lua/lexers/pony.lua

   1 -- Copyright 2017 Murray Calavera. See LICENSE.
   2 -- Pony LPeg lexer.
   3
   4 local l = require('lexer')
   5 local token = l.token
   6 local P, S = lpeg.P, lpeg.S
   7
   8 local function pword(words)
   9   return l.word_match(words, "'")
  10 end
  11
  12 local ws = token(l.WHITESPACE, l.space^1)
  13
  14 local comment_line = '//' * l.nonnewline^0
  15 local comment_block = l.nested_pair('/*', '*/')
  16 local comment = token(l.COMMENT, comment_line + comment_block)
  17
  18 local annotation = token(l.PREPROCESSOR, l.delimited_range('\\', false, true))
  19
  20 local lit_bool = token(l.CONSTANT, pword{'true', 'false'})
  21
  22 local lit_str = token(l.STRING,
  23   l.delimited_range('"') -- this covers triple quoted strings
  24   + l.delimited_range("'")
  25 )
  26
  27 local function num(digit)
  28   return digit * (digit^0 * P'_')^0 * digit^1 + digit
  29 end
  30
  31 local int = num(l.digit)
  32 local frac = P('.') * int
  33 local exp = S('eE') * (P('-') + P('+'))^-1 * int
  34
  35 local lit_num = token(l.NUMBER,
  36   P('0x') * num(l.xdigit)
  37   + P('0b') * num(S('01'))
  38   + int * frac^-1 * exp^-1
  39 )
  40
  41 local keyword = token(l.KEYWORD, pword{
  42   'actor', 'as', 'be', 'break', 'class', 'compile_error', 'compile_intrinsic',
  43   'continue', 'consume', 'do', 'else', 'elseif', 'embed', 'end', 'error',
  44   'for', 'fun', 'if', 'ifdef', 'iftype', 'in', 'interface', 'is', 'isnt',
  45   'lambda', 'let', 'match', 'new', 'object', 'primitive', 'recover', 'repeat',
  46   'return', 'struct', 'then', 'this', 'trait', 'try', 'type', 'until', 'use',
  47   'var', 'where', 'while', 'with'})
  48 local capability = token(l.LABEL, pword{
  49   'box', 'iso', 'ref', 'tag', 'trn', 'val'})
  50 local qualifier = token(l.LABEL,
  51   P'#' * pword{'read', 'send', 'share', 'any', 'alias'})
  52
  53 local operator = token(l.OPERATOR,
  54   pword{'and', 'or', 'xor', 'not', 'addressof', 'digestof'}
  55   + lpeg.Cmt(S('+-*/%<>=!~')^1, function(input, index, op)
  56       local ops = {
  57         ['+'] = true, ['-'] = true, ['*'] = true, ['/'] = true, ['%'] = true,
  58         ['+~'] = true, ['-~'] = true, ['*~'] = true, ['/~'] = true,
  59         ['%~'] = true, ['<<'] = true, ['>>'] = true, ['<<~'] = true,
  60         ['>>~'] = true, ['=='] = true, ['!='] = true, ['<'] = true,
  61         ['<='] = true, ['>='] = true, ['>'] = true, ['==~'] = true,
  62         ['!=~'] = true, ['<~'] = true, ['<=~'] = true, ['>=~'] = true,
  63         ['>~'] = true
  64       }
  65       return ops[op] and index or nil
  66     end)
  67 )
  68
  69 -- there is no suitable token name for this, change this if ever one is added
  70 local punctuation = token(l.OPERATOR,
  71   P'=>' + P'.>' + P'<:' + P'->'
  72   + S('=.,:;()[]{}!?~^&|_@'))
  73
  74 -- highlight functions with syntax sugar at declaration
  75 local func
  76   = token(l.KEYWORD, pword{'fun', 'new', 'be'}) * ws^-1
  77   * annotation^-1 * ws^-1
  78   * capability^-1 * ws^-1
  79   * token(l.FUNCTION, pword{
  80     'create', 'dispose', '_final', 'apply', 'update',
  81     'add', 'sub', 'mul', 'div', 'mod', 'add_unsafe', 'sub_unsafe',
  82     'mul_unsafe', 'div_unsafe', 'mod_unsafe', 'shl', 'shr', 'shl_unsafe',
  83     'shr_unsafe', 'op_and', 'op_or', 'op_xor', 'eq', 'ne', 'lt', 'le', 'ge',
  84     'gt', 'eq_unsafe', 'ne_unsafe', 'lt_unsafe', 'le_unsafe', 'ge_unsafe',
  85     'gt_unsafe', 'neg', 'neg_unsafe', 'op_not',
  86     'has_next', 'next',
  87     '_serialise_space', '_serialise', '_deserialise'})
  88
  89 local id_suffix = (l.alnum + P("'") + P('_'))^0
  90 local type = token(l.TYPE, P('_')^-1 * l.upper * id_suffix)
  91 local identifier = token(l.IDENTIFIER, P('_')^-1 * l.lower * id_suffix)
  92 local tuple_lookup = token(l.IDENTIFIER, P('_') * l.digit^1)
  93
  94 local M = {_NAME = 'pony'}
  95
  96 M._rules = {
  97   {'whitespace', ws},
  98   {'comment', comment},
  99   {'annotation', annotation},
 100   {'boolean', lit_bool},
 101   {'number', lit_num},
 102   {'string', lit_str},
 103   {'function', func},
 104   {'keyword', keyword},
 105   {'capability', capability},
 106   {'qualifier', qualifier},
 107   {'operator', operator},
 108   {'type', type},
 109   {'identifier', identifier},
 110   {'lookup', tuple_lookup},
 111   {'punctuation', punctuation}
 112 }
 113
 114 return M