1 //
2 // c32rtomb.cpp
3 //
4 // Copyright (c) Microsoft Corporation. All rights reserved.
5 //
6
7 #include <corecrt_internal_mbstring.h>
8 #include <corecrt_internal_ptd_propagation.h>
9 #include <stdint.h>
10 #include <uchar.h>
11
12 using namespace __crt_mbstring;
13
c32rtomb(char * s,char32_t c32,mbstate_t * ps)14 extern "C" size_t __cdecl c32rtomb(char* s, char32_t c32, mbstate_t* ps)
15 {
16 // TODO: Bug 13307590 says this is always assuming UTF-8.
17 __crt_cached_ptd_host ptd;
18 return __c32rtomb_utf8(s, c32, ps, ptd);
19 }
20
__c32rtomb_utf8(char * s,char32_t c32,mbstate_t * ps,__crt_cached_ptd_host & ptd)21 size_t __cdecl __crt_mbstring::__c32rtomb_utf8(char* s, char32_t c32, mbstate_t* ps, __crt_cached_ptd_host& ptd)
22 {
23 if (!s)
24 {
25 // Equivalent to c32rtomb(buf, U'\0', ps) for some internal buffer buf
26 *ps = {};
27 return 1;
28 }
29
30 if (c32 == U'\0')
31 {
32 *s = '\0';
33 *ps = {};
34 return 1;
35 }
36
37 // Fast path for ASCII
38 if ((c32 & ~0x7f) == 0)
39 {
40 *s = static_cast<char>(c32);
41 return 1;
42 }
43
44 // Figure out how many trail bytes we need
45 size_t trail_bytes;
46 uint8_t lead_byte;
47 if ((c32 & ~0x7ff) == 0)
48 {
49 trail_bytes = 1;
50 lead_byte = 0xc0;
51 }
52 else if ((c32 & ~0xffff) == 0)
53 {
54 // high/low surrogates are only valid in UTF-16 encoded data
55 if (0xd800 <= c32 && c32 <= 0xdfff)
56 {
57 return return_illegal_sequence(ps, ptd);
58 }
59 trail_bytes = 2;
60 lead_byte = 0xe0;
61 }
62 else if ((c32 & ~0x001fffff) == 0)
63 {
64 // Unicode's max code point is 0x10ffff
65 if (0x10ffff < c32)
66 {
67 return return_illegal_sequence(ps, ptd);
68 }
69 trail_bytes = 3;
70 lead_byte = 0xf0;
71 }
72 else
73 {
74 return return_illegal_sequence(ps, ptd);
75 }
76 _ASSERTE(1 <= trail_bytes && trail_bytes <= 3);
77
78 // Put six bits into each of the trail bytes
79 // Lowest bits are in the last UTF-8 byte.
80 // Filling back to front.
81 for (size_t i = trail_bytes; i > 0; --i)
82 {
83 s[i] = (c32 & 0x3f) | 0x80;
84 c32 >>= 6;
85 }
86
87 // The first byte needs the upper (trail_bytes + 1) bits to store the length
88 // And the lower (7 - trail_bytes) to store the upper bits of the code point
89 _ASSERTE(c32 < (1u << (7 - trail_bytes)));
90 s[0] = static_cast<uint8_t>(c32) | lead_byte;
91
92 return reset_and_return(trail_bytes + 1, ps);
93 }
94