lexers/csharp.lua

   1 -- Copyright 2006-2016 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 = 'csharp'}
   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 = l.delimited_range("'", true)
  20 local dq_str = l.delimited_range('"', true)
  21 local ml_str = P('@')^-1 * l.delimited_range('"', false, true)
  22 local string = token(l.STRING, sq_str + dq_str + ml_str)
  23
  24 -- Numbers.
  25 local number = token(l.NUMBER, (l.float + l.integer) * S('lLdDfFMm')^-1)
  26
  27 -- Preprocessor.
  28 local preproc_word = word_match{
  29   'define', 'elif', 'else', 'endif', 'error', 'if', 'line', 'undef', 'warning',
  30   'region', 'endregion'
  31 }
  32 local preproc = token(l.PREPROCESSOR,
  33                       l.starts_line('#') * S('\t ')^0 * preproc_word *
  34                       (l.nonnewline_esc^1 + l.space * l.nonnewline_esc^0))
  35
  36 -- Keywords.
  37 local keyword = token(l.KEYWORD, word_match{
  38   'class', 'delegate', 'enum', 'event', 'interface', 'namespace', 'struct',
  39   'using', 'abstract', 'const', 'explicit', 'extern', 'fixed', 'implicit',
  40   'internal', 'lock', 'out', 'override', 'params', 'partial', 'private',
  41   'protected', 'public', 'ref', 'sealed', 'static', 'readonly', 'unsafe',
  42   'virtual', 'volatile', 'add', 'as', 'assembly', 'base', 'break', 'case',
  43   'catch', 'checked', 'continue', 'default', 'do', 'else', 'finally', 'for',
  44   'foreach', 'get', 'goto', 'if', 'in', 'is', 'new', 'remove', 'return', 'set',
  45   'sizeof', 'stackalloc', 'super', 'switch', 'this', 'throw', 'try', 'typeof',
  46   'unchecked', 'value', 'void', 'while', 'yield',
  47   'null', 'true', 'false'
  48 })
  49
  50 -- Types.
  51 local type = token(l.TYPE, word_match{
  52   'bool', 'byte', 'char', 'decimal', 'double', 'float', 'int', 'long', 'object',
  53   'operator', 'sbyte', 'short', 'string', 'uint', 'ulong', 'ushort'
  54 })
  55
  56 -- Identifiers.
  57 local identifier = token(l.IDENTIFIER, l.word)
  58
  59 -- Operators.
  60 local operator = token(l.OPERATOR, S('~!.,:;+-*/<>=\\^|&%?()[]{}'))
  61
  62 M._rules = {
  63   {'whitespace', ws},
  64   {'keyword', keyword},
  65   {'type', type},
  66   {'identifier', identifier},
  67   {'string', string},
  68   {'comment', comment},
  69   {'number', number},
  70   {'preproc', preproc},
  71   {'operator', operator},
  72 }
  73
  74 M._foldsymbols = {
  75   _patterns = {'%l+', '[{}]', '/%*', '%*/', '//'},
  76   [l.PREPROCESSOR] = {
  77     region = 1, endregion = -1,
  78     ['if'] = 1, ifdef = 1, ifndef = 1, endif = -1
  79   },
  80   [l.OPERATOR] = {['{'] = 1, ['}'] = -1},
  81   [l.COMMENT] = {['/*'] = 1, ['*/'] = -1, ['//'] = l.fold_line_comments('//')}
  82 }
  83
  84 return M