lua/json/decode/util.lua

   1 --[[
   2         Licensed according to the included 'LICENSE' document
   3         Author: Thomas Harning Jr <harningt@gmail.com>
   4 ]]
   5 local lpeg = require("lpeg")
   6 local select = select
   7 local pairs, ipairs = pairs, ipairs
   8 local tonumber = tonumber
   9 local string_char = string.char
  10
  11 local error = error
  12 local setmetatable = setmetatable
  13
  14 module("json.decode.util")
  15
  16 -- 09, 0A, 0B, 0C, 0D, 20
  17 ascii_space = lpeg.S("\t\n\v\f\r ")
  18 do
  19         local chr = string_char
  20         local u_space = ascii_space
  21         -- \u0085 \u00A0
  22         u_space = u_space + lpeg.P(chr(0xC2)) * lpeg.S(chr(0x85) .. chr(0xA0))
  23         -- \u1680 \u180E
  24         u_space = u_space + lpeg.P(chr(0xE1)) * (lpeg.P(chr(0x9A, 0x80)) + chr(0xA0, 0x8E))
  25         -- \u2000 - \u200A, also 200B
  26         local spacing_end = ""
  27         for i = 0x80,0x8b do
  28                 spacing_end = spacing_end .. chr(i)
  29         end
  30         -- \u2028 \u2029 \u202F
  31         spacing_end = spacing_end .. chr(0xA8) .. chr(0xA9) .. chr(0xAF)
  32         u_space = u_space + lpeg.P(chr(0xE2, 0x80)) * lpeg.S(spacing_end)
  33         -- \u205F
  34         u_space = u_space + lpeg.P(chr(0xE2, 0x81, 0x9F))
  35         -- \u3000
  36         u_space = u_space + lpeg.P(chr(0xE3, 0x80, 0x80))
  37         -- BOM \uFEFF
  38         u_space = u_space + lpeg.P(chr(0xEF, 0xBB, 0xBF))
  39         _M.unicode_space = u_space
  40 end
  41
  42 identifier = lpeg.R("AZ","az","__") * lpeg.R("AZ","az", "__", "09") ^0
  43
  44 hex = lpeg.R("09","AF","af")
  45 hexpair = hex * hex
  46
  47 comments = {
  48         cpp = lpeg.P("//") * (1 - lpeg.P("\n"))^0 * lpeg.P("\n"),
  49         c = lpeg.P("/*") * (1 - lpeg.P("*/"))^0 * lpeg.P("*/")
  50 }
  51
  52 comment = comments.cpp + comments.c
  53
  54 ascii_ignored = (ascii_space + comment)^0
  55
  56 unicode_ignored = (unicode_space + comment)^0
  57
  58 local types = setmetatable({false}, {
  59         __index = function(self, k)
  60                 error("Unknown type: " .. k)
  61         end
  62 })
  63
  64 function register_type(name)
  65         types[#types + 1] = name
  66         types[name] = #types
  67         return #types
  68 end
  69
  70 _M.types = types
  71
  72 function append_grammar_item(grammar, name, capture)
  73         local id = types[name]
  74         local original = grammar[id]
  75         if original then
  76                 grammar[id] = original + capture
  77         else
  78                 grammar[id] = capture
  79         end
  80 end
  81
  82 -- Parse the lpeg version skipping patch-values
  83 -- LPEG <= 0.7 have no version value... so 0.7 is value
  84 DecimalLpegVersion = lpeg.version and tonumber(lpeg.version():match("^(%d+%.%d+)")) or 0.7
  85
  86 function get_invalid_character_info(input, index)
  87         local parsed = input:sub(1, index)
  88         local bad_character = input:sub(index, index)
  89         local _, line_number = parsed:gsub('\n',{})
  90         local last_line = parsed:match("\n([^\n]+.)$") or parsed
  91         return line_number, #last_line, bad_character, last_line
  92 end