|
| 1 | +<?php |
| 2 | + |
| 3 | +declare(strict_types=1); |
| 4 | + |
| 5 | +/** |
| 6 | + * Derafu: XML - Library for XML manipulation. |
| 7 | + * |
| 8 | + * Copyright (c) 2026 Esteban De La Fuente Rubio / Derafu <https://www.derafu.dev> |
| 9 | + * Licensed under the MIT License. |
| 10 | + * See LICENSE file for more details. |
| 11 | + */ |
| 12 | + |
| 13 | +namespace Derafu\TestsXml; |
| 14 | + |
| 15 | +use Derafu\Xml\Service\XmlDecoder; |
| 16 | +use Derafu\Xml\XmlDocument; |
| 17 | +use Derafu\Xml\XmlHelper; |
| 18 | +use Derafu\Xml\XPathQuery; |
| 19 | +use PHPUnit\Framework\Attributes\CoversClass; |
| 20 | +use PHPUnit\Framework\TestCase; |
| 21 | + |
| 22 | +#[CoversClass(XmlDocument::class)] |
| 23 | +#[CoversClass(XmlDecoder::class)] |
| 24 | +#[CoversClass(XmlHelper::class)] |
| 25 | +#[CoversClass(XPathQuery::class)] |
| 26 | +class XmlSerializationTest extends TestCase |
| 27 | +{ |
| 28 | + /** |
| 29 | + * ISO-8859-1 encoded XML bytes loaded from the comprehensive fixture. |
| 30 | + */ |
| 31 | + private string $iso88591Xml; |
| 32 | + |
| 33 | + /** |
| 34 | + * Expected values after decoding the fixture. Always UTF-8 PHP strings, |
| 35 | + * regardless of the source encoding. |
| 36 | + */ |
| 37 | + private array $expectedValues = [ |
| 38 | + 'vocales_min' => 'á é í ó ú à è ì ò ù', |
| 39 | + 'vocales_may' => 'Á É Í Ó Ú À È Ì Ò Ù', |
| 40 | + 'enie' => 'ñ Ñ', |
| 41 | + 'dieresis' => 'ü Ü ö Ö', |
| 42 | + 'puntuacion' => '¿Hola? ¡Mundo!', |
| 43 | + 'simbolos' => '© ® ° ½ ¼ ¾', |
| 44 | + 'frase' => 'Fabricación de Ñoños en Güemes', |
| 45 | + ]; |
| 46 | + |
| 47 | + protected function setUp(): void |
| 48 | + { |
| 49 | + $this->iso88591Xml = require __DIR__ . '/../fixtures/encoding/iso88591-comprehensive.php'; |
| 50 | + } |
| 51 | + |
| 52 | + // ------------------------------------------------------------------------- |
| 53 | + // Case 1: UTF-8 document serialize/unserialize |
| 54 | + // ------------------------------------------------------------------------- |
| 55 | + |
| 56 | + /** |
| 57 | + * Verifica que un XmlDocument UTF-8 se puede serializar y deserializar |
| 58 | + * preservando la codificación y el contenido. |
| 59 | + */ |
| 60 | + public function testUtf8DocumentPreservesContentAfterSerializeUnserialize(): void |
| 61 | + { |
| 62 | + $doc = new XmlDocument(); |
| 63 | + $doc->loadXml( |
| 64 | + '<?xml version="1.0" encoding="UTF-8"?><root><element>Valor</element></root>' |
| 65 | + ); |
| 66 | + |
| 67 | + /** @var XmlDocument $restored */ |
| 68 | + $restored = unserialize(serialize($doc)); |
| 69 | + |
| 70 | + $this->assertInstanceOf(XmlDocument::class, $restored); |
| 71 | + $this->assertSame('UTF-8', $restored->getEncoding()); |
| 72 | + $this->assertSame('root', $restored->getName()); |
| 73 | + $this->assertSame('Valor', $restored->query('//element')); |
| 74 | + } |
| 75 | + |
| 76 | + /** |
| 77 | + * Verifica que el XML byte-a-byte es idéntico antes y después de un ciclo |
| 78 | + * serialize/unserialize en un documento UTF-8 con caracteres especiales. |
| 79 | + * |
| 80 | + * - ñ (U+00F1) debe estar como secuencia UTF-8 (0xC3 0xB1), nunca como |
| 81 | + * byte ISO-8859-1 (0xF1). |
| 82 | + * - á (U+00E1) debe estar como secuencia UTF-8 (0xC3 0xA1), nunca como |
| 83 | + * byte ISO-8859-1 (0xE1). |
| 84 | + */ |
| 85 | + public function testUtf8DocumentWithSpecialCharactersPreservesBytes(): void |
| 86 | + { |
| 87 | + $doc = new XmlDocument(); |
| 88 | + $doc->loadXml( |
| 89 | + '<?xml version="1.0" encoding="UTF-8"?><root><nombre>áéíóú ñÑ Güemes</nombre></root>' |
| 90 | + ); |
| 91 | + |
| 92 | + $xmlBefore = $doc->saveXml(); |
| 93 | + |
| 94 | + /** @var XmlDocument $restored */ |
| 95 | + $restored = unserialize(serialize($doc)); |
| 96 | + $xmlAfter = $restored->saveXml(); |
| 97 | + |
| 98 | + $this->assertSame($xmlBefore, $xmlAfter); |
| 99 | + |
| 100 | + // ñ como secuencia UTF-8 (0xC3 0xB1), nunca como byte ISO-8859-1 (0xF1). |
| 101 | + $this->assertStringContainsString("\xC3\xB1", $xmlAfter); |
| 102 | + $this->assertStringNotContainsString("\xF1", $xmlAfter); |
| 103 | + |
| 104 | + // á como secuencia UTF-8 (0xC3 0xA1), nunca como byte ISO-8859-1 (0xE1). |
| 105 | + $this->assertStringContainsString("\xC3\xA1", $xmlAfter); |
| 106 | + $this->assertStringNotContainsString("\xE1", $xmlAfter); |
| 107 | + } |
| 108 | + |
| 109 | + // ------------------------------------------------------------------------- |
| 110 | + // Case 2: ISO-8859-1 document serialize/unserialize |
| 111 | + // ------------------------------------------------------------------------- |
| 112 | + |
| 113 | + /** |
| 114 | + * Verifica que un XmlDocument cargado desde bytes ISO-8859-1 reales |
| 115 | + * preserva la declaración de codificación después de serialize/unserialize. |
| 116 | + * |
| 117 | + * - ñ debe estar como byte ISO-8859-1 (0xF1), nunca como secuencia |
| 118 | + * UTF-8 (0xC3 0xB1). |
| 119 | + * - á debe estar como byte ISO-8859-1 (0xE1), nunca como secuencia |
| 120 | + * UTF-8 (0xC3 0xA1). |
| 121 | + */ |
| 122 | + public function testIso88591DocumentPreservesEncodingAfterSerializeUnserialize(): void |
| 123 | + { |
| 124 | + $doc = new XmlDocument(); |
| 125 | + $doc->loadXml($this->iso88591Xml); |
| 126 | + |
| 127 | + /** @var XmlDocument $restored */ |
| 128 | + $restored = unserialize(serialize($doc)); |
| 129 | + |
| 130 | + // La declaración de codificación se preserva. |
| 131 | + $this->assertSame('ISO-8859-1', $restored->getEncoding()); |
| 132 | + |
| 133 | + $restoredXml = $restored->saveXml(); |
| 134 | + |
| 135 | + // El header declara ISO-8859-1. |
| 136 | + $this->assertStringContainsString('encoding="ISO-8859-1"', $restoredXml); |
| 137 | + |
| 138 | + // ñ como byte ISO-8859-1 (0xF1), nunca como secuencia UTF-8 (0xC3 0xB1). |
| 139 | + $this->assertStringContainsString("\xF1", $restoredXml); |
| 140 | + $this->assertStringNotContainsString("\xC3\xB1", $restoredXml); |
| 141 | + |
| 142 | + // á como byte ISO-8859-1 (0xE1), nunca como secuencia UTF-8 (0xC3 0xA1). |
| 143 | + $this->assertStringContainsString("\xE1", $restoredXml); |
| 144 | + $this->assertStringNotContainsString("\xC3\xA1", $restoredXml); |
| 145 | + } |
| 146 | + |
| 147 | + /** |
| 148 | + * Verifica que el contenido decodificado de un documento ISO-8859-1 es |
| 149 | + * idéntico antes y después de serialize/unserialize. |
| 150 | + * |
| 151 | + * DOMDocument almacena los valores internamente en UTF-8, por lo que los |
| 152 | + * valores decodificados deben ser siempre strings UTF-8 válidos. |
| 153 | + */ |
| 154 | + public function testIso88591DocumentPreservesDecodedContentAfterSerializeUnserialize(): void |
| 155 | + { |
| 156 | + $doc = new XmlDocument(); |
| 157 | + $doc->loadXml($this->iso88591Xml); |
| 158 | + |
| 159 | + /** @var XmlDocument $restored */ |
| 160 | + $restored = unserialize(serialize($doc)); |
| 161 | + |
| 162 | + $decoded = (new XmlDecoder())->decode($restored); |
| 163 | + |
| 164 | + foreach ($this->expectedValues as $field => $expected) { |
| 165 | + $this->assertSame( |
| 166 | + $expected, |
| 167 | + $decoded['documento'][$field], |
| 168 | + "El campo '{$field}' tiene un valor inesperado tras serialize/unserialize." |
| 169 | + ); |
| 170 | + } |
| 171 | + } |
| 172 | + |
| 173 | + // ------------------------------------------------------------------------- |
| 174 | + // Case 3: Multiple serialize/unserialize cycles |
| 175 | + // ------------------------------------------------------------------------- |
| 176 | + |
| 177 | + /** |
| 178 | + * Verifica que múltiples ciclos de serialize/unserialize no acumulan |
| 179 | + * corrupción en un documento ISO-8859-1. |
| 180 | + * |
| 181 | + * Un único ciclo ya lo cubre testIso88591DocumentPreservesEncoding*. |
| 182 | + * Este test corre tres ciclos adicionales para confirmar que no hay |
| 183 | + * deriva en la declaración de codificación, los bytes o los valores |
| 184 | + * decodificados. |
| 185 | + */ |
| 186 | + public function testMultipleCyclesDoNotCorruptIso88591Document(): void |
| 187 | + { |
| 188 | + $doc = new XmlDocument(); |
| 189 | + $doc->loadXml($this->iso88591Xml); |
| 190 | + |
| 191 | + for ($i = 0; $i < 3; $i++) { |
| 192 | + /** @var XmlDocument $doc */ |
| 193 | + $doc = unserialize(serialize($doc)); |
| 194 | + } |
| 195 | + |
| 196 | + // La declaración de codificación sigue siendo ISO-8859-1 tras 3 ciclos. |
| 197 | + $this->assertSame('ISO-8859-1', $doc->getEncoding()); |
| 198 | + |
| 199 | + $xml = $doc->saveXml(); |
| 200 | + |
| 201 | + // ñ sigue siendo el byte ISO-8859-1 0xF1, no la secuencia UTF-8 0xC3 0xB1. |
| 202 | + $this->assertStringContainsString("\xF1", $xml); |
| 203 | + $this->assertStringNotContainsString("\xC3\xB1", $xml); |
| 204 | + |
| 205 | + // Los valores decodificados coinciden con los originales. |
| 206 | + $decoded = (new XmlDecoder())->decode($doc); |
| 207 | + foreach ($this->expectedValues as $field => $expected) { |
| 208 | + $this->assertSame($expected, $decoded['documento'][$field]); |
| 209 | + } |
| 210 | + } |
| 211 | + |
| 212 | + /** |
| 213 | + * Verifica que múltiples ciclos de serialize/unserialize en un documento |
| 214 | + * UTF-8 producen siempre el mismo XML byte-a-byte. |
| 215 | + */ |
| 216 | + public function testMultipleCyclesProduceSameXmlForUtf8Document(): void |
| 217 | + { |
| 218 | + $doc = new XmlDocument(); |
| 219 | + $doc->loadXml( |
| 220 | + '<?xml version="1.0" encoding="UTF-8"?><root><frase>Fabricación de Ñoños</frase></root>' |
| 221 | + ); |
| 222 | + |
| 223 | + $xmlOriginal = $doc->saveXml(); |
| 224 | + |
| 225 | + for ($i = 0; $i < 3; $i++) { |
| 226 | + /** @var XmlDocument $doc */ |
| 227 | + $doc = unserialize(serialize($doc)); |
| 228 | + } |
| 229 | + |
| 230 | + $this->assertSame($xmlOriginal, $doc->saveXml()); |
| 231 | + } |
| 232 | + |
| 233 | + // ------------------------------------------------------------------------- |
| 234 | + // Case 4: Methods work correctly after unserialize (XPath, namespace, schema) |
| 235 | + // ------------------------------------------------------------------------- |
| 236 | + |
| 237 | + /** |
| 238 | + * Verifica que los métodos de XmlDocument funcionan correctamente tras |
| 239 | + * la deserialización, incluidas las consultas XPath (xPathQuery lazy). |
| 240 | + */ |
| 241 | + public function testAllMethodsWorkAfterUnserialize(): void |
| 242 | + { |
| 243 | + $doc = new XmlDocument(); |
| 244 | + $doc->loadXml(<<<XML |
| 245 | + <root xmlns="http://example.com" |
| 246 | + xsi:schemaLocation="http://example.com schema.xsd" |
| 247 | + xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"> |
| 248 | + <child>Hola</child> |
| 249 | + <nested><value>42</value></nested> |
| 250 | + </root> |
| 251 | + XML); |
| 252 | + |
| 253 | + /** @var XmlDocument $restored */ |
| 254 | + $restored = unserialize(serialize($doc)); |
| 255 | + |
| 256 | + $this->assertSame('root', $restored->getName()); |
| 257 | + $this->assertSame('http://example.com', $restored->getNamespace()); |
| 258 | + $this->assertSame('schema.xsd', $restored->getSchema()); |
| 259 | + |
| 260 | + // XPath query (fuerza la inicialización lazy de xPathQuery). |
| 261 | + $this->assertSame('Hola', $restored->query('//child')); |
| 262 | + $this->assertSame('42', $restored->query('//value')); |
| 263 | + } |
| 264 | + |
| 265 | + // ------------------------------------------------------------------------- |
| 266 | + // Case 5: Serialized form uses the 'xml' key |
| 267 | + // ------------------------------------------------------------------------- |
| 268 | + |
| 269 | + /** |
| 270 | + * Verifica que la forma serializada de PHP contiene la clave 'xml' definida |
| 271 | + * en __serialize(), confirmando que el mecanismo correcto está en uso. |
| 272 | + */ |
| 273 | + public function testSerializedStringContainsXmlKey(): void |
| 274 | + { |
| 275 | + $doc = new XmlDocument(); |
| 276 | + $doc->loadXml('<root><element>Value</element></root>'); |
| 277 | + |
| 278 | + $serialized = serialize($doc); |
| 279 | + |
| 280 | + // La clave 'xml' del array devuelto por __serialize() debe aparecer |
| 281 | + // en la representación serializada de PHP. |
| 282 | + $this->assertStringContainsString('s:3:"xml";', $serialized); |
| 283 | + } |
| 284 | +} |
0 commit comments