Add Walk and Moonwalk generators

lib/DOM/Comment.php

@@ -3,7 +3,7 @@ declare(strict_types=1);
 namespace dW\HTML5;
 
 class Comment extends \DOMComment {
-    use Ancestor;
+    use Moonwalk;
 
     public function __toString(): string {
         # Append the literal string "<!--" (U+003C LESS-THAN SIGN, U+0021 EXCLAMATION

lib/DOM/DOMException.php

@@ -4,16 +4,18 @@ namespace dW\HTML5;
 
 class DOMException extends \Exception {
     // From PHP's DOMException; keeping error codes consistent
+    const WRONG_DOCUMENT = 4;
     const NO_MODIFICATION_ALLOWED = 7;
 
-    const DOCUMENT_DOCUMENTFRAG_EXPECTED = 100;
-    const STRING_OR_CLOSURE_EXPECTED = 101;
+    const DOCUMENT_ELEMENT_DOCUMENTFRAG_EXPECTED = 100;
+    const STRING_EXPECTED = 101;
     const OUTER_HTML_FAILED_NOPARENT = 102;
 
     protected static $messages = [
+          4 => 'Supplied node does not belong to this document',
           7 => 'Modification not allowed here',
-        100 => 'Element, Document, or DOMDocumentFragment expected; found %s',
-        101 => 'The first argument must either be an instance of \DOMNode, a string, or a closure; found %s',
+        100 => 'Document, Element, or DocumentFragment expected; found %s',
+        101 => 'The "%s" argument should be a string; found %s',
         102 => 'Failed to set the "outerHTML" property; the element does not have a parent node'
     ];
 
@@ -36,7 +38,7 @@ class DOMException extends \Exception {
 
         // Count the number of replacements needed in the message.
         preg_match_all('/(\%(?:\d+\$)?s)/', $message, $matches);
-        $count = count(array_unique($matches[1]));
+        $count = count($matches[1]);
 
         // If the number of replacements don't match the arguments then oops.
         if (count($args) !== $count) {

lib/DOM/Document.php

@@ -3,7 +3,7 @@ declare(strict_types=1);
 namespace dW\HTML5;
 
 class Document extends \DOMDocument {
-    use Descendant, Serialize, EscapeString;
+    use EscapeString, Moonwalk, Serialize, Walk;
 
     // Quirks mode constants
     public const NO_QUIRKS_MODE = 0;
@@ -31,23 +31,32 @@ class Document extends \DOMDocument {
         $this->registerNodeClass('DOMText', '\dW\HTML5\Text');
     }
 
-    public function load($source, $options = null, ?string $encodingOrContentType = null): bool {
-        $data = Parser::fetchFile($source, $encodingOrContentType);
-        if (!$data) {
-            return false;
+    public function createAttribute($name) {
+        try {
+            return parent::createAttribute($name);
+        } catch (\DOMException $e) {
+            // The element name is invalid for XML
+            // Replace any offending characters with "UHHHHHH" where H are the
+            //   uppercase hexadecimal digits of the character's code point
+            $this->mangledAttributes = true;
+            $name = $this->coerceName($name);
+            return parent::createAttribute($name);
         }
-        [$data, $encodingOrContentType] = $data;
-        Parser::parse($data, $this, $encodingOrContentType, null, (string) $source);
-        return true;
     }
 
-    public function loadHTML($source, $options = null, ?string $encodingOrContentType = null): bool {
-        Parser::parse((string)$source, $this, $encodingOrContentType);
-        return true;
+    public function createAttributeNS($namespaceURI, $qualifiedName) {
+        try {
+            return parent::createAttributeNS($namespaceURI, $qualifiedName);
+        } catch (\DOMException $e) {
+            // The element name is invalid for XML
+            // Replace any offending characters with "UHHHHHH" where H are the
+            //   uppercase hexadecimal digits of the character's code point
+            $this->mangledAttributes = true;
+            $qualifiedName = $this->coerceName($qualifiedName);
+            return parent::createAttributeNS($namespaceURI, $qualifiedName);
+        }
     }
 
-    public function saveHTMLFile($filename) {}
-
     public function createElement($name, $value = "") {
         try {
             $e = parent::createElement($name, $value);
@@ -84,30 +93,50 @@ class Document extends \DOMDocument {
         }
     }
 
-    public function createAttribute($name) {
-        try {
-            return parent::createAttribute($name);
-        } catch (\DOMException $e) {
-            // The element name is invalid for XML
-            // Replace any offending characters with "UHHHHHH" where H are the
-            //   uppercase hexadecimal digits of the character's code point
-            $this->mangledAttributes = true;
-            $name = $this->coerceName($name);
-            return parent::createAttribute($name);
+    public function load($source, $options = null, ?string $encodingOrContentType = null): bool {
+        $data = Parser::fetchFile($source, $encodingOrContentType);
+        if (!$data) {
+            return false;
         }
+        [$data, $encodingOrContentType] = $data;
+        Parser::parse($data, $this, $encodingOrContentType, null, (string) $source);
+        return true;
     }
 
-    public function createAttributeNS($namespaceURI, $qualifiedName) {
-        try {
-            return parent::createAttributeNS($namespaceURI, $qualifiedName);
-        } catch (\DOMException $e) {
-            // The element name is invalid for XML
-            // Replace any offending characters with "UHHHHHH" where H are the
-            //   uppercase hexadecimal digits of the character's code point
-            $this->mangledAttributes = true;
-            $qualifiedName = $this->coerceName($qualifiedName);
-            return parent::createAttributeNS($namespaceURI, $qualifiedName);
+    public function loadHTML($source, $options = null, ?string $encodingOrContentType = null): bool {
+        assert(is_string($source), new DOMException(DOMException::STRING_EXPECTED, 'source', gettype($source)));
+        Parser::parse($source, $this, $encodingOrContentType);
+        return true;
+    }
+
+    public function save($filename, $options = 0): string {
+        return file_put_contents($filename, $this->serialize());
+    }
+
+    public function saveHTML(\DOMNode $node = null): string {
+        if ($node === null) {
+            $node = $this;
+        } elseif ($node->ownerDocument !== $this) {
+            throw new DOMException(DOMException::WRONG_DOCUMENT);
         }
+
+        return $node->serialize();
+    }
+
+    public function saveHTMLFile($filename): int {
+        return $this->save($filename);
+    }
+
+    public function saveXML(?\DOMNode $node = null, $options = null) {
+        return false;
+    }
+
+    public function validate(): bool {
+        return true;
+    }
+
+    public function xinclude($options = null): bool {
+        return false;
     }
 
     public function __toString() {

lib/DOM/DocumentFragment.php

@@ -3,7 +3,7 @@ declare(strict_types=1);
 namespace dW\HTML5;
 
 class DocumentFragment extends \DOMDocumentFragment {
-    use Descendant, Serialize;
+    use Moonwalk, Serialize;
 
     public function __toString() {
         return $this->serialize();

lib/DOM/Element.php

@@ -3,9 +3,7 @@ declare(strict_types=1);
 namespace dW\HTML5;
 
 class Element extends \DOMElement {
-    use Ancestor, Descendant, EscapeString, Serialize {
-        Ancestor::compare insteadof Descendant;
-    }
+    use EscapeString, Moonwalk, Serialize, Walk;
 
     // Used for template elements
     public $content = null;

lib/DOM/ProcessingInstruction.php

@@ -3,7 +3,7 @@ declare(strict_types=1);
 namespace dW\HTML5;
 
 class ProcessingInstruction extends \DOMProcessingInstruction {
-    use Ancestor;
+    use Moonwalk;
 
     public function __toString(): string {
         # Append the literal string "<?" (U+003C LESS-THAN SIGN, U+003F QUESTION MARK),

lib/DOM/Text.php

@@ -3,7 +3,7 @@ declare(strict_types=1);
 namespace dW\HTML5;
 
 class Text extends \DOMText {
-    use Ancestor, EscapeString;
+    use EscapeString, Moonwalk;
 
     function __toString(): string {
         # If the parent of current node is a style, script, xmp, iframe, noembed,

lib/DOM/traits/Ancestor.php

@@ -1,27 +0,0 @@
-<?php
-declare(strict_types=1);
-namespace dW\HTML5;
-
-trait Ancestor {
-    use Compare;
-
-    public function getAncestor($needle): Element {
-        return $this->ancestor($needle, true);
-    }
-
-    public static function hasAncestor($needle): bool {
-        return $this->ancestor($needle, false);
-    }
-
-    protected function ancestor($needle, bool $returnNode = true) {
-        $context = $this->parentNode;
-        do {
-            $result = self::compare($needle, $context);
-            if (!is_null($result)) {
-                return ($returnNode === true) ? $result : true;
-            }
-        } while ($context = $context->parentNode);
-
-        return ($returnNode === true) ? null : false;
-    }
-}

lib/DOM/traits/Compare.php

@@ -1,25 +0,0 @@
-<?php
-declare(strict_types=1);
-namespace dW\HTML5;
-
-trait Compare {
-    protected function compare($needle, Element $context): \DOMNode {
-        if (is_string($needle)) {
-            if ($context->nodeName == $needle) {
-                return $this;
-            }
-        } elseif ($needle instanceof \DOMNode) {
-            if ($context->isSameNode($needle)) {
-                return $context;
-            }
-        } elseif ($needle instanceof \Closure) {
-            if ($needle($context) === true) {
-                return $context;
-            }
-        } else {
-            throw new DOMException(DOMException::STRING_OR_CLOSURE_EXPECTED, gettype($needle));
-        }
-
-        return null;
-    }
-}

lib/DOM/traits/Descendant.php

@@ -1,37 +0,0 @@
-<?php
-declare(strict_types=1);
-namespace dW\HTML5;
-
-trait Descendant {
-    use Compare;
-
-    public function getDescendant($needle): \DOMNode {
-        return self::descendant($needle, true);
-    }
-
-    public function hasDescendant($needle): bool {
-        return self::descendant($needle, false);
-    }
-
-    protected function descendant($needle, bool $returnNode = true): \DOMNode {
-        if ($this->hasChildNodes() === false) {
-            return ($returnNode === true) ? null : false;
-        }
-
-        $context = $this->firstChild;
-
-        do {
-            $result = $this->compare($needle, $context);
-            if (!is_null($result)) {
-                return ($returnNode === true) ? $result : true;
-            }
-
-            $result = $this->descendant($needle, $context);
-            if (!is_null($result)) {
-                return ($returnNode === true) ? $result : true;
-            }
-        } while ($context = $context->nextSibling);
-
-        return ($returnNode === true) ? null : false;
-    }
-}

lib/DOM/traits/Moonwalk.php

@@ -0,0 +1,17 @@
+<?php
+declare(strict_types=1);
+namespace dW\HTML5;
+
+trait Moonwalk {
+    public function moonwalk(\Closure $filter): \Generator {
+        return $this->moonwalkGenerator($this, $filter);
+    }
+    
+    private function moonwalkGenerator(\DOMNode $node, \Closure $filter) {
+        do {
+            if ($filter($node)) {
+                yield $node;
+            }
+        } while ($node = $node->parentNode);
+    }
+}

lib/DOM/traits/Serialize.php

@@ -18,7 +18,7 @@ trait Serialize {
         }
 
         if (!$node instanceof Element && !$node instanceof Document && !$node instanceof DocumentFragment) {
-            throw new DOMException(DOMException::DOCUMENT_DOCUMENTFRAG_EXPECTED, gettype($node));
+            throw new DOMException(DOMException::DOCUMENT_ELEMENT_DOCUMENTFRAG_EXPECTED, gettype($node));
         }
 
         # 13.3. Serializing HTML fragments

lib/DOM/traits/Walk.php

@@ -0,0 +1,22 @@
+<?php
+declare(strict_types=1);
+namespace dW\HTML5;
+
+trait Walk {
+    public function walk(\Closure $filter): \Generator {
+        return $this->walkGenerator($this, $filter);
+    }
+    
+    private function walkGenerator(\DOMNode $node, \Closure $filter) {
+        if ($filter($node)) {
+            yield $node;
+        }
+
+        if ($node->hasChildNodes()) {
+            $children = $node->childNodes;
+            foreach ($children as $c) {
+                yield from $this->walkGenerator($c, $filter);
+            }
+        }
+    }
+}

lib/Exception.php

@@ -3,68 +3,68 @@ declare(strict_types=1);
 namespace dW\HTML5;
 
 class Exception extends \Exception {
-    const INVALID_CODE = 10000;
-    const UNKNOWN_ERROR = 10001;
-    const INCORRECT_PARAMETERS_FOR_MESSAGE = 10002;
-    const UNREACHABLE_CODE = 10003;
-
-    const PARSER_NONEMPTY_DOCUMENT = 10101;
-
-    const STACK_INVALID_INDEX = 10201;
-    const STACK_ELEMENT_DOCUMENT_DOCUMENTFRAG_EXPECTED = 10202;
-    const STACK_ELEMENT_STRING_ARRAY_EXPECTED = 10203;
-    const STACK_STRING_ARRAY_EXPECTED = 10204;
-    const STACK_INCORRECTLY_EMPTY = 10205;
-    const STACK_INVALID_STATE = 10206;
-    const STACK_NO_CONTEXT_EXISTS = 10207;
-    const STACK_INVALID_VALUE = 10208;
-    const STACK_INVALID_OFFSET = 10209;
-    const STACK_ROOT_ELEMENT_DELETE = 10210;
-
-    const DATA_NODATA = 10301;
-    const DATA_INVALID_DATA_CONSUMPTION_LENGTH = 10302;
-
-    const TOKENIZER_INVALID_STATE = 10401;
-    const TOKENIZER_INVALID_CHARACTER_REFERENCE_STATE = 10402;
-
-    const TREEBUILDER_FORMELEMENT_EXPECTED = 10501;
-    const TREEBUILDER_DOCUMENTFRAG_ELEMENT_DOCUMENT_DOCUMENTFRAG_EXPECTED = 10502;
-    const TREEBUILDER_UNEXPECTED_END_OF_FILE = 10503;
-    const TREEBUILDER_NON_EMPTY_TARGET_DOCUMENT = 10504;
-    const TREEBUILDER_INVALID_TOKEN_CLASS = 10505;
-    const TREEBUILDER_INVALID_INSERTION_LOCATION = 10506;
+    const INVALID_CODE = 100;
+    const UNKNOWN_ERROR = 101;
+    const INCORRECT_PARAMETERS_FOR_MESSAGE = 102;
+    const UNREACHABLE_CODE = 103;
+
+    const PARSER_NONEMPTY_DOCUMENT = 201;
+
+    const STACK_INVALID_INDEX = 301;
+    const STACK_ELEMENT_DOCUMENT_DOCUMENTFRAG_EXPECTED = 302;
+    const STACK_ELEMENT_STRING_ARRAY_EXPECTED = 303;
+    const STACK_STRING_ARRAY_EXPECTED = 304;
+    const STACK_INCORRECTLY_EMPTY = 305;
+    const STACK_INVALID_STATE = 306;
+    const STACK_NO_CONTEXT_EXISTS = 307;
+    const STACK_INVALID_VALUE = 308;
+    const STACK_INVALID_OFFSET = 309;
+    const STACK_ROOT_ELEMENT_DELETE = 310;
+
+    const DATA_NODATA = 401;
+    const DATA_INVALID_DATA_CONSUMPTION_LENGTH = 402;
+
+    const TOKENIZER_INVALID_STATE = 501;
+    const TOKENIZER_INVALID_CHARACTER_REFERENCE_STATE = 502;
+
+    const TREEBUILDER_FORMELEMENT_EXPECTED = 601;
+    const TREEBUILDER_DOCUMENTFRAG_ELEMENT_DOCUMENT_DOCUMENTFRAG_EXPECTED = 602;
+    const TREEBUILDER_UNEXPECTED_END_OF_FILE = 603;
+    const TREEBUILDER_NON_EMPTY_TARGET_DOCUMENT = 604;
+    const TREEBUILDER_INVALID_TOKEN_CLASS = 605;
+    const TREEBUILDER_INVALID_INSERTION_LOCATION = 606;
 
     protected static $messages = [
-        10000 => 'Invalid error code',
-        10001 => 'Unknown error; escaping',
-        10002 => 'Incorrect number of parameters for Exception message; %s expected',
-        10003 => 'Unreachable code',
-
-        10101 => 'Non-empty Document supplied as argument for Parser',
-
-        10201 => 'Invalid Stack index at %s',
-        10202 => 'Element, Document, or DOMDocumentFragment expected for fragment context',
-        10203 => 'Element, string, or array expected',
-        10204 => 'String or array expected',
-        10205 => 'Stack is incorrectly empty',
-        10206 => 'Stack is in an invalid state; dump: %s',
-        10207 => 'No %s context exists in stack',
-        10208 => 'Stack value is invalid',
-        10209 => 'Invalid stack offset; offset must be %s',
-        10210 => 'Root element cannot be deleted from the stack',
-
-        10301 => 'Data string expected; found %s',
-        10302 => '%s is an invalid data consumption length; a value of 1 or above is expected',
-
-        10401 => 'The Tokenizer has entered an invalid state: %s',
-        10402 => 'Invalid character reference consumption state: %s',
-
-        10501 => 'Form element expected, found %s',
-        10502 => 'Element, Document, or DOMDocumentFragment expected; found %s',
-        10503 => 'Unexpected end of file',
-        10504 => 'Target document is not empty',
-        10505 => 'Invalid token class: %s',
-        10506 => 'Invalid insertion location'
+        100 => 'Invalid error code',
+        101 => 'Unknown error; escaping',
+        102 => 'Incorrect number of parameters for Exception message; %s expected',
+        103 => 'Unreachable code',
+
+        201 => 'Non-empty Document supplied as argument for Parser',
+
+        301 => 'Invalid Stack index at %s',
+        302 => 'Element, Document, or DOMDocumentFragment expected for fragment context',
+        303 => 'Element, string, or array expected',
+        304 => 'String or array expected',
+        305 => 'Stack is incorrectly empty',
+        306 => 'Stack is in an invalid state; dump: %s',
+        307 => 'No %s context exists in stack',
+        308 => 'Stack value is invalid',
+        309 => 'Invalid stack offset; offset must be %s',
+        310 => 'Root element cannot be deleted from the stack',
+
+        401 => 'Data string expected; found %s',
+        402 => '%s is an invalid data consumption length; a value of 1 or above is expected',
+
+        501 => 'The Tokenizer has entered an invalid state: %s',
+        502 => 'Invalid character reference consumption state: %s',
+
+        601 => 'Form element expected, found %s',
+        602 => 'Element, Document, or DOMDocumentFragment expected; found %s',
+        603 => 'Unexpected end of file',
+        604 => 'Target document is not empty',
+        605 => 'Invalid token class: %s',
+        606 => 'Invalid insertion location'
     ];
 
     public function __construct(int $code, ...$args) {