From 6ff9476c0b22d8951a9a502a39902328f8b10e55 Mon Sep 17 00:00:00 2001 From: Timothy J Warren Date: Tue, 12 Mar 2019 09:51:38 -0400 Subject: [PATCH] Move XML codec from Hummingbird --- src/XML.php | 279 ++++++++++++++++++++ tests/XMLTest.php | 89 +++++++ tests/test_data/XML/minifiedXmlTestFile.xml | 2 + tests/test_data/XML/xmlTestFile.xml | 22 ++ 4 files changed, 392 insertions(+) create mode 100644 src/XML.php create mode 100644 tests/XMLTest.php create mode 100644 tests/test_data/XML/minifiedXmlTestFile.xml create mode 100644 tests/test_data/XML/xmlTestFile.xml diff --git a/src/XML.php b/src/XML.php new file mode 100644 index 0000000..9a8ad78 --- /dev/null +++ b/src/XML.php @@ -0,0 +1,279 @@ + + * @copyright 2015 - 2018 Timothy J. Warren + * @license http://www.opensource.org/licenses/mit-license.html MIT License + * @version 2.4.1 + * @link https://git.timshomepage.net/timw4mail/ion + */ + +namespace Aviat\Ion; + +use DOMDocument, DOMNode, DOMNodeList, InvalidArgumentException; + +/** + * XML <=> PHP Array codec + */ +final class XML { + + /** + * XML representation of the data + * + * @var string + */ + private $xml; + + /** + * PHP array version of the data + * + * @var array + */ + private $data; + + /** + * XML constructor + * + * @param string $xml + * @param array $data + */ + public function __construct(string $xml = '', array $data = []) + { + $this->setXML($xml)->setData($data); + } + + /** + * Serialize the data to an xml string + * + * @return string + */ + public function __toString(): string + { + return static::toXML($this->getData()); + } + + /** + * Get the data parsed from the XML + * + * @return array + */ + public function getData(): array + { + return $this->data; + } + + /** + * Set the data to create xml from + * + * @param array $data + * @return self + */ + public function setData(array $data): self + { + $this->data = $data; + return $this; + } + + /** + * Get the xml created from the data + * + * @return string + */ + public function getXML(): string + { + return $this->xml; + } + + /** + * Set the xml to parse the data from + * + * @param string $xml + * @return self + */ + public function setXML(string $xml): self + { + $this->xml = $xml; + return $this; + } + + /** + * Parse an xml document string to a php array + * + * @param string $xml + * @return array + */ + public static function toArray(string $xml): array + { + $data = []; + + $xml = static::stripXMLWhitespace($xml); + + $dom = new DOMDocument(); + $hasLoaded = @$dom->loadXML($xml); + + if ( ! $hasLoaded) + { + throw new InvalidArgumentException('Failed to load XML'); + } + + $root = $dom->documentElement; + + $data[$root->tagName] = []; + + if ($root->hasChildNodes()) + { + static::childNodesToArray($data[$root->tagName], $root->childNodes); + } + + return $data; + } + + /** + * Transform the array into XML + * + * @param array $data + * @return string + */ + public static function toXML(array $data): string + { + $dom = new DOMDocument(); + $dom->encoding = 'UTF-8'; + + static::arrayPropertiesToXmlNodes($dom, $dom, $data); + + return $dom->saveXML(); + } + + /** + * Parse the xml document string to a php array + * + * @return array + */ + public function parse(): array + { + $xml = $this->getXML(); + $data = static::toArray($xml); + return $this->setData($data)->getData(); + } + + /** + * Transform the array into XML + * + * @return string + */ + public function createXML(): string + { + return static::toXML($this->getData()); + } + + /** + * Strip whitespace from raw xml to remove irrelevant text nodes + * + * @param string $xml + * @return string + */ + private static function stripXMLWhitespace(string $xml): string + { + // Get rid of unimportant text nodes by removing + // whitespace characters from between xml tags, + // except for the xml declaration tag, Which looks + // something like: + /* */ + return preg_replace('/([^?])>\s+<', $xml); + } + + /** + * Recursively create array structure based on xml structure + * + * @param array $root A reference to the current array location + * @param DOMNodeList $nodeList The current NodeList object + * @return void + */ + private static function childNodesToArray(array &$root, DOMNodelist $nodeList): void + { + $length = $nodeList->length; + for ($i = 0; $i < $length; $i++) + { + $el = $nodeList->item($i); + $current =& $root[$el->nodeName]; + + // It's a top level element! + if (( ! $el->hasChildNodes()) || ($el->childNodes->item(0) instanceof \DomText)) + { + $current = $el->textContent; + continue; + } + + // An empty value at the current root + if ($current === NULL) + { + $current = []; + static::childNodesToArray($current, $el->childNodes); + continue; + } + + $keys = array_keys($current); + + // Wrap the array in a containing array + // if there are only string keys + if ( ! is_numeric($keys[0])) + { + // But if there is only one key, don't wrap it in + // an array, just recurse to parse the child nodes + if (count($current) === 1) + { + static::childNodesToArray($current, $el->childNodes); + continue; + } + $current = [$current]; + } + + $current[] = []; + $index = count($current) - 1; + + static::childNodesToArray($current[$index], $el->childNodes); + } + } + + /** + * Recursively create xml nodes from array properties + * + * @param DOMDocument $dom The current DOM object + * @param DOMNode $parent The parent element to append children to + * @param array $data The data for the current node + * @return void + */ + private static function arrayPropertiesToXmlNodes(DOMDocument $dom, DOMNode $parent, array $data): void + { + foreach ($data as $key => $props) + { + // 'Flatten' the array as you create the xml + if (is_numeric($key)) + { + foreach ($props as $k => $p) + { + break; + } + } + + $node = $dom->createElement($key); + + if (\is_array($props)) + { + static::arrayPropertiesToXmlNodes($dom, $node, $props); + } else + { + $tNode = $dom->createTextNode((string)$props); + $node->appendChild($tNode); + } + + $parent->appendChild($node); + } + } +} \ No newline at end of file diff --git a/tests/XMLTest.php b/tests/XMLTest.php new file mode 100644 index 0000000..33ba504 --- /dev/null +++ b/tests/XMLTest.php @@ -0,0 +1,89 @@ + + * @copyright 2015 - 2018 Timothy J. Warren + * @license http://www.opensource.org/licenses/mit-license.html MIT License + * @version 2.4.1 + * @link https://git.timshomepage.net/timw4mail/ion + */ + +namespace Aviat\Ion\Tests; + +use Aviat\Ion\XML; +use PHPUnit\Framework\TestCase; + +class XMLTest extends TestCase { + + protected $xml; + protected $expectedXml; + protected $object; + protected $array; + + public function setUp() + { + $this->xml = file_get_contents(__DIR__ . '/test_data/XML/xmlTestFile.xml'); + $this->expectedXml = file_get_contents(__DIR__ . '/test_data/XML/minifiedXmlTestFile.xml'); + + $this->array = [ + 'entry' => [ + 'foo' => [ + 'bar' => [ + 'baz' => 42 + ] + ], + 'episode' => '11', + 'status' => 'watching', + 'score' => '7', + 'storage_type' => '1', + 'storage_value' => '2.5', + 'times_rewatched' => '1', + 'rewatch_value' => '3', + 'date_start' => '01152015', + 'date_finish' => '10232016', + 'priority' => '2', + 'enable_discussion' => '0', + 'enable_rewatching' => '1', + 'comments' => 'Should you say something?', + 'tags' => 'test tag, 2nd tag' + ] + ]; + + $this->object = new XML(); + } + + public function testToArray() + { + $this->assertEquals($this->array, XML::toArray($this->xml)); + } + + public function testParse() + { + $this->object->setXML($this->xml); + $this->assertEquals($this->array, $this->object->parse()); + } + + public function testToXML() + { + $this->assertEquals($this->expectedXml, XML::toXML($this->array)); + } + + public function testCreateXML() + { + $this->object->setData($this->array); + $this->assertEquals($this->expectedXml, $this->object->createXML()); + } + + public function testToString() + { + $this->object->setData($this->array); + $this->assertEquals($this->expectedXml, $this->object->__toString()); + $this->assertEquals($this->expectedXml, (string)$this->object); + } +} \ No newline at end of file diff --git a/tests/test_data/XML/minifiedXmlTestFile.xml b/tests/test_data/XML/minifiedXmlTestFile.xml new file mode 100644 index 0000000..c0490c7 --- /dev/null +++ b/tests/test_data/XML/minifiedXmlTestFile.xml @@ -0,0 +1,2 @@ + +4211watching712.5130115201510232016201Should you say something?test tag, 2nd tag diff --git a/tests/test_data/XML/xmlTestFile.xml b/tests/test_data/XML/xmlTestFile.xml new file mode 100644 index 0000000..b37cf96 --- /dev/null +++ b/tests/test_data/XML/xmlTestFile.xml @@ -0,0 +1,22 @@ + + + + + 42 + + + 11 + watching + 7 + 1 + 2.5 + 1 + 3 + 01152015 + 10232016 + 2 + 0 + 1 + Should you say something? + test tag, 2nd tag + \ No newline at end of file