ny-exam-platform/lib/XlsxReader.php

<?php
/**
 * 极简 XLSX 读取器（无第三方依赖）
 * 优先使用 ZipArchive，缺失时回退到纯 PHP 的 ZIP 解析（支持 stored / deflate）。
 */
declare(strict_types=1);

class XlsxReader
{
    /**
     * @return array<int, array<int, string>> 二维数组（行 => 列）
     */
    public static function read(string $file): array
    {
        if (!is_file($file)) {
            throw new RuntimeException('文件不存在：' . $file);
        }
        $entries = self::readZip($file);

        $shared = [];
        if (isset($entries['xl/sharedStrings.xml'])) {
            $shared = self::parseSharedStrings($entries['xl/sharedStrings.xml']);
        }

        $sheet = self::firstSheetContent($entries);
        if ($sheet === null) {
            throw new RuntimeException('未找到工作表内容');
        }
        return self::parseSheet($sheet, $shared);
    }

    // ---------------- ZIP ----------------
    private static function readZip(string $file): array
    {
        if (class_exists('ZipArchive')) {
            $out = [];
            $za = new ZipArchive();
            if ($za->open($file) === true) {
                for ($i = 0; $i < $za->numFiles; $i++) {
                    $name = $za->getNameIndex($i);
                    if ($name === false) {
                        continue;
                    }
                    $content = $za->getFromIndex($i);
                    $out[$name] = $content === false ? '' : $content;
                }
                $za->close();
                return $out;
            }
        }
        return self::parseZipRaw($file);
    }

    private static function parseZipRaw(string $file): array
    {
        $data = (string)@file_get_contents($file);
        $len  = strlen($data);
        if ($len < 22) {
            throw new RuntimeException('不是有效的 xlsx 文件');
        }
        $eocd = -1;
        for ($i = $len - 22; $i >= 0; $i--) {
            if (substr($data, $i, 4) === "\x50\x4b\x05\x06") {
                $eocd = $i;
                break;
            }
        }
        if ($eocd < 0) {
            throw new RuntimeException('不是有效的 xlsx/zip 文件（未找到目录结束标记）');
        }
        $count  = self::u16($data, $eocd + 10);
        $offset = self::u32($data, $eocd + 16);

        $out = [];
        $p   = $offset;
        for ($i = 0; $i < $count; $i++) {
            if (substr($data, $p, 4) !== "\x50\x4b\x01\x02") {
                break;
            }
            $method     = self::u16($data, $p + 10);
            $compSize   = self::u32($data, $p + 20);
            $nameLen    = self::u16($data, $p + 28);
            $extraLen   = self::u16($data, $p + 30);
            $commentLen = self::u16($data, $p + 32);
            $localOff   = self::u32($data, $p + 42);
            $name       = substr($data, $p + 46, $nameLen);

            $lNameLen  = self::u16($data, $localOff + 26);
            $lExtraLen = self::u16($data, $localOff + 28);
            $start     = $localOff + 30 + $lNameLen + $lExtraLen;
            $raw       = substr($data, $start, $compSize);

            if ($method === 0) {
                $content = $raw;
            } else {
                $content = @gzinflate($raw);
                if ($content === false) {
                    $content = @gzinflate(substr($raw, 2));
                }
            }

            $out[str_replace('\\', '/', $name)] = $content === false ? '' : $content;
            $p += 46 + $nameLen + $extraLen + $commentLen;
        }
        return $out;
    }

    private static function u16(string $d, int $off): int
    {
        $s = substr($d, $off, 2);
        return $s === '' || strlen($s) < 2 ? 0 : unpack('v', $s)[1];
    }

    private static function u32(string $d, int $off): int
    {
        $s = substr($d, $off, 4);
        return $s === '' || strlen($s) < 4 ? 0 : unpack('V', $s)[1];
    }

    // ---------------- XML ----------------
    private static function firstSheetContent(array $entries): ?string
    {
        if (isset($entries['xl/worksheets/sheet1.xml'])) {
            return $entries['xl/worksheets/sheet1.xml'];
        }
        $found = [];
        foreach (array_keys($entries) as $k) {
            if (preg_match('#^xl/worksheets/sheet\d+\.xml$#', $k)) {
                $found[] = $k;
            }
        }
        if ($found === []) {
            return null;
        }
        sort($found, SORT_NATURAL);
        return $entries[$found[0]];
    }

    private static function parseSharedStrings(string $xml): array
    {
        $out = [];
        if (preg_match_all('#<si>(.*?)</si>#s', $xml, $m) > 0) {
            foreach ($m[1] as $si) {
                if (preg_match_all('#<t[^>]*>(.*?)</t>#s', $si, $tm) > 0) {
                    $out[] = implode('', array_map(
                        static fn($x) => html_entity_decode($x, ENT_QUOTES | ENT_HTML5, 'UTF-8'),
                        $tm[1]
                    ));
                } else {
                    $out[] = '';
                }
            }
        }
        return $out;
    }

    private static function parseSheet(string $xml, array $shared): array
    {
        $rows = [];
        if (preg_match_all('#<row\b[^>]*>(.*?)</row>#s', $xml, $rm) === 0) {
            return $rows;
        }
        foreach ($rm[1] as $rowXml) {
            $cells = [];
            if (preg_match_all('#<c(?:\s([^>]*?))?(?:/>|>(.*?)</c>)#s', $rowXml, $cm, PREG_SET_ORDER) > 0) {
                $autoIdx = 0;
                foreach ($cm as $c) {
                    $attrs = $c[1] ?? '';
                    $inner = $c[2] ?? '';

                    $idx = $autoIdx;
                    if (preg_match('#\br="([A-Z]+)\d*"#', $attrs, $r)) {
                        $idx = self::colToIndex($r[1]);
                    }
                    $autoIdx = $idx + 1;

                    $type = preg_match('#\bt="([^"]+)"#', $attrs, $tm) ? $tm[1] : '';

                    if ($type === 'inlineStr') {
                        $val = '';
                        if (preg_match_all('#<t[^>]*>(.*?)</t>#s', $inner, $im) > 0) {
                            $val = implode('', array_map(
                                static fn($x) => html_entity_decode($x, ENT_QUOTES | ENT_HTML5, 'UTF-8'),
                                $im[1]
                            ));
                        }
                    } else {
                        $val = preg_match('#<v>(.*?)</v>#s', $inner, $vm)
                            ? html_entity_decode($vm[1], ENT_QUOTES | ENT_HTML5, 'UTF-8')
                            : '';
                        if ($type === 's') {
                            $i   = (int)$val;
                            $val = $shared[$i] ?? '';
                        } elseif ($type === 'b') {
                            $val = ((int)$val) === 1 ? 'TRUE' : 'FALSE';
                        }
                    }
                    $cells[$idx] = $val;
                }
            }
            if ($cells === []) {
                continue;
            }
            $max = max(array_keys($cells));
            $line = [];
            for ($i = 0; $i <= $max; $i++) {
                $line[] = $cells[$i] ?? '';
            }
            $rows[] = $line;
        }
        return $rows;
    }

    private static function colToIndex(string $letters): int
    {
        $n = 0;
        $len = strlen($letters);
        for ($i = 0; $i < $len; $i++) {
            $n = $n * 26 + (ord($letters[$i]) - 64);
        }
        return $n - 1;
    }
}