matecat / xml-dom-parser

src/XmlDomLoader.php

Spacing +37 added lines, -37 removed lines

@@ -35,72 +35,72 @@ 
      * @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, $allowDocumentType = false, $setRootElement = null, $schemaOrCallable = null ) {
-        if ( !extension_loaded( 'dom' ) ) {
-            throw new RuntimeException( 'Extension DOM is required.' );
+    public static function load($content, $allowDocumentType = false, $setRootElement = null, $schemaOrCallable = null) {
+        if (!extension_loaded('dom')) {
+            throw new RuntimeException('Extension DOM is required.');
         }
 
-        $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( $setRootElement ) && !empty( $setRootElement ) ) {
+        if (is_string($setRootElement) && !empty($setRootElement)) {
             $content = "<$setRootElement>$content</$setRootElement>";
         }
 
-        $res = $dom->loadXML( $content, LIBXML_NONET | LIBXML_NOBLANKS | ( defined( 'LIBXML_COMPACT' ) ? LIBXML_COMPACT : 0 ) );
+        $res = $dom->loadXML($content, LIBXML_NONET | LIBXML_NOBLANKS | (defined('LIBXML_COMPACT') ? LIBXML_COMPACT : 0));
 
-        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 === $child->nodeType && !$allowDocumentType ) {
-                throw new XmlParsingException( 'Document types are not allowed.' );
+        foreach ($dom->childNodes as $child) {
+            if (XML_DOCUMENT_TYPE_NODE === $child->nodeType && !$allowDocumentType) {
+                throw new XmlParsingException('Document types are not allowed.');
             }
         }
 
-        if ( null !== $schemaOrCallable ) {
-            $internalErrors = libxml_use_internal_errors( true );
+        if (null !== $schemaOrCallable) {
+            $internalErrors = libxml_use_internal_errors(true);
             libxml_clear_errors();
 
             $e = null;
-            if ( is_callable( $schemaOrCallable ) ) {
+            if (is_callable($schemaOrCallable)) {
                 try {
-                    $valid = call_user_func( $schemaOrCallable, $dom, $internalErrors );
-                } catch ( Exception $e ) {
+                    $valid = call_user_func($schemaOrCallable, $dom, $internalErrors);
+                } catch (Exception $e) {
                     $valid = false;
                 }
-            } elseif ( !is_array( $schemaOrCallable ) && is_file( (string)$schemaOrCallable ) ) {
-                $schemaSource = file_get_contents( (string)$schemaOrCallable );
-                $valid        = @$dom->schemaValidateSource( $schemaSource );
+            } elseif (!is_array($schemaOrCallable) && is_file((string)$schemaOrCallable)) {
+                $schemaSource = file_get_contents((string)$schemaOrCallable);
+                $valid        = @$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;
     }
@@ -110,14 +110,14 @@ 
      *
      * @return array
      */
-    private static function getXmlErrors( $internalErrors ) {
-        $errors = [];
-        foreach ( libxml_get_errors() as $error ) {
-            $errors[] = sprintf(
+    private static function getXmlErrors($internalErrors) {
+        $errors = [ ];
+        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
@@ -125,7 +125,7 @@ 
         }
 
         libxml_clear_errors();
-        libxml_use_internal_errors( $internalErrors );
+        libxml_use_internal_errors($internalErrors);
 
         return $errors;
     }

src/HtmlParser.php

Spacing +4 added lines, -4 removed lines

@@ -19,12 +19,12 @@
      */
     protected function getNodeListFromQueryPath() {
 
-        $xpath = new DOMXPath( $this->dom );
+        $xpath = new DOMXPath($this->dom);
 
-        if ( $this->isXmlFragment ) {
-            $htmlNodeList = $xpath->query( "/" . self::fragmentDocumentRoot );
+        if ($this->isXmlFragment) {
+            $htmlNodeList = $xpath->query("/" . self::fragmentDocumentRoot);
         } else {
-            $htmlNodeList = $xpath->query( "/html" );
+            $htmlNodeList = $xpath->query("/html");
         }
 
         return $htmlNodeList;

src/XmlParser.php

Spacing +5 added lines, -5 removed lines

@@ -17,14 +17,14 @@
     /**
      * @return DOMNodeList
      */
-    protected function getNodeListFromQueryPath(){
+    protected function getNodeListFromQueryPath() {
 
-        $xpath = new DOMXPath( $this->dom );
+        $xpath = new DOMXPath($this->dom);
 
-        if ( $this->isXmlFragment ) {
-            $xmlNodeList = $xpath->query( "/" . self::fragmentDocumentRoot );
+        if ($this->isXmlFragment) {
+            $xmlNodeList = $xpath->query("/" . self::fragmentDocumentRoot);
         } else {
-            $xmlNodeList = $xpath->query( "*" );
+            $xmlNodeList = $xpath->query("*");
         }
 
         return $xmlNodeList;

src/AbstractParser.php

Spacing +40 added lines, -40 removed lines

@@ -76,10 +76,10 @@ 
      * @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, true, ( $isXmlFragment ? self::fragmentDocumentRoot : null ) );
+        $this->dom           = XmlDomLoader::load($html, true, ($isXmlFragment ? self::fragmentDocumentRoot : null));
         $this->elements      = new ArrayObject();
     }
 
@@ -95,8 +95,8 @@ 
      * @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();
     }
@@ -109,36 +109,36 @@ 
      *
      * @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 ( $byte[ 0 ] == '0' ) {
+            foreach ($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';
                 }
 
-                $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);
                 }
 
             }
@@ -155,20 +155,20 @@ 
      *
      * @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()
             ];
 
         }
@@ -182,18 +182,18 @@ 
      *
      * @return array
      */
-    protected function getAttributes( DOMNode $element ) {
+    protected function getAttributes(DOMNode $element) {
 
-        if ( !$element->hasAttributes() ) {
-            return [];
+        if (!$element->hasAttributes()) {
+            return [ ];
         }
 
-        $attributesMap = [];
+        $attributesMap = [ ];
 
         /**
          * @var DOMAttr $attr
          */
-        foreach ( $element->attributes as $attr ) {
+        foreach ($element->attributes as $attr) {
             $attributesMap[ $attr->nodeName ] = $attr->nodeValue;
         }
 
@@ -209,15 +209,15 @@ 
 
         $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;