Merge dockerfiles and fix utf8 encoding for aspell and hunspell

.github/workflows/run-tests.yml

@@ -27,7 +27,7 @@ jobs:
         run: make pull
 
       - name: Define folders which changes within requires docker build step
-        run: echo "FILES_REQUIRING_DOCKER_BUILD=php${{ matrix.php }}/Dockerfile" >> $GITHUB_ENV
+        run: echo "FILES_REQUIRING_DOCKER_BUILD=php/Dockerfile" >> $GITHUB_ENV
 
       - name: Build docker images
         run: make build
@@ -48,6 +48,8 @@ jobs:
         run: |
           export WITH_COVERAGE=$(if [[ ("${{ matrix.php }}" = "8.0") && ("${{ matrix.stability }}" = "--prefer-stable") ]]; then echo "true"; else echo "false"; fi)
           echo "WITH_COVERAGE=${WITH_COVERAGE}" >> $GITHUB_ENV
+          export XDEBUG_VERSION=$(if [[ ("${{ matrix.php }}" = "7.2") then echo "2.9.3"; else echo "3.0.4"; fi)
+          echo "XDEBUG_VERSION=${XDEBUG_VERSION}" >> $GITHUB_ENV
           make vendor
           make tests
       

Makefile

@@ -1,19 +1,20 @@
 DOCKER_COMPOSE 	?= docker-compose
-EXEC_PHP      	= $(DOCKER_COMPOSE) run --rm -T php$(PHP_VERSION)
+EXEC_PHP      	= $(DOCKER_COMPOSE) run --rm -T php
 PHP_VERSION     ?= 8.0
+XDEBUG_VERSION  ?= 3.0.4
 DEPS_STRATEGY   ?= --prefer-stable
 COMPOSER      	= $(EXEC_PHP) composer
 WITH_COVERAGE   ?= "FALSE"
 EXAMPLES_DIR   ?= "examples"
 
 pull:
-	@$(DOCKER_COMPOSE) pull languagetools jamspell php$(PHP_VERSION)
+	@$(DOCKER_COMPOSE) pull languagetools jamspell php
 
 build:
-	$(DOCKER_COMPOSE) build --no-cache php$(PHP_VERSION)
+	$(DOCKER_COMPOSE) build --no-cache php
 
 push:
-	$(DOCKER_COMPOSE) push php$(PHP_VERSION)
+	$(DOCKER_COMPOSE) push php
 
 kill:
 	$(DOCKER_COMPOSE) kill

docker-compose.yml

@@ -1,7 +1,13 @@
 version: '3.4'
 
-x-php-common-conf:
-  - &php-common-conf
+services:
+  php:
+    image: tigitz/phpspellchecker:${PHP_VERSION:-8.0}
+    build:
+      context: docker/php
+      args:
+        PHP_VERSION: ${PHP_VERSION:-8.0}
+        XDEBUG_VERSION: ${XDEBUG_VERSION:-3.0.4}
     volumes:
       - .:/usr/src/myapp
       - ./cache:/root/composer/cache
@@ -12,19 +18,6 @@ x-php-common-conf:
       - languagetools
       - jamspell
 
-services:
-  php7.2:
-    image: tigitz/phpspellchecker:7.2
-    build:
-      context: docker/php7.2
-    <<: *php-common-conf
-
-  php8.0:
-    image: tigitz/phpspellchecker:8.0
-    build:
-      context: docker/php8.0
-    <<: *php-common-conf
-
   jamspell:
     image: tigitz/jamspell
 

docker/php8.0/Dockerfile → docker/php/Dockerfile

@@ -1,10 +1,12 @@
-FROM php:8.0-cli-buster
+ARG PHP_VERSION
+
+FROM php:${PHP_VERSION}-cli-buster
 
 ENV XDEBUG_MODE=coverage
 RUN echo "memory_limit=-1" > "$PHP_INI_DIR/conf.d/memory-limit.ini" \
  && echo "date.timezone=${PHP_TIMEZONE:-UTC}" > "$PHP_INI_DIR/conf.d/date_timezone.ini"
 
-ARG XDEBUG_VERSION=3.0.2
+ARG XDEBUG_VERSION
 
 RUN apt-get update \
     && apt install -y \
@@ -16,10 +18,13 @@ RUN apt-get update \
      libzip-dev \
      ispell \
      iamerican \
+     irussian \
      hunspell \
      hunspell-en-us \
+     hunspell-ru \
      aspell \
      aspell-en \
+     aspell-ru \
      libpspell-dev \
     && pecl install xdebug-${XDEBUG_VERSION} \
     && docker-php-ext-configure pspell \

src/Spellchecker/Aspell.php

@@ -27,7 +27,8 @@ public function check(string $text, array $languages = [], array $context = []):
     {
         Assert::maxCount($languages, 1, 'Aspell spellchecker doesn\'t support multiple languages check');
 
-        $cmd = $this->binaryPath->addArg('-a');
+        $cmd = $this->binaryPath->addArgs(['--encoding', 'utf-8']);
+        $cmd = $cmd->addArg('-a');
 
         if (!empty($languages)) {
             $cmd = $cmd->addArg('--lang=' . implode(',', $languages));

tests/Fixtures/Hunspell/check.txt

@@ -1,5 +1,6 @@
 @(#) International Ispell Version 3.2.06 (but really Hunspell 1.7.0)
 & Tigr 6 0: Ti gr, Ti-gr, Tiger, Trig, Tier, Tigris
+# страх 21
 
 & theforests 6 3: the forests, the-forests, therefore, afforests, forests, forefathers
 

tests/Spellchecker/AspellTest.php

@@ -40,14 +40,17 @@ public function testSpellcheckFromRealBinaries(): void
     /**
      * @group integration
      */
-    public function testGetSupportedLanguagesFromRealBinaries(): void
+    public function testSpellcheckFromRealBinariesUTF8(): void
     {
-        $this->assertWorkingSupportedLanguages(self::realBinaryPath());
+        $this->assertWorkingSpellcheckRUText(self::realBinaryPath());
     }
 
-    public function getTextInput(): string
+    /**
+     * @group integration
+     */
+    public function testGetSupportedLanguagesFromRealBinaries(): void
     {
-        return TextTest::CONTENT_STUB;
+        $this->assertWorkingSupportedLanguages(self::realBinaryPath());
     }
 
     public function getFakeDicts(): array
@@ -77,7 +80,7 @@ private function assertWorkingSpellcheck(string $binaries): void
         /** @var Misspelling[] $misspellings */
         $misspellings = iterator_to_array(
             $aspell->check(
-                $this->getTextInput(),
+                TextTest::CONTENT_STUB,
                 ['en_US'],
                 ['ctx']
             )
@@ -95,4 +98,24 @@ private function assertWorkingSpellcheck(string $binaries): void
         $this->assertSame(2, $misspellings[1]->getLineNumber());
         $this->assertNotEmpty($misspellings[1]->getSuggestions());
     }
+
+    /**
+     * @param array|string $binaries
+     */
+    private function assertWorkingSpellcheckRUText($binaries): void
+    {
+        $aspell = new Aspell(new CommandLine($binaries));
+        /** @var Misspelling[] $misspellings */
+        $misspellings = iterator_to_array($aspell->check(TextTest::CONTENT_STUB_RU, ['ru'], ['ctx']));
+
+        $this->assertSame(['ctx'], $misspellings[0]->getContext());
+        $this->assertSame('граматических', $misspellings[0]->getWord());
+        $this->assertSame(1, $misspellings[0]->getLineNumber());
+        $this->assertSame(54, $misspellings[0]->getOffset());
+
+        $this->assertSame(['ctx'], $misspellings[1]->getContext());
+        $this->assertSame('англиских', $misspellings[1]->getWord());
+        $this->assertSame(1, $misspellings[1]->getLineNumber());
+        $this->assertSame(94, $misspellings[1]->getOffset());
+    }
 }

tests/Spellchecker/HunspellTest.php

@@ -13,10 +13,10 @@ class HunspellTest extends TestCase
 {
     private const FAKE_BINARIES_PATH = [PHP_BINARY, __DIR__ . '/../Fixtures/Hunspell/bin/hunspell.php'];
 
-//    public function testSpellcheckFromFakeBinaries(): void
-//    {
-//        $this->assertWorkingSpellcheck(self::FAKE_BINARIES_PATH);
-//    }
+    public function testSpellcheckFromFakeBinaries(): void
+    {
+        $this->assertWorkingSpellcheckENText(self::FAKE_BINARIES_PATH, TextTest::CONTENT_STUB, ['en_US']);
+    }
 
     public function testGetSupportedLanguagesFromFakeBinaries(): void
     {
@@ -44,20 +44,23 @@ public function testSpellcheckFromRealBinariesLanguage(): void
      */
     public function testSpellcheckFromRealBinaries(): void
     {
-        $this->assertWorkingSpellcheck(self::realBinaryPath());
+        $this->assertWorkingSpellcheckENText(self::realBinaryPath(), TextTest::CONTENT_STUB, ['en_US']);
     }
 
     /**
      * @group integration
      */
-    public function testGetSupportedLanguagesFromRealBinaries(): void
+    public function testSpellcheckFromRealBinariesUTF8(): void
     {
-        $this->assertWorkingSupportedLanguages(self::realBinaryPath());
+        $this->assertWorkingSpellcheckRUText(self::realBinaryPath(), TextTest::CONTENT_STUB_RU, ['ru_RU']);
     }
 
-    public function getTextInput(): string
+    /**
+     * @group integration
+     */
+    public function testGetSupportedLanguagesFromRealBinaries(): void
     {
-        return TextTest::CONTENT_STUB;
+        $this->assertWorkingSupportedLanguages(self::realBinaryPath());
     }
 
     public function getFakeDicts(): array
@@ -87,11 +90,11 @@ public static function realBinaryPath(): string
     /**
      * @param array|string $binaries
      */
-    private function assertWorkingSpellcheck($binaries): void
+    private function assertWorkingSpellcheckENText($binaries, string $textInput, array $locales): void
     {
         $hunspell = new Hunspell(new CommandLine($binaries));
         /** @var Misspelling[] $misspellings */
-        $misspellings = iterator_to_array($hunspell->check($this->getTextInput(), ['en_US'], ['ctx']));
+        $misspellings = iterator_to_array($hunspell->check($textInput, $locales, ['ctx']));
 
         $this->assertSame(['ctx'], $misspellings[0]->getContext());
         $this->assertSame('Tigr', $misspellings[0]->getWord());
@@ -101,7 +104,27 @@ private function assertWorkingSpellcheck($binaries): void
 
         $this->assertSame(['ctx'], $misspellings[1]->getContext());
         $this->assertSame('страх', $misspellings[1]->getWord());
-//        $this->assertSame(3, $misspellings[1]->getOffset());
-//        $this->assertSame(2, $misspellings[1]->getLineNumber());
+        $this->assertSame(21, $misspellings[1]->getOffset());
+        $this->assertSame(1, $misspellings[1]->getLineNumber());
+    }
+
+    /**
+     * @param array|string $binaries
+     */
+    private function assertWorkingSpellcheckRUText($binaries, string $textInput, array $locales): void
+    {
+        $hunspell = new Hunspell(new CommandLine($binaries));
+        /** @var Misspelling[] $misspellings */
+        $misspellings = iterator_to_array($hunspell->check($textInput, $locales, ['ctx']));
+
+        $this->assertSame(['ctx'], $misspellings[0]->getContext());
+        $this->assertSame('граматических', $misspellings[0]->getWord());
+        $this->assertSame(1, $misspellings[0]->getLineNumber());
+        $this->assertSame(54, $misspellings[0]->getOffset());
+
+        $this->assertSame(['ctx'], $misspellings[1]->getContext());
+        $this->assertSame('англиских', $misspellings[1]->getWord());
+        $this->assertSame(1, $misspellings[1]->getLineNumber());
+        $this->assertSame(94, $misspellings[1]->getOffset());
     }
 }

tests/Spellchecker/IspellTest.php

@@ -38,6 +38,14 @@ public function testSpellcheckFromRealBinaries(): void
         $this->assertWorkingSpellcheck(self::realBinaryPath());
     }
 
+//    /**
+//     * @group integration
+//     */
+//    public function testSpellcheckFromRealBinariesUTF8(): void
+//    {
+//        $this->assertWorkingSpellcheckRUText(self::realBinaryPath());
+//    }
+
     /**
      * @group integration
      */
@@ -109,4 +117,24 @@ private function assertWorkingSpellcheck(string $binaries): void
         $this->assertSame($misspellings[1]->getLineNumber(), 2);
         $this->assertNotEmpty($misspellings[1]->getSuggestions());
     }
+
+//    /**
+//     * @param array|string $binaries
+//     */
+//    private function assertWorkingSpellcheckRUText($binaries): void
+//    {
+//        $ispell = new Ispell(new CommandLine($binaries));
+//        /** @var Misspelling[] $misspellings */
+//        $misspellings = iterator_to_array($ispell->check(TextTest::CONTENT_STUB_RU, ['russian'], ['ctx']));
+//
+//        $this->assertSame(['ctx'], $misspellings[0]->getContext());
+//        $this->assertSame('граматических', $misspellings[0]->getWord());
+//        $this->assertSame(1, $misspellings[0]->getLineNumber());
+//        $this->assertSame(54, $misspellings[0]->getOffset());
+//
+//        $this->assertSame(['ctx'], $misspellings[1]->getContext());
+//        $this->assertSame('англиских', $misspellings[1]->getWord());
+//        $this->assertSame(1, $misspellings[1]->getLineNumber());
+//        $this->assertSame(94, $misspellings[1]->getOffset());
+//    }
 }

tests/Spellchecker/LanguageToolTest.php

@@ -93,7 +93,7 @@ public function testSpellcheckMultiBytesStringFromRealAPI(): void
     {
         $misspellings = iterator_to_array(
             (new LanguageTool(new LanguageToolApiClient(self::realAPIEndpoint())))->check(
-                TextTest::CONTENT_STUB_MULTIBYTE,
+                TextTest::CONTENT_STUB_JP,
                 ['ja-JP'],
                 ['ctx' => 'ctx']
             )

tests/TextTest.php

@@ -15,12 +15,19 @@ class TextTest extends TestCase
 CCould frame thy fearful symmetry?
 TEXT;
 
-    public const CONTENT_STUB_MULTIBYTE = <<<TEXT
+    public const CONTENT_STUB_JP = <<<TEXT
 さよなら解決なる
 
 さよなら
 
 解決なる
+TEXT;
+
+    public const CONTENT_STUB_RU = <<<TEXT
+Используйте этот инструмент для обнаружения опечаток, граматических и стилистических ошибок в англиских текстах.
+Пример (с ошибками:
+наведите мышь на подсвечиваемые слова, чтобы просмотреть опсание и, при наличии, варианта исправления ошибки.
+Для проверки собственного текста, щёлкните на текстовое поле , вставьте свой текст и нажмите на кнопку "Отправить".
 TEXT;
 
     public function testContextOverridingMerge(): void