Parsing Characters

The consumption of input is central to parsing. In parsley, the only way of consuming input is by consuming characters, as customised token streams are not supported. However, as parsley is based on Scala, which relies on 16-bit Basic-Multilingual Plane UTF16 characters, there are two modules for input consumption: character, for 16-bit Char, and unicode, handling 32-bit Ints.

Character Primitives

There are three key input consumption primitives in parsley: satisfy, which takes a predicate and parses any character for which that predicate returns true; char, which reads a specific character; and string, which reads a specific sequence of characters. Technically, char and string can be built in terms of satisfy, but would then lose their enhanced error message characteristics.

The oneOf and noneOf are based on satisfy but with better error messages, based on the kind of range of characters they are provided with. The strings and trie combinators can be used similarly to match a set of strings: they are likely to parse these more efficiently than a manual construction and may improve further in future.

String Building

It is possible to efficiently construct a String using the stringOfMany and stringOfSome combinators. This can be used with either a character predicate (which turns out to be called takeWhileP in megaparsec), or can be given a parser that consumes a character instead. The construction of the underlying string is done efficiently with a StringBuilder. Note that, with the version that takes a parser, the characters incorporated into the string are the ones returned by the parser and not necessarily the ones the parser consumed.

Pre-Built Character Parsers

Some pre-built parsers are bunched into character and unicode, which parse specific sets of letter, with good error messages. Each will document the specific set of characters they recognise.