Parsimmon
(by @jayferd and @laughinghan)
Parsimmon is a small library for writing big parsers made up of lots of little parsers. The API is inspired by parsec and Promises/A.
Quick Example
var regex = Parsimmon.regex;
var string = Parsimmon.string;
var optWhitespace = Parsimmon.optWhitespace;
var lazy = Parsimmon.lazy;
function lexeme(p) { return p.skip(optWhitespace); }
var lparen = lexeme(string('('));
var rparen = lexeme(string(')'));
var expr = lazy(function() { return form.or(atom) });
var number = lexeme(regex(/[0-9]+/).map(parseInt));
var id = lexeme(regex(/[a-z_]\w*/i));
var atom = number.or(id);
var form = lparen.then(expr.many()).skip(rparen);
expr.parse('3').value
expr.parse('(add (mul 10 (add 3 4)) (add 7 8))').value
Explanation
A Parsimmon parser is an object that represents an action on a stream
of text, and the promise of either an object yielded by that action on
success or a message in case of failure. For example, string('foo')
yields the string 'foo'
if the beginning of the stream is 'foo'
,
and otherwise fails.
The combinator method .map
is used to transform the yielded value.
For example,
string('foo').map(function(x) { return x + 'bar'; })
will yield 'foobar'
if the stream starts with 'foo'
. The parser
digits.map(function(x) { return parseInt(x) * 2; })
will yield the number 24 when it encounters the string '12'. The method
.result
can be used to set a constant result.
Calling .parse(str)
on a parser parses the string, and returns an
object with a status
flag, indicating whether the parse succeeded.
If it succeeded, the value
attribute will contain the yielded value.
Otherwise, the index
and expected
attributes will contain the
index of the parse error, and a message indicating what was expected.
The error object can be passed along with the original source to
Parsimmon.formatError(source, error)
to obtain a human-readable
error string.
Full API
Included parsers / parser generators:
Parsimmon.string("my-string")
is a parser that expects to find
"my-string"
, and will yield the same.Parsimmon.regex(/myregex/)
is a parser that expects the stream
to match the given regex.Parsimmon.succeed(result)
is a parser that doesn't consume any of
the string, and yields result
.Parsimmon.seq(p1, p2, ... pn)
accepts a variable number of parsers
that it expects to find in order, yielding an array of the results.Parsimmon.alt(p1, p2, ... pn)
accepts a variable number of parsers,
and yields the value of the first one that succeeds, backtracking in between.Parsimmon.lazy(f)
accepts a function that returns a parser, which
is evaluated the first time the parser is used. This is useful for
referencing parsers that haven't yet been defined.Parsimmon.fail(message)
Parsimmon.letter
is equivalent to Parsimmon.regex(/[a-z]/i)
Parsimmon.letters
is equivalent to Parsimmon.regex(/[a-z]*/i)
Parsimmon.digit
is equivalent to Parsimmon.regex(/[0-9]/)
Parsimmon.digits
is equivalent to Parsimmon.regex(/[0-9]*/)
Parsimmon.whitespace
is equivalent to Parsimmon.regex(/\s+/)
Parsimmon.optWhitespace
is equivalent to Parsimmon.regex(/\s*/)
Parsimmon.any
consumes and yields the next character of the stream.Parsimmon.all
consumes and yields the entire remainder of the stream.Parsimmon.eof
expects the end of the stream.Parsimmon.index
is a parser that yields the current index of the parse.
Parser methods
parser.or(otherParser)
:
returns a new parser which tries parser
, and if it fails uses otherParser
.parser.chain(function(result) { return anotherParser; })
:
returns a new parser which tries parser
, and on success calls the
given function with the result of the parse, which is expected to
return another parser, which will be tried next. This allows you
to dynamically decide how to continue the parse, which is impossible
with the other combinators.parser.then(anotherParser)
:
expects anotherParser
to follow parser
, and yields the result
of anotherParser
. NB: the result of parser
here is ignored.parser.map(function(result) { return anotherResult; })
:
transforms the output of parser
with the given function.parser.skip(otherParser)
expects otherParser
after parser
, but preserves the yield value
of parser
.parser.result(aResult)
:
returns a new parser with the same behavior, but which yields aResult
.parser.many()
:
expects parser
zero or more times, and yields an array of the results.parser.times(n)
:
expects parser
exactly n
times, and yields an array of the results.parser.times(min, max)
:
expects parser
between min
and max
times, and yields an array
of the results.parser.atMost(n)
:
expects parser
at most n
times. Yields an array of the results.parser.atLeast(n)
:
expects parser
at least n
times. Yields an array of the results.parser.mark()
yields an object with start
, value
, and end
keys, where
value
is the original value yielded by the parser, and start
and end
are
the indices in the stream that contain the parsed text.
Tips and patterns
These apply to most parsers for traditional langauges - it's possible
you may need to do something different for yours!
For most parsers, the following format is helpful:
-
define a lexeme
function to skip all the stuff you don't care
about (whitespace, comments, etc). You may need multiple types of lexemes.
For example,
var ignore = whitespace.or(comment.many());
function lexeme(p) { return p.skip(ignore); }
-
Define all your lexemes first. These should yield native javascript values.
var lparen = lexeme(string('('));
var rparen = lexeme(string(')'));
var number = lexeme(regex(/[0-9]+/)).map(parseInt);
-
Forward-declare one or more top-level expressions with lazy
,
referring to parsers that have not yet been defined. Generally, this
takes the form of a large .alt()
call
var expr = lazy(function() { return Parsimmon.alt(p1, p2, ...); });
-
Then build your parsers from the inside out - these should return
AST nodes or other objects specific to your domain.
var p1 = ...
var p2 = ...
-
Finally, export your top-level parser. Remember to skip ignored
stuff at the beginning.
return ignore.then(expr.many());
Fantasyland
Parsimmon is also compatible with fantasyland. It is a Semigroup, an Applicative Functor and a Monad.