prosody
a4ba5819bf50
util/json.lua

Find changesets by keywords by author, files, the commit message, revision number or hash, or revset expression.

Software / code / prosody

Comparison

util/json.lua @ 5436:a4ba5819bf50

util.json: Convert \uXXXX to UTF-8 when decoding
author Matthew Wild <mwild1@gmail.com>
date Sat, 06 Apr 2013 12:20:31 +0100
parent 5395:ec33d72a08b6
child 5516:9733836629f9
comparison
equal deleted inserted replaced
5435:f56e449a63e3 5436:a4ba5819bf50
1 -- Prosody IM
2 -- Copyright (C) 2008-2010 Matthew Wild
3 -- Copyright (C) 2008-2010 Waqas Hussain
4 --
5 -- utf8char copyright (C) 2007 Rici Lake
6 --
7 -- This project is MIT/X11 licensed. Please see the
8 -- COPYING file in the source package for more information.
9 --
1 10
2 local type = type; 11 local type = type;
3 local t_insert, t_concat, t_remove, t_sort = table.insert, table.concat, table.remove, table.sort; 12 local t_insert, t_concat, t_remove, t_sort = table.insert, table.concat, table.remove, table.sort;
4 local s_char = string.char; 13 local s_char = string.char;
5 local tostring, tonumber = tostring, tonumber; 14 local tostring, tonumber = tostring, tonumber;
26 b = "\b", f = "\f", n = "\n", r = "\r", t = "\t"}; 35 b = "\b", f = "\f", n = "\n", r = "\r", t = "\t"};
27 for i=0,31 do 36 for i=0,31 do
28 local ch = s_char(i); 37 local ch = s_char(i);
29 if not escapes[ch] then escapes[ch] = ("\\u%.4X"):format(i); end 38 if not escapes[ch] then escapes[ch] = ("\\u%.4X"):format(i); end
30 end 39 end
40
41 local function utf8char(i)
42 if i >= 0 then
43 i = i - i%1
44 if i < 128 then
45 return s_char(i)
46 else
47 local c1 = i % 64
48 i = (i - c1) / 64
49 if i < 32 then
50 return s_char(0xC0+i, 0x80+c1)
51 else
52 local c2 = i % 64
53 i = (i - c2) / 64
54 if i < 16 and (i ~= 13 or c2 < 32) then
55 return s_char(0xE0+i, 0x80+c2, 0x80+c1)
56 elseif i >= 16 and i < 0x110 then
57 local c3 = i % 64
58 i = (i - c3) / 64
59 return s_char(0xF0+i, 0x80+c3, 0x80+c2, 0x80+c1)
60 end
61 end
62 end
63 end
64 end
65
31 66
32 local valid_types = { 67 local valid_types = {
33 number = true, 68 number = true,
34 string = true, 69 string = true,
35 table = true, 70 table = true,
247 next(); 282 next();
248 if not ch then error("unexpected eof in string"); end 283 if not ch then error("unexpected eof in string"); end
249 if not ch:match("[0-9a-fA-F]") then error("invalid unicode escape sequence in string"); end 284 if not ch:match("[0-9a-fA-F]") then error("invalid unicode escape sequence in string"); end
250 seq = seq..ch; 285 seq = seq..ch;
251 end 286 end
252 s = s..s.char(tonumber(seq, 16)); -- FIXME do proper utf-8 287 s = s..utf8char(tonumber(seq, 16));
253 next(); 288 next();
254 else error("invalid escape sequence in string"); end 289 else error("invalid escape sequence in string"); end
255 end 290 end
256 if ch == "\"" then 291 if ch == "\"" then
257 next(); 292 next();