<?php
/**
 * 数据组装：卷王答卷 -> 学号 / 姓名 / 成绩 / 提交时间 / 考试时长
 */
declare(strict_types=1);

class DataService
{
    /** 容器型题目（分页、题组、说明）不参与取值 */
    private const CONTAINER_TYPES = [
        'pagination', 'page', 'group', 'survey', 'divider', 'tree',
    ];

    /** 题目标题可能是富文本，去掉标签便于展示与匹配 */
    private static function cleanTitle(string $t): string
    {
        $t = strip_tags($t);
        $t = html_entity_decode($t, ENT_QUOTES | ENT_HTML5, 'UTF-8');
        return trim(preg_replace('/\s+/u', ' ', $t) ?? '');
    }

    /**
     * 展平题目树
     * @return array<int,array{id:string,title:string,type:string,path:string}>
     */
    public static function flattenQuestions(mixed $node, string $path = ''): array
    {
        $out = [];
        if (!is_array($node)) {
            return $out;
        }

        // children 在根或子节点
        $self = $node;
        $children = (isset($node['children']) && is_array($node['children'])) ? $node['children'] : null;

        $id    = isset($self['id']) ? (string)$self['id'] : '';
        $title = isset($self['title']) ? self::cleanTitle((string)$self['title']) : '';
        $type  = isset($self['type']) ? (string)$self['type'] : '';

        $isRoot = ($path === '');
        if (!$isRoot && $id !== '' && !in_array(strtolower($type), self::CONTAINER_TYPES, true)) {
            $out[] = [
                'id'    => $id,
                'title' => $title !== '' ? $title : $id,
                'type'  => $type,
                'path'  => $path,
            ];
        }

        if ($children !== null) {
            foreach ($children as $idx => $child) {
                $subPath = $path === '' ? (string)($idx + 1) : $path . '-' . ($idx + 1);
                foreach (self::flattenQuestions($child, $subPath) as $q) {
                    $out[] = $q;
                }
            }
        }
        return $out;
    }

    /** 根据关键词猜测"姓名"题 */
    public static function guessNameQuestion(array $questions, string $keywords): array
    {
        $kws = array_values(array_filter(array_map('trim', preg_split('/[,，]/u', $keywords) ?: [])));
        $candidates = [];
        foreach ($questions as $q) {
            $t = (string)$q['title'];
            foreach ($kws as $kw) {
                if ($kw !== '' && mb_stripos($t, $kw) !== false) {
                    $candidates[] = $q;
                    break;
                }
            }
        }
        return $candidates;
    }

    /**
     * 组装结果
     */
    public static function build(array $cfg, string $projectId, string $nameQid = ''): array
    {
        $client = new SurveyKingClient($cfg);

        $students = Roster::all();
        // 姓名 -> 学号 / 班级（规范化）
        $studentInfo = [];
        foreach ($students as $s) {
            $k = ny_norm_name((string)$s['name']);
            if ($k !== '') {
                $studentInfo[$k] = [
                    'sid' => (string)$s['sid'],
                    'cls' => (string)($s['cls'] ?? ''),
                    'grp' => (string)($s['grp'] ?? ''),
                ];
            }
        }

        $project   = $client->getProject($projectId);
        $survey    = $project['survey'] ?? null;
        if (is_string($survey) && $survey !== '') {
            $decoded = json_decode($survey, true);
            if (is_array($decoded)) {
                $survey = $decoded;
            }
        }
        $questions = self::flattenQuestions($survey);

        $keywords      = (string)($cfg['name_keywords'] ?? '姓名');
        $nameCandidates = self::guessNameQuestion($questions, $keywords);
        $autoQid = $nameCandidates[0]['id'] ?? '';

        // 指定优先
        $pickedQid = $nameQid !== '' ? $nameQid : $autoQid;
        $candidateIds = [];
        if ($pickedQid !== '') {
            $candidateIds[] = $pickedQid;
        }
        foreach ($nameCandidates as $q) {
            if (!in_array($q['id'], $candidateIds, true)) {
                $candidateIds[] = $q['id'];
            }
        }

        $answers = $client->listAnswers($projectId);

        $includeTemp = !empty($cfg['include_temp_save']);
        $dedupe      = !empty($cfg['dedupe']);

        $rawRows   = [];   // 已匹配学号的原始 row（每条 = 一次提交）
        $unmatched = [];

        foreach ($answers as $a) {
            $tempSave = isset($a['tempSave']) ? (int)$a['tempSave'] : 1;
            if (!$includeTemp && $tempSave === 0) {
                continue; // 跳过暂存
            }
            $answerMap = isset($a['answer']) && is_array($a['answer']) ? $a['answer'] : [];

            // 姓名：优先关键词命中的题目 -> 全量扫描能匹配名单的答案 -> 登录用户名
            $name = '';
            foreach ($candidateIds as $qid) {
                if (isset($answerMap[$qid])) {
                    $v = ny_flatten_value($answerMap[$qid]);
                    if ($v !== '') {
                        $name = $v;
                        break;
                    }
                }
            }
            if ($name === '') {
                foreach ($questions as $q) {
                    $qid = (string)$q['id'];
                    if (!array_key_exists($qid, $answerMap)) {
                        continue;
                    }
                    $v = ny_flatten_value($answerMap[$qid]);
                    if ($v !== '' && isset($nameMap[ny_norm_name($v)])) {
                        $name = $v;
                        break;
                    }
                }
            }
            if ($name === '' && $questions === []) {
                foreach ($answerMap as $v) {
                    $s = ny_flatten_value($v);
                    if ($s !== '' && isset($nameMap[ny_norm_name($s)])) {
                        $name = $s;
                        break;
                    }
                }
            }
            if ($name === '') {
                $name = ny_str($a['createByName'] ?? '');
            }
            if ($name === '') {
                $name = '（未填写姓名）';
            }

            // 成绩
            $score = null;
            if (isset($a['examScore']) && is_numeric($a['examScore'])) {
                $score = (float)$a['examScore'];
            }

            // 提交时间
            $ts = ny_to_timestamp($a['createAt'] ?? null);
            if ($ts === 0) {
                $ts = ny_to_timestamp($a['updateAt'] ?? null);
            }

            // 考试时长
            $ai    = $a['metaInfo']['answerInfo'] ?? [];
            $start = is_array($ai) ? (int)($ai['startTime'] ?? 0) : 0;
            $end   = is_array($ai) ? (int)($ai['endTime'] ?? 0) : 0;
            $dur   = null;
            if ($start > 0 && $end > $start) {
                $dur = (int)round(($end - $start) / 1000);
            }

            $key = ny_norm_name($name);
            $info = $studentInfo[$key] ?? null;
            $sid = $info['sid'] ?? '';
            $cls = $info['cls'] ?? '';
            $grp = $info['grp'] ?? '';

            $row = [
                'sid'         => $sid,
                'name'        => $name,
                'cls'         => $cls,
                'grp'         => $grp,
                'score'       => $score,
                'submit_ts'   => $ts,
                'submit_at'   => $ts > 0 ? date('Y-m-d H:i:s', $ts) : '-',
                'duration'    => $dur,
                'duration_txt'=> ny_format_duration($dur),
                'answer_id'   => ny_str($a['id'] ?? ''),
                'matched'     => $sid !== '',
                'temp_save'   => $tempSave === 0,
            ];

            if ($sid === '') {
                $unmatched[] = $row;
                continue;
            }

            $rawRows[] = $row;
        }

        // 按学号聚合：dedupe=true 仅保留最新一份；false 则每名一行（代表=最高分）
        // 并附带 attempts(次数) / attempts_list(每次提交明细)，供看板展示"多次刷题"
        $agg           = self::aggregateByStudent($rawRows, $dedupe);
        $rows          = $agg['rows'];
        $attemptsTotal = $agg['attempts_total'];

        // 成绩明细默认按提交时间降序（最新在前），学号作为次级排序
        // 重要：排序键必须用「最近一次提交时间 last_submit_ts」，而不是代表行的 submit_ts。
        // 否则当 dedupe=false（代表成绩=最高分）时，学生重考且分数更低，代表行仍是最高分那次，
        // submit_ts 不变 → 该行会停在原来的位置只高亮，不会置顶，教师登记时看不到新提交。
        $sortTs = static fn(array $r): int => (int)($r['last_submit_ts'] ?? 0) ?: (int)($r['submit_ts'] ?? 0);
        usort($rows, static function (array $x, array $y) use ($sortTs): int {
            $d = $sortTs($y) <=> $sortTs($x);
            if ($d !== 0) {
                return $d;
            }
            $a = (string)$x['sid'];
            $b = (string)$y['sid'];
            if (is_numeric($a) && is_numeric($b)) {
                return $a <=> $b;
            }
            return strnatcmp($a, $b);
        });

        // 缺考名单
        $submittedSids = [];
        foreach ($rows as $r) {
            $submittedSids[(string)$r['sid']] = true;
        }
        $absent = [];
        foreach ($students as $s) {
            if (!isset($submittedSids[(string)$s['sid']])) {
                $absent[] = [
                    'sid'  => (string)$s['sid'],
                    'name' => (string)$s['name'],
                    'cls'  => (string)($s['cls'] ?? ''),
                    'grp'  => (string)($s['grp'] ?? ''),
                ];
            }
        }
        usort($absent, static function (array $x, array $y): int {
            $a = (string)$x['sid'];
            $b = (string)$y['sid'];
            if (is_numeric($a) && is_numeric($b)) {
                return $a <=> $b;
            }
            return strnatcmp($a, $b);
        });

        // 统计
        $scores = array_values(array_filter(array_map(static fn($r) => $r['score'], $rows), static fn($v) => $v !== null));
        $durs   = array_values(array_filter(array_map(static fn($r) => $r['duration'], $rows), static fn($v) => $v !== null));
        $stats = [
            'roster_total'  => count($students),
            'submitted'     => count($rows),
            'absent'        => count($absent),
            'unmatched'     => count($unmatched),
            'submit_rate'   => count($students) > 0 ? round(count($rows) / count($students) * 100, 1) : 0,
            'avg_score'     => $scores !== [] ? round(array_sum($scores) / count($scores), 2) : null,
            'max_score'     => $scores !== [] ? max($scores) : null,
            'min_score'     => $scores !== [] ? min($scores) : null,
            'avg_duration'  => $durs !== [] ? (int)round(array_sum($durs) / count($durs)) : null,
            'answer_total'  => count($answers),
            'attempts_total'=> $attemptsTotal,
        ];

        // 异常检测（防作弊线索，仅供教师人工判断，不做自动判定）
        $anomalies = self::detectAnomalies($rows, $stats);

        return [
            'project'    => [
                'id'   => ny_str($project['id'] ?? $projectId),
                'name' => ny_str($project['name'] ?? ''),
                'mode' => ny_str($project['mode'] ?? ''),
            ],
            'questions'  => $questions,
            'name_qid'   => $pickedQid,
            'name_candidates' => array_map(static fn($q) => ['id' => $q['id'], 'title' => $q['title']], $nameCandidates),
            'rows'       => $rows,
            'absent'     => $absent,
            'unmatched'  => $unmatched,
            'stats'      => $stats,
            'anomalies'  => $anomalies,
            'roster'     => $students,
            'roster_count' => count($students),
            'updated_at' => date('Y-m-d H:i:s'),
        ];
    }

    /**
     * 异常检测（防作弊线索）：基于聚合后的学生行，输出三类线索供教师人工复核。
     * 1) 短时长高得分：用时远低于平均（<平均 40% 且 <180s）却接近满分；
     * 2) 高频重考：考试次数明显多于多数人（>1 且 >= 平均次数+3）；
     * 3) 雷同成绩：多人分数完全相同（疑似互抄，需满足该分数下人数 >=3）。
     *
     * 返回值：['fast_high'=>[], 'frequent'=>[], 'identical'=>[['score'=>x,'count'=>n,'names'=>[...]]]]
     */
    public static function detectAnomalies(array $rows, array $stats): array
    {
        $avgDur   = (int)($stats['avg_duration'] ?? 0);
        $fullHint = max(1, (float)($stats['max_score'] ?? 100));

        $fastHigh  = [];
        $frequent  = [];
        $attemptsAll = [];

        foreach ($rows as $r) {
            $dur  = $r['duration'] ?? null;
            $score = $r['score'] ?? null;
            $attempts = (int)($r['attempts'] ?? 1);
            $attemptsAll[] = $attempts;

            // 短时长高得分：有时长记录、明显快于平均、分数 >= 满分 85%
            if ($dur !== null && $avgDur > 0 && $dur < max(60, (int)($avgDur * 0.4))
                && $score !== null && $score >= $fullHint * 0.85) {
                $fastHigh[] = [
                    'sid'      => (string)($r['sid'] ?? ''),
                    'name'     => (string)($r['name'] ?? ''),
                    'cls'      => (string)($r['cls'] ?? ''),
                    'score'    => $score,
                    'duration' => $dur,
                    'duration_txt' => (string)($r['duration_txt'] ?? ''),
                ];
            }

            // 高频重考：考试次数明显偏多
            if ($attempts > 1) {
                $frequent[] = [
                    'sid'      => (string)($r['sid'] ?? ''),
                    'name'     => (string)($r['name'] ?? ''),
                    'cls'      => (string)($r['cls'] ?? ''),
                    'attempts' => $attempts,
                    'best_score' => $r['best_score'] ?? null,
                ];
            }
        }

        // 高频重考阈值：平均次数 + 3（且至少 4 次才够"高频"）
        $avgAttempts = $attemptsAll !== [] ? array_sum($attemptsAll) / count($attemptsAll) : 0;
        $freqThreshold = max(4, (int)ceil($avgAttempts + 3));
        $frequent = array_values(array_filter($frequent, static fn($f) => $f['attempts'] >= $freqThreshold));
        usort($frequent, static fn($a, $b) => $b['attempts'] <=> $a['attempts']);

        // 雷同成绩：按分数分组，出现 >=3 人同分（且分数不为 null）视为疑似
        $scoreGroups = [];
        foreach ($rows as $r) {
            $score = $r['score'] ?? null;
            if ($score === null) {
                continue;
            }
            $key = (string)$score;
            $scoreGroups[$key][] = [
                'sid'  => (string)($r['sid'] ?? ''),
                'name' => (string)($r['name'] ?? ''),
                'cls'  => (string)($r['cls'] ?? ''),
            ];
        }
        $identical = [];
        foreach ($scoreGroups as $score => $members) {
            if (count($members) >= 3) {
                $identical[] = ['score' => (float)$score, 'count' => count($members), 'members' => $members];
            }
        }
        usort($identical, static fn($a, $b) => $b['count'] <=> $a['count']);

        return [
            'fast_high' => $fastHigh,
            'frequent'  => $frequent,
            'identical' => $identical,
        ];
    }

    /**
     * 把聚合后的学生行展开为"每次考试一行"，用于导出全部考试记录 / 明细内全部展示。
     * 每行携带该次考试的成绩、提交时间、时长，以及该学生的总考试次数 attempts 与第几次 idx。
     *
     * @param array $rows 聚合后的行（每行含 attempts_list）
     * @param bool  $desc 是否按提交时间降序（新 -> 旧）
     * @return array
     */
    public static function expandAttempts(array $rows, bool $desc = true): array
    {
        $out = [];
        foreach ($rows as $r) {
            $list = (array)($r['attempts_list'] ?? []);
            if ($list === []) {
                // 没有明细时退回代表行本身
                $one = $r;
                $one['idx']     = 1;
                $one['is_best'] = true;
                $out[] = $one;
                continue;
            }
            foreach ($list as $a) {
                $out[] = [
                    'sid'          => (string)($r['sid'] ?? ''),
                    'name'         => (string)($r['name'] ?? ''),
                    'cls'          => (string)($r['cls'] ?? ''),
                    'grp'          => (string)($r['grp'] ?? ''),
                    'score'        => $a['score'] ?? null,
                    'submit_ts'    => (int)($a['submit_ts'] ?? 0),
                    'submit_at'    => (string)($a['submit_at'] ?? '-'),
                    'duration'     => $a['duration'] ?? null,
                    'duration_txt' => (string)($a['duration_txt'] ?? '-'),
                    'answer_id'    => (string)($a['answer_id'] ?? ''),
                    'temp_save'    => !empty($a['temp_save']),
                    'attempts'     => (int)($r['attempts'] ?? 1),
                    'best_score'   => $r['best_score'] ?? null,
                    'idx'          => (int)($a['idx'] ?? 1),
                    'is_best'      => !empty($a['is_best']),
                    'last_submit_ts' => (int)($r['last_submit_ts'] ?? 0),
                    'last_submit_at' => (string)($r['last_submit_at'] ?? '-'),
                ];
            }
        }
        usort($out, static function (array $x, array $y) use ($desc): int {
            // 注意：x<=>y 是升序，降序要用 y<=>x
            $d = ((int)($y['submit_ts'] ?? 0)) <=> ((int)($x['submit_ts'] ?? 0));
            if (!$desc) {
                $d = -$d;
            }
            if ($d !== 0) {
                return $d;
            }
            // 同时间：学号升序
            $a = (string)$x['sid'];
            $b = (string)$y['sid'];
            if (is_numeric($a) && is_numeric($b)) {
                return $a <=> $b;
            }
            return strnatcmp($a, $b);
        });
        return $out;
    }

    /**
     * 导出/展示用的排序：
     *  - time（默认）：按提交时间降序，同时间按学号升序
     *  - sid：按学号升序，同一学号内多次考试再按提交时间降序
     *
     * @param array  $rows expandAttempts 展开后的行
     * @param string $mode 'time' | 'sid'
     */
    public static function sortForExport(array $rows, string $mode = 'time'): array
    {
        $out = $rows;
        if ($mode === 'sid') {
            usort($out, static function ($x, $y): int {
                $a = (string)($x['sid'] ?? '');
                $b = (string)($y['sid'] ?? '');
                if (is_numeric($a) && is_numeric($b)) {
                    $d = $a <=> $b;
                } else {
                    $d = strnatcmp($a, $b);
                }
                if ($d !== 0) {
                    return $d;
                }
                return ((int)($y['submit_ts'] ?? 0)) <=> ((int)($x['submit_ts'] ?? 0));
            });
            return $out;
        }
        usort($out, static function ($x, $y): int {
            $d = ((int)($y['submit_ts'] ?? 0)) <=> ((int)($x['submit_ts'] ?? 0));
            if ($d !== 0) {
                return $d;
            }
            $a = (string)($x['sid'] ?? '');
            $b = (string)($y['sid'] ?? '');
            if (is_numeric($a) && is_numeric($b)) {
                return $a <=> $b;
            }
            return strnatcmp($a, $b);
        });
        return $out;
    }

    /**
     * 按学号聚合一次提交的答卷记录，解决"同人多次刷卷"在看板中被当成多个人的问题。
     *
     * 关键语义（修复"考试次数=1 / 历史只有一条"的 bug）：
     *  - attempts(真实考试次数) 与 attempts_list(每次提交明细) **始终等于全部提交**，与 dedupe 无关；
     *    这样"考试次数列"与"点击查看历史"永远展示完整数据。
     *  - dedupe 仅决定主明细表"代表成绩"那一行取哪次考试的成绩：
     *      dedupe=true  → 取提交时间最新的一份（默认，看板更干净）
     *      dedupe=false  → 取最高分（同分时取最早达到该分数的那次）
     *
     * @param array $rows 已匹配学号的原始 row（每条 = 一次提交）
     * @return array{rows:array, attempts_total:int}
     */
    public static function aggregateByStudent(array $rows, bool $dedupe): array
    {
        $bySid = [];
        foreach ($rows as $r) {
            $bySid[(string)($r['sid'] ?? '')][] = $r;
        }
        $agg = [];
        $attemptsTotal = 0;
        foreach ($bySid as $sid => $list) {
            $attemptsTotal += count($list);
            // 升序：旧 -> 新
            $sorted = $list;
            usort($sorted, static function (array $x, array $y): int {
                return ($x['submit_ts'] ?? 0) <=> ($y['submit_ts'] ?? 0);
            });
            // 最佳成绩（用于标记每次考试是否为最高分）
            $maxScore = null;
            foreach ($sorted as $a) {
                if ($a['score'] !== null && ($maxScore === null || $a['score'] > $maxScore)) {
                    $maxScore = $a['score'];
                }
            }
            // 最近一次提交（升序排序后的最后一份）
            $lastSubmitTs = (int)($sorted[count($sorted) - 1]['submit_ts'] ?? 0);
            // 标记是否已给"最佳"打过标（同分时只标最早达到该分数的那次）
            $bestMarked = false;
            $idx = 0;
            $attemptsList = array_map(static function (array $a) use (&$idx, $maxScore, &$bestMarked): array {
                $idx++;
                $isBest = (!$bestMarked && $a['score'] !== null && $maxScore !== null && (float)$a['score'] === (float)$maxScore);
                if ($isBest) {
                    $bestMarked = true;
                }
                return [
                    'idx'          => $idx,
                    'is_best'      => $isBest,
                    'submit_at'    => $a['submit_at'],
                    'submit_ts'    => $a['submit_ts'],
                    'score'        => $a['score'],
                    'duration'     => $a['duration'],
                    'duration_txt' => $a['duration_txt'],
                    'answer_id'    => $a['answer_id'],
                    'temp_save'    => $a['temp_save'],
                ];
            }, $sorted);

            // 代表成绩选择（仅影响主明细表那一行的成绩，不影响考试次数 / 历史明细）
            if ($dedupe) {
                // 代表成绩 = 提交时间最新的一份（默认，看板更干净）
                $row = $sorted[count($sorted) - 1];
            } else {
                // 代表成绩 = 最高分（null 成绩不参与比较，全部为 null 时回退为最近一份）
                $best = null;
                $bestScore = null;
                foreach ($sorted as $a) {
                    if ($a['score'] !== null && ($bestScore === null || $a['score'] > $bestScore)) {
                        $bestScore = $a['score'];
                        $best = $a;
                    }
                }
                if ($best === null) {
                    $best = $sorted[count($sorted) - 1];
                }
                $row = $best;
            }

            // 关键修复：考试次数与历史明细始终使用真实全部提交，与去重无关
            // （之前 dedupe=true 时把 attempts 写成 1、attempts_list 只留最新一份，导致"考试次数"和
            //   历史弹窗都只显示 1 条，多次刷题的数据被吞掉）
            $row['attempts']      = count($list);
            $row['best_score']    = $maxScore;   // 真实最高分（跨全部尝试），用于历史弹窗标题展示
            $row['attempts_list'] = $attemptsList;
            // 最近一次提交时间（与"代表成绩"可能是不同的一次考试）：
            // 用于"按提交时间排序"和"新提交高亮"，保证重考（哪怕分数更低）也能被发现
            $row['last_submit_ts'] = $lastSubmitTs;
            $row['last_submit_at'] = $lastSubmitTs > 0 ? date('Y-m-d H:i:s', $lastSubmitTs) : '-';
            $agg[] = $row;
        }
        return ['rows' => $agg, 'attempts_total' => $attemptsTotal];
    }
}