matecat / xml-dom-parser

src/XmlDomLoader.php

Spacing +36 added lines, -36 removed lines

@@ -33,76 +33,76 @@ 
      * @throws InvalidXmlException When parsing of XML with schema or callable produces any errors unrelated to the XML parsing itself
      * @throws XmlParsingException When parsing of XML file returns error
      */
-    public static function load( $content, Config $config = null ) {
-        if ( !extension_loaded( 'dom' ) ) {
-            throw new RuntimeException( 'Extension DOM is required.' );
+    public static function load($content, Config $config = null) {
+        if (!extension_loaded('dom')) {
+            throw new RuntimeException('Extension DOM is required.');
         }
 
-        if ( is_null( $config ) ) {
+        if (is_null($config)) {
             $config = new Config();
         }
 
-        $internalErrors  = libxml_use_internal_errors( true );
+        $internalErrors  = libxml_use_internal_errors(true);
         $disableEntities = libxml_disable_entity_loader();
         libxml_clear_errors();
 
-        $dom                  = new DOMDocument( '1.0', 'UTF-8' );
+        $dom                  = new DOMDocument('1.0', 'UTF-8');
         $dom->validateOnParse = true;
 
-        if ( is_string( $config->getSetRootElement() ) && !empty( $config->getSetRootElement() ) ) {
+        if (is_string($config->getSetRootElement()) && !empty($config->getSetRootElement())) {
             $content = "<{$config->getSetRootElement()}>$content</{$config->getSetRootElement()}>";
         }
 
-        $res = $dom->loadXML( $content, $config->getXML_OPTIONS() );
+        $res = $dom->loadXML($content, $config->getXML_OPTIONS());
 
-        if ( !$res ) {
-            libxml_disable_entity_loader( $disableEntities );
+        if (!$res) {
+            libxml_disable_entity_loader($disableEntities);
 
-            throw new XmlParsingException( implode( "\n", static::getXmlErrors( $internalErrors ) ) );
+            throw new XmlParsingException(implode("\n", static::getXmlErrors($internalErrors)));
         }
 
         $dom->normalizeDocument();
 
-        libxml_use_internal_errors( $internalErrors );
-        libxml_disable_entity_loader( $disableEntities );
+        libxml_use_internal_errors($internalErrors);
+        libxml_disable_entity_loader($disableEntities);
 
-        foreach ( $dom->childNodes as $child ) {
-            if ( XML_DOCUMENT_TYPE_NODE === $child->nodeType && !$config->getAllowDocumentType() ) {
-                throw new XmlParsingException( 'Document types are not allowed.' );
+        foreach ($dom->childNodes as $child) {
+            if (XML_DOCUMENT_TYPE_NODE === $child->nodeType && !$config->getAllowDocumentType()) {
+                throw new XmlParsingException('Document types are not allowed.');
             }
         }
 
-        if ( null !== $config->getSchemaOrCallable() ) {
-            $internalErrors = libxml_use_internal_errors( true );
+        if (null !== $config->getSchemaOrCallable()) {
+            $internalErrors = libxml_use_internal_errors(true);
             libxml_clear_errors();
 
             $e = null;
-            if ( is_callable( $config->getSchemaOrCallable() ) ) {
+            if (is_callable($config->getSchemaOrCallable())) {
                 try {
-                    $valid = call_user_func( $config->getSchemaOrCallable(), $dom, $internalErrors );
-                } catch ( Exception $e ) {
+                    $valid = call_user_func($config->getSchemaOrCallable(), $dom, $internalErrors);
+                } catch (Exception $e) {
                     $valid = false;
                 }
-            } elseif ( !is_array( $config->getSchemaOrCallable() ) && is_file( (string)$config->getSchemaOrCallable() ) ) {
-                $schemaSource = file_get_contents( (string)$config->getSchemaOrCallable() );
-                $valid        = @$dom->schemaValidateSource( $schemaSource );
+            } elseif (!is_array($config->getSchemaOrCallable()) && is_file((string) $config->getSchemaOrCallable())) {
+                $schemaSource = file_get_contents((string) $config->getSchemaOrCallable());
+                $valid        = @$dom->schemaValidateSource($schemaSource);
             } else {
-                libxml_use_internal_errors( $internalErrors );
+                libxml_use_internal_errors($internalErrors);
 
-                throw new XmlParsingException( 'The schemaOrCallable argument has to be a valid path to XSD file or callable.' );
+                throw new XmlParsingException('The schemaOrCallable argument has to be a valid path to XSD file or callable.');
             }
 
-            if ( !$valid ) {
-                $messages = static::getXmlErrors( $internalErrors );
-                if ( empty( $messages ) ) {
-                    throw new InvalidXmlException( 'The XML is not valid.', 0, $e );
+            if (!$valid) {
+                $messages = static::getXmlErrors($internalErrors);
+                if (empty($messages)) {
+                    throw new InvalidXmlException('The XML is not valid.', 0, $e);
                 }
-                throw new XmlParsingException( implode( "\n", $messages ), 0, $e );
+                throw new XmlParsingException(implode("\n", $messages), 0, $e);
             }
         }
 
         libxml_clear_errors();
-        libxml_use_internal_errors( $internalErrors );
+        libxml_use_internal_errors($internalErrors);
 
         return $dom;
     }
@@ -112,14 +112,14 @@ 
      *
      * @return array
      */
-    private static function getXmlErrors( $internalErrors ) {
+    private static function getXmlErrors($internalErrors) {
         $errors = [];
-        foreach ( libxml_get_errors() as $error ) {
+        foreach (libxml_get_errors() as $error) {
             $errors[] = sprintf(
                     '[%s %s] %s (in %s - line %d, column %d)',
                     LIBXML_ERR_WARNING == $error->level ? 'WARNING' : 'ERROR',
                     $error->code,
-                    trim( $error->message ),
+                    trim($error->message),
                     $error->file ?: 'n/a',
                     $error->line,
                     $error->column
@@ -127,7 +127,7 @@ 
         }
 
         libxml_clear_errors();
-        libxml_use_internal_errors( $internalErrors );
+        libxml_use_internal_errors($internalErrors);
 
         return $errors;
     }

src/AbstractParser.php

Spacing +74 added lines, -74 removed lines

@@ -25,39 +25,39 @@ 
     const regexpEntity         = '/&#x([0-1]{0,1}[0-9A-F]{1,2})/u'; //  
     const regexpAscii          = '/([\x{00}-\x{1F}\x{7F}]{1})/u';
     protected static $asciiPlaceHoldMap = [
-            '00' => [ 'symbol' => 'NULL', 'placeHold' => '', 'numeral' => 0x00 ],
-            '01' => [ 'symbol' => 'SOH', 'placeHold' => '', 'numeral' => 0x01 ],
-            '02' => [ 'symbol' => 'STX', 'placeHold' => '', 'numeral' => 0x02 ],
-            '03' => [ 'symbol' => 'ETX', 'placeHold' => '', 'numeral' => 0x03 ],
-            '04' => [ 'symbol' => 'EOT', 'placeHold' => '', 'numeral' => 0x04 ],
-            '05' => [ 'symbol' => 'ENQ', 'placeHold' => '', 'numeral' => 0x05 ],
-            '06' => [ 'symbol' => 'ACK', 'placeHold' => '', 'numeral' => 0x06 ],
-            '07' => [ 'symbol' => 'BEL', 'placeHold' => '', 'numeral' => 0x07 ],
-            '08' => [ 'symbol' => 'BS', 'placeHold' => '', 'numeral' => 0x08 ],
-            '09' => [ 'symbol' => 'HT', 'placeHold' => '', 'numeral' => 0x09 ],
-            '0A' => [ 'symbol' => 'LF', 'placeHold' => '', 'numeral' => 0x0A ],
-            '0B' => [ 'symbol' => 'VT', 'placeHold' => '', 'numeral' => 0x0B ],
-            '0C' => [ 'symbol' => 'FF', 'placeHold' => '', 'numeral' => 0x0C ],
-            '0D' => [ 'symbol' => 'CR', 'placeHold' => '', 'numeral' => 0x0D ],
-            '0E' => [ 'symbol' => 'SO', 'placeHold' => '', 'numeral' => 0x0E ],
-            '0F' => [ 'symbol' => 'SI', 'placeHold' => '', 'numeral' => 0x0F ],
-            '10' => [ 'symbol' => 'DLE', 'placeHold' => '', 'numeral' => 0x10 ],
-            '11' => [ 'symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x11 ],
-            '12' => [ 'symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x12 ],
-            '13' => [ 'symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x13 ],
-            '14' => [ 'symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x14 ],
-            '15' => [ 'symbol' => 'NAK', 'placeHold' => '', 'numeral' => 0x15 ],
-            '16' => [ 'symbol' => 'SYN', 'placeHold' => '', 'numeral' => 0x16 ],
-            '17' => [ 'symbol' => 'ETB', 'placeHold' => '', 'numeral' => 0x17 ],
-            '18' => [ 'symbol' => 'CAN', 'placeHold' => '', 'numeral' => 0x18 ],
-            '19' => [ 'symbol' => 'EM', 'placeHold' => '', 'numeral' => 0x19 ],
-            '1A' => [ 'symbol' => 'SUB', 'placeHold' => '', 'numeral' => 0x1A ],
-            '1B' => [ 'symbol' => 'ESC', 'placeHold' => '', 'numeral' => 0x1B ],
-            '1C' => [ 'symbol' => 'FS', 'placeHold' => '', 'numeral' => 0x1C ],
-            '1D' => [ 'symbol' => 'GS', 'placeHold' => '', 'numeral' => 0x1D ],
-            '1E' => [ 'symbol' => 'RS', 'placeHold' => '', 'numeral' => 0x1E ],
-            '1F' => [ 'symbol' => 'US', 'placeHold' => '', 'numeral' => 0x1F ],
-            '7F' => [ 'symbol' => 'DEL', 'placeHold' => '', 'numeral' => 0x7F ],
+            '00' => ['symbol' => 'NULL', 'placeHold' => '', 'numeral' => 0x00],
+            '01' => ['symbol' => 'SOH', 'placeHold' => '', 'numeral' => 0x01],
+            '02' => ['symbol' => 'STX', 'placeHold' => '', 'numeral' => 0x02],
+            '03' => ['symbol' => 'ETX', 'placeHold' => '', 'numeral' => 0x03],
+            '04' => ['symbol' => 'EOT', 'placeHold' => '', 'numeral' => 0x04],
+            '05' => ['symbol' => 'ENQ', 'placeHold' => '', 'numeral' => 0x05],
+            '06' => ['symbol' => 'ACK', 'placeHold' => '', 'numeral' => 0x06],
+            '07' => ['symbol' => 'BEL', 'placeHold' => '', 'numeral' => 0x07],
+            '08' => ['symbol' => 'BS', 'placeHold' => '', 'numeral' => 0x08],
+            '09' => ['symbol' => 'HT', 'placeHold' => '', 'numeral' => 0x09],
+            '0A' => ['symbol' => 'LF', 'placeHold' => '', 'numeral' => 0x0A],
+            '0B' => ['symbol' => 'VT', 'placeHold' => '', 'numeral' => 0x0B],
+            '0C' => ['symbol' => 'FF', 'placeHold' => '', 'numeral' => 0x0C],
+            '0D' => ['symbol' => 'CR', 'placeHold' => '', 'numeral' => 0x0D],
+            '0E' => ['symbol' => 'SO', 'placeHold' => '', 'numeral' => 0x0E],
+            '0F' => ['symbol' => 'SI', 'placeHold' => '', 'numeral' => 0x0F],
+            '10' => ['symbol' => 'DLE', 'placeHold' => '', 'numeral' => 0x10],
+            '11' => ['symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x11],
+            '12' => ['symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x12],
+            '13' => ['symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x13],
+            '14' => ['symbol' => 'DC', 'placeHold' => '', 'numeral' => 0x14],
+            '15' => ['symbol' => 'NAK', 'placeHold' => '', 'numeral' => 0x15],
+            '16' => ['symbol' => 'SYN', 'placeHold' => '', 'numeral' => 0x16],
+            '17' => ['symbol' => 'ETB', 'placeHold' => '', 'numeral' => 0x17],
+            '18' => ['symbol' => 'CAN', 'placeHold' => '', 'numeral' => 0x18],
+            '19' => ['symbol' => 'EM', 'placeHold' => '', 'numeral' => 0x19],
+            '1A' => ['symbol' => 'SUB', 'placeHold' => '', 'numeral' => 0x1A],
+            '1B' => ['symbol' => 'ESC', 'placeHold' => '', 'numeral' => 0x1B],
+            '1C' => ['symbol' => 'FS', 'placeHold' => '', 'numeral' => 0x1C],
+            '1D' => ['symbol' => 'GS', 'placeHold' => '', 'numeral' => 0x1D],
+            '1E' => ['symbol' => 'RS', 'placeHold' => '', 'numeral' => 0x1E],
+            '1F' => ['symbol' => 'US', 'placeHold' => '', 'numeral' => 0x1F],
+            '7F' => ['symbol' => 'DEL', 'placeHold' => '', 'numeral' => 0x7F],
     ];
 
     /**
@@ -76,14 +76,14 @@ 
      * @throws InvalidXmlException
      * @throws XmlParsingException
      */
-    protected function __construct( $html, $isXmlFragment ) {
-        $html                = $this->removeNotPrintableChars( $html );
+    protected function __construct($html, $isXmlFragment) {
+        $html                = $this->removeNotPrintableChars($html);
         $this->isXmlFragment = $isXmlFragment;
 
         $this->dom = XmlDomLoader::load(
                 $html,
                 new Config(
-                        ( $isXmlFragment ? self::fragmentDocumentRoot : null ),
+                        ($isXmlFragment ? self::fragmentDocumentRoot : null),
                         true,
                         LIBXML_NONET | LIBXML_NOBLANKS
                 )
@@ -104,8 +104,8 @@ 
      * @throws InvalidXmlException
      * @throws XmlParsingException
      */
-    public static function parse( $html, $isXmlFragment = false ) {
-        $parser = new static( $html, $isXmlFragment );
+    public static function parse($html, $isXmlFragment = false) {
+        $parser = new static($html, $isXmlFragment);
 
         return $parser->extractNodes();
     }
@@ -118,36 +118,36 @@ 
      *
      * @return string
      */
-    protected function removeNotPrintableChars( $seg ) {
+    protected function removeNotPrintableChars($seg) {
 
-        preg_match_all( self::regexpAscii, $seg, $matches );
+        preg_match_all(self::regexpAscii, $seg, $matches);
 
-        if ( !empty( $matches[ 1 ] ) ) {
+        if (!empty($matches[1])) {
             $test_src = $seg;
-            foreach ( $matches[ 1 ] as $v ) {
-                $key      = sprintf( "%02X", ord( $v ) );
-                $hexNum   = sprintf( "/(\\x{%s})/u", $key );
-                $test_src = preg_replace( $hexNum, self::$asciiPlaceHoldMap[ $key ][ 'placeHold' ], $test_src, 1 );
+            foreach ($matches[1] as $v) {
+                $key      = sprintf("%02X", ord($v));
+                $hexNum   = sprintf("/(\\x{%s})/u", $key);
+                $test_src = preg_replace($hexNum, self::$asciiPlaceHoldMap[$key]['placeHold'], $test_src, 1);
             }
 
             $seg = $test_src;
         }
 
-        preg_match_all( self::regexpEntity, $seg, $matches );
+        preg_match_all(self::regexpEntity, $seg, $matches);
 
-        if ( !empty( $matches[ 1 ] ) ) {
+        if (!empty($matches[1])) {
             $test_src = $seg;
-            foreach ( $matches[ 1 ] as $v ) {
-                $byte = sprintf( "%02X", hexdec( $v ) );
-                if ( $byte[ 0 ] == '0' ) {
-                    $regexp = '/&#x([' . $byte[ 0 ] . ']?' . $byte[ 1 ] . ');/u';
+            foreach ($matches[1] as $v) {
+                $byte = sprintf("%02X", hexdec($v));
+                if ($byte[0] == '0') {
+                    $regexp = '/&#x(['.$byte[0].']?'.$byte[1].');/u';
                 } else {
-                    $regexp = '/&#x(' . $byte . ');/u';
+                    $regexp = '/&#x('.$byte.');/u';
                 }
 
-                $key = sprintf( "%02X", hexdec( $v ) );
-                if ( array_key_exists( $key, self::$asciiPlaceHoldMap ) ) {
-                    $test_src = preg_replace( $regexp, self::$asciiPlaceHoldMap[ $key ][ 'placeHold' ], $test_src );
+                $key = sprintf("%02X", hexdec($v));
+                if (array_key_exists($key, self::$asciiPlaceHoldMap)) {
+                    $test_src = preg_replace($regexp, self::$asciiPlaceHoldMap[$key]['placeHold'], $test_src);
                 }
 
             }
@@ -164,20 +164,20 @@ 
      *
      * @return ArrayObject
      */
-    protected function mapElements( DOMNodeList $elementList, ArrayObject $elements ) {
+    protected function mapElements(DOMNodeList $elementList, ArrayObject $elements) {
 
-        for ( $i = 0; $i < $elementList->length; $i++ ) {
+        for ($i = 0; $i < $elementList->length; $i++) {
 
-            $element = $elementList->item( $i );
+            $element = $elementList->item($i);
 
-            $elements[] = (object)[
-                    'node'         => $this->dom->saveXML( $element ),
+            $elements[] = (object) [
+                    'node'         => $this->dom->saveXML($element),
                     'tagName'      => $element->nodeName,
-                    'attributes'   => $this->getAttributes( $element ),
-                    'text'         => ( $element instanceof DOMText ) ? $element->textContent : null,
-                    'self_closed'  => ( $element instanceof DOMText ) ? null : !$element->hasChildNodes(),
-                    'has_children' => ( $element instanceof DOMText ) ? null : $element->hasChildNodes(),
-                    'inner_html'   => $element->hasChildNodes() ? $this->mapElements( $element->childNodes, new ArrayObject() ) : new ArrayObject()
+                    'attributes'   => $this->getAttributes($element),
+                    'text'         => ($element instanceof DOMText) ? $element->textContent : null,
+                    'self_closed'  => ($element instanceof DOMText) ? null : !$element->hasChildNodes(),
+                    'has_children' => ($element instanceof DOMText) ? null : $element->hasChildNodes(),
+                    'inner_html'   => $element->hasChildNodes() ? $this->mapElements($element->childNodes, new ArrayObject()) : new ArrayObject()
             ];
 
         }
@@ -191,9 +191,9 @@ 
      *
      * @return array
      */
-    protected function getAttributes( DOMNode $element ) {
+    protected function getAttributes(DOMNode $element) {
 
-        if ( !$element->hasAttributes() ) {
+        if (!$element->hasAttributes()) {
             return [];
         }
 
@@ -202,8 +202,8 @@ 
         /**
          * @var DOMAttr $attr
          */
-        foreach ( $element->attributes as $attr ) {
-            $attributesMap[ $attr->nodeName ] = $attr->nodeValue;
+        foreach ($element->attributes as $attr) {
+            $attributesMap[$attr->nodeName] = $attr->nodeValue;
         }
 
         return $attributesMap;
@@ -218,15 +218,15 @@ 
 
         $htmlNodeList = $this->getNodeListFromQueryPath();
 
-        if ( !$htmlNodeList instanceof DOMNodeList ) {
-            throw new DOMException( 'Bad DOMNodeList' );
+        if (!$htmlNodeList instanceof DOMNodeList) {
+            throw new DOMException('Bad DOMNodeList');
         }
 
-        if ( $this->isXmlFragment && $htmlNodeList->item( 0 )->nodeName == self::fragmentDocumentRoot ) {
+        if ($this->isXmlFragment && $htmlNodeList->item(0)->nodeName == self::fragmentDocumentRoot) {
             // there is a fake root node, skip the first element end start with child nodes
-            $this->mapElements( $htmlNodeList->item( 0 )->childNodes, $this->elements );
+            $this->mapElements($htmlNodeList->item(0)->childNodes, $this->elements);
         } else {
-            $this->mapElements( $htmlNodeList, $this->elements );
+            $this->mapElements($htmlNodeList, $this->elements);
         }
 
         return $this->elements;

src/Config.php

Spacing +2 added lines, -2 removed lines

@@ -31,8 +31,8 @@
      */
     protected $XML_OPTIONS = 0;
 
-    public function __construct( $setRootElement = null, $allowDocumentType = false, $XML_OPTIONS = 0, $schemaOrCallable = null ) {
-        $this->XML_OPTIONS       = $XML_OPTIONS | ( defined( 'LIBXML_COMPACT' ) ? LIBXML_COMPACT : 0 );
+    public function __construct($setRootElement = null, $allowDocumentType = false, $XML_OPTIONS = 0, $schemaOrCallable = null) {
+        $this->XML_OPTIONS       = $XML_OPTIONS | (defined('LIBXML_COMPACT') ? LIBXML_COMPACT : 0);
         $this->setRootElement    = $setRootElement;
         $this->allowDocumentType = $allowDocumentType;
         $this->schemaOrCallable  = $schemaOrCallable;