annotate tools/mkunicode/Unicode-after.cpp @ 401:ca5e4360f79a

Js: - Add support of constants map (js::Map<T>) - Add supports for vectors (std::vector<T>) - Add index based getProperty/putProperty
author David Demelier <markand@malikania.fr>
date Sat, 03 Oct 2015 11:27:49 +0200
parents b78d6d8f2872
children f083259de5e6
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
1 void encode(char32_t c, char res[5]) noexcept
352
David Demelier <markand@malikania.fr>
parents:
diff changeset
2 {
David Demelier <markand@malikania.fr>
parents:
diff changeset
3 switch (nbytesPoint(c)) {
David Demelier <markand@malikania.fr>
parents:
diff changeset
4 case 1:
David Demelier <markand@malikania.fr>
parents:
diff changeset
5 res[0] = c;
David Demelier <markand@malikania.fr>
parents:
diff changeset
6 res[1] = '\0';
David Demelier <markand@malikania.fr>
parents:
diff changeset
7 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
8 case 2:
David Demelier <markand@malikania.fr>
parents:
diff changeset
9 res[0] = 0xC0 | ((c >> 6) & 0x1F);
David Demelier <markand@malikania.fr>
parents:
diff changeset
10 res[1] = 0x80 | (c & 0x3F);
David Demelier <markand@malikania.fr>
parents:
diff changeset
11 res[2] = '\0';
David Demelier <markand@malikania.fr>
parents:
diff changeset
12 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
13 case 3:
David Demelier <markand@malikania.fr>
parents:
diff changeset
14 res[0] = 0xE0 | ((c >> 12) & 0xF );
David Demelier <markand@malikania.fr>
parents:
diff changeset
15 res[1] = 0x80 | ((c >> 6) & 0x3F);
David Demelier <markand@malikania.fr>
parents:
diff changeset
16 res[2] = 0x80 | (c & 0x3F);
David Demelier <markand@malikania.fr>
parents:
diff changeset
17 res[3] = '\0';
David Demelier <markand@malikania.fr>
parents:
diff changeset
18 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
19 case 4:
David Demelier <markand@malikania.fr>
parents:
diff changeset
20 res[0] = 0xF0 | ((c >> 18) & 0x7 );
David Demelier <markand@malikania.fr>
parents:
diff changeset
21 res[1] = 0x80 | ((c >> 12) & 0x3F);
David Demelier <markand@malikania.fr>
parents:
diff changeset
22 res[2] = 0x80 | ((c >> 6) & 0x3F);
David Demelier <markand@malikania.fr>
parents:
diff changeset
23 res[3] = 0x80 | (c & 0x3F);
David Demelier <markand@malikania.fr>
parents:
diff changeset
24 res[4] = '\0';
David Demelier <markand@malikania.fr>
parents:
diff changeset
25 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
26 default:
David Demelier <markand@malikania.fr>
parents:
diff changeset
27 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
28 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
29 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
30
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
31 void decode(char32_t &c, const char *res) noexcept
352
David Demelier <markand@malikania.fr>
parents:
diff changeset
32 {
David Demelier <markand@malikania.fr>
parents:
diff changeset
33 c = 0;
David Demelier <markand@malikania.fr>
parents:
diff changeset
34
David Demelier <markand@malikania.fr>
parents:
diff changeset
35 switch (nbytesUtf8(res[0])) {
David Demelier <markand@malikania.fr>
parents:
diff changeset
36 case 1:
David Demelier <markand@malikania.fr>
parents:
diff changeset
37 c = res[0];
David Demelier <markand@malikania.fr>
parents:
diff changeset
38 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
39 case 2:
David Demelier <markand@malikania.fr>
parents:
diff changeset
40 c = (res[0] & 0x1f) << 6;
David Demelier <markand@malikania.fr>
parents:
diff changeset
41 c |= (res[1] & 0x3f);
David Demelier <markand@malikania.fr>
parents:
diff changeset
42 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
43 case 3:
David Demelier <markand@malikania.fr>
parents:
diff changeset
44 c = (res[0] & 0x0f) << 12;
David Demelier <markand@malikania.fr>
parents:
diff changeset
45 c |= (res[1] & 0x3f) << 6;
David Demelier <markand@malikania.fr>
parents:
diff changeset
46 c |= (res[2] & 0x3f);
David Demelier <markand@malikania.fr>
parents:
diff changeset
47 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
48 case 4:
David Demelier <markand@malikania.fr>
parents:
diff changeset
49 c = (res[0] & 0x07) << 16;
David Demelier <markand@malikania.fr>
parents:
diff changeset
50 c |= (res[1] & 0x3f) << 12;
David Demelier <markand@malikania.fr>
parents:
diff changeset
51 c |= (res[2] & 0x3f) << 6;
David Demelier <markand@malikania.fr>
parents:
diff changeset
52 c |= (res[3] & 0x3f);
David Demelier <markand@malikania.fr>
parents:
diff changeset
53 default:
David Demelier <markand@malikania.fr>
parents:
diff changeset
54 break;
David Demelier <markand@malikania.fr>
parents:
diff changeset
55 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
56 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
57
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
58 int nbytesUtf8(char c) noexcept
352
David Demelier <markand@malikania.fr>
parents:
diff changeset
59 {
David Demelier <markand@malikania.fr>
parents:
diff changeset
60 if ((c & 0xE0) == 0xC0)
David Demelier <markand@malikania.fr>
parents:
diff changeset
61 return 2;
David Demelier <markand@malikania.fr>
parents:
diff changeset
62 if ((c & 0xF0) == 0xE0)
David Demelier <markand@malikania.fr>
parents:
diff changeset
63 return 3;
David Demelier <markand@malikania.fr>
parents:
diff changeset
64 if ((c & 0xF8) == 0xF0)
David Demelier <markand@malikania.fr>
parents:
diff changeset
65 return 4;
David Demelier <markand@malikania.fr>
parents:
diff changeset
66
David Demelier <markand@malikania.fr>
parents:
diff changeset
67 return 1;
David Demelier <markand@malikania.fr>
parents:
diff changeset
68 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
69
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
70 int nbytesPoint(char32_t c) noexcept
352
David Demelier <markand@malikania.fr>
parents:
diff changeset
71 {
David Demelier <markand@malikania.fr>
parents:
diff changeset
72 if (c <= 0x7F)
David Demelier <markand@malikania.fr>
parents:
diff changeset
73 return 1;
David Demelier <markand@malikania.fr>
parents:
diff changeset
74 if (c <= 0x7FF)
David Demelier <markand@malikania.fr>
parents:
diff changeset
75 return 2;
David Demelier <markand@malikania.fr>
parents:
diff changeset
76 if (c <= 0xFFFF)
David Demelier <markand@malikania.fr>
parents:
diff changeset
77 return 3;
David Demelier <markand@malikania.fr>
parents:
diff changeset
78 if (c <= 0x1FFFFF)
David Demelier <markand@malikania.fr>
parents:
diff changeset
79 return 4;
David Demelier <markand@malikania.fr>
parents:
diff changeset
80
David Demelier <markand@malikania.fr>
parents:
diff changeset
81 return -1;
David Demelier <markand@malikania.fr>
parents:
diff changeset
82 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
83
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
84 int length(const std::string &str)
352
David Demelier <markand@malikania.fr>
parents:
diff changeset
85 {
David Demelier <markand@malikania.fr>
parents:
diff changeset
86 int total = 0;
David Demelier <markand@malikania.fr>
parents:
diff changeset
87
David Demelier <markand@malikania.fr>
parents:
diff changeset
88 forEach(str, [&] (char32_t) {
David Demelier <markand@malikania.fr>
parents:
diff changeset
89 ++ total;
David Demelier <markand@malikania.fr>
parents:
diff changeset
90 });
David Demelier <markand@malikania.fr>
parents:
diff changeset
91
David Demelier <markand@malikania.fr>
parents:
diff changeset
92 return total;
David Demelier <markand@malikania.fr>
parents:
diff changeset
93 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
94
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
95 std::string toUtf8(const std::u32string &array)
352
David Demelier <markand@malikania.fr>
parents:
diff changeset
96 {
David Demelier <markand@malikania.fr>
parents:
diff changeset
97 std::string res;
David Demelier <markand@malikania.fr>
parents:
diff changeset
98
David Demelier <markand@malikania.fr>
parents:
diff changeset
99 for (size_t i = 0; i < array.size(); ++i) {
David Demelier <markand@malikania.fr>
parents:
diff changeset
100 char tmp[5];
David Demelier <markand@malikania.fr>
parents:
diff changeset
101 int size = nbytesPoint(array[i]);
David Demelier <markand@malikania.fr>
parents:
diff changeset
102
David Demelier <markand@malikania.fr>
parents:
diff changeset
103 if (size < 0) {
David Demelier <markand@malikania.fr>
parents:
diff changeset
104 throw std::invalid_argument("invalid sequence");
David Demelier <markand@malikania.fr>
parents:
diff changeset
105 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
106
David Demelier <markand@malikania.fr>
parents:
diff changeset
107 encode(array[i], tmp);
David Demelier <markand@malikania.fr>
parents:
diff changeset
108 res.insert(res.length(), tmp);
David Demelier <markand@malikania.fr>
parents:
diff changeset
109 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
110
David Demelier <markand@malikania.fr>
parents:
diff changeset
111 return res;
David Demelier <markand@malikania.fr>
parents:
diff changeset
112 }
David Demelier <markand@malikania.fr>
parents:
diff changeset
113
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
114 std::u32string toUtf32(const std::string &str)
352
David Demelier <markand@malikania.fr>
parents:
diff changeset
115 {
David Demelier <markand@malikania.fr>
parents:
diff changeset
116 std::u32string res;
David Demelier <markand@malikania.fr>
parents:
diff changeset
117
David Demelier <markand@malikania.fr>
parents:
diff changeset
118 forEach(str, [&] (char32_t code) {
David Demelier <markand@malikania.fr>
parents:
diff changeset
119 res.push_back(code);
David Demelier <markand@malikania.fr>
parents:
diff changeset
120 });
David Demelier <markand@malikania.fr>
parents:
diff changeset
121
David Demelier <markand@malikania.fr>
parents:
diff changeset
122 return res;
395
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
123 }
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
124
b78d6d8f2872 Unicode: remove class, use namespace
David Demelier <markand@malikania.fr>
parents: 352
diff changeset
125 } // !unicode