lua/lexers/reason.lua

   1 -- Copyright 2006-2017 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Reason (https://reasonml.github.io/) 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 = 'reason'}
   9
  10 -- Whitespace.
  11 local ws = token(l.WHITESPACE, l.space^1)
  12
  13 -- Comments.
  14 local comment = token(l.COMMENT, l.nested_pair('/*', '*/'))
  15
  16 -- Strings.
  17 local sq_str = l.delimited_range("'", true)
  18 local dq_str = l.delimited_range('"', true)
  19 local string = token(l.STRING, sq_str + dq_str)
  20
  21 -- Numbers.
  22 local number = token(l.NUMBER, l.float + l.integer)
  23
  24 -- Keywords.
  25 local keyword = token(l.KEYWORD, word_match{
  26   'and', 'as', 'asr', 'begin', 'class', 'closed', 'constraint', 'do', 'done',
  27   'downto', 'else', 'end', 'exception', 'external', 'failwith', 'false',
  28   'flush', 'for', 'fun', 'function', 'functor', 'if', 'in', 'include',
  29   'inherit',  'incr', 'land', 'let', 'load', 'los', 'lsl', 'lsr', 'lxor',
  30   'method', 'mod', 'module', 'mutable', 'new', 'not', 'of', 'open', 'option',
  31   'or', 'parser', 'private', 'ref', 'rec', 'raise', 'regexp', 'sig', 'struct',
  32   'stdout', 'stdin', 'stderr', 'switch', 'then', 'to', 'true', 'try', 'type',
  33   'val', 'virtual', 'when', 'while', 'with'
  34 })
  35
  36 -- Types.
  37 local type = token(l.TYPE, word_match{
  38   'int', 'float', 'bool', 'char', 'string', 'unit'
  39 })
  40
  41 -- Functions.
  42 local func = token(l.FUNCTION, word_match{
  43   'raise', 'invalid_arg', 'failwith', 'compare', 'min', 'max', 'succ', 'pred',
  44   'mod', 'abs', 'max_int', 'min_int', 'sqrt', 'exp', 'log', 'log10', 'cos',
  45   'sin', 'tan', 'acos', 'asin', 'atan', 'atan2', 'cosh', 'sinh', 'tanh', 'ceil',
  46   'floor', 'abs_float', 'mod_float', 'frexp', 'ldexp', 'modf', 'float',
  47   'float_of_int', 'truncate', 'int_of_float', 'infinity', 'nan', 'max_float',
  48   'min_float', 'epsilon_float', 'classify_float', 'int_of_char', 'char_of_int',
  49   'ignore', 'string_of_bool', 'bool_of_string', 'string_of_int',
  50   'int_of_string', 'string_of_float', 'float_of_string', 'fst', 'snd', 'stdin',
  51   'stdout', 'stderr', 'print_char', 'print_string', 'print_int', 'print_float',
  52   'print_endline', 'print_newline', 'prerr_char', 'prerr_string', 'prerr_int',
  53   'prerr_float', 'prerr_endline', 'prerr_newline', 'read_line', 'read_int',
  54   'read_float', 'open_out', 'open_out_bin', 'open_out_gen', 'flush',
  55   'flush_all', 'output_char', 'output_string', 'output', 'output_byte',
  56   'output_binary_int', 'output_value', 'seek_out', 'pos_out',
  57   'out_channel_length', 'close_out', 'close_out_noerr', 'set_binary_mode_out',
  58   'open_in', 'open_in_bin', 'open_in_gen', 'input_char', 'input_line', 'input',
  59   'really_input', 'input_byte', 'input_binary_int', 'input_value', 'seek_in',
  60   'pos_in', 'in_channel_length', 'close_in', 'close_in_noerr',
  61   'set_binary_mode_in', 'incr', 'decr', 'string_of_format', 'format_of_string',
  62   'exit', 'at_exit'
  63 })
  64
  65 -- Identifiers.
  66 local identifier = token(l.IDENTIFIER, l.word)
  67
  68 -- Operators.
  69 local operator = token(l.OPERATOR, S('=<>+-*/.,:;~!#%^&|?[](){}'))
  70
  71 M._rules = {
  72   {'whitespace', ws},
  73   {'keyword', keyword},
  74   {'type', type},
  75   {'function', func},
  76   {'identifier', identifier},
  77   {'string', string},
  78   {'comment', comment},
  79   {'number', number},
  80   {'operator', operator},
  81 }
  82
  83 return M