MwParserFromScratch

• HtmlTag.ToPlainTextCore: Write \n when formatting <br /> or <hr /> into plaintext.

• Fixed CloneCore implementation for various Node classes. (#18 by @FaFre)
• TemplateArgumentCollection.SetValue now accepts string as argument name.
• Added Wikitext constructor overload shorthands that accept string (as PlainText) or InlineNode instances.
• Dropped explicit support for .NET Framework. We are planning to drop support for lower .NET Standard version in the future in order to adopt new C# language features.

• Bug fix: NullReferenceException when calling Node.ToPlainText on HtmlTags with empty content. (#15)

• Add support for embedded image expression parsing. AST node type: WikiImageLink. (#13)

• Refactored Node.ToPlainText to receive an optional delegate for customizing how a Node should be converted into plain text. (#15)
  • Removed NodePlainTextOptions to this end.
• The following parser tags are not shown in ToPlainText output
  • math
  • ref
  • templatedata
  • templatestyles
  • To change this default behavior, refer to #15.

• Force to end all the unbalanced HTML tags at the end of WIKITEXT. (#14)
  • The old behavior is to leave tags as unparsed PLAIN_TEXT. The old behavior can cause performance issue when parsing unbalanced tags, such as #3.
• Take care of unbalanced <li> tag parsing. (#14)
• Unbalanced tags will have TagStyle set to TagStyle.NotClosed.

• Fixed #12. Now parser handles tags with uppercase letters properly.
• Prepare for the wikitext table expression support
  • Added TagAttributeCollection, which in the future will be used in TagNode as well as in TableContentNode
  • Converted InlineContainer abstract class to IInlineContainer interface
  • Added preliminary Table class.
  • The table parsing support is still yet to be implemented.

• (#6) Implemented a rudimentary closing mark inference feature.
• (#7) Migrated IWikitextSpanInfo from offset-based to line/column-based.
  • Renamed IWikitextSpanInfo to IWikitextLineInfo.
• Requires LIST_ITEM starts at the beginning of the line.
  • the * item in {{Template|* item}} will be parsed as plain-text rather than unordered list.
• WikitextParser is now thread-safe.
  • As long as you do not call Parse when changing the content of WikitextParserOptions.

The attached "debug" package is built in DEBUG mode, and may provide some extra runtime assertions (Debug.Assert).

• Fixed missing LineInfo of PlainText in Paragraph, especially when a \n is in the paragraph content.

• Now WikitextParser.Parse() supports CancellationToken.
• Fxied NullReferenceException when WikitextParser is instantiated with IWikitextParserLogger.

• Use MagicTemplateNames to take the place of VariableNames in WikitextParserOptions.
  • You can specifiy whether a variable or parser function is case-sensitive.
• Improved debugger view of NodeCollection.
• Fixed the bug of presence of multiple LineInfoAnnotation in parsed Node instances.
• Fixed the bug of ParentNode == null for HTML tag content Wikitext node.

Maybe it's time to publish a real release.

• The NuGet package now has two target platforms: .NET Framework 4.5 and .NET Standard 1.1.
• Template can now distingish parser functions, and variables, from normal template expressions.
  • You will be able to extract the first argument from {{#if:expr|yes|no}}, rather than getting a Template whose Name is #if:expr.
• Fixed the missing LineInfo for ExternalLink & TagAttribute nodes.
• NormalizeTitle & NormalizeTemplateArgumentName now accepts null values, and they will return null for such cases.

• Add NodeCollection.AddFirst & InlineContainer.Prepend methods.
• Bug fix: Incorrect implementation of INodeCollection.Remove.
• Bug fix: Incorrect decision of template argument name for positional arguments.

• Now NormalizeTitlePart recognizes all whitespace characters, including \r, \n, in addition to space, so that the leading and trailing line carry will be stripped in the normalization process.

Initial release.