3.0.0

Breaking changes

• Some tweaks to the tokenizer/attribute selector parsing mean that whitespace locations might be slightly different to the 2.x code.
• Better attribute selector parsing with more validation; postcss-selector-parser no longer uses regular expressions to parse attribute selectors.
• Added an async API (thanks to @jacobp100); the default process API is now async, and the sync API is now accessed through processSync instead.
• process() and processSync() now return a string instead of the Processor instance.
• Tweaks handling of Less interpolation (thanks to @jwilsson).
• Removes support for Node 0.12.

Other changes

• ast() and astSync() methods have been added to the Processor. These return the Root node of the selectors after processing them.
• transform() and transformSync() methods have been added to the Processor. These return the value returned by the processor callback after processing the selectors.
• Set the parent when inserting a node (thanks to @chriseppstein).
• Correctly adjust indices when using insertBefore/insertAfter (thanks to @tivac).
• Fixes handling of namespaces with qualified tag selectors.
• process, ast and transform (and their sync variants) now accept a postcss rule node. When provided, better errors are generated and selector processing is automatically set back to the rule selector (unless the updateSelector option is set to false.)
• Now more memory efficient when tokenizing selectors.

Upgrade hints

The pattern of:

rule.selector = processor.process(rule.selector).result.toString();

is now:

processor.processSync(rule)

2.2.3

• Resolves an issue where the parser would not reduce multiple spaces between an ampersand and another simple selector in lossy mode (thanks to @adam-26).

2.2.2

• No longer hangs on an unescaped semicolon; instead the parser will throw an exception for these cases.

2.2.1

• Allows a consumer to specify whitespace tokens when creating a new Node (thanks to @Semigradsky).

2.2.0

• Added a new option to normalize whitespace when parsing the selector string (thanks to @adam-26).

2.1.1

• Better unquoted value handling within attribute selectors (thanks to @evilebottnawi).

2.1.0

• Added: Use string constants for all node types & expose them on the main parser instance (thanks to @Aweary).

2.0.0

This release contains the following breaking changes:

• Renamed all eachInside iterators to walk. For example, eachTag is now walkTags, and eachInside is now walk.
• Renamed Node#removeSelf() to Node#remove().
• Renamed Container#remove() to Container#removeChild().
• Renamed Node#raw to Node#raws (thanks to @davidtheclark).
• Now parses & as the nesting selector, rather than a tag selector.
• Fixes misinterpretation of Sass interpolation (e.g. #{foo}) as an id selector (thanks to @davidtheclark).

and;

• Fixes parsing of attribute selectors with equals signs in them (e.g. [data-attr="foo=bar"]) (thanks to @montmanu).
• Adds quoted and raw.unquoted properties to attribute nodes (thanks to @davidtheclark).

1.3.3

• Fixes an infinite loop on ) and ] tokens when they had no opening pairs. Now postcss-selector-parser will throw when it encounters these lone tokens.