[2.5.0] - 2022-05-27 - Janosch Müller
Added
* Regexp::Expression::Base.construct and .token_class methods
o see the wiki for details
[2.4.0] - 2022-05-09 - Janosch Müller
Fixed
* fixed interpretation of + and ? after interval quantifiers ({n,n})
- they used to be treated as reluctant or possessive mode indicators
- however, Ruby does not support these modes for interval quantifiers
- they are now treated as chained quantifiers instead, as Ruby does it
- c.f. #3
* fixed Expression::Base#nesting_level for some tree rewrite cases
- e.g. the alternatives in /a|[b]/ had an inconsistent nesting_level
* fixed Scanner accepting invalid posix classes, e.g. [[:foo:]]
- they raise a SyntaxError when used in a Regexp, so could only be
passed as String
- they now raise a Regexp::Scanner::ValidationError in the Scanner
Added
* added Expression::Base#== for (deep) comparison of expressions
* added Expression::Base#parts
- returns the text elements and subexpressions of an expression
- e.g. parse(/(a)/)[0].parts # => ["(", #<Literal @text="a"...>, ")"]
* added Expression::Base#te (a.k.a. token end index)
- Expression::Subexpression always had #te, only terminal nodes
lacked it so far
* made some Expression::Base methods available on Quantifier instances, too
- #type, #type?, #is?, #one_of?, #options, #terminal?
- #base_length, #full_length, #starts_at, #te, #ts, #offset
- #conditional_level, #level, #nesting_level , #set_level
- this allows a more unified handling with Expression::Base instances
* allowed Quantifier#initialize to take a token and options Hash like
other nodes
* added a deprecation warning for initializing Quantifiers with 4+ arguments:
Calling Expression::Base#quantify or Quantifier.new with 4+
arguments is deprecated.
It will no longer be supported in regexp_parser v3.0.0.
Please pass a Regexp::Token instead, e.g. replace token, text, min,
max, mode with ::Regexp::Token.new(:quantifier, token, text). min,
max, and mode will be derived automatically.
Or do exp.quantifier = Quantifier.construct(token: token, text: str).
This is consistent with how Expression::Base instances are created.
[2.3.1] - 2022-04-24 - Janosch Müller
Fixed
* removed five inexistent unicode properties from Syntax#features
- these were never supported by Ruby or the Regexp::Scanner
- thanks to Markus Schirp for the report
[2.3.0] - 2022-04-08 - Janosch Müller
Added
* improved parsing performance through Syntax refactoring
- instead of fresh Syntax instances, pre-loaded constants are now re-used
- this approximately doubles the parsing speed for simple regexps
* added methods to Syntax classes to show relative feature sets
- e.g. Regexp::Syntax::V3_2_0.added_features
* support for new unicode properties of Ruby 3.2 / Unicode 14.0
875ab13421