Pinyin::romanize() - Code Metrics - mucts/pinyin - Measure and Improve Code Quality continuously with Scrutinizer

Pinyin::romanize() A
last analyzed 2020-05-29 10:04 UTC

↳ Parent: Pinyin

Complexity

Conditions	2
Paths	2

Size

Total Lines	15
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Importance

Changes	3
Bugs	0	Features	0

Metric	Value
eloc	7
c	3
b	0
f	0
dl	0
loc	15
rs	10
cc	2
nc	2
nop	2

<?php
/**
 * This file is part of the mucts/pinyin.
 *
 * This source file is subject to the MIT license that is bundled
 * with this source code in the file LICENSE.
 *
 * @version 1.0
 * @author herry<[email protected]>
 * @copyright © 2020 MuCTS.com All Rights Reserved.
 */

namespace MuCTS\Pinyin;

use Exception;
use MuCTS\Pinyin\Exceptions\InvalidArgumentException;
use MuCTS\Pinyin\Interfaces\DictLoader;
use MuCTS\Pinyin\Loaders\File;
use MuCTS\Pinyin\Loaders\GeneratorFile;
use MuCTS\Pinyin\Loaders\MemoryFile;
use MuCTS\Support\Arr;

class Pinyin
{
    public const DEFAULT = 4096;
    /** @var int UNICODE 式音调：měi hǎo */
    public const TONE = 2;
    /** @var int 无音调：mei hao */
    public const NO_TONE = 4;
    /** @var int 带数字式音调： mei3 hao3 */
    public const ASCII_TONE = 8;
    /** @var int 翻译姓名 */
    public const NAME = 16;
    /** @var int 保留数字 */
    public const KEEP_NUMBER = 32;
    /** @var int 保留英文 */
    public const KEEP_ENGLISH = 64;
    /** @var int 使用 v 代替 yu, 例如：吕 lyu 将会转为 lv */
    public const UMLAUT_V = 128;
    /** @var int  保留标点 */
    public const KEEP_PUNCTUATION = 256;
    /**
     * Dict loader.
     *
     * @var DictLoader|string
     */
    private $loader;
    /** @var string */
    private string $path;
    /** @var string[] */
    protected array $alias = [
        'file' => File::class,
        'generator' => GeneratorFile::class,
        'memory' => MemoryFile::class
    ];

    /**
     * Punctuations map.
     *
     * @var array
     */
    protected array $punctuations = [
        '，' => ',',
        '。' => '.',
        '！' => '!',
        '？' => '?',
        '：' => ':',
        '“' => '"',
        '”' => '"',
        '‘' => "'",
        '’' => "'",
        '_' => '_',
    ];

    /**
     * Constructor.
     *
     * @param string|null $loader
     * @param string|null $path
     * @throws Exception|InvalidArgumentException
     */
    public function __construct(?string $loader = null, ?string $path = null)
    {
        $this->setLoader($loader);
        $this->setPath($path);
    }

    /**
     * Convert string to pinyin.
     *
     * @param string $string
     * @param int $option
     *
     * @return array
     */
    public function convert(string $string, int $option = self::DEFAULT): array
    {
        $pinyin = $this->romanize($string, $option);

        return $this->splitWords($pinyin, $option);
    }

    /**
     * Convert string (person name) to pinyin.
     *
     * @param string $string
     * @param int $option
     *
     * @return array
     */
    public function name(string $string, int $option = self::NAME): array
    {
        $option = $option | self::NAME;

        $pinyin = $this->romanize($string, $option);

        return $this->splitWords($pinyin, $option);
    }

    /**
     * Return a pinyin permalink from string.
     *
     * @param string $string
     * @param string|int $delimiter
     * @param int $option
     *
     * @return string
     */
    public function permalink(string $string, $delimiter = '-', int $option = self::DEFAULT): string
    {
        if (is_int($delimiter)) {
            list($option, $delimiter) = [$delimiter, '-'];
        }

        if (!in_array($delimiter, ['_', '-', '.', ''], true)) {
            throw new InvalidArgumentException("Delimiter must be one of: '_', '-', '', '.'.");
        }

        return implode($delimiter, $this->convert($string, $option | self::KEEP_NUMBER | self::KEEP_ENGLISH));
    }

    /**
     * Return first letters.
     *
     * @param string $string
     * @param string|int $delimiter
     * @param int $option
     *
     * @return string
     */
    public function abbr(string $string, $delimiter = '', int $option = self::DEFAULT): string
    {
        if (is_int($delimiter)) {
            list($option, $delimiter) = [$delimiter, ''];
        }

        return implode($delimiter, array_map(function ($pinyin) {
            return is_numeric($pinyin) ? $pinyin : mb_substr($pinyin, 0, 1);
        }, $this->convert($string, $option)));
    }

    /**
     * Chinese phrase to pinyin.
     *
     * @param string $string
     * @param string|int $delimiter
     * @param int $option
     *
     * @return string
     */
    public function phrase(string $string, $delimiter = ' ', int $option = self::DEFAULT): string
    {
        if (is_int($delimiter)) {
            list($option, $delimiter) = [$delimiter, ' '];
        }

        return implode($delimiter, $this->convert($string, $option));
    }

    /**
     * Chinese to pinyin sentence.
     *
     * @param string|int $string
     * @param string|int $delimiter
     * @param int $option
     *
     * @return string
     */
    public function sentence($string, $delimiter = ' ', $option = self::NO_TONE)
    {
        if (is_int($delimiter)) {
            list($option, $delimiter) = [$delimiter, ' '];
        }

        return implode($delimiter, $this->convert($string, $option | self::KEEP_PUNCTUATION | self::KEEP_ENGLISH | self::KEEP_NUMBER));
    }

    /**
     * Determine if it's an alias
     *
     * @param string $name
     * @return bool
     */
    protected function isAlias(string $name): bool
    {
        return Arr::exists($this->alias, $name);
    }

    /**
     * Loader setter.
     *
     * @param DictLoader|string|null $loader
     *
     * @return $this
     * @throws Exception|InvalidArgumentException
     */
    public function setLoader($loader = null)
    {
        if (is_null($loader)) {
            $this->loader = File::class;
            return $this;
        }
        if ($loader instanceof DictLoader) {
            $this->loader = $loader;
            return $this;
        }
        if ($this->isAlias($loader)) {
            $this->loader = Arr::get($this->alias, $loader);
            return $this;
        }
        if (class_exists($loader) && in_array(DictLoader::class, class_implements($loader))) {
            $this->loader = $loader;
            return $this;
        }
        throw new InvalidArgumentException('This\'s not valid dict loader class.');
    }

    /**
     * Return dict loader,.
     *
     * @return DictLoader
     */
    public function getLoader(): DictLoader
    {
        if (!($this->loader instanceof DictLoader)) {
            $loaderName = $this->loader;
            $this->loader = new $loaderName($this->path);
        }

        return $this->loader;
    }

    /**
     * Set data path
     *
     * @param string|null $path
     * @return $this
     */
    public function setPath(?string $path): self
    {
        $path = $path ?? dirname(__DIR__) . '/data/';
        if (!is_dir($path)) {
            throw new InvalidArgumentException(sprintf('\'%s\' is not valid data path.', $path));
        }
        $this->path = $path;
        return $this;
    }

    /**
     * Convert Chinese to pinyin.
     *
     * @param string $string
     * @param int $option
     *
     * @return string
     */
    protected function romanize(string $string, int $option = self::DEFAULT): string
    {
        $string = $this->prepare($string, $option);

        $dictLoader = $this->getLoader();

        if ($this->hasOption($option, self::NAME)) {
            $string = $this->convertSurname($string, $dictLoader);
        }

        $dictLoader->map(function ($dictionary) use (&$string) {
            $string = strtr($string, $dictionary);
        });

        return $string;
    }

    /**
     * Convert Chinese Surname to pinyin.
     *
     * @param string $string
     * @param DictLoader $dictLoader
     *
     * @return string
     */
    protected function convertSurname(string $string, DictLoader $dictLoader): string
    {
        $dictLoader->mapSurname(function ($dictionary) use (&$string) {
            foreach ($dictionary as $surname => $pinyin) {
                if (0 === strpos($string, $surname)) {
                    $string = $pinyin . mb_substr($string, mb_strlen($surname, 'UTF-8'), mb_strlen($string, 'UTF-8') - 1, 'UTF-8');

                    break;
                }
            }
        });

        return $string;
    }

    /**
     * Split pinyin string to words.
     *
     * @param string $pinyin
     * @param int $option
     *
     * @return array
     */
    protected function splitWords(string $pinyin, int $option): array
    {
        $split = preg_split('/\s+/i', $pinyin);
        if (!is_array($split)) {
            throw new InvalidArgumentException(sprintf('\'%s\' is not valid pinyin.', $pinyin));
        }
        $split = array_filter($split);

        if (!$this->hasOption($option, self::TONE)) {
            foreach ($split as $index => $pinyin) {
                $split[$index] = $this->formatTone($pinyin, $option);
            }
        }

        return array_values($split);
    }

    /**
     * @param int $option
     * @param int $check
     *
     * @return bool
     */
    public function hasOption(int $option, int $check): bool
    {
        return ($option & $check) === $check;
    }

    /**
     * Pre-process.
     *
     * @param string $string
     * @param int $option
     *
     * @return string
     */
    protected function prepare(string $string, int $option = self::DEFAULT): string
    {
        $string = preg_replace_callback('~[a-z0-9_-]+~i', function ($matches) {
            return "\t" . $matches[0];
        }, $string);

        $regex = ['\p{Han}', '\p{Z}', '\p{M}', "\t"];

        if ($this->hasOption($option, self::KEEP_NUMBER)) {
            array_push($regex, '0-9');
        }

        if ($this->hasOption($option, self::KEEP_ENGLISH)) {
            array_push($regex, 'a-zA-Z');
        }

        if ($this->hasOption($option, self::KEEP_PUNCTUATION)) {
            $punctuations = array_merge($this->punctuations, ['  ' => ' ']);
            $string = trim(str_replace(array_keys($punctuations), $punctuations, $string));

            array_push($regex, preg_quote(implode(array_merge(array_keys($this->punctuations), $this->punctuations)), '~'));
        }

        return preg_replace(sprintf('~[^%s]~u', implode($regex)), '', $string);
    }

    /**
     * Format.
     *
     * @param string $pinyin
     * @param int $option
     *
     * @return string
     */
    protected function formatTone(string $pinyin, int $option = self::NO_TONE): string
    {
        $replacements = [
            'üē' => ['ue', 1], 'üé' => ['ue', 2], 'üě' => ['ue', 3], 'üè' => ['ue', 4],
            'ā' => ['a', 1], 'ē' => ['e', 1], 'ī' => ['i', 1], 'ō' => ['o', 1], 'ū' => ['u', 1], 'ǖ' => ['yu', 1],
            'á' => ['a', 2], 'é' => ['e', 2], 'í' => ['i', 2], 'ó' => ['o', 2], 'ú' => ['u', 2], 'ǘ' => ['yu', 2],
            'ǎ' => ['a', 3], 'ě' => ['e', 3], 'ǐ' => ['i', 3], 'ǒ' => ['o', 3], 'ǔ' => ['u', 3], 'ǚ' => ['yu', 3],
            'à' => ['a', 4], 'è' => ['e', 4], 'ì' => ['i', 4], 'ò' => ['o', 4], 'ù' => ['u', 4], 'ǜ' => ['yu', 4],
        ];

        foreach ($replacements as $unicode => $replacement) {
            if (false !== strpos($pinyin, $unicode)) {
                $umlaut = $replacement[0];

                if ($this->hasOption($option, self::UMLAUT_V) && 'yu' == $umlaut) {
                    $umlaut = 'v';
                }

                $pinyin = str_replace($unicode, $umlaut, $pinyin) . ($this->hasOption($option, self::ASCII_TONE) ? $replacement[1] : '');
            }
        }

        return $pinyin;
    }
}


1			<?php
2			/**
3			* This file is part of the mucts/pinyin.
4			*
5			* This source file is subject to the MIT license that is bundled
6			* with this source code in the file LICENSE.
7			*
8			* @version 1.0
9			* @author herry<[email protected]>
10			* @copyright © 2020 MuCTS.com All Rights Reserved.
11			*/
12
13			namespace MuCTS\Pinyin;
14
15			use Exception;
16			use MuCTS\Pinyin\Exceptions\InvalidArgumentException;
17			use MuCTS\Pinyin\Interfaces\DictLoader;
18			use MuCTS\Pinyin\Loaders\File;
19			use MuCTS\Pinyin\Loaders\GeneratorFile;
20			use MuCTS\Pinyin\Loaders\MemoryFile;
21			use MuCTS\Support\Arr;
22
23			class Pinyin
24			{
25			public const DEFAULT = 4096;
26			/** @var int UNICODE 式音调：měi hǎo */
27			public const TONE = 2;
28			/** @var int 无音调：mei hao */
29			public const NO_TONE = 4;
30			/** @var int 带数字式音调： mei3 hao3 */
31			public const ASCII_TONE = 8;
32			/** @var int 翻译姓名 */
33			public const NAME = 16;
34			/** @var int 保留数字 */
35			public const KEEP_NUMBER = 32;
36			/** @var int 保留英文 */
37			public const KEEP_ENGLISH = 64;
38			/** @var int 使用 v 代替 yu, 例如：吕 lyu 将会转为 lv */
39			public const UMLAUT_V = 128;
40			/** @var int 保留标点 */
41			public const KEEP_PUNCTUATION = 256;
42			/**
43			* Dict loader.
44			*
45			* @var DictLoader\|string
46			*/
47			private $loader;
48			/** @var string */
49			private string $path;
50			/** @var string[] */
51			protected array $alias = [
52			'file' => File::class,
53			'generator' => GeneratorFile::class,
54			'memory' => MemoryFile::class
55			];
56
57			/**
58			* Punctuations map.
59			*
60			* @var array
61			*/
62			protected array $punctuations = [
63			'，' => ',',
64			'。' => '.',
65			'！' => '!',
66			'？' => '?',
67			'：' => ':',
68			'“' => '"',
69			'”' => '"',
70			'‘' => "'",
71			'’' => "'",
72			'_' => '_',
73			];
74
75			/**
76			* Constructor.
77			*
78			* @param string\|null $loader
79			* @param string\|null $path
80			* @throws Exception\|InvalidArgumentException
81			*/
82			public function __construct(?string $loader = null, ?string $path = null)
83			{
84			$this->setLoader($loader);
85			$this->setPath($path);
86			}
87
88			/**
89			* Convert string to pinyin.
90			*
91			* @param string $string
92			* @param int $option
93			*
94			* @return array
95			*/
96			public function convert(string $string, int $option = self::DEFAULT): array
97			{
98			$pinyin = $this->romanize($string, $option);
99
100			return $this->splitWords($pinyin, $option);
101			}
102
103			/**
104			* Convert string (person name) to pinyin.
105			*
106			* @param string $string
107			* @param int $option
108			*
109			* @return array
110			*/
111			public function name(string $string, int $option = self::NAME): array
112			{
113			$option = $option \| self::NAME;
114
115			$pinyin = $this->romanize($string, $option);
116
117			return $this->splitWords($pinyin, $option);
118			}
119
120			/**
121			* Return a pinyin permalink from string.
122			*
123			* @param string $string
124			* @param string\|int $delimiter
125			* @param int $option
126			*
127			* @return string
128			*/
129			public function permalink(string $string, $delimiter = '-', int $option = self::DEFAULT): string
130			{
131			if (is_int($delimiter)) {
132			list($option, $delimiter) = [$delimiter, '-'];
133			}
134
135			if (!in_array($delimiter, ['_', '-', '.', ''], true)) {
136			throw new InvalidArgumentException("Delimiter must be one of: '_', '-', '', '.'.");
137			}
138
139			return implode($delimiter, $this->convert($string, $option \| self::KEEP_NUMBER \| self::KEEP_ENGLISH));
140			}
141
142			/**
143			* Return first letters.
144			*
145			* @param string $string
146			* @param string\|int $delimiter
147			* @param int $option
148			*
149			* @return string
150			*/
151			public function abbr(string $string, $delimiter = '', int $option = self::DEFAULT): string
152			{
153			if (is_int($delimiter)) {
154			list($option, $delimiter) = [$delimiter, ''];
155			}
156
157			return implode($delimiter, array_map(function ($pinyin) {
158			return is_numeric($pinyin) ? $pinyin : mb_substr($pinyin, 0, 1);
159			}, $this->convert($string, $option)));
160			}
161
162			/**
163			* Chinese phrase to pinyin.
164			*
165			* @param string $string
166			* @param string\|int $delimiter
167			* @param int $option
168			*
169			* @return string
170			*/
171			public function phrase(string $string, $delimiter = ' ', int $option = self::DEFAULT): string
172			{
173			if (is_int($delimiter)) {
174			list($option, $delimiter) = [$delimiter, ' '];
175			}
176
177			return implode($delimiter, $this->convert($string, $option));
178			}
179
180			/**
181			* Chinese to pinyin sentence.
182			*
183			* @param string\|int $string
184			* @param string\|int $delimiter
185			* @param int $option
186			*
187			* @return string
188			*/
189			public function sentence($string, $delimiter = ' ', $option = self::NO_TONE)
190			{
191			if (is_int($delimiter)) {
192			list($option, $delimiter) = [$delimiter, ' '];
193			}
194
195			return implode($delimiter, $this->convert($string, $option \| self::KEEP_PUNCTUATION \| self::KEEP_ENGLISH \| self::KEEP_NUMBER));
196			}
197
198			/**
199			* Determine if it's an alias
200			*
201			* @param string $name
202			* @return bool
203			*/
204			protected function isAlias(string $name): bool
205			{
206			return Arr::exists($this->alias, $name);
207			}
208
209			/**
210			* Loader setter.
211			*
212			* @param DictLoader\|string\|null $loader
213			*
214			* @return $this
215			* @throws Exception\|InvalidArgumentException
216			*/
217			public function setLoader($loader = null)
218			{
219			if (is_null($loader)) {
220			$this->loader = File::class;
221			return $this;
222			}
223			if ($loader instanceof DictLoader) {
224			$this->loader = $loader;
225			return $this;
226			}
227			if ($this->isAlias($loader)) {
228			$this->loader = Arr::get($this->alias, $loader);
229			return $this;
230			}
231			if (class_exists($loader) && in_array(DictLoader::class, class_implements($loader))) {
232			$this->loader = $loader;
233			return $this;
234			}
235			throw new InvalidArgumentException('This\'s not valid dict loader class.');
236			}
237
238			/**
239			* Return dict loader,.
240			*
241			* @return DictLoader
242			*/
243			public function getLoader(): DictLoader
244			{
245			if (!($this->loader instanceof DictLoader)) {
246			$loaderName = $this->loader;
247			$this->loader = new $loaderName($this->path);
248			}
249
250			return $this->loader;
251			}
252
253			/**
254			* Set data path
255			*
256			* @param string\|null $path
257			* @return $this
258			*/
259			public function setPath(?string $path): self
260			{
261			$path = $path ?? dirname(__DIR__) . '/data/';
262			if (!is_dir($path)) {
263			throw new InvalidArgumentException(sprintf('\'%s\' is not valid data path.', $path));
264			}
265			$this->path = $path;
266			return $this;
267			}
268
269			/**
270			* Convert Chinese to pinyin.
271			*
272			* @param string $string
273			* @param int $option
274			*
275			* @return string
276			*/
277			protected function romanize(string $string, int $option = self::DEFAULT): string
278			{
279			$string = $this->prepare($string, $option);
280
281			$dictLoader = $this->getLoader();
282
283			if ($this->hasOption($option, self::NAME)) {
284			$string = $this->convertSurname($string, $dictLoader);
285			}
286
287			$dictLoader->map(function ($dictionary) use (&$string) {
288			$string = strtr($string, $dictionary);
289			});
290
291			return $string;
292			}
293
294			/**
295			* Convert Chinese Surname to pinyin.
296			*
297			* @param string $string
298			* @param DictLoader $dictLoader
299			*
300			* @return string
301			*/
302			protected function convertSurname(string $string, DictLoader $dictLoader): string
303			{
304			$dictLoader->mapSurname(function ($dictionary) use (&$string) {
305			foreach ($dictionary as $surname => $pinyin) {
306			if (0 === strpos($string, $surname)) {
307			$string = $pinyin . mb_substr($string, mb_strlen($surname, 'UTF-8'), mb_strlen($string, 'UTF-8') - 1, 'UTF-8');
308
309			break;
310			}
311			}
312			});
313
314			return $string;
315			}
316
317			/**
318			* Split pinyin string to words.
319			*
320			* @param string $pinyin
321			* @param int $option
322			*
323			* @return array
324			*/
325			protected function splitWords(string $pinyin, int $option): array
326			{
327			$split = preg_split('/\s+/i', $pinyin);
328			if (!is_array($split)) {
329			throw new InvalidArgumentException(sprintf('\'%s\' is not valid pinyin.', $pinyin));
330			}
331			$split = array_filter($split);
332
333			if (!$this->hasOption($option, self::TONE)) {
334			foreach ($split as $index => $pinyin) {
335			$split[$index] = $this->formatTone($pinyin, $option);
336			}
337			}
338
339			return array_values($split);
340			}
341
342			/**
343			* @param int $option
344			* @param int $check
345			*
346			* @return bool
347			*/
348			public function hasOption(int $option, int $check): bool
349			{
350			return ($option & $check) === $check;
351			}
352
353			/**
354			* Pre-process.
355			*
356			* @param string $string
357			* @param int $option
358			*
359			* @return string
360			*/
361			protected function prepare(string $string, int $option = self::DEFAULT): string
362			{
363			$string = preg_replace_callback('~[a-z0-9_-]+~i', function ($matches) {
364			return "\t" . $matches[0];
365			}, $string);
366
367			$regex = ['\p{Han}', '\p{Z}', '\p{M}', "\t"];
368
369			if ($this->hasOption($option, self::KEEP_NUMBER)) {
370			array_push($regex, '0-9');
371			}
372
373			if ($this->hasOption($option, self::KEEP_ENGLISH)) {
374			array_push($regex, 'a-zA-Z');
375			}
376
377			if ($this->hasOption($option, self::KEEP_PUNCTUATION)) {
378			$punctuations = array_merge($this->punctuations, [' ' => ' ']);
379			$string = trim(str_replace(array_keys($punctuations), $punctuations, $string));
380
381			array_push($regex, preg_quote(implode(array_merge(array_keys($this->punctuations), $this->punctuations)), '~'));
382			}
383
384			return preg_replace(sprintf('~[^%s]~u', implode($regex)), '', $string);
385			}
386
387			/**
388			* Format.
389			*
390			* @param string $pinyin
391			* @param int $option
392			*
393			* @return string
394			*/
395			protected function formatTone(string $pinyin, int $option = self::NO_TONE): string
396			{
397			$replacements = [
398			'üē' => ['ue', 1], 'üé' => ['ue', 2], 'üě' => ['ue', 3], 'üè' => ['ue', 4],
399			'ā' => ['a', 1], 'ē' => ['e', 1], 'ī' => ['i', 1], 'ō' => ['o', 1], 'ū' => ['u', 1], 'ǖ' => ['yu', 1],
400			'á' => ['a', 2], 'é' => ['e', 2], 'í' => ['i', 2], 'ó' => ['o', 2], 'ú' => ['u', 2], 'ǘ' => ['yu', 2],
401			'ǎ' => ['a', 3], 'ě' => ['e', 3], 'ǐ' => ['i', 3], 'ǒ' => ['o', 3], 'ǔ' => ['u', 3], 'ǚ' => ['yu', 3],
402			'à' => ['a', 4], 'è' => ['e', 4], 'ì' => ['i', 4], 'ò' => ['o', 4], 'ù' => ['u', 4], 'ǜ' => ['yu', 4],
403			];
404
405			foreach ($replacements as $unicode => $replacement) {
406			if (false !== strpos($pinyin, $unicode)) {
407			$umlaut = $replacement[0];
408
409			if ($this->hasOption($option, self::UMLAUT_V) && 'yu' == $umlaut) {
410			$umlaut = 'v';
411			}
412
413			$pinyin = str_replace($unicode, $umlaut, $pinyin) . ($this->hasOption($option, self::ASCII_TONE) ? $replacement[1] : '');
414			}
415			}
416
417			return $pinyin;
418			}
419			}
420

mucts / pinyin

Pinyin::romanize() A last analyzed 2020-05-29 10:04 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

Pinyin::romanize() A
last analyzed 2020-05-29 10:04 UTC