v0.6.1

Improvements

• Better error reporting for some cases when parsing fails

Bugfixes

• Fixed a corner-case of separated and split combinators failing to parse valid input

v0.6.0

New

• Parsus uses a new "scannerless" parsing approach to avoid token declaration order problems
  • The tokens can be declared in any order now!
  • However, this could be breaking and require reordering alternatives in the parsers like p1 or p2 or p3
• Explicit parsing of ignored tokens is now possible.
  • This can be handy, for instance, to enforce whitespace between certain things.
• Unmatched-token errors are now user-friendly
  • They include a section of the input to pinpoint problems faster, and also previously matched token for context

Deprecated

• Using currentToken is deprecated, because it will always be null due to avoidance of the eager tokenization in the "scannerless" approach.

v0.5.5

Fixes

• Non-literal token declaration order is now honored, same as it already was for literal tokens.
  If there was a regex token declared before a literal token, and a literal token was a prefix of the regex, the parser would have always chosen the literal, which is not correct.

v0.5.4

New

• New parseOrElse and parseOrNull shorthand functions for Grammar
• Case-insensitive parsing of literal and regex tokens with ignoreCase=true parameter
• Case-insensitive grammars with Grammar<T>(ignoreCase = true) (makes all literal and regex tokens case-insensitive)
• Parsing with non-root parser is now possible with g.parse(g.someParser, text) for easier testing and debugging
• High level and combinator to type-safely parse sequences of parses up to Tuple7
• New maybe alias for optional

Breaking

• parser1 and parser2 now returns Parser<Tuple2<T1, T2>> instead of Parser<Pair<T1, T2>>

Deprecations

• Parsing functions that start with parseEntire* are deprecated in favor of shorter parse* functions.
• Unary operators on the ParsingContext are deprecated, because they have a different semantics for the same types compared to Grammar scope. Explicit functions like skip or has should be used instead.

v0.4.0

New Combinators

and, optional, repeated, zeroOrMore, oneOrMore, separated, leftAssociative, rightAssociative

Most utilities that were available inside the parser { ... } closure are now also accessible as actual parser combinators.

    val keyValue by str * -colon and ref(::jsonValue)
    val jsonObj by -lbrace * separated(keyValue, comma) * -rbrace map { Json.Obj(it.toMap()) }

v0.3.0

First release with Kotlin Multiplatform support.

Big thanks to @aSemy for making this happen!

v0.1.4

Publish the library to Maven Central. See README for details.