• Moved the ancestor and descendant methods into their own traits along with the compare method which they share.
• Made DocumentFragment use only the descendant methods and not the ancestor ones.
• Fixed error in README.
• Added an option to use Document::loadHTML or Document::load to parse a document.
• Made the DOM elements use dW\HTML5 namespace instead of dW\HTML5\DOM.
• Fixed where TreeBuilder wasn't being properly destructed when the parser is finished.
• Discovered a way to extend the existing DOM tools, so Parser now works with custom classes for the DOM.
• \dW\HTML5\DOM is no longer needed, and its functionality has been split between DOM\Document and DOM\Element.
• Printer's functionality is now handled by DOM\Document\saveHTML through a trait to make maintenance easier.
• Normalized Exception constants.
• Fixed various typing bugs.
• Oops. Didn't have the \ in front of the DOM classes.
• Forgot function keyword on Printer::escapeString.
• Finished out the spec on Printer::serialize.
• Fixed bug where Data::consumeWhile and Data::consumeUntil wouldn't move the pointer back one position if there were no matches.
• Changed DataStream to Data.
• Made each class have its own debug static property so each can print debug information separately.
• Removed dependency on ext-mbstring.
• Fixed bug where when using DataStream::consumeUntil and DataStream::consumeWhile where the column would be advanced whether the pointer was moved or not.
• Changed DataStream from using an internal array storage to using /mensbeam/intl, much more efficient and in line with the specification.
• DataStream now calculates the line and column as it goes instead of relying upon a weird array of positions.
• Removed dependency on ext-intl.
• The document was being rewritten when tree building and therefore not being output when the parser completed.
• Allowed DOM to be instanced, containing an implementation and document so the tree builder can create a document when a doctype is found.
• PHP's DOM returns null when the namespace isn't specified, so it was causing the dispatcher to parse tokens in foreign content when it wasn't supposed to.
• Added debugging output for TreeBuilder, showing the insertion mode and foreign if content is foreign.
• Fixed syntax errors that weren't caught in last commit
• Made the token insertion methods static in Parser so that ActiveFormattingElementsList::reconstruct can insert methods without having to access the Parser instance itself. Might do it differently later, but this works for now.
• Since the fragment context that is passed to Stack can be null it is necessary to check whether the argument is valid before continuing.
• Parser::instance is now protected as all classes are now decoupled.
Changing static:: to self:: makes constant de-referencing a
compile-time operation rather than run-time, potentially improving
performance. As constants cannot be overridden by extending classes,
there is no advantage to using static:: for these constants
• Changed the name of the parser instance variable from Parser::$self to Parser::$instance
• Added parse errors for entities into ParseError.
• Moved Parser::fixDOM to DOM::fixIdAttributes.
• Added an exception for when the tokenizer enters an invalid state (infinite looping).
• Made ParseError use Parser::$instance->data instead of a passed around DataStream object.
• Made names of storage property in Stack and ActiveFormattingElementsList consistent to elsewhere
• Parser has its own public static variable that is an intance of itself, so ActiveFormattingElementsList can use that instead of passing an instance around.
• Added separate exception for ActiveFormattingElementsList
• Removed html5.php; shouldn't have been there to begin with.
• Fixed bug where when feeding ParseError::trigger the wrong number of parameters it wouldn't have the correct exception to throw.
• Changes to the spec since the last edit required a rewrite of the tree building algorithm.
• Searching the stack should search from reverse by default because the spec works that way.
• Rewrote StartTagToken because the token attributes need to be easily editable as per the spec foreign attributes are edited before the token goes through the element creation process and not after.
• Yes, there's a goto. Sue me.
