lexers/scala.lua

   1 -- Copyright 2006-2013 JMS. See LICENSE.
   2 -- Scala 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 = 'scala'}
   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 symbol = "'" * l.word
  20 local dq_str = l.delimited_range('"', true)
  21 local tq_str = '"""' * (l.any - '"""')^0 * P('"""')^-1
  22 local string = token(l.STRING, tq_str + symbol + 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', 'case', 'catch', 'class', 'def', 'do', 'else', 'extends', 'false',
  30   'final', 'finally', 'for', 'forSome', 'if', 'implicit', 'import', 'lazy',
  31   'match', 'new', 'null', 'object', 'override', 'package', 'private',
  32   'protected', 'return', 'sealed', 'super', 'this', 'throw', 'trait', 'try',
  33   'true', 'type', 'val', 'var', 'while', 'with', 'yield'
  34 })
  35
  36 -- Types.
  37 local type = token(l.TYPE, word_match{
  38   'Array', 'Boolean', 'Buffer', 'Byte', 'Char', 'Collection', 'Double', 'Float',
  39   'Int', 'Iterator', 'LinkedList', 'List', 'Long', 'Map', 'None', 'Option',
  40   'Set', 'Short', 'SortedMap', 'SortedSet', 'String', 'TreeMap', 'TreeSet'
  41 })
  42
  43 -- Identifiers.
  44 local identifier = token(l.IDENTIFIER, l.word)
  45
  46 -- Operators.
  47 local operator = token(l.OPERATOR, S('+-/*%<>!=^&|?~:;.()[]{}'))
  48
  49 -- Functions.
  50 local func = token(l.FUNCTION, l.word) * #P('(')
  51
  52 -- Classes.
  53 local class_sequence = token(l.KEYWORD, P('class')) * ws^1 *
  54                        token(l.CLASS, l.word)
  55
  56 M._rules = {
  57   {'whitespace', ws},
  58   {'class', class_sequence},
  59   {'keyword', keyword},
  60   {'type', type},
  61   {'function', func},
  62   {'identifier', identifier},
  63   {'string', string},
  64   {'comment', comment},
  65   {'number', number},
  66   {'operator', operator},
  67 }
  68
  69 M._foldsymbols = {
  70   _patterns = {'[{}]', '/%*', '%*/', '//'},
  71   [l.OPERATOR] = {['{'] = 1, ['}'] = -1},
  72   [l.COMMENT] = {['/*'] = 1, ['*/'] = -1, ['//'] = l.fold_line_comments('//')}
  73 }
  74
  75 return M