<?php
/**
 * 汉字 → 拼音首字母（用户 2026-10-03 第 4 条）。
 *
 * 用途：名册搜索支持「姓名首字母」——
 *   输入 y / yy / yyq 都能搜到「杨永全」。
 *
 * 两条路（**必须都有**，因为本地与线上环境不同）：
 *
 *   ① `Intl` 扩展的 `Collator('zh-Hans-CN-u-co-pinyin')`：拿一张
 *      「a..z 各挑一个锚字」的表做二分，就能判断任意汉字落在哪个字母段。
 *      线上 PHP 7.4/8.1 都有 intl，走这条最准、零数据。
 *   ② 没有 intl 时的兜底：**常用姓氏/人名用字表**。
 *      本地开发环境没有 intl（实测 `intl: 无`），没有兜底就等于
 *      「这个功能本地永远测不了」—— 那是最糟的情况，只能上线才发现问题。
 *
 * ⚠️ 兜底表覆盖不到的生僻字会**原样保留首字符参与匹配**（不产字母），
 *    也就是说：搜不到时是"少匹配"，绝不会错匹配。
 */

declare(strict_types=1);

namespace App;

final class Pinyin
{
    /** 拼音排序用的锚字（a..z 各一个；没有 i/u/v 开头的常用字，所以是 23 个） */
    private const ANCHORS = '阿八擦搭娥发嘎哈击喀垃妈拿哦啪期然撒塌挖昔压匝';
    private const LETTERS = 'abcdefghjklmnopqrstwxyz';

    /**
     * 常用姓氏 + 常见人名字的**首字母表**（兜底用）。
     * 只列"作为名字第一个字"出现频率高的；查不到就交给下一档处理。
     * 这一份不可能穷尽，所以它只是兜底 —— 线上有 intl 时走的是锚字二分。
     */
    private const TABLE = [
        // 姓氏（覆盖绝大多数中国姓氏）
        '赵a' => 1, '钱q' => 1, '孙s' => 1, '李l' => 1, '周z' => 1, '吴w' => 1, '郑z' => 1,
        '王w' => 1, '冯f' => 1, '陈c' => 1, '褚c' => 1, '卫w' => 1, '蒋j' => 1, '沈s' => 1,
        '韩h' => 1, '杨y' => 1, '朱z' => 1, '秦q' => 1, '尤y' => 1, '许x' => 1, '何h' => 1,
        '吕l' => 1, '施s' => 1, '张z' => 1, '孔k' => 1, '曹c' => 1, '严y' => 1, '华h' => 1,
        '金j' => 1, '魏w' => 1, '陶t' => 1, '姜j' => 1, '戚q' => 1, '谢x' => 1, '邹z' => 1,
        '喻y' => 1, '柏b' => 1, '水s' => 1, '窦d' => 1, '章z' => 1, '云y' => 1, '苏s' => 1,
        '潘p' => 1, '葛g' => 1, '奚x' => 1, '范f' => 1, '彭p' => 1, '郎l' => 1, '鲁l' => 1,
        '韦w' => 1, '昌c' => 1, '马m' => 1, '苗m' => 1, '凤f' => 1, '花h' => 1, '方f' => 1,
        '俞y' => 1, '任r' => 1, '袁y' => 1, '柳l' => 1, '酆f' => 1, '鲍b' => 1, '史s' => 1,
        '唐t' => 1, '费f' => 1, '廉l' => 1, '岑c' => 1, '薛x' => 1, '雷l' => 1, '贺h' => 1,
        '倪n' => 1, '汤t' => 1, '滕t' => 1, '殷y' => 1, '罗l' => 1, '毕b' => 1, '郝h' => 1,
        '邬w' => 1, '安a' => 1, '常c' => 1, '乐y' => 1, '于y' => 1, '时s' => 1, '傅f' => 1,
        '皮p' => 1, '卞b' => 1, '齐q' => 1, '康k' => 1, '伍w' => 1, '余y' => 1, '元y' => 1,
        '卜b' => 1, '顾g' => 1, '孟m' => 1, '平p' => 1, '黄h' => 1, '和h' => 1, '穆m' => 1,
        '萧x' => 1, '尹y' => 1, '姚y' => 1, '邵s' => 1, '湛z' => 1, '汪w' => 1, '祁q' => 1,
        '毛m' => 1, '禹y' => 1, '狄d' => 1, '米m' => 1, '贝b' => 1, '明m' => 1, '臧z' => 1,
        '计j' => 1, '伏f' => 1, '成c' => 1, '戴d' => 1, '谈t' => 1, '宋s' => 1, '茅m' => 1,
        '庞p' => 1, '熊x' => 1, '纪j' => 1, '舒s' => 1, '屈q' => 1, '项x' => 1, '祝z' => 1,
        '董d' => 1, '梁l' => 1, '杜d' => 1, '阮r' => 1, '蓝l' => 1, '闵m' => 1, '席x' => 1,
        '季j' => 1, '麻m' => 1, '强q' => 1, '贾j' => 1, '路l' => 1, '娄l' => 1, '危w' => 1,
        '江j' => 1, '童t' => 1, '颜y' => 1, '郭g' => 1, '梅m' => 1, '盛s' => 1, '林l' => 1,
        '刁d' => 1, '钟z' => 1, '徐x' => 1, '邱q' => 1, '骆l' => 1, '高g' => 1, '夏x' => 1,
        '蔡c' => 1, '田t' => 1, '樊f' => 1, '胡h' => 1, '凌l' => 1, '霍h' => 1, '虞y' => 1,
        '万w' => 1, '支z' => 1, '柯k' => 1, '昝z' => 1, '管g' => 1, '卢l' => 1, '莫m' => 1,
        '房f' => 1, '裘q' => 1, '缪m' => 1, '干g' => 1, '解x' => 1, '应y' => 1, '宗z' => 1,
        '丁d' => 1, '宣x' => 1, '邓d' => 1, '单s' => 1, '杭h' => 1, '洪h' => 1, '包b' => 1,
        '诸z' => 1, '左z' => 1, '石s' => 1, '崔c' => 1, '吉j' => 1, '钮n' => 1, '龚g' => 1,
        '程c' => 1, '嵇j' => 1, '邢x' => 1, '滑h' => 1, '裴p' => 1, '陆l' => 1, '荣r' => 1,
        '翁w' => 1, '荀x' => 1, '羊y' => 1, '於y' => 1, '惠h' => 1, '甄z' => 1, '曲q' => 1,
        '封f' => 1, '芮r' => 1, '羿y' => 1, '储c' => 1, '靳j' => 1, '汲j' => 1, '邴b' => 1,
        '糜m' => 1, '松s' => 1, '井j' => 1, '段d' => 1, '富f' => 1, '巫w' => 1, '乌w' => 1,
        '焦j' => 1, '巴b' => 1, '弓g' => 1, '牧m' => 1, '隗k' => 1, '山s' => 1, '谷g' => 1,
        '车c' => 1, '侯h' => 1, '宓m' => 1, '蓬p' => 1, '全q' => 1, '郗x' => 1, '班b' => 1,
        '仰y' => 1, '秋q' => 1, '仲z' => 1, '伊y' => 1, '宫g' => 1, '宁n' => 1, '仇q' => 1,
        '栾l' => 1, '暴b' => 1, '甘g' => 1, '厉l' => 1, '戎r' => 1, '祖z' => 1, '武w' => 1,
        '符f' => 1, '刘l' => 1, '景j' => 1, '詹z' => 1, '束s' => 1, '龙l' => 1, '叶y' => 1,
        '幸x' => 1, '司s' => 1, '韶s' => 1, '郜g' => 1, '黎l' => 1, '蓟j' => 1, '薄b' => 1,
        '印y' => 1, '宿s' => 1, '白b' => 1, '怀h' => 1, '蒲p' => 1, '邰t' => 1, '从c' => 1,
        '鄂e' => 1, '索s' => 1, '咸x' => 1, '籍j' => 1, '赖l' => 1, '卓z' => 1, '蔺l' => 1,
        '屠t' => 1, '蒙m' => 1, '池c' => 1, '乔q' => 1, '阴y' => 1, '郁y' => 1, '胥x' => 1,
        '能n' => 1, '苍c' => 1, '双s' => 1, '闻w' => 1, '莘s' => 1, '党d' => 1, '翟z' => 1,
        '谭t' => 1, '贡g' => 1, '劳l' => 1, '逄p' => 1, '姬j' => 1, '申s' => 1, '扶f' => 1,
        '堵d' => 1, '冉r' => 1, '宰z' => 1, '郦l' => 1, '雍y' => 1, '郤x' => 1, '璩q' => 1,
        '桑s' => 1, '桂g' => 1, '濮p' => 1, '牛n' => 1, '寿s' => 1, '通t' => 1, '边b' => 1,
        '扈h' => 1, '燕y' => 1, '冀j' => 1, '郏j' => 1, '浦p' => 1, '尚s' => 1, '农n' => 1,
        '温w' => 1, '别b' => 1, '庄z' => 1, '晏y' => 1, '柴c' => 1, '瞿q' => 1, '阎y' => 1,
        '充c' => 1, '慕m' => 1, '连l' => 1, '茹r' => 1, '习x' => 1, '宦h' => 1, '艾a' => 1,
        '鱼y' => 1, '容r' => 1, '向x' => 1, '古g' => 1, '易y' => 1, '慎s' => 1, '戈g' => 1,
        '廖l' => 1, '庾y' => 1, '终z' => 1, '暨j' => 1, '居j' => 1, '衡h' => 1, '步b' => 1,
        '都d' => 1, '耿g' => 1, '满m' => 1, '弘h' => 1, '匡k' => 1, '国g' => 1, '文w' => 1,
        '寇k' => 1, '广g' => 1, '禄l' => 1, '阙q' => 1, '东d' => 1, '欧o' => 1, '殳s' => 1,
        '沃w' => 1, '利l' => 1, '蔚w' => 1, '越y' => 1, '夔k' => 1, '隆l' => 1, '师s' => 1,
        '巩g' => 1, '厍s' => 1, '聂n' => 1, '晁c' => 1, '勾g' => 1, '敖a' => 1, '融r' => 1,
        '冷l' => 1, '訾z' => 1, '辛x' => 1, '阚k' => 1, '那n' => 1, '简j' => 1, '饶r' => 1,
        '空k' => 1, '曾z' => 1, '毋w' => 1, '沙s' => 1, '乜n' => 1, '养y' => 1, '鞠j' => 1,
        '须x' => 1, '丰f' => 1, '巢c' => 1, '关g' => 1, '蒯k' => 1, '相x' => 1, '查z' => 1,
        '后h' => 1, '荆j' => 1, '红h' => 1, '游y' => 1, '竺z' => 1, '权q' => 1, '逯l' => 1,
        '盖g' => 1, '益y' => 1, '桓h' => 1, '公g' => 1,
        // 名字里高频的字
        '永y' => 1, '全q' => 1, '博b' => 1, '为w' => 1, '明m' => 1, '华h' => 1, '建j' => 1,
        '国g' => 1, '文w' => 1, '军j' => 1, '平p' => 1, '志z' => 1, '强q' => 1, '磊l' => 1,
        '洋y' => 1, '勇y' => 1, '艳y' => 1, '杰j' => 1, '娟j' => 1, '涛t' => 1, '超c' => 1,
        '敏m' => 1, '静j' => 1, '丽l' => 1, '秀x' => 1, '英y' => 1, '娜n' => 1, '丹d' => 1,
        '玲l' => 1, '芳f' => 1, '燕y' => 1, '红h' => 1, '霞x' => 1, '梅m' => 1, '兰l' => 1,
        '琴q' => 1, '云y' => 1, '飞f' => 1, '鹏p' => 1, '宇y' => 1, '浩h' => 1, '鑫x' => 1,
        '亮l' => 1, '刚g' => 1, '辉h' => 1, '峰f' => 1, '龙l' => 1, '波b' => 1, '斌b' => 1,
        '春c' => 1, '雨y' => 1, '晨c' => 1, '阳y' => 1, '雪x' => 1, '悦y' => 1, '欣x' => 1,
        '怡y' => 1, '佳j' => 1, '思s' => 1, '雨y' => 1, '子z' => 1, '小x' => 1, '大d' => 1,
        '老l' => 1, '爸b' => 1, '妈m' => 1, '哥g' => 1, '姐j' => 1, '弟d' => 1, '妹m' => 1,
        '爷y' => 1, '奶n' => 1, '叔s' => 1, '姨y' => 1, '舅j' => 1, '嫂s' => 1, '儿e' => 1,
        '女n' => 1, '宝b' => 1, '乐l' => 1, '安a' => 1, '康k' => 1, '福f' => 1, '寿s' => 1,
        '喜x' => 1, '顺s' => 1, '和h' => 1, '美m' => 1, '好h' => 1, '新x' => 1, '中z' => 1,
        '正z' => 1, '长c' => 1, '海h' => 1, '江j' => 1, '河h' => 1, '山s' => 1, '林l' => 1,
        '木m' => 1, '金j' => 1, '玉y' => 1, '石s' => 1, '田t' => 1, '力l' => 1, '学x' => 1,
        '生s' => 1, '成c' => 1, '立l' => 1, '德d' => 1, '仁r' => 1, '义y' => 1, '礼l' => 1,
        '智z' => 1, '信x' => 1, '忠z' => 1, '孝x' => 1, '友y' => 1, '爱a' => 1, '慈c' => 1,
        // 家谱称呼里高频的那批字（父/母/伯/叔/姑/舅/姨/岳…）——
        // 名册搜索经常会拿这些字当关键词，兜底表里缺了就会出现
        // "本地搜 s 搜不到『张三』"这种与环境相关的怪现象。
        '父f' => 1, '母m' => 1, '祖z' => 1, '孙s' => 1, '外w' => 1, '表b' => 1, '堂t' => 1,
        '侄z' => 1, '甥s' => 1, '婿x' => 1, '媳x' => 1, '伯b' => 1, '姑g' => 1, '岳y' => 1,
        '公g' => 1, '婆p' => 1, '妻q' => 1, '夫f' => 1, '丈z' => 1, '娘n' => 1, '爹d' => 1,
        '三s' => 1, '四s' => 1, '五w' => 1, '六l' => 1, '七q' => 1, '八b' => 1, '九j' => 1,
        '十s' => 1, '百b' => 1, '千q' => 1, '万w' => 1, '一y' => 1, '二e' => 1, '零l' => 1,
        '不b' => 1, '的d' => 1, '了l' => 1, '在z' => 1, '有y' => 1, '是s' => 1, '我w' => 1,
        '人r' => 1, '家j' => 1, '门m' => 1, '户h' => 1, '先x' => 1, '后h' => 1, '晚w' => 1,
        '长c' => 1, '幼y' => 1, '大d' => 1, '小x' => 1, '多d' => 1, '少s' => 1, '高g' => 1,
        '矮a' => 1, '胖p' => 1, '瘦s' => 1, '黑h' => 1, '白b' => 1, '花h' => 1, '草c' => 1,
        '树s' => 1, '叶y' => 1, '果g' => 1, '苗m' => 1, '阳y' => 1, '光g' => 1, '月y' => 1,
        '星x' => 1, '风f' => 1, '雪x' => 1, '雨y' => 1, '雷l' => 1, '电d' => 1, '冰b' => 1,
        '州z' => 1, '城c' => 1, '村c' => 1, '镇z' => 1, '县x' => 1, '市s' => 1, '省s' => 1,
        '华h' => 1, '青q' => 1, '岛d' => 1, '京j' => 1, '津j' => 1, '冀j' => 1, '鲁l' => 1,
        '思s' => 1, '念n' => 1, '想x' => 1, '记j' => 1, '忘w' => 1, '知z' => 1, '道d' => 1,
        '说s' => 1, '话h' => 1, '听t' => 1, '看k' => 1, '走z' => 1, '来l' => 1, '去q' => 1,
        '回h' => 1, '到d' => 1, '给g' => 1, '拿n' => 1, '做z' => 1, '用y' => 1, '买m' => 1,
        '卖m' => 1, '吃c' => 1, '喝h' => 1, '睡s' => 1, '起q' => 1, '坐z' => 1, '站z' => 1,
        '跑p' => 1, '跳t' => 1, '唱c' => 1, '写x' => 1, '读d' => 1, '画h' => 1, '算s' => 1,
        '作z' => 1, '课k' => 1, '班b' => 1, '校x' => 1, '师s' => 1, '医y' => 1, '院y' => 1,
        '药y' => 1, '病b' => 1, '健j' => 1, '体t' => 1, '身s' => 1, '心x' => 1, '手s' => 1,
        '脚j' => 1, '眼y' => 1, '耳e' => 1, '鼻b' => 1, '口k' => 1, '头t' => 1, '发f' => 1,
    ];

    /** 首字母查表：'杨' => 'y'。首次调用时从 TABLE 摊平，之后走内存缓存。 */
    private static ?array $map = null;

    public static function initialMap(): array
    {
        if (self::$map === null) {
            self::$map = [];
            foreach (self::TABLE as $kv => $_) {
                $k = (string)$kv;
                if (mb_strlen($k, 'UTF-8') >= 1) {
                    $ch = mb_substr($k, 0, 1, 'UTF-8');
                    $ch = mb_substr($k, 0, mb_strlen($k, 'UTF-8') - 1, 'UTF-8');
                    $letter = mb_substr((string)$kv, mb_strlen((string)$kv, 'UTF-8') - 1, 1, 'UTF-8');
                    if ($ch !== '' && $letter !== '') {
                        self::$map[$ch] = strtolower($letter);
                    }
                }
            }
        }
        return self::$map;
    }

    /**
     * 一个字的拼音首字母（小写）。认不出时返回 ''。
     * 有 intl 走锚字二分；没有就走内置表。
     */
    public static function initial(string $ch): string
    {
        if ($ch === '') {
            return '';
        }
        // 已经是 ASCII 字母 → 直接用
        if (preg_match('/^[a-zA-Z]$/', $ch)) {
            return strtolower($ch);
        }
        if (preg_match('/^[0-9]$/', $ch)) {
            return $ch;
        }
        // ① intl 拼音排序（线上有）
        $viaIntl = self::initialViaIntl($ch);
        if ($viaIntl !== null) {
            return $viaIntl;
        }
        // ② 内置表兜底（本地无 intl 时）
        return self::initialMap()[$ch] ?? '';
    }

    /** 用 Intl 的拼音 collation + 锚字二分定位；没有 intl 返回 null */
    private static function initialViaIntl(string $ch): ?string
    {
        static $coll = false;
        if ($coll === false) {
            $coll = null;
            if (class_exists('\\Collator')) {
                $c = new \Collator('zh-Hans-CN-u-co-pinyin');
                // 排序没真的生效时（compare('阿','匝') 不 < 0）就别用，免得给出错结果
                if ($c->compare('阿', '匝') < 0) {
                    $coll = $c;
                }
            }
        }
        if ($coll === null) {
            return null;
        }
        $anchors = self::ANCHORS;
        $len = mb_strlen($anchors, 'UTF-8');
        $lo = 0;
        $hi = $len - 1;
        $best = -1;
        while ($lo <= $hi) {
            $mid = intdiv($lo + $hi, 2);
            $a = mb_substr($anchors, $mid, 1, 'UTF-8');
            if ($coll->compare($ch, $a) >= 0) {
                $best = $mid;
                $lo = $mid + 1;
            } else {
                $hi = $mid - 1;
            }
        }
        if ($best < 0) {
            return null;
        }
        return mb_substr(self::LETTERS, $best, 1, 'UTF-8');
    }

    /**
     * 一串字 → 首字母串（小写）。认不出的字**跳过**（不产字母，也不报错）。
     * 例：'杨永全' → 'yyq'。
     */
    public static function initials(string $s): string
    {
        $out = '';
        $n = mb_strlen($s, 'UTF-8');
        for ($i = 0; $i < $n; $i++) {
            $out .= self::initial(mb_substr($s, $i, 1, 'UTF-8'));
        }
        return $out;
    }

    /**
     * q 是否是 s 的**子序列**（用于"输入 y 也能搜到 yyq"这类前缀式检索）。
     *
     * 为什么用子序列而不是前缀：
     *   用户可能只记得中间某个字（比如"永全" → yq），或干脆跳着打；
     *   名册这种小列表宁可宽松一点，也别让人以为搜索坏了。
     */
    public static function isSubsequence(string $q, string $s): bool
    {
        $q = strtolower(trim($q));
        $s = strtolower($s);
        if ($q === '') {
            return true;
        }
        $i = 0;
        $qn = strlen($q);
        for ($j = 0, $sn = strlen($s); $j < $sn && $i < $qn; $j++) {
            if ($s[$j] === $q[$i]) {
                $i++;
            }
        }
        return $i === $qn;
    }

    /**
     * 这个人是否匹配搜索词。
     *
     * 覆盖：姓名 / 称呼 / 邮箱 / 手机号 的**模糊**包含，以及姓名首字母的**子序列**匹配。
     *
     * @param array  $row  名册行（含 name / email / phone_plain，可选 my_nickname）
     * @param string $q    搜索词
     */
    public static function matches(array $row, string $q): bool
    {
        $q = trim($q);
        if ($q === '') {
            return true;
        }
        $needle = mb_strtolower($q, 'UTF-8');
        foreach (['name', 'my_nickname', 'nickname', 'email', 'phone_plain'] as $k) {
            $v = mb_strtolower(trim((string)($row[$k] ?? '')), 'UTF-8');
            if ($v !== '' && mb_strpos($v, $needle) !== false) {
                return true;
            }
        }
        // 首字母：只对**字母/数字**查询词做子序列匹配，
        // 中文查询词走上面的"包含"（否则"永全"会被当成首字母串去比对，永远匹配不上）
        if (preg_match('/^[a-z0-9]+$/', $needle)) {
            $ini = self::initials((string)($row['name'] ?? ''));
            if ($ini !== '' && self::isSubsequence($needle, $ini)) {
                return true;
            }
            // 称呼也参与首字母（比如称呼是"小宝" → xb）
            $nickIni = self::initials((string)($row['my_nickname'] ?? ''));
            if ($nickIni !== '' && self::isSubsequence($needle, $nickIni)) {
                return true;
            }
        }
        return false;
    }
}