lua/json/decode/strings.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 util = require("json.decode.util")
   7 local merge = require("json.util").merge
   8
   9 local tonumber = tonumber
  10 local string = string
  11 local string_char = string.char
  12 local floor = math.floor
  13 local table_concat = table.concat
  14
  15 local error = error
  16 module("json.decode.strings")
  17 local function get_error(item)
  18         local fmt_string = item .. " in string [%q] @ %i:%i"
  19         return lpeg.P(function(data, index)
  20                 local line, line_index, bad_char, last_line = util.get_invalid_character_info(data, index)
  21                 local err = fmt_string:format(bad_char, line, line_index)
  22                 error(err)
  23         end) * 1
  24 end
  25
  26 local bad_unicode   = get_error("Illegal unicode escape")
  27 local bad_hex       = get_error("Illegal hex escape")
  28 local bad_character = get_error("Illegal character")
  29 local bad_escape    = get_error("Illegal escape")
  30
  31 local knownReplacements = {
  32         ["'"] = "'",
  33         ['"'] = '"',
  34         ['\\'] = '\\',
  35         ['/'] = '/',
  36         b = '\b',
  37         f = '\f',
  38         n = '\n',
  39         r = '\r',
  40         t = '\t',
  41         v = '\v',
  42         z = '\z'
  43 }
  44
  45 -- according to the table at http://da.wikipedia.org/wiki/UTF-8
  46 local function utf8DecodeUnicode(code1, code2)
  47         code1, code2 = tonumber(code1, 16), tonumber(code2, 16)
  48         if code1 == 0 and code2 < 0x80 then
  49                 return string_char(code2)
  50         end
  51         if code1 < 0x08 then
  52                 return string_char(
  53                         0xC0 + code1 * 4 + floor(code2 / 64),
  54                         0x80 + code2 % 64)
  55         end
  56         return string_char(
  57                 0xE0 + floor(code1 / 16),
  58                 0x80 + (code1 % 16) * 4 + floor(code2 / 64),
  59                 0x80 + code2 % 64)
  60 end
  61
  62 local function decodeX(code)
  63         code = tonumber(code, 16)
  64         return string_char(code)
  65 end
  66
  67 local doSimpleSub = lpeg.C(lpeg.S("'\"\\/bfnrtvz")) / knownReplacements
  68 local doUniSub = lpeg.P('u') * (lpeg.C(util.hexpair) * lpeg.C(util.hexpair) + bad_unicode)
  69 local doXSub = lpeg.P('x') * (lpeg.C(util.hexpair) + bad_hex)
  70
  71 local defaultOptions = {
  72         badChars = '',
  73         additionalEscapes = false, -- disallow untranslated escapes
  74         escapeCheck = #lpeg.S('bfnrtv/\\"xu\'z'), -- no check on valid characters
  75         decodeUnicode = utf8DecodeUnicode,
  76         strict_quotes = false
  77 }
  78
  79 default = nil -- Let the buildCapture optimization take place
  80
  81 strict = {
  82         badChars = '\b\f\n\r\t\v',
  83         additionalEscapes = false, -- no additional escapes
  84         escapeCheck = #lpeg.S('bfnrtv/\\"u'), --only these chars are allowed to be escaped
  85         strict_quotes = true
  86 }
  87
  88 local function buildCaptureString(quote, badChars, escapeMatch)
  89         local captureChar = (1 - lpeg.S("\\" .. badChars .. quote)) + (lpeg.P("\\") / "" * escapeMatch)
  90         -- During error, force end
  91         local captureString = captureChar^0 + (-#lpeg.P(quote) * bad_character + -1)
  92         return lpeg.P(quote) * lpeg.Cs(captureString) * lpeg.P(quote)
  93 end
  94
  95 local function buildCapture(options)
  96         options = options and merge({}, defaultOptions, options) or defaultOptions
  97         local quotes = { '"' }
  98         if not options.strict_quotes then
  99                 quotes[#quotes + 1] = "'"
 100         end
 101         local escapeMatch = doSimpleSub
 102         escapeMatch = escapeMatch + doXSub / decodeX
 103         escapeMatch = escapeMatch + doUniSub / options.decodeUnicode
 104         if options.additionalEscapes then
 105                 escapeMatch = escapeMatch + options.additionalEscapes
 106         end
 107         if options.escapeCheck then
 108                 escapeMatch = options.escapeCheck * escapeMatch + bad_escape
 109         end
 110         local captureString
 111         for i = 1, #quotes do
 112                 local cap = buildCaptureString(quotes[i], options.badChars, escapeMatch)
 113                 if captureString == nil then
 114                         captureString = cap
 115                 else
 116                         captureString = captureString + cap
 117                 end
 118         end
 119         return captureString
 120 end
 121
 122 function register_types()
 123         util.register_type("STRING")
 124 end
 125
 126 function load_types(options, global_options, grammar)
 127         local capture = buildCapture(options)
 128         local string_id = util.types.STRING
 129         grammar[string_id] = capture
 130         util.append_grammar_item(grammar, "VALUE", lpeg.V(string_id))
 131 end