lexer.lua

   1 require'lpeg'
   2 local l = lpeg
   3 local P, S, R, V, Cmt = l.P, l.S, l.R, l.V, l.Cmt
   4
   5 -- Adapted from scintillua's lexer.lua:
   6 -- Copyright 2006-2017 Mitchell mitchell.att.foicica.com.
   7
   8 local M = {}
   9
  10 M.any = P(1)
  11 M.alpha = R('AZ', 'az')
  12 M.digit = R('09')
  13 M.alnum = R('AZ', 'az', '09')
  14 M.xdigit = R('09', 'AF', 'af')
  15
  16 M.dec_num = M.digit^1
  17 M.hex_num = '0' * S('xX') * M.xdigit^1
  18 M.oct_num = '0' * R('07')^1
  19 M.integer = S('+-')^-1 * (M.hex_num + M.oct_num + M.dec_num)
  20 M.float = S('+-')^-1 *
  21           ((M.digit^0 * '.' * M.digit^1 + M.digit^1 * '.' * M.digit^0) *
  22            (S('eE') * S('+-')^-1 * M.digit^1)^-1 +
  23            (M.digit^1 * S('eE') * S('+-')^-1 * M.digit^1))
  24
  25 function M.delimited_range(chars, single_line, no_escape, balanced)
  26   local s = chars:sub(1, 1)
  27   local e = #chars == 2 and chars:sub(2, 2) or s
  28   local range
  29   local b = balanced and s or ''
  30   local n = single_line and '\n' or ''
  31   if no_escape then
  32     local invalid = S(e..n..b)
  33     range = M.any - invalid
  34   else
  35     local invalid = S(e..n..b) + '\\'
  36     range = M.any - invalid + '\\' * M.any
  37   end
  38   if balanced and s ~= e then
  39     return P{s * (range + V(1))^0 * e}
  40   else
  41     return s * range^0 * P(e)^-1
  42   end
  43 end
  44
  45 function M.word_match(words, word_chars)
  46   local word_list = {}
  47   for i = 1, #words do
  48     word_list[words[i]] = true
  49   end
  50   local chars = M.alnum + '_'
  51   if word_chars then chars = chars + S(word_chars) end
  52   return Cmt(chars^1, function(_, index, word)
  53     return word_list[word] and index or nil
  54   end)
  55 end
  56
  57 return M