lexers/ansi_c.lua

   1 -- Copyright 2006-2015 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- C 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 = 'ansi_c'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Comments.
  14 local line_comment = '//' * l.nonnewline_esc^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 sq_str = P('L')^-1 * l.delimited_range("'", true)
  20 local dq_str = P('L')^-1 * l.delimited_range('"', true)
  21 local string = token(l.STRING, sq_str + dq_str)
  22
  23 -- Numbers.
  24 local number = token(l.NUMBER, l.float + l.integer)
  25
  26 -- Preprocessor.
  27 local preproc_word = word_match{
  28   'define', 'elif', 'else', 'endif', 'if', 'ifdef', 'ifndef', 'include', 'line',
  29   'pragma', 'undef'
  30 }
  31 local preproc = token(l.PREPROCESSOR,
  32                       l.starts_line('#') * S('\t ')^0 * preproc_word)
  33
  34 -- Keywords.
  35 local keyword = token(l.KEYWORD, word_match{
  36   'auto', 'break', 'case', 'const', 'continue', 'default', 'do', 'else',
  37   'extern', 'for', 'goto', 'if', 'inline', 'register', 'restrict', 'return',
  38   'sizeof', 'static', 'switch', 'typedef', 'volatile', 'while'
  39 })
  40
  41 -- Types.
  42 local type = token(l.TYPE, word_match{
  43   'char', 'double', 'enum', 'float', 'int', 'long', 'short', 'signed', 'struct',
  44   'union', 'unsigned', 'void', '_Bool', '_Complex', '_Imaginary'
  45 })
  46
  47 -- Identifiers.
  48 local identifier = token(l.IDENTIFIER, l.word)
  49
  50 -- Operators.
  51 local operator = token(l.OPERATOR, S('+-/*%<>~!=^&|?~:;,.()[]{}'))
  52
  53 M._rules = {
  54   {'whitespace', ws},
  55   {'keyword', keyword},
  56   {'type', type},
  57   {'identifier', identifier},
  58   {'string', string},
  59   {'comment', comment},
  60   {'number', number},
  61   {'preproc', preproc},
  62   {'operator', operator},
  63 }
  64
  65 M._foldsymbols = {
  66   _patterns = {'%l+', '[{}]', '/%*', '%*/', '//'},
  67   [l.PREPROCESSOR] = {['if'] = 1, ifdef = 1, ifndef = 1, endif = -1},
  68   [l.OPERATOR] = {['{'] = 1, ['}'] = -1},
  69   [l.COMMENT] = {['/*'] = 1, ['*/'] = -1, ['//'] = l.fold_line_comments('//')}
  70 }
  71
  72 return M