lexers/cpp.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 = 'cpp'}
   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', 'error', 'if', 'ifdef', 'ifndef', 'import',
  29   'include', 'line', 'pragma', 'undef', 'using', 'warning'
  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   'asm', 'auto', 'break', 'case', 'catch', 'class', 'const', 'const_cast',
  37   'continue', 'default', 'delete', 'do', 'dynamic_cast', 'else', 'explicit',
  38   'export', 'extern', 'false', 'for', 'friend', 'goto', 'if', 'inline',
  39   'mutable', 'namespace', 'new', 'operator', 'private', 'protected', 'public',
  40   'register', 'reinterpret_cast', 'return', 'sizeof', 'static', 'static_cast',
  41   'switch', 'template', 'this', 'throw', 'true', 'try', 'typedef', 'typeid',
  42   'typename', 'using', 'virtual', 'volatile', 'while',
  43   -- Operators
  44   'and', 'and_eq', 'bitand', 'bitor', 'compl', 'not', 'not_eq', 'or', 'or_eq',
  45   'xor', 'xor_eq',
  46   -- C++11
  47   'alignas', 'alignof', 'constexpr', 'decltype', 'final', 'noexcept',
  48   'override', 'static_assert', 'thread_local'
  49 })
  50
  51 -- Types.
  52 local type = token(l.TYPE, word_match{
  53   'bool', 'char', 'double', 'enum', 'float', 'int', 'long', 'short', 'signed',
  54   'struct', 'union', 'unsigned', 'void', 'wchar_t',
  55   -- C++11
  56   'char16_t', 'char32_t', 'nullptr'
  57 })
  58
  59 -- Identifiers.
  60 local identifier = token(l.IDENTIFIER, l.word)
  61
  62 -- Operators.
  63 local operator = token(l.OPERATOR, S('+-/*%<>!=^&|?~:;,.()[]{}'))
  64
  65 M._rules = {
  66   {'whitespace', ws},
  67   {'keyword', keyword},
  68   {'type', type},
  69   {'identifier', identifier},
  70   {'string', string},
  71   {'comment', comment},
  72   {'number', number},
  73   {'preproc', preproc},
  74   {'operator', operator},
  75 }
  76
  77 M._foldsymbols = {
  78   _patterns = {'%l+', '[{}]', '/%*', '%*/', '//'},
  79   [l.PREPROCESSOR] = {
  80     region = 1, endregion = -1,
  81     ['if'] = 1, ifdef = 1, ifndef = 1, endif = -1
  82   },
  83   [l.OPERATOR] = {['{'] = 1, ['}'] = -1},
  84   [l.COMMENT] = {['/*'] = 1, ['*/'] = -1, ['//'] = l.fold_line_comments('//')}
  85 }
  86
  87 return M