author | Matthew Wild <mwild1@gmail.com> |
Wed, 03 Dec 2008 13:06:23 +0000 | |
changeset 518 | 127eb78cbf4a |
parent 509 | 32899c8a6fe5 |
child 519 | cccd610a0ef9 |
permissions | -rw-r--r-- |
509
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
1 |
--- WARNING! --- |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
2 |
-- This file contains a mix of encodings below. |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
3 |
-- Many editors will unquestioningly convert these for you. |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
4 |
-- Please be careful :( (I recommend Scite) |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
5 |
--------------------------------- |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
6 |
|
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
7 |
local gmatch = string.gmatch; |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
8 |
local t_concat, t_insert = table.concat, table.insert; |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
9 |
local to_byte, to_char = string.byte, string.char; |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
10 |
|
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
11 |
local function _latin1toutf8(str) |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
12 |
if not str then return str; end |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
13 |
local p = {}; |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
14 |
for ch in gmatch(str, ".") do |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
15 |
ch = to_byte(ch); |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
16 |
if (ch < 0x80) then |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
17 |
t_insert(p, to_char(ch)); |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
18 |
elseif (ch < 0xC0) then |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
19 |
t_insert(p, to_char(0xC2, ch)); |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
20 |
else |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
21 |
t_insert(p, to_char(0xC3, ch - 64)); |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
22 |
end |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
23 |
end |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
24 |
return t_concat(p); |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
25 |
end |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
26 |
|
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
27 |
function latin1toutf8() |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
28 |
local function assert_utf8(latin, utf8) |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
29 |
assert_equal(_latin1toutf8(latin), utf8, "Incorrect UTF8 from Latin1: "..tostring(latin)); |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
30 |
end |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
31 |
|
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
32 |
assert_utf8("", "") |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
33 |
assert_utf8("test", "test") |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
34 |
assert_utf8(nil, nil) |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
35 |
assert_utf8("foobar.r�kat.se", "foobar.råkat.se") |
32899c8a6fe5
Add test for latin1toutf8 (which passes)
Matthew Wild <mwild1@gmail.com>
parents:
diff
changeset
|
36 |
end |