lexers/nemerle.lua

   1 -- Copyright 2006-2015 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Nemerle 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 = 'nemerle'}
   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', 'endregion', 'error', 'if', 'ifdef',
  29   'ifndef', 'line', 'pragma', 'region', '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   '_', 'abstract', 'and', 'array', 'as', 'base', 'catch', 'class', 'def', 'do',
  37   'else', 'extends', 'extern', 'finally', 'foreach', 'for',  'fun', 'if',
  38   'implements', 'in', 'interface', 'internal', 'lock', 'macro', 'match',
  39   'module', 'mutable', 'namespace', 'new', 'out', 'override', 'params',
  40   'private', 'protected', 'public', 'ref', 'repeat', 'sealed', 'static',
  41   'struct', 'syntax', 'this', 'throw', 'try', 'type', 'typeof', 'unless',
  42   'until', 'using', 'variant', 'virtual', 'when', 'where', 'while',
  43   -- Values.
  44   'null', 'true', 'false'
  45 })
  46
  47 -- Types.
  48 local type = token(l.TYPE, word_match{
  49   'bool', 'byte', 'char', 'decimal', 'double', 'float', 'int', 'list', 'long',
  50   'object', 'sbyte', 'short', 'string', 'uint', 'ulong', 'ushort', 'void'
  51 })
  52
  53 -- Identifiers.
  54 local identifier = token(l.IDENTIFIER, l.word)
  55
  56 -- Operators.
  57 local operator = token(l.OPERATOR, S('+-/*%<>!=^&|?~:;.()[]{}'))
  58
  59 M._rules = {
  60   {'whitespace', ws},
  61   {'keyword', keyword},
  62   {'type', type},
  63   {'identifier', identifier},
  64   {'string', string},
  65   {'comment', comment},
  66   {'number', number},
  67   {'preproc', preproc},
  68   {'operator', operator},
  69 }
  70
  71 M._foldsymbols = {
  72   _patterns = {'%l+', '[{}]', '/%*', '%*/', '//'},
  73   [l.PREPROCESSOR] = {
  74     region = 1, endregion = -1,
  75     ['if'] = 1, ifdef = 1, ifndef = 1, endif = -1
  76   },
  77   [l.OPERATOR] = {['{'] = 1, ['}'] = -1},
  78   [l.COMMENT] = {['/*'] = 1, ['*/'] = -1, ['//'] = l.fold_line_comments('//')}
  79 }
  80
  81 return M