lexers/pure.lua

   1 -- Copyright 2015-2016 David B. Lamkins <david@lamkins.net>. See LICENSE.
   2 -- pure LPeg lexer, see http://purelang.bitbucket.org/
   3
   4 local l = require('lexer')
   5 local token, word_match = l.token, l.word_match
   6 local P, R, S = lpeg.P, lpeg.R, lpeg.S
   7
   8 local M = {_NAME = 'pure'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Comments.
  14 local line_comment = '//' * l.nonnewline^0
  15 local block_comment = '/*' * (l.any - '*/')^0 * P('*/')^-1
  16 local comment = token(l.COMMENT, line_comment + block_comment)
  17
  18 -- Strings.
  19 local string = token(l.STRING, l.delimited_range('"', true))
  20
  21 -- Numbers.
  22 local bin = '0' * S('Bb') * S('01')^1
  23 local hex = '0' * S('Xx') * (R('09') + R('af') + R('AF'))^1
  24 local dec = R('09')^1
  25 local int = (bin + hex + dec) * P('L')^-1
  26 local rad = P('.') - P('..')
  27 local exp = (S('Ee') * S('+-')^-1 * int)^-1
  28 local flt = int * (rad * dec)^-1 * exp + int^-1 * rad * dec * exp
  29 local number = token(l.NUMBER, flt + int)
  30
  31 -- Keywords.
  32 local keyword = token(l.KEYWORD, word_match{
  33   'namespace', 'with', 'end', 'using', 'interface', 'extern', 'let', 'const',
  34   'def', 'type', 'public', 'private', 'nonfix', 'outfix', 'infix', 'infixl',
  35   'infixr', 'prefix', 'postfix', 'if', 'otherwise', 'when', 'case', 'of',
  36   'then', 'else'
  37 })
  38
  39 -- Identifiers.
  40 local identifier = token(l.IDENTIFIER, l.word)
  41
  42 -- Operators.
  43 local punct = S('+-/*%<>~!=^&|?~:;,.()[]{}@#$`\\\'')
  44 local dots = P('..')
  45 local operator = token(l.OPERATOR, dots + punct)
  46
  47 -- Pragmas.
  48 local hashbang = l.starts_line('#!') * (l.nonnewline - P('//'))^0
  49 local pragma = token(l.PREPROCESSOR, hashbang)
  50
  51 M._rules = {
  52   {'whitespace', ws},
  53   {'comment', comment},
  54   {'pragma', pragma},
  55   {'keyword', keyword},
  56   {'number', number},
  57   {'operator', operator},
  58   {'identifier', identifier},
  59   {'string', string},
  60 }
  61
  62 return M