lexers/forth.lua

   1 -- Copyright 2006-2015 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Forth LPeg lexer.
   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 = 'forth'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Comments.
  14 local line_comment = S('|\\') * 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 s_str = 's' * l.delimited_range('"', true, true)
  20 local dot_str = '.' * l.delimited_range('"', true, true)
  21 local f_str = 'f' * l.delimited_range('"', true, true)
  22 local dq_str = l.delimited_range('"', true, true)
  23 local string = token(l.STRING, s_str + dot_str + f_str + dq_str)
  24
  25 -- Numbers.
  26 local number = token(l.NUMBER, P('-')^-1 * l.digit^1 * (S('./') * l.digit^1)^-1)
  27
  28 -- Keywords.
  29 local keyword = token(l.KEYWORD, word_match({
  30   'swap', 'drop', 'dup', 'nip', 'over', 'rot', '-rot', '2dup', '2drop', '2over',
  31   '2swap', '>r', 'r>',
  32   'and', 'or', 'xor', '>>', '<<', 'not', 'negate', 'mod', '/mod', '1+', '1-',
  33   'base', 'hex', 'decimal', 'binary', 'octal',
  34   '@', '!', 'c@', 'c!', '+!', 'cell+', 'cells', 'char+', 'chars',
  35   'create', 'does>', 'variable', 'variable,', 'literal', 'last', '1,', '2,',
  36   '3,', ',', 'here', 'allot', 'parse', 'find', 'compile',
  37   -- Operators.
  38   'if', '=if', '<if', '>if', '<>if', 'then', 'repeat', 'until', 'forth', 'macro'
  39 }, '2><1-@!+3,='))
  40
  41 -- Identifiers.
  42 local identifier = token(l.IDENTIFIER, (l.alnum + S('+-*=<>.?/\'%,_$'))^1)
  43
  44 -- Operators.
  45 local operator = token(l.OPERATOR, S(':;<>+*-/()[]'))
  46
  47 M._rules = {
  48   {'whitespace', ws},
  49   {'keyword', keyword},
  50   {'string', string},
  51   {'identifier', identifier},
  52   {'comment', comment},
  53   {'number', number},
  54   {'operator', operator},
  55 }
  56
  57 return M