fix:修复BUG/升级1.1.6版本

vendor/overtrue/pinyin/bin/build

@@ -0,0 +1,96 @@
+#!/usr/bin/env php
+<?php
+
+require __DIR__ . '/utils.php';
+
+$polyphones = explode(',', file_get_contents(__DIR__ . '/../sources/polyphones.txt'));
+$charsSouce = __DIR__ . '/../sources/chars.txt';
+$charsPathes = __DIR__.'/../sources/pathes/chars.txt';
+$wordsSouce = __DIR__ . '/../sources/words.txt';
+$wordsPathes = __DIR__ . '/../sources/pathes/words.txt';
+$surnamesSource = file(__DIR__.'/../sources/surnames.txt');
+
+
+if (!file_exists($charsSouce)) {
+    file_put_contents($charsSouce, file_get_contents('https://raw.githubusercontent.com/mozillazg/pinyin-data/master/pinyin.txt'));
+}
+
+if (!file_exists($wordsSouce)) {
+    file_put_contents($wordsSouce, file_get_contents('https://raw.githubusercontent.com/mozillazg/phrase-pinyin-data/master/large_pinyin.txt'));
+}
+
+
+// ------------------------------------------------
+$surnames = [];
+foreach ($surnamesSource as $line) {
+    [$surname, $pinyin] = explode(',', trim($line));
+
+    $surnames[trim($surname)] = join("\t", ["", ...preg_split('/\s+/', trim($pinyin)), ""]);
+}
+
+
+// ------------------------------------------------
+
+// 单字，带多音
+$charWithPolyphones = [];
+// 单字，不带多音
+$chars = [];
+
+foreach (parse_chars($charsSouce) as $char => $pinyin) {
+    $charWithPolyphones[$char] = $pinyin;
+    $chars[$char] = "\t{$pinyin[0]}\t";
+}
+
+// 补丁部分
+foreach (parse_chars($charsPathes, fn ($p) => "\t{$p[0]}\t") as $char => $pinyin) {
+    $chars[$char] = $pinyin;
+}
+
+// ------------------------------------------------
+
+$words = [];
+
+foreach (parse_words($wordsSouce) as $word => $pinyin) {
+    $wordChars = preg_split('//u', $word, -1, PREG_SPLIT_NO_EMPTY);
+
+    try {
+        $pinyinSegments = array_combine($wordChars, $pinyin);
+    } catch (Throwable $e) {
+        throw new Exception("行解析错误：$line");
+    }
+
+    // 多音字处理
+    $polyphoneChars = array_intersect_key($wordChars, $polyphones);
+
+    foreach ($polyphoneChars as $char) {
+        // 如果词里的任何一个多音字在词里的读音和常用读音不一致，则需要加入词典，否则抛弃该词
+        if (isset($charWithPolyphones[$char]) && $pinyinSegments[$char] != $charWithPolyphones[$char][0]) {
+            $words[$word] = join("\t", ["", ...$pinyin, ""]);
+            break;
+        }
+    }
+}
+
+foreach (parse_words($wordsPathes) as $word => $pinyin) {
+    $words[$word] = join("\t", ["", ...$pinyin, ""]);
+}
+
+// 清理
+exec('rm -rf ' . __DIR__ . '/../data/*');
+
+// 姓氏
+file_put_contents(__DIR__ . '/../data/surnames.php', "<?php\nreturn ".var_export($surnames, true).";\n");
+echo count($surnames)." surnames saved.\n";
+
+// 单字：带多音
+file_put_contents(__DIR__ . '/../data/chars.php', "<?php\nreturn ".var_export($charWithPolyphones, true).";\n");
+echo count($chars)." chars saved.\n";
+
+// 词：从长到短 + 单字
+$words = array_merge($words, $chars);
+uksort($words, fn ($a, $b) => strlen($b) <=> strlen($a));
+
+foreach (array_chunk($words, 8000, true) as $index => $group) {
+    file_put_contents(__DIR__ . "/../data/words-{$index}.php", "<?php\nreturn ".var_export($group, true).";\n");
+    echo count($group)." words saved in ".__DIR__ . "/../data/words-{$index}.php \n";
+}

vendor/overtrue/pinyin/bin/pinyin

@@ -0,0 +1,84 @@
+#!/usr/bin/env php
+<?php
+
+require __DIR__ . '/../vendor/autoload.php';
+
+use Overtrue\Pinyin\Pinyin;
+
+$input = $argv[1] ?? null;
+$methods = explode(',', 'name,phrase,permalink,polyphones,chars,nameAbbr,abbr,sentence');
+$method = 'sentence';
+$inputOptions = [];
+$help = <<<"HELP"
+Usage:
+    ./pinyin [chinese] [method] [options]
+Options:
+    -j, --json               输出 JSON 格式.
+    -c, --compact            不格式化输出 JSON.
+    -m, --method=[method]    转换方式，可选：name/phrase/permalink/polyphones/chars/nameAbbr/abbr/sentence.
+    --no-tone                不使用音调.
+    --tone-style=[style]     音调风格，可选值：default/none/number.
+    -h, --help               显示帮助.
+
+HELP;
+
+foreach ($argv as $i => $arg) {
+    if ($i === 0) {
+        continue;
+    }
+    if (in_array($arg, $methods)) {
+        $method = $arg;
+    } elseif (str_starts_with($arg, '-')) {
+        [$key, $value] = array_pad(array_map('trim', explode('=', $arg, 2)), 2, null);
+        $inputOptions[$key] = $value;
+    }
+}
+
+function has_option($option, $alias = null): bool
+{
+    global $inputOptions;
+
+    if ($alias) {
+        return array_key_exists($option, $inputOptions) || array_key_exists($alias, $inputOptions);
+    }
+
+    return array_key_exists($option, $inputOptions);
+}
+
+function get_option($option, $default = null, $alias = null): ?string
+{
+    global $inputOptions;
+
+    if ($alias) {
+        return $inputOptions[$option] ?? $inputOptions[$alias] ?? $default;
+    }
+
+    return $inputOptions[$option] ?? $default;
+}
+
+if (empty($input) || has_option('--help', '-h')) {
+    echo $help;
+    exit(0);
+}
+
+if (has_option('--method', '-m')) {
+    $method = get_option('--method');
+}
+
+$toneStyle = has_option('--no-tone') ? 'none' : get_option('--tone-style', 'default');
+
+$result = Pinyin::$method($input, $method === 'permalink' ? '-' : $toneStyle);
+
+$toJson = has_option('--json', '-j') || in_array($method, ['polyphones']);
+
+if ($toJson) {
+    $options = JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT;
+
+    if (has_option('--compact', '-c')) {
+        $options = 0;
+    }
+
+    $result = json_encode($result, $options);
+}
+
+echo $result, "\n";

vendor/overtrue/pinyin/bin/utils.php

@@ -0,0 +1,44 @@
+<?php
+
+/**
+ * @example
+ * <pre>
+ * // U+4E2D: zhōng,zhòng  # 中
+ * </pre>
+ * @throws Exception
+ */
+function parse_chars(string $path, callable $fn = null): Generator
+{
+    $fn ??= fn ($p) => $p;
+
+    foreach (file($path) as $line) {
+        preg_match('/^U\+(?<code>[0-9A-Z]+):\s+(?<pinyin>\S+)\s+#\s*(?<char>\S+)/', $line, $matched);
+
+        if ($matched && !empty($matched['pinyin'])) {
+            yield $matched['char'] => $fn(explode(',', $matched['pinyin']));
+        } elseif (!str_starts_with($line, '#')) {
+            throw new Exception("行解析错误：$line");
+        }
+    }
+}
+
+/**
+ * @example
+ * <pre>
+ * // 㞎㞎: bǎ ba # 注释
+ * </pre>
+ *
+ * @throws Exception
+ */
+function parse_words(string $path, callable $fn = null): Generator
+{
+    $fn ??= fn ($p) => $p;
+
+    foreach (file($path) as $line) {
+        preg_match('/^(?<word>[^#\s]+):\s+(?<pinyin>[\p{L} ]+)#?/u', $line, $matched);
+
+        if ($matched && !empty($matched['pinyin'])) {
+            yield $matched['word'] => $fn(explode(' ', trim($matched['pinyin'])));
+        }
+    }
+}