commonmark

Added

• Added ability to invoke Converter as a function (#233, #239)
• Added new advanceBySpaceOrTab convenience method to Cursor

Changed

• Bumped spec target version to 0.25
• Adjusted how tabs are handled by the Cursor (#234)
• Made a couple small micro-optimizations to heavily used functions (#240)
• Updated URLs in docblocks to use HTTPS where possible (#238)

Changed

• Refactored EmphasisParser::parse() to simplify it (#223)
• Updated dev dependencies (#218 & #220)

Fixed

• Fixed invalid regex generated when no inline parsers are defined (#224)
• Fixed logic bug with blank line after empty list item (#230)
• Fixed some incorrect code comments

Removed

• Removed unused variables (#223)

Added

• Added AST document processors (#210)
• Added optional Environment parameter to CommonMarkConverter constructor

Changed

• Renamed "header" things to "heading" for spec consistency
  • Header => Heading
  • ATXHeaderParser => ATXHeadingParser
  • SetExtHeaderParser => SetExtHeadingParser
  • HeaderRenderer => HeadingRenderer
• Renamed "HorizontalRule" to "ThematicBreak" for spec consistency
  • HorizontalRule => ThematicBreak
  • HorizontalRuleParser => ThematicBreakParser
  • HorizontalRuleRenderer => ThematicBreakRenderer
  • HorizontalRuleRendererTest => ThematicBreakRendererTest
  • RegexHelper::getHRuleRegex() => RegexHelper::getThematicBreakRegex()
• Renamed inline "Html" and "RawHtml" to "HtmlInline" for consistency
  • Html => HtmlInline
  • RawHtmlParser => HtmlInlineParser
  • RawHtmlRenderer => HtmlInlineRenderer
• Don't allow whitespace between link text and link label of a reference link (spec change)
• Don't allow spaces in link destinations, even in <>
• Allow multiline setext header content
  • The Heading constructor now allows $contents to be a string (old behavior) or string[] (new)

Fixed

• Fixed several list issues and regressions (jgm/commonmark.js#59)

Removed

• Removed schema whitelist from autolink regex
• Moved SmartPunct functionality into new league/commonmark-extras package

Added

• Added ability to configure characters and disable emphasis/strong (#135)
• Added new ConfigurationAwareInterface support for all parsers, processors, and renderers (#201)
• Added HTML safe mode to handle untrusted input (#200, #201)
  • Safe mode is disabled by default for backwards-compatibility
  • To enable it, set the safe option to true
• Added AppVeyor integration for automated unit/functional testing on Windows (#195)

Changed

• AbstractBlock::finalize() now reqires a second parameter, $endLineNumber
• RegexHelper::REGEX_ENTITY no longer includes the starting / or the ending /i (#194)
• Node::setParent() now accepts null values (#203)

Fixed

• Fixed incorrect endLine positions (#187)
• Fixed DocParser::preProcessInput dropping up to 2 ending newlines instead of just one
• Fixed EntityParser not checking for ampersands at the start of the current position (#192, #194)

Removed

• Removed protected function Context::addChild()
  • It was a duplicate of the Context::addBlock() method
• Disabled STDIN reading on bin/commonmark for Windows due to PHP issues (#189, #195)

Fixed

• Reset container after closing containing lists (#183; jgm/commonmark.js#67)
  • The temporary fix from 0.11.2 was reverted

Fixed

• Fixed parser checking acceptsLines on the wrong element (#183)

Changed

• Tightened up some loose comparisons

Fixed

• Fixed missing "bin" directive in composer.json
• Updated a docblock to match recent changes to method parameters

Removed

• Removed unused variable from within QuoteProcessor's closure

This release contains several major internal changes. It will likely break compatibility with custom elements, parsers, and renderers. Simple Markdown parsing is unaffected.

Added

• Added new Node class, which both AbstractBlock and AbstractInline extend from (#169)
• Added a NodeWalker and NodeWalkerEvent to traverse the AST without using recursion
• Added new InlineContainer interface for blocks
• Added new getContainer() and getReferenceMap() methods to InlineParserContext
• Added iframe to whitelist of HTML block tags (as per spec)
• Added ./bin/commonmark for converting Markdown at the command line

Changed

• Bumped spec target version to 0.22
• Revised AST to use a double-linked list (#169)
• AbstractBlock and AbstractInline both extend from Node
  • Sub-classes must implement new isContainer() method
• Other major changes to AbstractBlock:
  • getParent() is now parent()
  • setParent() now expects a Node instead of an AbstractBlock
  • getChildren() is now children()
  • getLastChild() is now lastChild()
  • addChild() is now appendChild()
• InlineParserContext is constructed using the container AbstractBlock and the document's RefereceMap
  • The constructor will automatically create the Cursor using the container's string contents
• InlineParserEngine::parse now requires the Node container and the document's ReferenceMap instead of a ContextInterface and Cursor
• Changed Delimiter to reference the actual inline Node instead of the position
  • The int $pos protected member and constructor arg is now Node $node
  • Use getInlineNode() and setInlineNode() instead of getPos() and setPos()
• Changed DocParser::processInlines to use a NodeWalker to iterate through inlines
  • Walker passed as second argument instead of AbstractBlock
  • Uses a while loop instead of recursion to traverse the AST
• Image and Link now only accept a string as their second argument
• Refactored how CloseBracketParser::parse() works internally
• CloseBracketParser::createInline no longer accepts label inlines
• Disallow list item starting with multiple blank lines (see jgm/CommonMark#332)
• Modified AbstractBlock::setLastLineBlank()
  • Functionality moved to AbstractBlock::shouldLastLineBeBlank() and new DocParser::setAndPropagateLastLineBlank() method
  • AbstractBlock::setLastLineBlank() is now a setter method for AbstractBlock::$lastLineBlank
• AbstractBlock::handleRemainingContents() is no longer abstract
  • A default implementation is provided
  • Removed duplicate code from sub-classes which used the default implementation - they'll just use the parent method from now on

Fixed

• Fixed logic error in calculation of offset (see jgm/commonmark.js@94053a8)
• Fixed bug where DocParser checked the wrong method to determine remainder handling behavior
• Fixed bug where HorizontalRuleParser failed to advance the cursor beyond the parsed horizontal rule characters
• Fixed DocParser not ignoring the final newline of the input (like the reference parser does)

Removed

• Removed Block\Element\AbstractInlineContainer
  • Extend AbstractBlock and implement InlineContainer instead
  • Use child methods instead of getInlines and setInlines
• Removed AbstractBlock::replaceChild()
  • Call Node::replaceWith() directly the child node instead
• Removed the getInlines() method from InlineParserContext
  • Add parsed inlines using $inlineContext->getContainer()->appendChild() instead of $inlineContext->getInlines()->add()
• Removed the ContextInterface argument from AbstractInlineParser::parse() and InlineParserEngine::parseCharacter
• Removed the first ArrayCollection $inlines argument from InlineProcessorInterface::processInlines()
• Removed CloseBracketParser::nullify()
• Removed pre from rule 6 of HTML blocks (see jgm/CommonMark#355)

Added

• Added parent references to inline elements (#124)
• Added smart punctuation extension (#134)
• Added HTML block types
• Added indentation caching to the cursor
• Added automated code style checks (#133)
• Added support for tag attributes in renderers (#101, #165)

Changed

• Bumped spec target version to 0.21
• Revised HTML block parsing to conform to new spec (jgm/commonmark.js@99bd473)
• Imposed 9-digit limit on ordered list markers, per spec
• Allow non-initial hyphens in html tag names (jgm/CommonMark#239)
• Updated list of block tag names
• Changed tab/indentation handling to meet the new spec behavior
• Modified spec tests to show spaces and tabs in test results
• Replaced HtmlRendererInterface with ElementRendererInterface (#141)
• Removed the unnecessary trim() and string cast from ListItemRenderer

Fixed

• Fixed link reference definition edge case (#120)
• Allow literal (non-escaping) backslashes in link destinations (#118)
• Allow backslash-escaped backslashes in link labels (#119)
• Allow link labels up to 999 characters (per the spec)
• Properly split on whitespace when determining code block class (jgm/commonmark.js#54)
• Fixed code style issues (#132, #133, #151, #152)
• Fixed wording for invalid inline exception (#136)

Removed

• Removed the advance-by-one optimization due to added cursor complexity

Added

• Added public $data array to block elements (#95)
• Added isIndented helper method to Cursor
• Added a new Converter base class which CommonMarkConverter extends from (#105)

Changed

• Bumped spec target version to 0.20 (#112)
• Renamed ListBlock::$data and ListItem::$data to $listData
• Require link labels to contain non-whitespace (jgm/CommonMark#322)
• Use U+FFFD for entities resolving to 0 (jgm/CommonMark#323)
• Moved IndentedCodeParser::CODE_INDENT_LEVEL to Cursor::INDENT_LEVEL
• Changed arrays to short syntax (#116)
• Improved efficiency of DelimiterStack iteration (jgm/commonmark.js#43)

Fixed

• Fixed open block tag followed by newline not being recognized (jgm/CommonMark#324)
• Fixed indented lists sometimes being parsed incorrectly (jgm/commonmark.js#42)

Added

• Allow swapping built-in renderers without using their fully qualified names (#84)
• Lots of unit tests (for existing code)
• Ability to include arbitrary functional tests in addition to spec-based tests

Changed

• Dropped support for PHP 5.3 (#64 and #76)
• Bumped spec target version to 0.19
• Made the AbstractInlineContainer be abstract
• Moved environment config. logic into separate class

Fixed

• Fixed underscore emphasis to conform to spec changes (jgm/CommonMark#317)

Removed

• Removed PHP 5.3 workaround (see commit 5747822)
• Removed unused AbstractWebResource::setUrl() method
• Removed unnecessary check for hrule when parsing lists (#85)

Changed

• Bumped spec target version to 0.18

Fixed

• Fixed broken parsing of emphasized text ending with a '0' character (#81)

Added

• All references can now be obtained from the ReferenceMap via listReferences() (#73)
• Test against PHP 7.0 (nightly) but allow failures

Changed

• ListData::$start now defaults to null instead of 0 (#74)
• Replace references to HtmlRenderer with new HtmlRendererInterface

Fixed

• Fixed 0-based ordered lists starting at 1 instead of 0 (#74)
• Fixed errors parsing multi-byte characters (#78 and #79)

Now with 50% more speed!

Added

• More unit tests to increase code coverage

Changed

• Enabled the InlineParserEngine to parse several non-special characters at once (performance boost)
• NewlineParser no longer attempts to parse spaces; look-behind is used instead (major performance boost)
• Moved closeUnmatchedBlocks into its own class
• Image and link elements now extend AbstractInlineContainer; label data is stored via $inlineContents instead
• Renamed AbstractInlineContainer::$inlineContents and its getter/setter

Removed

• Removed the InlineCollection class
• Removed the unused ArrayCollection::splice() method
• Removed impossible-to-reach code in Cursor::advanceToFirstNonSpace
• Removed unnecessary test from the InlineParserEngine
• Removed unnecessary/unused RegexHelper::getMainRegex() method

Changed

• Bumped spec target version to 0.17
• Updated emphasis parsing for underscores to prevent intra-word emphasis
• Defered closing of fenced code blocks

Added

• Bulk registration of parsers/renderers via extensions (#45)
• Proper UTF-8 support, especially in the Cursor; mbstring extension is now required (#49)
• Environment is now configurable; options can be accessed in its parsers/renderers (#56)
• Added some unit tests

Changed

• Bumped spec target version to 0.15 (#50)
• Parsers/renderers are now lazy-initialized (#52)
• Some private elements are now protected for easier extending, especially on Element classes (#53)
• Renderer option names changed from underscore_case to camelCase (#56)
• Moved CommonMark parser/render definitions into CommonMarkCoreExtension

Fixed

• Improved parsing of emphasis around punctuation
• Improved regexes for CDATA and HTML comments
• Fixed issue with HTML content that is considered false in loose comparisons, like '0' (#55)
• Fixed DocParser trying to add empty strings to closed containers (#58)
• Fixed incorrect use of a null parameter value in the HtmlElementTest

Removed

• Removed unused ReferenceDefinition* classes (#51)
• Removed UnicodeCaseFolder in favor of mb_strtoupper

Fixed

• Fixed infinite loop and link-in-link-in-image parsing (#37)

Removed

• Removed hard dependency on mbstring extension; workaround used if not installed (#38)

Added

• Support for custom directives, parsers, and renderers

Changed

• Major refactoring to de-couple directives from the parser, support custom directive functionality, and reduce complexity
• Updated references to stmd.js in README and docblocks
• Modified CHANGELOG formatting
• Improved travis configuration
• Put tests in autoload-dev

Fixed

• Fixed CommonMarkConverter re-creating object each time new text is converted (#26)

Removed

• Removed dependency on symfony/options-resolver (fixes #20)

• Changed namespace to League\CommonMark
• Made compatible with spec version 0.13
• Moved delimiter stack functionality into seperate class
• Fixed regex which caused HHVM tests to fail
• Added some missing copyright info