From c7f0b94da971084718251b2f83f1974ad626061d Mon Sep 17 00:00:00 2001
From: Jan Pecha <janpecha@email.cz>
Date: Thu, 18 Apr 2024 08:45:10 +0200
Subject: [PATCH] Processors: added EntityProcessor

---
 src/Processors/EntityProcessor.php    | 60 +++++++++++++++++++++++++++
 src/Processors/TagProcessor.php       |  2 +-
 tests/Translator/EntityProcessor.phpt | 30 ++++++++++++++
 tests/Translator/HtmlTranslator.phpt  |  7 ++++
 tests/Translator/TextTranslator.phpt  |  7 ++++
 5 files changed, 105 insertions(+), 1 deletion(-)
 create mode 100644 src/Processors/EntityProcessor.php
 create mode 100644 tests/Translator/EntityProcessor.phpt

diff --git a/src/Processors/EntityProcessor.php b/src/Processors/EntityProcessor.php
new file mode 100644
index 0000000..4d0b318
--- /dev/null
+++ b/src/Processors/EntityProcessor.php
@@ -0,0 +1,60 @@
+<?php
+
+	namespace Inteve\Translator\Processors;
+
+	use Inteve\Translator\Ast;
+	use Inteve\Translator\MessageProcessor;
+
+
+	class EntityProcessor implements MessageProcessor
+	{
+		/** @var MessageProcessor|NULL */
+		private $textProcessor;
+
+
+		public function __construct(MessageProcessor $textProcessor = NULL)
+		{
+			$this->textProcessor = $textProcessor;
+		}
+
+
+		public function processMessage($messageText)
+		{
+			if ($this->textProcessor === NULL) {
+				return new Ast\MessageText([$this->decodeEntities($messageText)]);
+			}
+
+			$messageText = $this->textProcessor->processMessage($messageText);
+			$parts = [];
+
+			foreach ($messageText->getChildren() as $child) {
+				if (is_string($child)) {
+					if ($child !== '') {
+						$parts[] = $this->decodeEntities($child);
+					}
+
+				} else {
+					$parts[] = $child;
+				}
+			}
+
+			return new Ast\MessageText($parts);
+		}
+
+
+		/**
+		 * @param  non-empty-string $messageText
+		 * @return non-empty-string
+		 */
+		private function decodeEntities($messageText)
+		{
+			$messageText = html_entity_decode($messageText, ENT_QUOTES | ENT_SUBSTITUTE | ENT_XHTML, 'UTF-8');
+			// ENT_XHTML - allows 'all' entities (like ENT_XML) + &apos;
+
+			if ($messageText === '') {
+				throw new \Inteve\Translator\InvalidStateException('Decoded string is empty.');
+			}
+
+			return $messageText;
+		}
+	}
diff --git a/src/Processors/TagProcessor.php b/src/Processors/TagProcessor.php
index eb23cd2..5350389 100644
--- a/src/Processors/TagProcessor.php
+++ b/src/Processors/TagProcessor.php
@@ -23,7 +23,7 @@ public function __construct(
 			array $emptyElements = ['br']
 		)
 		{
-			$this->textPartsProcessor = $textPartsProcessor !== NULL ? $textPartsProcessor : new ParametersProcessor;
+			$this->textPartsProcessor = $textPartsProcessor !== NULL ? $textPartsProcessor : new EntityProcessor(new ParametersProcessor);
 			$this->emptyElements = [];
 
 			foreach ($emptyElements as $emptyElement) {
diff --git a/tests/Translator/EntityProcessor.phpt b/tests/Translator/EntityProcessor.phpt
new file mode 100644
index 0000000..690609a
--- /dev/null
+++ b/tests/Translator/EntityProcessor.phpt
@@ -0,0 +1,30 @@
+<?php
+
+use Inteve\Translator\Ast;
+use Inteve\Translator\Processors\EntityProcessor;
+use Inteve\Translator\Processors\ParametersProcessor;
+use Tester\Assert;
+
+require __DIR__ . '/../bootstrap.php';
+
+test('Only text', function () {
+	$processor = new EntityProcessor;
+	Assert::equal(new Ast\MessageText(['lorem ipsum']), $processor->processMessage('lorem ipsum'));
+});
+
+
+test('Named & codeded entity', function () {
+	$processor = new EntityProcessor;
+	Assert::equal(new Ast\MessageText([
+		"lorem \xC2\xA0 \" ' ' \r \x20",
+	]), $processor->processMessage('lorem &nbsp; &#34; &#39; &apos; &#13; &#32;'));
+});
+
+
+test('With parameters', function () {
+	$processor = new EntityProcessor(new ParametersProcessor);
+	Assert::equal(new Ast\MessageText([
+		'{$test}',
+		new Ast\Parameter('name'),
+	]), $processor->processMessage('&#x7B;&#x24;test}{$name}'));
+});
diff --git a/tests/Translator/HtmlTranslator.phpt b/tests/Translator/HtmlTranslator.phpt
index daa6ef0..1d3aa12 100644
--- a/tests/Translator/HtmlTranslator.phpt
+++ b/tests/Translator/HtmlTranslator.phpt
@@ -19,6 +19,7 @@ function createTranslator($lang)
 					'homepage.hello' => 'Hello, <b>{$name}</b>!',
 					'homepage.unknow' => 'Hello, <muted>lorem <i>{$name|lower}!</i></muted>',
 					'homepage.br' => 'Hello<br>friends',
+					'homepage.entity' => 'A&nbsp;entity',
 				]),
 			]),
 			new Translator\Processors\TagProcessor
@@ -61,6 +62,12 @@ test('MessageId', function () {
 });
 
 
+test('Entities', function () {
+	$translator = createTranslator('en');
+	Assert::same("A\xC2\xA0entity", (string) $translator->translate('homepage.entity'));
+});
+
+
 test('Prefix', function () {
 	$translator = createTranslator('en')->prefix('homepage');
 	Assert::same('Hello, <b>John</b>!', (string) $translator->translate('hello', ['name' => 'John']));
diff --git a/tests/Translator/TextTranslator.phpt b/tests/Translator/TextTranslator.phpt
index cd9fc29..746f620 100644
--- a/tests/Translator/TextTranslator.phpt
+++ b/tests/Translator/TextTranslator.phpt
@@ -23,6 +23,7 @@ function createTranslator($lang)
 					'homepage.hello' => 'Hello!',
 					'homepage.hello2' => 'Hello, <b>{$name}</b>!',
 					'homepage.br' => 'Hello<br>friends',
+					'homepage.entity' => 'A&nbsp;entity',
 				]),
 			]),
 			new Translator\Processors\TagProcessor
@@ -76,6 +77,12 @@ test('MessageId', function () {
 });
 
 
+test('Entities', function () {
+	$translator = createTranslator('en');
+	Assert::same("A\xC2\xA0entity", (string) $translator->translate('homepage.entity'));
+});
+
+
 test('Prefix', function () {
 	$translator = createTranslator('en')->prefix('homepage');
 	Assert::same('Hello, John!', $translator->translate('hello2', ['name' => 'John']));