feat: full sentence

.gitattributes

@@ -3,6 +3,7 @@
 /tests export-ignore
 /vendor export-ignore
 /sources export-ignore
+/benchmark export-ignore
 /bin/build export-ignore
 /bin/utils.php export-ignore
 .gitattributes export-ignore

README.md

@@ -72,8 +72,8 @@ echo Pinyin::sentence('带着希望去旅行，比到达终点更美好');
 echo Pinyin::sentence('带着希望去旅行，比到达终点更美好', 'none');
 // dai zhe xi wang qu lv xing ， bi dao da zhong dian geng mei hao
 
-// 保留所有其他字符
-echo Pinyin::sentenceFull('ル是片假名，π是希腊字母', 'none');
+// 保留所有非汉字字符
+echo Pinyin::fullSentence('ル是片假名，π是希腊字母', 'none');
 // ル shi pian jia ming ，π shi xi la zi mu
 ```
 

tests/benchmark.php → benchmark/run.php

@@ -1,16 +1,18 @@
 <?php
 
-require __DIR__.'/../vendor/autoload.php';
+require __DIR__ . '/../vendor/autoload.php';
 
 use Overtrue\Pinyin\Pinyin;
 
 use function Termwind\{render};
 
 $totalStart = microtime(true);
-$text = file_get_contents(__DIR__.'/input.txt');
+$text = file_get_contents(__DIR__ . '/input.txt');
 
 $html = [];
-foreach (['name', 'phrase', 'permalink', 'polyphones', 'chars', 'nameAbbr', 'abbr', 'sentence'] as $method) {
+$methods = ['sentence','fullSentence','name','passportName','phrase','permalink','polyphones','chars','abbr','nameAbbr'];
+
+foreach ($methods as $method) {
     $start = microtime(true);
     $result = call_user_func(Pinyin::class.'::'.$method, $text);
     $usage = round(microtime(true) - $start, 5) * 1000;

bin/pinyin

@@ -2,38 +2,27 @@
 <?php
 
 require __DIR__ . '/../vendor/autoload.php';
+require __DIR__ . '/utils.php';
 
 use Overtrue\Pinyin\Pinyin;
 
-$input = $argv[1] ?? null;
-$methods = explode(',', 'name,phrase,permalink,polyphones,chars,nameAbbr,abbr,sentence');
+$methods = ['sentence','fullSentence','name','passportName','phrase','permalink','polyphones','chars','abbr','nameAbbr'];
 $method = 'sentence';
 $inputOptions = [];
+$methodAsString = implode('/', $methods);
+
 $help = <<<"HELP"
 Usage:
     ./pinyin [chinese] [method] [options]
 Options:
     -j, --json               输出 JSON 格式.
     -c, --compact            不格式化输出 JSON.
-    -m, --method=[method]    转换方式，可选：sentence/sentenceFull/permalink/abbr/nameAbbr/name/passportName/phrase/polyphones/chars.
+    -m, --method=[method]    转换方式，可选：{$methodAsString}.
     --no-tone                不使用音调.
     --tone-style=[style]     音调风格，可选值：symbol/none/number, default: none.
     -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 $inputOptions;
@@ -56,25 +45,30 @@ function get_option($option, $default = null, $alias = null): ?string
     return $inputOptions[$option] ?? $default;
 }
 
-if (empty($input) || has_option('--help', '-h')) {
+$inputOptions = parse_options($argv);
+$input = $inputOptions[0] ?? null;
+
+if (empty($input) || has_option('help', 'h')) {
     echo $help;
     exit(0);
 }
 
-if (has_option('--method', '-m')) {
-    $method = get_option('--method');
-}
+$method = get_option('method', $method, 'm');
+$toneStyle = has_option('no-tone') ? 'none' : get_option('tone-style', 'none');
 
-$toneStyle = has_option('--no-tone') ? 'none' : get_option('--tone-style', 'none');
+if (! in_array($method, $methods)) {
+    echo "Method '{$method}' is not supported.\n";
+    exit(1);
+}
 
 $result = Pinyin::$method($input, $method === 'permalink' ? '-' : $toneStyle);
 
-$toJson = has_option('--json', '-j') || in_array($method, ['polyphones']);
+$toJson = has_option('json', 'j') || in_array($method, ['polyphones']);
 
 if ($toJson) {
     $options = JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT;
 
-    if (has_option('--compact', '-c')) {
+    if (has_option('compact', 'c')) {
         $options = 0;
     }
 

bin/utils.php

@@ -43,3 +43,53 @@ function parse_words(string $path, callable $fn = null): Generator
         }
     }
 }
+
+function parse_options(array $argv): array
+{
+    $inputOptions = [];
+    $currentOption = null;
+
+    for ($i = 1; $i < count($argv); $i++) {
+        $arg = $argv[$i];
+
+        if (str_starts_with($arg, '--')) {
+            // 长选项
+            $option = substr($arg, 2);
+            if (str_contains($option, '=')) {
+                [$option, $value] = explode('=', $option, 2);
+                $inputOptions[$option] = $value;
+            } else {
+                $inputOptions[$option] = true;
+                $currentOption = $option;
+            }
+        } elseif (str_starts_with($arg, '-')) {
+            // 短选项
+            $option = substr($arg, 1);
+            if (strlen($option) > 1 && $option[1] !== '=') {
+                // 多个短选项，如 -abc
+                for ($j = 0; $j < strlen($option); $j++) {
+                    $inputOptions[$option[$j]] = true;
+                }
+            } else {
+                // 单个短选项，如 -a 或 -a=value
+                if (str_contains($option, '=')) {
+                    [$option, $value] = explode('=', $option, 2);
+                    $inputOptions[$option] = $value;
+                } else {
+                    $inputOptions[$option] = true;
+                    $currentOption = $option;
+                }
+            }
+        } else {
+            // 参数值
+            if ($currentOption !== null) {
+                $inputOptions[$currentOption] = $arg;
+                $currentOption = null;
+            } else {
+                $inputOptions[] = $arg;
+            }
+        }
+    }
+
+    return $inputOptions;
+}

composer.json

@@ -61,7 +61,8 @@
     "pint": "vendor/bin/pint ./src ./tests",
     "fix-style": "vendor/bin/pint ./src ./tests",
     "test": "vendor/bin/phpunit --colors=always",
-    "build": "php ./bin/build"
+    "build": "php ./bin/build",
+    "benchmark": "php ./benchmark/run.php"
   },
   "scripts-descriptions": {
     "test": "Run all tests.",

src/Converter.php

@@ -26,7 +26,7 @@ class Converter
 
     protected bool $noWords = false;
 
-    protected bool $keepOtherCharacters = false;
+    protected bool $cleanup = true;
 
     protected string $yuTo = 'v';
 
@@ -79,9 +79,9 @@ public function noWords(): static
         return $this;
     }
 
-    public function keepOtherCharacters(): static
+    public function noCleanup(): static
     {
-        $this->keepOtherCharacters = true;
+        $this->cleanup = false;
 
         return $this;
     }
@@ -174,7 +174,7 @@ public function convert(string $string, callable $beforeSplit = null): Collectio
         }, $string);
 
         // 过滤掉不保留的字符
-        if (! $this->keepOtherCharacters) {
+        if ($this->cleanup) {
             $string = \preg_replace(\sprintf('~[^%s]~u', \implode($this->regexps)), '', $string);
         }
 

src/Pinyin.php

@@ -10,6 +10,7 @@
  * @method static Converter onlyHans()
  * @method static Converter noAlpha()
  * @method static Converter noNumber()
+ * @method static Converter noCleanup()
  * @method static Converter noPunctuation()
  * @method static Converter noTone()
  * @method static Converter useNumberTone()
@@ -41,9 +42,9 @@ public static function sentence(string $string, string $toneStyle = Converter::T
         return self::withToneStyle($toneStyle)->convert($string);
     }
 
-    public static function sentenceFull(string $string, string $toneStyle = Converter::TONE_STYLE_SYMBOL): Collection
+    public static function fullSentence(string $string, string $toneStyle = Converter::TONE_STYLE_SYMBOL): Collection
     {
-        return self::keepOtherCharacters()->withToneStyle($toneStyle)->convert($string);
+        return self::noCleanup()->withToneStyle($toneStyle)->convert($string);
     }
 
     public static function polyphones(string $string, string $toneStyle = Converter::TONE_STYLE_SYMBOL, bool $asList = false): Collection

tests/ConverterTest.php

@@ -67,6 +67,11 @@ public function test_noPunctuation()
         $this->assertPinyin('123 dài zhe xī 456 wàng qù lǚ xíng 789', Converter::make()->noPunctuation()->convert('123带着希456望去旅行789'));
     }
 
+    public function test_noCleanup()
+    {
+        $this->assertPinyin('dài zhe （ xī wàng ） qù lǚ xíng ， bǐ dào dá zhōng diǎn gèng kuài lè 。ル shì piàn jiǎ míng ，π shì xī là zì mǔ 。', Converter::make()->noCleanup()->convert('带着（希望）去旅行，比到达终点更快乐。ル是片假名，π是希腊字母。'));
+    }
+
     public function test_tone_style()
     {
         $this->assertPinyin('chóng qìng', Converter::make()->convert('重庆'));

tests/PinyinTest.php

@@ -268,9 +268,9 @@ public function test_sentence()
         $this->assertPinyin('java gōng chéng shī', Pinyin::sentence('java工程师'));
     }
 
-    public function test_sentenceFull()
+    public function test_fullSentence()
     {
-        $this->assertPinyin('ル shì piàn jiǎ míng ，π shì xī là zì mǔ', Pinyin::sentenceFull('ル是片假名，π是希腊字母'));
+        $this->assertPinyin('ル shì piàn 。 jiǎ # míng ，π shì xī là zì …… mǔ', Pinyin::fullSentence('ル是片。假#名，π是希腊字……母'));
     }
 
     public function test_issues()