lexers/java.lua

   1 -- Copyright 2006-2015 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Java LPeg lexer.
   3 -- Modified by Brian Schott.
   4
   5 local l = require('lexer')
   6 local token, word_match = l.token, l.word_match
   7 local P, R, S = lpeg.P, lpeg.R, lpeg.S
   8
   9 local M = {_NAME = 'java'}
  10
  11 -- Whitespace.
  12 local ws = token(l.WHITESPACE, l.space^1)
  13
  14 -- Comments.
  15 local line_comment = '//' * l.nonnewline_esc^0
  16 local block_comment = '/*' * (l.any - '*/')^0 * P('*/')^-1
  17 local comment = token(l.COMMENT, line_comment + block_comment)
  18
  19 -- Strings.
  20 local sq_str = l.delimited_range("'", true)
  21 local dq_str = l.delimited_range('"', true)
  22 local string = token(l.STRING, sq_str + dq_str)
  23
  24 -- Numbers.
  25 local number = token(l.NUMBER, (l.float + l.integer) * S('LlFfDd')^-1)
  26
  27 -- Keywords.
  28 local keyword = token(l.KEYWORD, word_match{
  29   'abstract', 'assert', 'break', 'case', 'catch', 'class', 'const', 'continue',
  30   'default', 'do', 'else', 'enum', 'extends', 'final', 'finally', 'for', 'goto',
  31   'if', 'implements', 'import', 'instanceof', 'interface', 'native', 'new',
  32   'package', 'private', 'protected', 'public', 'return', 'static', 'strictfp',
  33   'super', 'switch', 'synchronized', 'this', 'throw', 'throws', 'transient',
  34   'try', 'while', 'volatile',
  35   -- Literals.
  36   'true', 'false', 'null'
  37 })
  38
  39 -- Types.
  40 local type = token(l.TYPE, word_match{
  41   'boolean', 'byte', 'char', 'double', 'float', 'int', 'long', 'short', 'void',
  42   'Boolean', 'Byte', 'Character', 'Double', 'Float', 'Integer', 'Long', 'Short',
  43   'String'
  44 })
  45
  46 -- Identifiers.
  47 local identifier = token(l.IDENTIFIER, l.word)
  48
  49 -- Operators.
  50 local operator = token(l.OPERATOR, S('+-/*%<>!=^&|?~:;.()[]{}'))
  51
  52 -- Annotations.
  53 local annotation = token('annotation', '@' * l.word)
  54
  55 -- Functions.
  56 local func = token(l.FUNCTION, l.word) * #P('(')
  57
  58 -- Classes.
  59 local class_sequence = token(l.KEYWORD, P('class')) * ws^1 *
  60                        token(l.CLASS, l.word)
  61
  62 M._rules = {
  63   {'whitespace', ws},
  64   {'class', class_sequence},
  65   {'keyword', keyword},
  66   {'type', type},
  67   {'function', func},
  68   {'identifier', identifier},
  69   {'string', string},
  70   {'comment', comment},
  71   {'number', number},
  72   {'annotation', annotation},
  73   {'operator', operator},
  74 }
  75
  76 M._tokenstyles = {
  77   annotation = l.STYLE_PREPROCESSOR
  78 }
  79
  80 M._foldsymbols = {
  81   _patterns = {'[{}]', '/%*', '%*/', '//'},
  82   [l.OPERATOR] = {['{'] = 1, ['}'] = -1},
  83   [l.COMMENT] = {['/*'] = 1, ['*/'] = -1, ['//'] = l.fold_line_comments('//')}
  84 }
  85
  86 return M