2 Commits
Author SHA1 Message Date
kor-elf 8ce9900afb Merge pull request 'Version 1.1.0' (#2) from develop into main
Reviewed-on: #2
2024-10-11 00:53:29 +05:00
kor-elf 2af65cb5db Merge pull request 'Version 1.0.0' (#1) from develop into main
Reviewed-on: #1
2024-09-16 22:37:07 +05:00
17 changed files with 137 additions and 607 deletions
-3
View File
@@ -1,3 +0,0 @@
/.git export-ignore
/.gitignore export-ignore
/.gitattributes export-ignore
+7 -36
View File
@@ -5,10 +5,7 @@
# Параметры в **.env**: # Параметры в **.env**:
### TRANSLATE_SERVICE ### TRANSLATE_SERVICE
По умолчанию через какой сервис обращаться за переводом. На данный момент доступен только сервисы - **yandex**, **log**. По умолчанию через какой сервис обращаться за переводом. На данный момент доступен только один сервис - **yandex**.
**yandex** - https://aistudio.yandex.ru/docs/ru/translate/concepts/<br>
**log** - для проверки и внедрения в свой продукт.
### TRANSLATE_YANDEX_FOLDER_ID ### TRANSLATE_YANDEX_FOLDER_ID
ID folder. Код можно увидеть в адресе console.yandex.cloud/folders/**{тут будет код}**. ID folder. Код можно увидеть в адресе console.yandex.cloud/folders/**{тут будет код}**.
@@ -26,16 +23,6 @@ ID folder. Код можно увидеть в адресе console.yandex.cloud
### TRANSLATE_YANDEX_LIMIT_MAX_SYMBOLS ### TRANSLATE_YANDEX_LIMIT_MAX_SYMBOLS
Максимальное количество символов за один запрос. Если превышает, то делится на две части и делает по очереди два запроса. По умолчанию стоит 9000 символов. **На данный момент работает только перевод через систему очередей.** Максимальное количество символов за один запрос. Если превышает, то делится на две части и делает по очереди два запроса. По умолчанию стоит 9000 символов. **На данный момент работает только перевод через систему очередей.**
### TRANSLATE_LOG_LIMIT_MAX_REQUEST
Максимальное количество запросов в период, который указан в TRANSLATE_YANDEX_LIMIT_RATE_SECONDS. По умолчанию 20 запросов в секунду. **На данный момент работает только перевод через систему очередей.**
### TRANSLATE_LOG_LIMIT_RATE_SECONDS
Период в котором считается максимальное количество запросов. По умолчанию стоит секунда. **На данный момент работает только перевод через систему очередей.**
### TRANSLATE_LOG_LIMIT_MAX_SYMBOLS
Максимальное количество символов за один запрос. Если превышает, то делится на две части и делает по очереди два запроса. По умолчанию стоит 9000 символов. **На данный момент работает только перевод через систему очередей.**
# Методы # Методы
### Перевести обычный текст ### Перевести обычный текст
> \KorElf\TranslateLaravel\Facades\Translate::translateText( > \KorElf\TranslateLaravel\Facades\Translate::translateText(
@@ -70,12 +57,12 @@ ID folder. Код можно увидеть в адресе console.yandex.cloud
### Перевести с помощью очередей (Queues) ### Перевести с помощью очередей (Queues)
> \KorElf\TranslateLaravel\Facades\Translate::runJob( > \KorElf\TranslateLaravel\Facades\Translate::runJob(
> <br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;\KorElf\TranslateLaravel\DTO\RunTranslateDto &nbsp;&nbsp;\$params, > <br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;\KorElf\TranslateLaravel\DTO\RunTranslateDto &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;\$params,
> <br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;\KorElf\TranslateLaravel\DTO\AfterTranslateDto &nbsp;\$afterTranslateDto > <br>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;\KorElf\TranslateLaravel\Contracts\TranslationCompletedListener \$completedListener
<br>): \Illuminate\Foundation\Bus\PendingDispatch <br>): \Illuminate\Foundation\Bus\PendingDispatch
**\$params** - параметры перевода. **\$params** - параметры перевода.
<br>**\$afterTranslateDto** - после завершения перевода отправляет результат в объект \$afterTranslateDto->\$className. Объект должен соблюдать контракт **\KorElf\TranslateLaravel\Contracts\TranslationCompletedListener**. <br>**\$completedListener** - после завершения перевода отправляет результат в этот объект. Объект должен соблюдать контракт **\KorElf\TranslateLaravel\Contracts\TranslationCompletedListener**.
**Пример:** **Пример:**
> Создаём файл TranslationListener например в папке app/Services > Создаём файл TranslationListener например в папке app/Services
@@ -89,41 +76,25 @@ ID folder. Код можно увидеть в адресе console.yandex.cloud
> >
> final class TranslationListener implements TranslationCompletedListener > final class TranslationListener implements TranslationCompletedListener
> { > {
> /** > public function onTranslationCompleted(array $translatedText): void
> * Объекты создаются и пробрасываются автоматически через контейнер `$application->make`,
> * благодаря механизму автоматического внедрения зависимостей (Dependency Injection).
> * Вы можете передавать свои собственные классы, поместив их в качестве параметров конструктора.
> * Контейнер создаст эти объекты автоматически, основываясь на разрешении типа (type-hint).
> */
> // public function __construct(
> // private readonly SaveContentCommand $saveContentCommand
> // ) { }
>
> public function onTranslationCompleted(array $translatedText, array $data = []): void
> { > {
> Log::info($data);
> foreach ($translatedText as $translatedTextKey => $translatedTextValue) { > foreach ($translatedText as $translatedTextKey => $translatedTextValue) {
> Log::info($translatedTextKey . ': ' . $translatedTextValue); > Log::info($translatedTextKey . ': ' . $translatedTextValue);
> } > }
> // $this->saveContentCommand->execute($data['contentId'], $translatedText);
> } > }
> } > }
> >
> Потом например в каком-то контроллере (а лучше конечно в сервисе) пишем такой метод > Потом например в каком-то контроллере (а лучше конечно в сервисе) пишем такой метод
> >
> public function sendingForTranslation(): View > public function sendingForTranslation(\App\Services\TranslationListener $translationListener): View
> { > {
> // Вначале создаём объект с параметрами > // Вначале создаём объект с параметрами
> $params = (new \KorElf\TranslateLaravel\DTO\RunTranslateDto) > $params = (new \KorElf\TranslateLaravel\DTO\RunTranslateDto)
> ->addParamText('title', 'Заголовок', 'en', 'ru') > ->addParamText('title', 'Заголовок', 'en', 'ru')
> ->addParamHtml('content', '<p>Привет, Мир!</p>', 'en', 'ru'); > ->addParamHtml('content', '<p>Привет, Мир!</p>', 'en', 'ru');
> >
> $translationCompletedListener = \App\Services\TranslationListener::class;
> $data = ['contentId' => 1];
> $afterTranslateDto = new \KorElf\TranslateLaravel\DTO\AfterTranslateDto($translationCompletedListener, $data);
>
> // Отправляем на очередь > // Отправляем на очередь
> \KorElf\TranslateLaravel\Facades\Translate::runJob($params, $afterTranslateDto); > \KorElf\TranslateLaravel\Facades\Translate::runJob($params, $translationListener);
> >
> return view('success'); > return view('success');
> } > }
+3 -4
View File
@@ -19,12 +19,11 @@
], ],
"require": { "require": {
"php": "^8.2", "php": "^8.2",
"illuminate/support": "^10.0|^11.0|^12.0|^13.0", "illuminate/support": "^10.0|^11.0",
"guzzlehttp/guzzle": "^7.0.1", "guzzlehttp/guzzle": "^7.0.1",
"web-token/jwt-framework": "^3.0", "web-token/jwt-framework": "^3.0",
"ext-libxml": "*", "ext-libxml": "^2.11.7",
"ext-dom": "*", "ext-dom": "^2.11.7"
"ext-mbstring": "*"
}, },
"extra": { "extra": {
"laravel": { "laravel": {
-11
View File
@@ -35,16 +35,5 @@ return [
], ],
], ],
], ],
'log' => [
'driver' => '\KorElf\TranslateLaravel\Translate\LogDriver',
'config' => [
'limit' => [
'max_request' => (int) env('TRANSLATE_LOG_LIMIT_MAX_REQUEST', 20),
'rate_seconds' => (int) env('TRANSLATE_LOG_LIMIT_RATE_SECONDS', 1),
'max_symbols' => (int) env('TRANSLATE_LOG_LIMIT_MAX_SYMBOLS', 9000),
],
],
],
], ],
]; ];
@@ -4,5 +4,5 @@ namespace KorElf\TranslateLaravel\Contracts;
interface TranslationCompletedListener interface TranslationCompletedListener
{ {
public function onTranslationCompleted(array $translatedText, array $data = []): void; public function onTranslationCompleted(array $translatedText): void;
} }
-32
View File
@@ -1,32 +0,0 @@
<?php declare(strict_types=1);
namespace KorElf\TranslateLaravel\DTO;
final readonly class AfterTranslateDto
{
/**
* @param string $className The fully qualified class name (FQN) of a class that implements the TranslationCompletedListener interface.
* @param array $data Additional data associated with the translation process.
*/
public function __construct(
private string $className,
private array $data = [],
) { }
/**
* Get the fully qualified class name (FQN) of a class that implements the TranslationCompletedListener interface.
*
* @return class-string<\KorElf\TranslateLaravel\Contracts\TranslationCompletedListener>
*/
public function getClassName(): string
{
return $this->className;
}
public function getData(): array
{
return $this->data;
}
}
-40
View File
@@ -1,40 +0,0 @@
<?php declare(strict_types=1);
namespace KorElf\TranslateLaravel\DTO;
final class PartText
{
private array $texts = [];
private array $beforeTexts = [];
private array $afterTexts = [];
private int $part = 0;
public function add(string $text, ?string $beforeText = null, ?string $afterText = null): void
{
$this->texts[$this->part] = $text;
$this->beforeTexts[$this->part] = $beforeText;
$this->afterTexts[$this->part] = $afterText;
$this->part++;
}
public function getTextsForTranslation(): array
{
return $this->texts;
}
public function getTextsAfterTranslation(array $texts): string
{
$result = '';
foreach ($texts as $key => $text) {
if ($this->beforeTexts[$key] !== null) {
$result .= $this->beforeTexts[$key];
}
$result .= $text;
if ($this->afterTexts[$key] !== null) {
$result .= $this->afterTexts[$key];
}
}
return $result;
}
}
+2 -2
View File
@@ -10,7 +10,7 @@ final readonly class ProcessTranslateDto
private string $groupName, private string $groupName,
private string $key, private string $key,
private int $part, private int $part,
private PartText $text, private string $text,
private TextType $textType, private TextType $textType,
private string $targetLanguageCode, private string $targetLanguageCode,
private ?string $sourceLanguageCode = null, private ?string $sourceLanguageCode = null,
@@ -32,7 +32,7 @@ final readonly class ProcessTranslateDto
return $this->part; return $this->part;
} }
public function getText(): PartText public function getText(): string
{ {
return $this->text; return $this->text;
} }
+2 -2
View File
@@ -3,7 +3,7 @@
namespace KorElf\TranslateLaravel\Facades; namespace KorElf\TranslateLaravel\Facades;
use Illuminate\Support\Facades\Facade; use Illuminate\Support\Facades\Facade;
use KorElf\TranslateLaravel\DTO\AfterTranslateDto; use KorElf\TranslateLaravel\Contracts\TranslationCompletedListener;
use KorElf\TranslateLaravel\DTO\RunTranslateDto; use KorElf\TranslateLaravel\DTO\RunTranslateDto;
/** /**
@@ -11,7 +11,7 @@ use KorElf\TranslateLaravel\DTO\RunTranslateDto;
* @method static \KorElf\TranslateLaravel\Contracts\Translate resolve(string $name) * @method static \KorElf\TranslateLaravel\Contracts\Translate resolve(string $name)
* @method static string|array translateText(string|array $text, string $targetLanguageCode, ?string $sourceLanguageCode = null) * @method static string|array translateText(string|array $text, string $targetLanguageCode, ?string $sourceLanguageCode = null)
* @method static string|array translateHtml(string|array $text, string $targetLanguageCode, ?string $sourceLanguageCode = null) * @method static string|array translateHtml(string|array $text, string $targetLanguageCode, ?string $sourceLanguageCode = null)
* @method static \Illuminate\Foundation\Bus\PendingDispatch runJob(RunTranslateDto $params, AfterTranslateDto $afterTranslateDto) * @method static \Illuminate\Foundation\Bus\PendingDispatch runJob(RunTranslateDto $params, TranslationCompletedListener $completedListener)
* @method static \KorElf\TranslateLaravel\DTO\Languages listLanguages() * @method static \KorElf\TranslateLaravel\DTO\Languages listLanguages()
* @method static string getDefaultDriver() * @method static string getDefaultDriver()
* @method static void setDefaultDriver(string $name) * @method static void setDefaultDriver(string $name)
+9 -11
View File
@@ -5,11 +5,10 @@ namespace KorElf\TranslateLaravel\Jobs;
use Illuminate\Bus\Queueable; use Illuminate\Bus\Queueable;
use Illuminate\Contracts\Queue\ShouldBeEncrypted; use Illuminate\Contracts\Queue\ShouldBeEncrypted;
use Illuminate\Contracts\Queue\ShouldQueue; use Illuminate\Contracts\Queue\ShouldQueue;
use Illuminate\Foundation\Application;
use Illuminate\Foundation\Bus\Dispatchable; use Illuminate\Foundation\Bus\Dispatchable;
use Illuminate\Queue\InteractsWithQueue; use Illuminate\Queue\InteractsWithQueue;
use Illuminate\Support\Facades\Cache; use Illuminate\Support\Facades\Cache;
use KorElf\TranslateLaravel\DTO\AfterTranslateDto; use KorElf\TranslateLaravel\Contracts\TranslationCompletedListener;
use KorElf\TranslateLaravel\DTO\Translated; use KorElf\TranslateLaravel\DTO\Translated;
use KorElf\TranslateLaravel\Exceptions\AfterTranslateException; use KorElf\TranslateLaravel\Exceptions\AfterTranslateException;
@@ -22,11 +21,14 @@ final class AfterTranslate implements ShouldQueue, ShouldBeEncrypted
*/ */
public function __construct( public function __construct(
private readonly string $groupName, private readonly string $groupName,
private readonly TranslationCompletedListener $listener,
private readonly Translated $translated, private readonly Translated $translated,
private readonly AfterTranslateDto $afterTranslateDto
) { } ) { }
public function handle(Application $application): void /**
* Execute the job.
*/
public function handle(): void
{ {
$translated = []; $translated = [];
$data = Cache::get($this->groupName, []); $data = Cache::get($this->groupName, []);
@@ -41,15 +43,11 @@ final class AfterTranslate implements ShouldQueue, ShouldBeEncrypted
$translated[$key] = implode(' ', $data[$key]); $translated[$key] = implode(' ', $data[$key]);
} }
if (empty($data) || !empty($errors)) { if (empty($data)) {
throw new AfterTranslateException('Part or all of the text has not been translated. Keys: ' . implode(', ', $errors)); throw new AfterTranslateException('Part or all of the text has not been translated. Keys: ' . implode(', ', $errors));
} }
$objectAfterTranslate = $application->make($this->afterTranslateDto->getClassName());
if (!$objectAfterTranslate instanceof \KorElf\TranslateLaravel\Contracts\TranslationCompletedListener) {
throw new AfterTranslateException('The class must implement the \KorElf\TranslateLaravel\Contracts\TranslationCompletedListener interface.');
}
$objectAfterTranslate->onTranslationCompleted($translated, $this->afterTranslateDto->getData());
Cache::forget($this->groupName); Cache::forget($this->groupName);
$this->listener->onTranslationCompleted($translated);
} }
} }
+7 -41
View File
@@ -59,62 +59,28 @@ final class ProcessTranslate implements ShouldQueue, ShouldBeEncrypted, ProcessT
{ {
$param = $this->param; $param = $this->param;
$groupName = $param->getGroupName(); $groupName = $param->getGroupName();
$key = $param->getKey();
$part = $param->getPart();
$translated = Cache::get($groupName, []); $translated = Cache::get($groupName, []);
if (!isset($translated[$param->getKey()])) { if (!isset($translated[$param->getKey()])) {
$translated[$param->getKey()] = []; $translated[$param->getKey()] = [];
} }
$translated[$key][$part] = $this->translate();
Cache::put($groupName, $translated, 86400);
}
private function translate(): string
{
$param = $this->param;
$translate = Translate::service($param->getDriver()); $translate = Translate::service($param->getDriver());
$function = $param->getTextType()->functionName(); $function = $param->getTextType()->functionName();
$key = $param->getKey();
$part = $param->getPart();
$partText = $param->getText(); $translated[$key][$part] = $param->getText();
$originalTexts = $partText->getTextsForTranslation(); if (\trim($param->getText()) !== '') {
$textsForTranslation = []; $translated[$key][$part] = $translate->{$function}(
$originalKeysByTranslationIndex = []; $param->getText(),
foreach ($originalTexts as $originalKey => $text) {
if (\trim($text) === '') {
continue;
}
$translationIndex = count($textsForTranslation);
$textsForTranslation[$translationIndex] = $text;
$originalKeysByTranslationIndex[$translationIndex] = $originalKey;
}
if ($textsForTranslation === []) {
return $partText->getTextsAfterTranslation($originalTexts);
}
$translatedTexts = $translate->{$function}(
$textsForTranslation,
$param->getTargetLanguageCode(), $param->getTargetLanguageCode(),
$param->getSourceLanguageCode() $param->getSourceLanguageCode()
); );
foreach ($translatedTexts as $translationIndex => $translatedText) {
if (!isset($originalKeysByTranslationIndex[$translationIndex])) {
continue;
} }
$originalKey = $originalKeysByTranslationIndex[$translationIndex]; Cache::put($groupName, $translated, 86400);
$originalTexts[$originalKey] = $translatedText;
}
return $partText->getTextsAfterTranslation($originalTexts);
} }
} }
+3 -3
View File
@@ -10,7 +10,7 @@ use Illuminate\Queue\InteractsWithQueue;
use Illuminate\Queue\SerializesModels; use Illuminate\Queue\SerializesModels;
use Illuminate\Support\Facades\Bus; use Illuminate\Support\Facades\Bus;
use Illuminate\Support\Str; use Illuminate\Support\Str;
use KorElf\TranslateLaravel\DTO\AfterTranslateDto; use KorElf\TranslateLaravel\Contracts\TranslationCompletedListener;
use KorElf\TranslateLaravel\DTO\ProcessTranslateDto; use KorElf\TranslateLaravel\DTO\ProcessTranslateDto;
use korElf\TranslateLaravel\DTO\RunTranslateDto; use korElf\TranslateLaravel\DTO\RunTranslateDto;
use KorElf\TranslateLaravel\DTO\Translated; use KorElf\TranslateLaravel\DTO\Translated;
@@ -25,7 +25,7 @@ final class RunTranslate implements ShouldQueue, ShouldBeEncrypted
*/ */
public function __construct( public function __construct(
private readonly RunTranslateDto $runTranslateDto, private readonly RunTranslateDto $runTranslateDto,
private readonly AfterTranslateDto $afterTranslateDto, private readonly TranslationCompletedListener $translationCompletedListener,
) { } ) { }
/** /**
@@ -55,7 +55,7 @@ final class RunTranslate implements ShouldQueue, ShouldBeEncrypted
} }
$translated->add($key, count($texts)); $translated->add($key, count($texts));
} }
$chains[] = new AfterTranslate($groupName, $translated, $this->afterTranslateDto); $chains[] = new AfterTranslate($groupName, $this->translationCompletedListener, $translated);
Bus::chain($chains)->dispatch(); Bus::chain($chains)->dispatch();
} }
-50
View File
@@ -1,50 +0,0 @@
<?php declare(strict_types=1);
namespace KorElf\TranslateLaravel\Translate;
use Illuminate\Contracts\Container\BindingResolutionException;
use Illuminate\Contracts\Foundation\Application;
use Illuminate\Support\Facades\Log;
use KorElf\TranslateLaravel\Contracts\Translate;
use KorElf\TranslateLaravel\DTO\Languages;
final class LogDriver implements Translate
{
/**
* @throws BindingResolutionException
*/
public static function init(Application $app, array $config = []): Translate
{
return new self(
$app->make(Log::class),
);
}
public function __construct(
private readonly Log $log
) { }
public function translateText(array|string $text, string $targetLanguageCode, ?string $sourceLanguageCode = null): string|array
{
$this->log::info('Translate text', ['text' => $text, 'targetLanguageCode' => $targetLanguageCode, 'sourceLanguageCode' => $sourceLanguageCode]);
return $text;
}
public function translateHtml(array|string $text, string $targetLanguageCode, ?string $sourceLanguageCode = null): string|array
{
$this->log::info('Translate html', ['text' => $text, 'targetLanguageCode' => $targetLanguageCode, 'sourceLanguageCode' => $sourceLanguageCode]);
return $text;
}
public function listLanguages(): Languages
{
$languages = new Languages();
$languages->add('ru', 'Russian');
$languages->add('en', 'English');
$languages->add('de', 'Deutsch');
return $languages;
}
}
@@ -1,270 +0,0 @@
<?php declare(strict_types=1);
namespace KorElf\TranslateLaravel\Translate\SplitTextIntoParts;
use DOMNodeList;
use KorElf\TranslateLaravel\DTO\PartText;
use DOMNode;
use DOMDocument;
final class SplitHTMLCommand
{
private array $parts = [];
private int $currentPart = 0;
private int $currentLength = 0;
private string $currentText = '';
private string $currentBeforeText = '';
private string $currentAfterText = '';
public function __construct(private readonly int $maxLength)
{
}
/**
* @param string $html
* @return array<int, PartText>
*/
public function execute(string $html): array
{
$this->currentPart = 0;
$this->currentLength = 0;
$this->currentText = '';
$this->parts = [
0 => new PartText(),
];
libxml_use_internal_errors(true);
$dom = new DOMDocument();
$dom->loadHTML(
mb_convert_encoding('<body>' . $html . '</body>', 'HTML-ENTITIES', 'UTF-8'),
LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD
);
$this->processNodes($dom, $dom->documentElement->childNodes);
if ($this->currentLength > 0) {
$this->addTextToPart();
$this->currentText = '';
$this->currentBeforeText = '';
$this->currentAfterText = '';
$this->currentLength = 0;
$this->currentPart++;
}
libxml_clear_errors();
return $this->parts;
}
private function processNodes(DOMDocument $dom, DOMNodeList $nodes): void
{
/** @var DOMNode $node */
foreach ($nodes as $node) {
if ($node->nodeType === XML_TEXT_NODE || $node->nodeType === XML_CDATA_SECTION_NODE) {
$this->processText($node->textContent);
continue;
}
if ($node->nodeType === XML_ELEMENT_NODE) {
$this->processHtml($dom, $node);
continue;
}
}
}
private function processText(string $text): void
{
$currentLength = mb_strlen($text);
if ($this->currentLength + $currentLength <= $this->maxLength) {
$this->currentText .= $text;
$this->currentLength += $currentLength;
return;
}
if ($this->currentLength > 0) {
$this->addTextToPart();
$this->newPart();
}
if ($currentLength <= $this->maxLength) {
$this->currentText .= $text;
$this->currentLength += $currentLength;
return;
}
$this->splitLongText($text);
}
private function splitLongText(string $text): void
{
// Common expression for searching for sentences.
$sentenceEndings = '/(?<=[.?!])[ \t]+(?=[A-ZА-Я])/u';
//Dividing the text into sentences
$sentences = preg_split($sentenceEndings, $text, -1, PREG_SPLIT_NO_EMPTY);
foreach ($sentences as $sentence) {
$currentLength = mb_strlen($sentence);
if ($this->currentLength > 0) {
$currentLength += 1;
}
// If adding a sentence does not exceed the limit, add it to the current part.
if ($this->currentLength + $currentLength <= $this->maxLength) {
$this->currentText .= (empty($this->currentText) ? '' : ' ') . $sentence;
$this->currentLength += $currentLength;
continue;
}
$this->addTextToPart();
$this->newPart();
$this->currentText = $sentence;
$this->currentLength = $currentLength;
}
}
private function processHtml(DOMDocument $dom, DOMNode $node): void
{
$html = $dom->saveHTML($node);
$currentLength = mb_strlen($html);
if ($this->currentLength + $currentLength <= $this->maxLength) {
$this->currentText .= $html;
$this->currentLength += $currentLength;
return;
}
if ($this->currentLength > 0) {
$this->addTextToPart();
$this->newPart();
}
if ($currentLength <= $this->maxLength) {
$this->currentText .= $html;
$this->currentLength += $currentLength;
return;
}
if ($node->hasChildNodes()) {
$tag = strtolower($node->nodeName);
if ($tag === 'table') {
$this->splitTable($dom, $node);
return;
}
$this->currentBeforeText .= '<' . $tag . $this->attributesToString($node) . '>';
$this->processNodes($dom, $node->childNodes);
$this->currentAfterText .= '</' . $tag . '>';
$this->addTextToPart();
$this->newPart();
return;
}
}
private function splitTable(DOMDocument $dom, DOMNode $node): void
{
if ($this->currentLength > 0) {
$this->addTextToPart();
$this->newPart();
}
$this->currentBeforeText .= '<table' . $this->attributesToString($node) . '>';
if ($node->hasChildNodes()) {
foreach ($node->childNodes as $childNode) {
if ($childNode->nodeType === XML_TEXT_NODE || $childNode->nodeType === XML_CDATA_SECTION_NODE) {
continue;
}
if ($childNode->nodeType !== XML_ELEMENT_NODE) {
continue;
}
$this->processTableSection($dom, $childNode);
}
}
$this->currentAfterText .= '</table>';
$this->addTextToPart();
$this->newPart();
}
private function processTableSection(DOMDocument $dom, DOMNode $node): void
{
$tag = strtolower($node->nodeName);
if ($tag === 'tr') {
$this->processTableTr($dom, $node);
}
if (in_array($tag, ['thead', 'tbody', 'tfoot'], true)) {
$this->currentBeforeText .= '<' . $tag . $this->attributesToString($node) . '>';
foreach ($node->childNodes as $childNode) {
$tagNode = strtolower($childNode->nodeName);
if ($tagNode === 'tr') {
$this->processTableTr($dom, $childNode);
}
}
$this->currentAfterText .= '</' . $tag . '>';
}
}
private function processTableTr(DOMDocument $dom, DOMNode $node): void
{
$this->currentBeforeText .= '<tr' . $this->attributesToString($node) . '>';
foreach ($node->childNodes as $childNode) {
$tagNode = strtolower($childNode->nodeName);
if ($tagNode === 'td') {
$this->processTableTd($dom, $childNode);
}
}
$this->currentAfterText .= '</tr>';
$this->addTextToPart();
}
private function processTableTd(DOMDocument $dom, DOMNode $node): void
{
$this->currentBeforeText .= '<td' . $this->attributesToString($node) . '>';
$this->processNodes($dom, $node->childNodes);
$this->currentAfterText .= '</td>';
$this->addTextToPart();
}
private function attributesToString(DOMNode $node): string
{
if (!$node->hasAttributes()) {
return '';
}
$parts = [];
foreach ($node->attributes as $attr) {
$parts[] = sprintf(
' %s="%s"',
$attr->nodeName,
htmlspecialchars($attr->nodeValue ?? '', ENT_QUOTES | ENT_SUBSTITUTE, 'UTF-8')
);
}
return implode('', $parts);
}
private function addTextToPart(): void
{
$this->parts[$this->currentPart]->add($this->currentText, $this->currentBeforeText, $this->currentAfterText);
$this->currentText = '';
$this->currentBeforeText = '';
$this->currentAfterText = '';
}
private function newPart(): void
{
$this->currentText = '';
$this->currentBeforeText = '';
$this->currentAfterText = '';
$this->currentLength = 0;
$this->currentPart++;
$this->parts[$this->currentPart] = new PartText();
}
}
@@ -1,50 +0,0 @@
<?php declare(strict_types=1);
namespace KorElf\TranslateLaravel\Translate\SplitTextIntoParts;
use KorElf\TranslateLaravel\DTO\PartText;
final readonly class SplitTextCommand
{
/**
* @param string $text
* @param int $maxLength
* @return array<int, PartText>
*/
public function execute(string $text, int $maxLength): array
{
// Common expression for searching for sentences.
$sentenceEndings = '/(?<=[.?!])[ \t]+(?=[A-ZА-Я])/u';
//Dividing the text into sentences
$sentences = preg_split($sentenceEndings, $text, -1, PREG_SPLIT_NO_EMPTY);
$parts = [];
$currentPart = '';
foreach ($sentences as $sentence) {
$part = new PartText();
// If adding a sentence does not exceed the limit, add it to the current part.
if (mb_strlen($currentPart . ' ' . $sentence) <= $maxLength) {
$currentPart .= (empty($currentPart) ? '' : ' ') . $sentence;
continue;
}
// Otherwise, save the current part and start a new one.
if (!empty($currentPart)) {
$part->add($currentPart);
$parts[] = $part;
}
$currentPart = $sentence;
}
if (!empty($currentPart)) {
$part = new PartText();
$part->add($currentPart);
$parts[] = $part;
}
return $parts;
}
}
+88 -30
View File
@@ -2,56 +2,114 @@
namespace KorElf\TranslateLaravel\Translate; namespace KorElf\TranslateLaravel\Translate;
use KorElf\TranslateLaravel\DTO\PartText;
use KorElf\TranslateLaravel\Enums\TextType; use KorElf\TranslateLaravel\Enums\TextType;
use KorElf\TranslateLaravel\Facades\Translate; use KorElf\TranslateLaravel\Facades\Translate;
use KorElf\TranslateLaravel\Translate\SplitTextIntoParts\SplitHTMLCommand; use DOMDocument;
use KorElf\TranslateLaravel\Translate\SplitTextIntoParts\SplitTextCommand;
final readonly class SplitTextIntoPartsCommand final readonly class SplitTextIntoPartsCommand
{ {
/**
* @param string $text
* @param TextType $textType
* @param string|null $driver
* @return array<int, PartText>
*/
public function execute(string $text, TextType $textType, ?string $driver): array public function execute(string $text, TextType $textType, ?string $driver): array
{ {
$maxLength = Translate::getLimit($driver); $maxLength = Translate::getLimit($driver);
$maxLength = $maxLength['max_symbols'] ?? null; $maxLength = $maxLength['max_symbols'] ?? null;
if ($maxLength === null || $maxLength > mb_strlen($text)) { if ($maxLength === null || $maxLength > mb_strlen($text)) {
$part = new PartText(); return [$text];
$part->add($text);
return [$part];
} }
if ($textType === TextType::Html) { if ($textType === TextType::Html) {
return $this->splitHtml($text, $maxLength); return $this->splitHtmlText($text, $maxLength);
} }
return $this->splitText($text, $maxLength); return $this->splitTextBySentences($text, $maxLength);
} }
/** private function splitTextBySentences(string $text, int $maxLength): array
* @param string $text
* @param int $maxLength
* @return array<int, PartText>
*/
private function splitText(string $text, int $maxLength): array
{ {
$splitTextCommand = new SplitTextCommand(); // Common expression for searching for sentences.
return $splitTextCommand->execute($text, $maxLength); $sentenceEndings = '/(?<=[.?!])\s+(?=[A-ZА-Я])/u';
//Dividing the text into sentences
$sentences = preg_split($sentenceEndings, $text, -1, PREG_SPLIT_NO_EMPTY);
$parts = [];
$currentPart = '';
foreach ($sentences as $sentence) {
// If adding a sentence does not exceed the limit, add it to the current part.
if (mb_strlen($currentPart . ' ' . $sentence) <= $maxLength) {
$currentPart .= (empty($currentPart) ? '' : ' ') . $sentence;
continue;
} }
/** // Otherwise, save the current part and start a new one.
* @param string $html if (!empty($currentPart)) {
* @param int $maxLength $parts[] = $currentPart;
* @return array<int, PartText> $currentPart = '';
*/ }
private function splitHtml(string $html, int $maxLength): array $currentPart = $sentence;
}
if (!empty($currentPart)) {
$parts[] = $currentPart;
}
return $parts;
}
private function splitHtmlText(string $html, int $maxLength): array
{ {
$splitHTMLCommand = new SplitHTMLCommand($maxLength); libxml_use_internal_errors(true);
return $splitHTMLCommand->execute($html);
$dom = new DOMDocument();
$dom->loadHTML(mb_convert_encoding('<body>' . $html . '</body>', 'HTML-ENTITIES', 'UTF-8'), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
$parts = [];
$currentPart = new DOMDocument();
$currentLength = 0;
// Iterate through all child nodes of the root element
foreach ($dom->documentElement->childNodes as $node) {
if ($node->nodeType === XML_TEXT_NODE || $node->nodeType === XML_CDATA_SECTION_NODE) {
// Dividing text nodes into sentences
$sentences = preg_split('/(?<=[.?!])\s+(?=[A-ZА-Я])/', $node->textContent, -1, PREG_SPLIT_NO_EMPTY);
foreach ($sentences as $sentence) {
$sentenceHtml = htmlspecialchars($sentence);
if ($currentLength + mb_strlen($sentenceHtml) > $maxLength) {
$parts[] = $currentPart->saveHTML();
// Start new part
$currentPart = new DOMDocument();
$currentLength = 0;
}
$textNode = $currentPart->createTextNode($sentence . ' ');
$currentPart->appendChild($textNode);
$currentLength += mb_strlen($sentenceHtml);
}
continue;
}
$nodeHtml = $dom->saveHTML($node);
if ($currentLength + mb_strlen($nodeHtml) > $maxLength) {
$parts[] = $currentPart->saveHTML();
// Start new part
$currentPart = new DOMDocument();
$currentLength = mb_strlen($nodeHtml);
}
$currentPart->appendChild($currentPart->importNode($node, true));
$currentLength += mb_strlen($nodeHtml);
}
if ($currentPart->hasChildNodes()) {
$parts[] = $currentPart->saveHTML();
}
libxml_clear_errors();
return $parts;
} }
} }
+4 -10
View File
@@ -4,10 +4,10 @@ namespace KorElf\TranslateLaravel\Translate;
use Illuminate\Contracts\Foundation\Application; use Illuminate\Contracts\Foundation\Application;
use Illuminate\Foundation\Bus\PendingDispatch; use Illuminate\Foundation\Bus\PendingDispatch;
use KorElf\TranslateLaravel\Contracts\CallbackAfterTranslated;
use KorElf\TranslateLaravel\Contracts\Translate;
use InvalidArgumentException; use InvalidArgumentException;
use KorElf\TranslateLaravel\Contracts\TranslationCompletedListener; use KorElf\TranslateLaravel\Contracts\TranslationCompletedListener;
use KorElf\TranslateLaravel\Contracts\Translate;
use KorElf\TranslateLaravel\DTO\AfterTranslateDto;
use KorElf\TranslateLaravel\DTO\RunTranslateDto; use KorElf\TranslateLaravel\DTO\RunTranslateDto;
use KorElf\TranslateLaravel\Jobs\RunTranslate; use KorElf\TranslateLaravel\Jobs\RunTranslate;
@@ -68,15 +68,9 @@ final class TranslateManager
/** /**
* Run through queues. * Run through queues.
*/ */
public function runJob(RunTranslateDto $params, AfterTranslateDto $afterTranslateDto): PendingDispatch public function runJob(RunTranslateDto $params, TranslationCompletedListener $completedListener): PendingDispatch
{ {
$className = $afterTranslateDto->getClassName(); return RunTranslate::dispatch($params, $completedListener);
if (!is_subclass_of($className, TranslationCompletedListener::class)) {
throw new InvalidArgumentException('The class ' . $className . ' must implement the \KorElf\TranslateLaravel\Contracts\Translate\TranslationCompletedListener interface.');
}
unset($className);
return RunTranslate::dispatch($params, $afterTranslateDto);
} }
public function getLimit(?string $driver = null): array public function getLimit(?string $driver = null): array