parse

• JS: add variable scope to switch statement
• JS: parse async(a) as possible expression or arrow func
• JS: fix lexical variable renaming for descoped
• JS: fix import statement/call parsing at module level

• JS: add expression level maximum for expression suffices as well (from fuzz)

• JS: fix import statement/call parsing at module level

• CSS: proper unicode range parsing with wildcards

• JS: accept shebang comments at the start of the file

JS: add support for class field definitions (see https://github.com/tc39/proposal-class-fields)

JS: add &&= ||= ??= operators

• JS: register directive prologues ("use strict") as a separate statement

• JS: support ellipsis operator anywhere in arguments

• JS: fix declarations with the same name as in the for initializer, see #69
• JS: fix var declarations not adding their usage in all the scopes between its block scope and the function scope

• JS: put limits to the number of levels in statements and in expressions
• JS: fix import/export statements with one element in list

• JS: fix errors messages that contain % symbols

Fix position of error in error context with unicode characters

• JS: add var decl as undeclared in block scopes between the current scope and the function scope, fixes tdewolff/minify#334

• DataURI: encode \r, \n and \t using URL encoding, fixes tdewolff/minify#329
• DecodeURL: don't decode when percent is followed by non-hexadecimal character
• JS: remove trailing commas in var [a,,]
• JS: variable declaration using BindingPattern must be followed by an assignment
• JS: improve parsing of in operator in for statements and arrow function bodies

• JS: mark number of declared variables in for initializers in Scope

• JS: allow parsing in parameters in group expressions (that couldn't have been an arrow function)
• JS: store all comments in the head of the document (for license comments)
• JS: fix parsing identifiers in arrow functions, it used to assume they were part of the parameter list causing nil errors

• Add JS (ECMAScript 2020) lexer and parser.
• Add parse.Input and use that instead of io.Reader for input to lexers.
• Use predefined byte-slices instead of inline ones.

For users, there is a slight breakage of function signatures in this release: instead of calling for example p := css.NewParser(r, false) you should call

input := parse.NewInput(r)
defer input.Restore()
p := css.NewParser(input, false)

or if the byte-slice behind r is not used anywhere else (which is generally the case), then you could just call

p := css.NewParser(parse.NewInput(r), false)

This is for all lexers/parsers.

Update URLEncodingTable and add DataURIEncodingTable for better URL/URI encoding.

• CSS: fix slow execution of selector with many errors
• Fix panic on MediaType that ends in a space
• Fix parsing + or - as integers, and don't parse the exponent if there is no number
• Fix parsing e as a float
• Moving hash tables/entities maps to minify repository
• JS: improved performance