tl

• Fixed a bug that caused the parser to parse closing tags incorrectly. See y21/tl#37 and y21/tl#38 for more details.

• Restructure internals (mainly SIMD functions)
• Add fuzzing targets for internals
• Optimize stable parser (adds stable alternatives when the simd feature isn't set)

• Fixed HTMLTag::raw() returning one byte less than it should have. See y21/tl#31.

• Add Attributes::contains(key) to check if an attribute exists.
• Add Attributes::remove(key) to remove an attribute.
• Add Attributes::remove_value(key) to delete the value of a given attribute key.

• Version bump in README.md

Warning: This release contains breaking changes

• ⚠ Function signature of Attributes::insert has changed:
  • It now takes two generic parameters K, V instead of just one. Prior to this version, this meant that the key and value type had to match. See y21/tl#27 for more details.
• Added a TryFrom<String> for Bytes implementation for convenience to create owned Bytes.
• Added HTMLTag::boundaries method for obtaining the start and end position of a tag in the source string.
• Fixed a panic when source string abruptly ends with <tag/

• Fixed a bug where Attributes::class() returned its id attribute instead of class. See y21/tl#26

• Fixed a bug where the slash in slash tags (<br />) is interpreted as > and causes the next > to be interpreted as a text node on its own.

• Fixed an off-by-one error in the QueryIterable trait implementation for HTMLTag that caused query selectors on HTML tags to return one node less than they should.

Warning: This release contains breaking changes

• ⚠ Removed deprecated method VDom::find_node
  • Alternative: use VDom::nodes().iter().find(...) instead
• ⚠ Attributes::get() now returns a reference to Bytes instead of cloning.
  • Prior to this version, it wasn't necessary to return a reference as the Bytes type was just an immutable &[u8]. Now it can hold owned data.
• ⚠ HTMLTag::children() no longer returns an iterator, and instead returns a wrapper struct around the children of the HTML tag. This wrapper struct makes it easy to obtain direct children of the tag (Children::top()), or all children (including their children, etc...) (Children::all()).
• ⚠ Node::children() no longer returns an iterator (see above).
• ⚠ HTMLTag::name() now returns a reference to Bytes instead of cloning (see above).
• Ability to create/parse query selectors independent of any parser (tl::parse_query_selector)
• Ability to reuse query selectors
• Ability to apply query selectors on HTMLTags (see #18)
• queryselector module is now public
• InnerNodeHandle is now u32
• Remove unused max_depth parser option
• Add convenience PartialEq<str> and PartialEq<[u8]> impls for Bytes

Warning: This release contains breaking changes

• Allow Bytes to store owned data through Bytes::set()
  • ⚠ The maximum length for Bytes is u32::MAX
• ⚠ tl::parse() now returns Result<VDom<'a>, ParseError>
• ⚠ Attributes fields are no longer public, instead use one of the provided methods
• ⚠ HTMLTag::inner_html() now takes a &Parser and no longer directly returns the substring
  • Node mutations to the tag or any of its subnodes means inner_html needs to be recomputed
  • Consider using HTMLTag::raw() if you never mutate any nodes

• Parse unquoted attribute values properly (<a href=foo>) [#12]
• Parse valueless attributes properly (<iframe allowfullscreen>) [#11]

• Add optional simd feature flag for SIMD-accelerated parsing

• Keep track of recursion depth when parsing to avoid overflowing the stack

• Fixed an infinite loop bug in parse_single and a panic that occurs when trying to parse incomplete markup tags (<!)

• Remove AsBytes trait in favor of From<T> for Bytes
• Add VDom::query_selector()
• Introduce InnerHandleNode to hide implementation details of NodeHandle
• Make Attributes::is_class_member() generic over T: Into<Bytes<'a>>
• Add Attributes::get_attribute()
• Make HTMLTag::name() and HTMLTag::inner_html() return owned Bytes

• Deprecate VDom::find_node in favor of VDom::nodes().iter().find()