lexers/icon.lua

   1 -- Copyright 2006-2016 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- LPeg lexer for the Icon programming language.
   3 -- http://www.cs.arizona.edu/icon
   4 -- Contributed by Carl Sturtivant.
   5
   6 local l = require('lexer')
   7 local token, word_match = l.token, l.word_match
   8 local P, R, S = lpeg.P, lpeg.R, lpeg.S
   9
  10 local M = {_NAME = 'icon'}
  11
  12 -- Whitespace.
  13 local ws = token(l.WHITESPACE, l.space^1)
  14
  15 --Comments
  16 local line_comment = '#' * l.nonnewline_esc^0
  17 local comment = token(l.COMMENT, line_comment)
  18
  19 -- Strings.
  20 local cset = l.delimited_range("'")
  21 local str = l.delimited_range('"')
  22 local string = token(l.STRING, cset + str)
  23
  24 -- Numbers.
  25 local radix_literal = P('-')^-1 * l.dec_num * S('rR') * l.alnum^1
  26 local number = token(l.NUMBER, radix_literal + l.float + l.integer)
  27
  28 -- Preprocessor.
  29 local preproc_word = word_match{
  30   'include', 'line', 'define', 'undef', 'ifdef', 'ifndef', 'else', 'endif',
  31   'error'
  32 }
  33 local preproc = token(l.PREPROCESSOR, S(' \t')^0 * P('$') * preproc_word)
  34
  35 -- Keywords.
  36 local keyword = token(l.KEYWORD, word_match{
  37   'break', 'by', 'case', 'create', 'default', 'do', 'else', 'end', 'every',
  38   'fail', 'global', 'if', 'initial', 'invocable', 'link', 'local', 'next',
  39   'not', 'of', 'procedure', 'record', 'repeat', 'return', 'static', 'suspend',
  40   'then', 'to', 'until', 'while'
  41 })
  42
  43 -- Icon Keywords: unique to Icon; use l.TYPE, as Icon is dynamically typed
  44 local type = token(l.TYPE, P('&') * word_match{
  45   'allocated', 'ascii', 'clock', 'collections', 'cset', 'current', 'date',
  46   'dateline', 'digits', 'dump', 'e', 'error', 'errornumber', 'errortext',
  47   'errorvalue', 'errout', 'fail', 'features', 'file', 'host', 'input', 'lcase',
  48   'letters', 'level', 'line', 'main', 'null', 'output', 'phi', 'pi', 'pos',
  49   'progname', 'random', 'regions', 'source', 'storage', 'subject', 'time',
  50   'trace', 'ucase', 'version'
  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   {'comment', comment},
  65   {'string', string},
  66   {'number', number},
  67   {'preproc', preproc},
  68   {'operator', operator},
  69 }
  70
  71 M._foldsymbols = {
  72   _patterns = {'%l+', '#'},
  73   [l.PREPROCESSOR] = {ifdef = 1, ifndef = 1, endif = -1},
  74   [l.KEYWORD] = { procedure = 1, ['end'] = -1},
  75   [l.COMMENT] = {['#'] = l.fold_line_comments('#')}
  76 }
  77
  78 return M