lua/lexers/crystal.lua

   1 -- Copyright 2006-2017 Mitchell mitchell.att.foicica.com. See LICENSE.
   2 -- Copyright 2017 Michel Martens.
   3 -- Crystal LPeg lexer (based on Ruby).
   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 = 'crystal'}
  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 comment = token(l.COMMENT, line_comment)
  17
  18 local delimiter_matches = {['('] = ')', ['['] = ']', ['{'] = '}'}
  19 local literal_delimitted = P(function(input, index)
  20   local delimiter = input:sub(index, index)
  21   if not delimiter:find('[%w\r\n\f\t ]') then -- only non alpha-numerics
  22     local match_pos, patt
  23     if delimiter_matches[delimiter] then
  24       -- Handle nested delimiter/matches in strings.
  25       local s, e = delimiter, delimiter_matches[delimiter]
  26       patt = l.delimited_range(s..e, false, false, true)
  27     else
  28       patt = l.delimited_range(delimiter)
  29     end
  30     match_pos = lpeg.match(patt, input, index)
  31     return match_pos or #input + 1
  32   end
  33 end)
  34
  35 -- Strings.
  36 local cmd_str = l.delimited_range('`')
  37 local sq_str = l.delimited_range("'")
  38 local dq_str = l.delimited_range('"')
  39 local heredoc = '<<' * P(function(input, index)
  40   local s, e, indented, _, delimiter =
  41     input:find('(%-?)(["`]?)([%a_][%w_]*)%2[\n\r\f;]+', index)
  42   if s == index and delimiter then
  43     local end_heredoc = (#indented > 0 and '[\n\r\f]+ *' or '[\n\r\f]+')
  44     local _, e = input:find(end_heredoc..delimiter, e)
  45     return e and e + 1 or #input + 1
  46   end
  47 end)
  48 -- TODO: regex_str fails with `obj.method /patt/` syntax.
  49 local regex_str = #P('/') * l.last_char_includes('!%^&*([{-=+|:;,?<>~') *
  50                   l.delimited_range('/', true, false) * S('iomx')^0
  51 local string = token(l.STRING, (sq_str + dq_str + heredoc + cmd_str) *
  52                                S('f')^-1) +
  53                token(l.REGEX, regex_str)
  54
  55 local word_char = l.alnum + S('_!?')
  56
  57 -- Numbers.
  58 local dec = l.digit^1 * ('_' * l.digit^1)^0 * S('ri')^-1
  59 local bin = '0b' * S('01')^1 * ('_' * S('01')^1)^0
  60 local integer = S('+-')^-1 * (bin + l.hex_num + l.oct_num + dec)
  61 -- TODO: meta, control, etc. for numeric_literal.
  62 local numeric_literal = '?' * (l.any - l.space) * -word_char
  63 local number = token(l.NUMBER, l.float * S('ri')^-1 + integer + numeric_literal)
  64
  65 -- Keywords.
  66 local keyword = token(l.KEYWORD, word_match({
  67   'alias', 'begin', 'break', 'case', 'class', 'def', 'defined?', 'do', 'else',
  68   'elsif', 'end', 'ensure', 'false', 'for', 'if', 'in', 'module', 'next', 'nil',
  69   'not', 'redo', 'rescue', 'retry', 'return', 'self', 'super', 'then', 'true',
  70   'undef', 'unless', 'until', 'when', 'while', 'yield', '__FILE__', '__LINE__'
  71 }, '?!'))
  72
  73 -- Functions.
  74 local func = token(l.FUNCTION, word_match({
  75   'abort', 'at_exit', 'caller', 'delay', 'exit', 'fork', 'future',
  76   'get_stack_top', 'gets', 'lazy', 'loop', 'main', 'p', 'print', 'printf',
  77   'puts', 'raise', 'rand', 'read_line', 'require', 'sleep', 'spawn', 'sprintf',
  78   'system', 'with_color',
  79   -- Macros
  80   'assert_responds_to', 'debugger', 'parallel', 'pp', 'record', 'redefine_main'
  81 }, '?!')) * -S('.:|')
  82
  83 -- Identifiers.
  84 local word = (l.alpha + '_') * word_char^0
  85 local identifier = token(l.IDENTIFIER, word)
  86
  87 -- Variables.
  88 local global_var = '$' * (word + S('!@L+`\'=~/\\,.;<>_*"$?:') + l.digit + '-' *
  89                    S('0FadiIKlpvw'))
  90 local class_var = '@@' * word
  91 local inst_var = '@' * word
  92 local variable = token(l.VARIABLE, global_var + class_var + inst_var)
  93
  94 -- Symbols.
  95 local symbol = token('symbol', ':' * P(function(input, index)
  96   if input:sub(index - 2, index - 2) ~= ':' then return index end
  97 end) * (word_char^1 + sq_str + dq_str))
  98
  99 -- Operators.
 100 local operator = token(l.OPERATOR, S('!%^&*()[]{}-=+/|:;.,?<>~'))
 101
 102 M._rules = {
 103   {'whitespace', ws},
 104   {'keyword', keyword},
 105   {'function', func},
 106   {'identifier', identifier},
 107   {'comment', comment},
 108   {'string', string},
 109   {'number', number},
 110   {'variable', variable},
 111   {'symbol', symbol},
 112   {'operator', operator},
 113 }
 114
 115 M._tokenstyles = {
 116   symbol = l.STYLE_CONSTANT
 117 }
 118
 119 local function disambiguate(text, pos, line, s)
 120   return line:sub(1, s - 1):match('^%s*$') and
 121          not text:sub(1, pos - 1):match('\\[ \t]*\r?\n$') and 1 or 0
 122 end
 123
 124 M._foldsymbols = {
 125   _patterns = {'%l+', '[%(%)%[%]{}]', '#'},
 126   [l.KEYWORD] = {
 127     begin = 1, class = 1, def = 1, ['do'] = 1, ['for'] = 1, ['module'] = 1,
 128     case = 1,
 129     ['if'] = disambiguate, ['while'] = disambiguate,
 130     ['unless'] = disambiguate, ['until'] = disambiguate,
 131     ['end'] = -1
 132   },
 133   [l.OPERATOR] = {
 134     ['('] = 1, [')'] = -1, ['['] = 1, [']'] = -1, ['{'] = 1, ['}'] = -1
 135   },
 136   [l.COMMENT] = {
 137     ['#'] = l.fold_line_comments('#')
 138   }
 139 }
 140
 141 return M