $arr['correct_answer'] ?? ''),
            'solution' => $solutionStored,
            'difficulty' => $difficulty,
            'source' => 'questions_tem_review',
            'tags' => is_string($arr['tags'] ?? null) ? $arr['tags'] : null,
            'meta' => [
                'imported_from' => 'questions_tem',
                'questions_tem_id' => $temId,
            ],
        ];

        if (isset($arr['textbook_id'])) {
            $payload['textbook_id'] = (int) $arr['textbook_id'];
        }

        try {
            $question = Question::query()->create($payload);

            $updates = [];
            if (Schema::hasColumn('questions', 'audit_status')) {
                $updates['audit_status'] = 0;
            }
            if (Schema::hasColumn('questions', 'grade') && isset($arr['grade'])) {
                $updates['grade'] = (int) $arr['grade'];
            }
            if ($updates !== []) {
                DB::table('questions')->where('id', $question->id)->update($updates);
            }

            return [
                'ok' => true,
                'message' => '已入库',
                'question_id' => (int) $question->id,
            ];
        } catch (\Throwable $e) {
            return [
                'ok' => false,
                'message' => '入库失败：'.$e->getMessage(),
                'question_id' => null,
            ];
        }
    }

    /**
     * 批量将 questions_tem 行写入 questions（每行逻辑与 importTemRowToQuestions 相同）
     *
     * @param  list<int>  $temIds
     * @return array{imported: int, skipped: int, failed: int, lines: list<string>}
     */
    /**
     * @return array{imported: int, skipped: int, failed: int, lines: list<string>, imported_question_ids: list<int>}
     */
    public function importTemIdsToQuestions(array $temIds): array
    {
        $imported = 0;
        $skipped = 0;
        $failed = 0;
        $lines = [];
        $importedQuestionIds = [];

        foreach (array_unique(array_filter(array_map('intval', $temIds))) as $id) {
            if ($id <= 0) {
                continue;
            }

            $r = $this->importTemRowToQuestions($id);
            if ($r['ok']) {
                $imported++;
                if (! empty($r['question_id'])) {
                    $importedQuestionIds[] = (int) $r['question_id'];
                }

                continue;
            }

            $msg = $r['message'];
            if (
                str_contains($msg, '正式库已存在')
                || str_contains($msg, '题干或知识点为空')
            ) {
                $skipped++;
            } else {
                $failed++;
            }

            if (count($lines) < 30) {
                $lines[] = "#{$id}: {$msg}";
            }
        }

        return [
            'imported' => $imported,
            'skipped' => $skipped,
            'failed' => $failed,
            'lines' => $lines,
            'imported_question_ids' => $importedQuestionIds,
        ];
    }

    /**
     * 年级下拉项来自教材章节-知识点关联链路（非 questions_tem.grade）。
     *
     * @return list<int>
     */
    public function catalogGradeOptions(?int $semesterFilter = null): array
    {
        if (! Schema::hasTable('textbooks') || ! Schema::hasTable('textbook_catalog_nodes') || ! Schema::hasTable('textbook_chapter_knowledge_relation')) {
            return [];
        }

        $cacheKey = 'qtr:catalog-grades:' . ($semesterFilter ?? 'all');
        return Cache::remember($cacheKey, now()->addMinutes(10), function () use ($semesterFilter) {
            return DB::table('textbooks as t')
                ->join('textbook_catalog_nodes as n', 'n.textbook_id', '=', 't.id')
                ->join('textbook_chapter_knowledge_relation as r', 'r.catalog_chapter_id', '=', 'n.id')
                ->whereNotNull('t.grade')
                ->where('n.node_type', 'section')
                ->when($semesterFilter !== null, fn ($q) => $q->where('t.semester', $semesterFilter))
                ->when(Schema::hasColumn('r', 'is_deleted'), fn ($q) => $q->where(function ($x) {
                    $x->where('r.is_deleted', 0)->orWhereNull('r.is_deleted');
                }))
                ->distinct()
                ->orderBy('t.grade')
                ->pluck('t.grade')
                ->map(fn ($g) => (int) $g)
                ->filter(fn ($g) => $g > 0)
                ->values()
                ->all();
        });
    }

    /**
     * 指定年级在教材章节关联中出现过的 kp_code（含子知识点扩展）。
     *
     * @return list<string>
     */
    /** @return list<int> */
    public function catalogSemesterOptions(?int $gradeFilter = null): array
    {
        if (! Schema::hasTable('textbooks') || ! Schema::hasTable('textbook_catalog_nodes') || ! Schema::hasTable('textbook_chapter_knowledge_relation')) {
            return [];
        }

        $cacheKey = 'qtr:catalog-semesters:' . ($gradeFilter ?? 'all');
        return Cache::remember($cacheKey, now()->addMinutes(10), function () use ($gradeFilter) {
            return DB::table('textbooks as t')
                ->join('textbook_catalog_nodes as n', 'n.textbook_id', '=', 't.id')
                ->join('textbook_chapter_knowledge_relation as r', 'r.catalog_chapter_id', '=', 'n.id')
                ->whereNotNull('t.semester')
                ->where('n.node_type', 'section')
                ->when($gradeFilter !== null, fn ($q) => $q->where('t.grade', $gradeFilter))
                ->when(Schema::hasColumn('r', 'is_deleted'), fn ($q) => $q->where(function ($x) {
                    $x->where('r.is_deleted', 0)->orWhereNull('r.is_deleted');
                }))
                ->distinct()
                ->orderBy('t.semester')
                ->pluck('t.semester')
                ->map(fn ($s) => (int) $s)
                ->filter(fn ($s) => $s > 0)
                ->values()
                ->all();
        });
    }

    private function kpCodesFromCatalogFilter(?int $grade, ?int $semester): array
    {
        if (! Schema::hasTable('textbooks') || ! Schema::hasTable('textbook_catalog_nodes') || ! Schema::hasTable('textbook_chapter_knowledge_relation')) {
            return [];
        }
        $cacheKey = sprintf('qtr:catalog-kps:g%s:s%s', $grade ?? 'all', $semester ?? 'all');
        return Cache::remember($cacheKey, now()->addMinutes(10), function () use ($grade, $semester) {
            $q = DB::table('textbooks as t')
                ->join('textbook_catalog_nodes as n', 'n.textbook_id', '=', 't.id')
                ->join('textbook_chapter_knowledge_relation as r', 'r.catalog_chapter_id', '=', 'n.id')
                ->where('n.node_type', 'section')
                ->whereNotNull('r.kp_code')
                ->where('r.kp_code', '!=', '')
                ->when($grade !== null && $grade > 0, fn ($x) => $x->where('t.grade', $grade))
                ->when($semester !== null && $semester > 0, fn ($x) => $x->where('t.semester', $semester))
                ->when(Schema::hasColumn('r', 'is_deleted'), fn ($x) => $x->where(function ($w) {
                    $w->where('r.is_deleted', 0)->orWhereNull('r.is_deleted');
                }));

            return $q->distinct()->pluck('r.kp_code')
                ->map(fn ($v) => trim((string) $v))
                ->filter(fn ($v) => $v !== '')
                ->values()
                ->all();
        });
    }

    private function normalizeQuestionTypeForDb(mixed $raw): string
    {
        $t = strtolower(trim((string) $raw));
        if (str_contains($t, 'choice') || str_contains($t, '选择')) {
            return 'choice';
        }
        if (str_contains($t, 'fill') || str_contains($t, 'blank') || str_contains($t, '填空')) {
            return 'fill';
        }

        return 'answer';
    }
}