Software / code / prosody
Changeset
5517:9d7349bbe4d2
util.json: New, improved, fixed codepoint to UTF-8 conversion.
| author | Waqas Hussain <waqas20@gmail.com> |
|---|---|
| date | Tue, 23 Apr 2013 15:55:49 -0400 |
| parents | 5516:9733836629f9 |
| children | 5518:0220093e34fa |
| files | util/json.lua |
| diffstat | 1 files changed, 11 insertions(+), 26 deletions(-) [+] |
line wrap: on
line diff
--- a/util/json.lua Tue Apr 23 14:41:52 2013 -0400 +++ b/util/json.lua Tue Apr 23 15:55:49 2013 -0400 @@ -2,8 +2,6 @@ -- Copyright (C) 2008-2010 Matthew Wild -- Copyright (C) 2008-2010 Waqas Hussain -- --- utf8char copyright (C) 2007 Rici Lake --- -- This project is MIT/X11 licensed. Please see the -- COPYING file in the source package for more information. -- @@ -41,32 +39,19 @@ if not escapes[ch] then escapes[ch] = ("\\u%.4X"):format(i); end end -local function utf8char(i) - if i >= 0 then - i = i - i%1 - if i < 128 then - return s_char(i) - else - local c1 = i % 64 - i = (i - c1) / 64 - if i < 32 then - return s_char(0xC0+i, 0x80+c1) - else - local c2 = i % 64 - i = (i - c2) / 64 - if i < 16 and (i ~= 13 or c2 < 32) then - return s_char(0xE0+i, 0x80+c2, 0x80+c1) - elseif i >= 16 and i < 0x110 then - local c3 = i % 64 - i = (i - c3) / 64 - return s_char(0xF0+i, 0x80+c3, 0x80+c2, 0x80+c1) - end - end - end +local function codepoint_to_utf8(code) + if code < 0x80 then return s_char(code); end + local bits0_6 = code % 64; + if code < 0x800 then + local bits6_5 = (code - bits0_6) / 64; + return s_char(0x80 + 0x40 + bits6_5, 0x80 + bits0_6); end + local bits0_12 = code % 4096; + local bits6_6 = (bits0_12 - bits0_6) / 64; + local bits12_4 = (code - bits0_12) / 4096; + return s_char(0x80 + 0x40 + 0x20 + bits12_4, 0x80 + bits6_6, 0x80 + bits0_6); end - local valid_types = { number = true, string = true, @@ -292,7 +277,7 @@ if not ch:match("[0-9a-fA-F]") then error("invalid unicode escape sequence in string"); end seq = seq..ch; end - s = s..utf8char(tonumber(seq, 16)); + s = s..codepoint_to_utf8(tonumber(seq, 16)); next(); else error("invalid escape sequence in string"); end end
