xref: /freebsd/crypto/heimdal/lib/wind/utf8.c (revision ae77177087c655fc883075af4f425b37e032cd05)
1*ae771770SStanislav Sedov /*
2*ae771770SStanislav Sedov  * Copyright (c) 2004, 2006, 2007, 2008 Kungliga Tekniska Högskolan
3*ae771770SStanislav Sedov  * (Royal Institute of Technology, Stockholm, Sweden).
4*ae771770SStanislav Sedov  * All rights reserved.
5*ae771770SStanislav Sedov  *
6*ae771770SStanislav Sedov  * Redistribution and use in source and binary forms, with or without
7*ae771770SStanislav Sedov  * modification, are permitted provided that the following conditions
8*ae771770SStanislav Sedov  * are met:
9*ae771770SStanislav Sedov  *
10*ae771770SStanislav Sedov  * 1. Redistributions of source code must retain the above copyright
11*ae771770SStanislav Sedov  *    notice, this list of conditions and the following disclaimer.
12*ae771770SStanislav Sedov  *
13*ae771770SStanislav Sedov  * 2. Redistributions in binary form must reproduce the above copyright
14*ae771770SStanislav Sedov  *    notice, this list of conditions and the following disclaimer in the
15*ae771770SStanislav Sedov  *    documentation and/or other materials provided with the distribution.
16*ae771770SStanislav Sedov  *
17*ae771770SStanislav Sedov  * 3. Neither the name of the Institute nor the names of its contributors
18*ae771770SStanislav Sedov  *    may be used to endorse or promote products derived from this software
19*ae771770SStanislav Sedov  *    without specific prior written permission.
20*ae771770SStanislav Sedov  *
21*ae771770SStanislav Sedov  * THIS SOFTWARE IS PROVIDED BY THE INSTITUTE AND CONTRIBUTORS ``AS IS'' AND
22*ae771770SStanislav Sedov  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23*ae771770SStanislav Sedov  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24*ae771770SStanislav Sedov  * ARE DISCLAIMED.  IN NO EVENT SHALL THE INSTITUTE OR CONTRIBUTORS BE LIABLE
25*ae771770SStanislav Sedov  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26*ae771770SStanislav Sedov  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27*ae771770SStanislav Sedov  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28*ae771770SStanislav Sedov  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29*ae771770SStanislav Sedov  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30*ae771770SStanislav Sedov  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31*ae771770SStanislav Sedov  * SUCH DAMAGE.
32*ae771770SStanislav Sedov  */
33*ae771770SStanislav Sedov 
34*ae771770SStanislav Sedov #include <config.h>
35*ae771770SStanislav Sedov #include "windlocl.h"
36*ae771770SStanislav Sedov 
37*ae771770SStanislav Sedov static int
38*ae771770SStanislav Sedov utf8toutf32(const unsigned char **pp, uint32_t *out)
39*ae771770SStanislav Sedov {
40*ae771770SStanislav Sedov     const unsigned char *p = *pp;
41*ae771770SStanislav Sedov     unsigned c = *p;
42*ae771770SStanislav Sedov 
43*ae771770SStanislav Sedov     if (c & 0x80) {
44*ae771770SStanislav Sedov 	if ((c & 0xE0) == 0xC0) {
45*ae771770SStanislav Sedov 	    const unsigned c2 = *++p;
46*ae771770SStanislav Sedov 	    if ((c2 & 0xC0) == 0x80) {
47*ae771770SStanislav Sedov 		*out =  ((c  & 0x1F) << 6)
48*ae771770SStanislav Sedov 		    | (c2 & 0x3F);
49*ae771770SStanislav Sedov 	    } else {
50*ae771770SStanislav Sedov 		return WIND_ERR_INVALID_UTF8;
51*ae771770SStanislav Sedov 	    }
52*ae771770SStanislav Sedov 	} else if ((c & 0xF0) == 0xE0) {
53*ae771770SStanislav Sedov 	    const unsigned c2 = *++p;
54*ae771770SStanislav Sedov 	    if ((c2 & 0xC0) == 0x80) {
55*ae771770SStanislav Sedov 		const unsigned c3 = *++p;
56*ae771770SStanislav Sedov 		if ((c3 & 0xC0) == 0x80) {
57*ae771770SStanislav Sedov 		    *out =   ((c  & 0x0F) << 12)
58*ae771770SStanislav Sedov 			| ((c2 & 0x3F) << 6)
59*ae771770SStanislav Sedov 			|  (c3 & 0x3F);
60*ae771770SStanislav Sedov 		} else {
61*ae771770SStanislav Sedov 		    return WIND_ERR_INVALID_UTF8;
62*ae771770SStanislav Sedov 		}
63*ae771770SStanislav Sedov 	    } else {
64*ae771770SStanislav Sedov 		return WIND_ERR_INVALID_UTF8;
65*ae771770SStanislav Sedov 	    }
66*ae771770SStanislav Sedov 	} else if ((c & 0xF8) == 0xF0) {
67*ae771770SStanislav Sedov 	    const unsigned c2 = *++p;
68*ae771770SStanislav Sedov 	    if ((c2 & 0xC0) == 0x80) {
69*ae771770SStanislav Sedov 		const unsigned c3 = *++p;
70*ae771770SStanislav Sedov 		if ((c3 & 0xC0) == 0x80) {
71*ae771770SStanislav Sedov 		    const unsigned c4 = *++p;
72*ae771770SStanislav Sedov 		    if ((c4 & 0xC0) == 0x80) {
73*ae771770SStanislav Sedov 			*out =   ((c  & 0x07) << 18)
74*ae771770SStanislav Sedov 			    | ((c2 & 0x3F) << 12)
75*ae771770SStanislav Sedov 			    | ((c3 & 0x3F) <<  6)
76*ae771770SStanislav Sedov 			    |  (c4 & 0x3F);
77*ae771770SStanislav Sedov 		    } else {
78*ae771770SStanislav Sedov 			return WIND_ERR_INVALID_UTF8;
79*ae771770SStanislav Sedov 		    }
80*ae771770SStanislav Sedov 		} else {
81*ae771770SStanislav Sedov 		    return WIND_ERR_INVALID_UTF8;
82*ae771770SStanislav Sedov 		}
83*ae771770SStanislav Sedov 	    } else {
84*ae771770SStanislav Sedov 		return WIND_ERR_INVALID_UTF8;
85*ae771770SStanislav Sedov 	    }
86*ae771770SStanislav Sedov 	} else {
87*ae771770SStanislav Sedov 	    return WIND_ERR_INVALID_UTF8;
88*ae771770SStanislav Sedov 	}
89*ae771770SStanislav Sedov     } else {
90*ae771770SStanislav Sedov 	*out = c;
91*ae771770SStanislav Sedov     }
92*ae771770SStanislav Sedov 
93*ae771770SStanislav Sedov     *pp = p;
94*ae771770SStanislav Sedov 
95*ae771770SStanislav Sedov     return 0;
96*ae771770SStanislav Sedov }
97*ae771770SStanislav Sedov 
98*ae771770SStanislav Sedov /**
99*ae771770SStanislav Sedov  * Convert an UTF-8 string to an UCS4 string.
100*ae771770SStanislav Sedov  *
101*ae771770SStanislav Sedov  * @param in an UTF-8 string to convert.
102*ae771770SStanislav Sedov  * @param out the resulting UCS4 strint, must be at least
103*ae771770SStanislav Sedov  * wind_utf8ucs4_length() long.  If out is NULL, the function will
104*ae771770SStanislav Sedov  * calculate the needed space for the out variable (just like
105*ae771770SStanislav Sedov  * wind_utf8ucs4_length()).
106*ae771770SStanislav Sedov  * @param out_len before processing out_len should be the length of
107*ae771770SStanislav Sedov  * the out variable, after processing it will be the length of the out
108*ae771770SStanislav Sedov  * string.
109*ae771770SStanislav Sedov  *
110*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
111*ae771770SStanislav Sedov  * @ingroup wind
112*ae771770SStanislav Sedov  */
113*ae771770SStanislav Sedov 
114*ae771770SStanislav Sedov int
115*ae771770SStanislav Sedov wind_utf8ucs4(const char *in, uint32_t *out, size_t *out_len)
116*ae771770SStanislav Sedov {
117*ae771770SStanislav Sedov     const unsigned char *p;
118*ae771770SStanislav Sedov     size_t o = 0;
119*ae771770SStanislav Sedov     int ret;
120*ae771770SStanislav Sedov 
121*ae771770SStanislav Sedov     for (p = (const unsigned char *)in; *p != '\0'; ++p) {
122*ae771770SStanislav Sedov 	uint32_t u;
123*ae771770SStanislav Sedov 
124*ae771770SStanislav Sedov 	ret = utf8toutf32(&p, &u);
125*ae771770SStanislav Sedov 	if (ret)
126*ae771770SStanislav Sedov 	    return ret;
127*ae771770SStanislav Sedov 
128*ae771770SStanislav Sedov 	if (out) {
129*ae771770SStanislav Sedov 	    if (o >= *out_len)
130*ae771770SStanislav Sedov 		return WIND_ERR_OVERRUN;
131*ae771770SStanislav Sedov 	    out[o] = u;
132*ae771770SStanislav Sedov 	}
133*ae771770SStanislav Sedov 	o++;
134*ae771770SStanislav Sedov     }
135*ae771770SStanislav Sedov     *out_len = o;
136*ae771770SStanislav Sedov     return 0;
137*ae771770SStanislav Sedov }
138*ae771770SStanislav Sedov 
139*ae771770SStanislav Sedov /**
140*ae771770SStanislav Sedov  * Calculate the length of from converting a UTF-8 string to a UCS4
141*ae771770SStanislav Sedov  * string.
142*ae771770SStanislav Sedov  *
143*ae771770SStanislav Sedov  * @param in an UTF-8 string to convert.
144*ae771770SStanislav Sedov  * @param out_len the length of the resulting UCS4 string.
145*ae771770SStanislav Sedov  *
146*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
147*ae771770SStanislav Sedov  * @ingroup wind
148*ae771770SStanislav Sedov  */
149*ae771770SStanislav Sedov 
150*ae771770SStanislav Sedov int
151*ae771770SStanislav Sedov wind_utf8ucs4_length(const char *in, size_t *out_len)
152*ae771770SStanislav Sedov {
153*ae771770SStanislav Sedov     return wind_utf8ucs4(in, NULL, out_len);
154*ae771770SStanislav Sedov }
155*ae771770SStanislav Sedov 
156*ae771770SStanislav Sedov static const char first_char[4] =
157*ae771770SStanislav Sedov     { 0x00, 0xC0, 0xE0, 0xF0 };
158*ae771770SStanislav Sedov 
159*ae771770SStanislav Sedov /**
160*ae771770SStanislav Sedov  * Convert an UCS4 string to a UTF-8 string.
161*ae771770SStanislav Sedov  *
162*ae771770SStanislav Sedov  * @param in an UCS4 string to convert.
163*ae771770SStanislav Sedov  * @param in_len the length input array.
164*ae771770SStanislav Sedov 
165*ae771770SStanislav Sedov  * @param out the resulting UTF-8 strint, must be at least
166*ae771770SStanislav Sedov  * wind_ucs4utf8_length() + 1 long (the extra char for the NUL).  If
167*ae771770SStanislav Sedov  * out is NULL, the function will calculate the needed space for the
168*ae771770SStanislav Sedov  * out variable (just like wind_ucs4utf8_length()).
169*ae771770SStanislav Sedov 
170*ae771770SStanislav Sedov  * @param out_len before processing out_len should be the length of
171*ae771770SStanislav Sedov  * the out variable, after processing it will be the length of the out
172*ae771770SStanislav Sedov  * string.
173*ae771770SStanislav Sedov  *
174*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
175*ae771770SStanislav Sedov  * @ingroup wind
176*ae771770SStanislav Sedov  */
177*ae771770SStanislav Sedov 
178*ae771770SStanislav Sedov int
179*ae771770SStanislav Sedov wind_ucs4utf8(const uint32_t *in, size_t in_len, char *out, size_t *out_len)
180*ae771770SStanislav Sedov {
181*ae771770SStanislav Sedov     uint32_t ch;
182*ae771770SStanislav Sedov     size_t i, len, o;
183*ae771770SStanislav Sedov 
184*ae771770SStanislav Sedov     for (o = 0, i = 0; i < in_len; i++) {
185*ae771770SStanislav Sedov 	ch = in[i];
186*ae771770SStanislav Sedov 
187*ae771770SStanislav Sedov 	if (ch < 0x80) {
188*ae771770SStanislav Sedov 	    len = 1;
189*ae771770SStanislav Sedov 	} else if (ch < 0x800) {
190*ae771770SStanislav Sedov 	    len = 2;
191*ae771770SStanislav Sedov 	} else if (ch < 0x10000) {
192*ae771770SStanislav Sedov 	    len = 3;
193*ae771770SStanislav Sedov 	} else if (ch <= 0x10FFFF) {
194*ae771770SStanislav Sedov 	    len = 4;
195*ae771770SStanislav Sedov 	} else
196*ae771770SStanislav Sedov 	    return WIND_ERR_INVALID_UTF32;
197*ae771770SStanislav Sedov 
198*ae771770SStanislav Sedov 	o += len;
199*ae771770SStanislav Sedov 
200*ae771770SStanislav Sedov 	if (out) {
201*ae771770SStanislav Sedov 	    if (o >= *out_len)
202*ae771770SStanislav Sedov 		return WIND_ERR_OVERRUN;
203*ae771770SStanislav Sedov 
204*ae771770SStanislav Sedov 	    switch(len) {
205*ae771770SStanislav Sedov 	    case 4:
206*ae771770SStanislav Sedov 		out[3] = (ch | 0x80) & 0xbf;
207*ae771770SStanislav Sedov 		ch = ch << 6;
208*ae771770SStanislav Sedov 	    case 3:
209*ae771770SStanislav Sedov 		out[2] = (ch | 0x80) & 0xbf;
210*ae771770SStanislav Sedov 		ch = ch << 6;
211*ae771770SStanislav Sedov 	    case 2:
212*ae771770SStanislav Sedov 		out[1] = (ch | 0x80) & 0xbf;
213*ae771770SStanislav Sedov 		ch = ch << 6;
214*ae771770SStanislav Sedov 	    case 1:
215*ae771770SStanislav Sedov 		out[0] = ch | first_char[len - 1];
216*ae771770SStanislav Sedov 	    }
217*ae771770SStanislav Sedov 	}
218*ae771770SStanislav Sedov 	out += len;
219*ae771770SStanislav Sedov     }
220*ae771770SStanislav Sedov     if (out) {
221*ae771770SStanislav Sedov 	if (o + 1 >= *out_len)
222*ae771770SStanislav Sedov 	    return WIND_ERR_OVERRUN;
223*ae771770SStanislav Sedov 	*out = '\0';
224*ae771770SStanislav Sedov     }
225*ae771770SStanislav Sedov     *out_len = o;
226*ae771770SStanislav Sedov     return 0;
227*ae771770SStanislav Sedov }
228*ae771770SStanislav Sedov 
229*ae771770SStanislav Sedov /**
230*ae771770SStanislav Sedov  * Calculate the length of from converting a UCS4 string to an UTF-8 string.
231*ae771770SStanislav Sedov  *
232*ae771770SStanislav Sedov  * @param in an UCS4 string to convert.
233*ae771770SStanislav Sedov  * @param in_len the length of UCS4 string to convert.
234*ae771770SStanislav Sedov  * @param out_len the length of the resulting UTF-8 string.
235*ae771770SStanislav Sedov  *
236*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
237*ae771770SStanislav Sedov  * @ingroup wind
238*ae771770SStanislav Sedov  */
239*ae771770SStanislav Sedov 
240*ae771770SStanislav Sedov int
241*ae771770SStanislav Sedov wind_ucs4utf8_length(const uint32_t *in, size_t in_len, size_t *out_len)
242*ae771770SStanislav Sedov {
243*ae771770SStanislav Sedov     return wind_ucs4utf8(in, in_len, NULL, out_len);
244*ae771770SStanislav Sedov }
245*ae771770SStanislav Sedov 
246*ae771770SStanislav Sedov /**
247*ae771770SStanislav Sedov  * Read in an UCS2 from a buffer.
248*ae771770SStanislav Sedov  *
249*ae771770SStanislav Sedov  * @param ptr The input buffer to read from.
250*ae771770SStanislav Sedov  * @param len the length of the input buffer.
251*ae771770SStanislav Sedov  * @param flags Flags to control the behavior of the function.
252*ae771770SStanislav Sedov  * @param out the output UCS2, the array must be at least out/2 long.
253*ae771770SStanislav Sedov  * @param out_len the output length
254*ae771770SStanislav Sedov  *
255*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise.
256*ae771770SStanislav Sedov  * @ingroup wind
257*ae771770SStanislav Sedov  */
258*ae771770SStanislav Sedov 
259*ae771770SStanislav Sedov int
260*ae771770SStanislav Sedov wind_ucs2read(const void *ptr, size_t len, unsigned int *flags,
261*ae771770SStanislav Sedov 	      uint16_t *out, size_t *out_len)
262*ae771770SStanislav Sedov {
263*ae771770SStanislav Sedov     const unsigned char *p = ptr;
264*ae771770SStanislav Sedov     int little = ((*flags) & WIND_RW_LE);
265*ae771770SStanislav Sedov     size_t olen = *out_len;
266*ae771770SStanislav Sedov 
267*ae771770SStanislav Sedov     /** if len is zero, flags are unchanged */
268*ae771770SStanislav Sedov     if (len == 0) {
269*ae771770SStanislav Sedov 	*out_len = 0;
270*ae771770SStanislav Sedov 	return 0;
271*ae771770SStanislav Sedov     }
272*ae771770SStanislav Sedov 
273*ae771770SStanislav Sedov     /** if len is odd, WIND_ERR_LENGTH_NOT_MOD2 is returned */
274*ae771770SStanislav Sedov     if (len & 1)
275*ae771770SStanislav Sedov 	return WIND_ERR_LENGTH_NOT_MOD2;
276*ae771770SStanislav Sedov 
277*ae771770SStanislav Sedov     /**
278*ae771770SStanislav Sedov      * If the flags WIND_RW_BOM is set, check for BOM. If not BOM is
279*ae771770SStanislav Sedov      * found, check is LE/BE flag is already and use that otherwise
280*ae771770SStanislav Sedov      * fail with WIND_ERR_NO_BOM. When done, clear WIND_RW_BOM and
281*ae771770SStanislav Sedov      * the LE/BE flag and set the resulting LE/BE flag.
282*ae771770SStanislav Sedov      */
283*ae771770SStanislav Sedov     if ((*flags) & WIND_RW_BOM) {
284*ae771770SStanislav Sedov 	uint16_t bom = (p[0] << 8) + p[1];
285*ae771770SStanislav Sedov 	if (bom == 0xfffe || bom == 0xfeff) {
286*ae771770SStanislav Sedov 	    little = (bom == 0xfffe);
287*ae771770SStanislav Sedov 	    p += 2;
288*ae771770SStanislav Sedov 	    len -= 2;
289*ae771770SStanislav Sedov 	} else if (((*flags) & (WIND_RW_LE|WIND_RW_BE)) != 0) {
290*ae771770SStanislav Sedov 	    /* little already set */
291*ae771770SStanislav Sedov 	} else
292*ae771770SStanislav Sedov 	    return WIND_ERR_NO_BOM;
293*ae771770SStanislav Sedov 	*flags = ((*flags) & ~(WIND_RW_BOM|WIND_RW_LE|WIND_RW_BE));
294*ae771770SStanislav Sedov 	*flags |= little ? WIND_RW_LE : WIND_RW_BE;
295*ae771770SStanislav Sedov     }
296*ae771770SStanislav Sedov 
297*ae771770SStanislav Sedov     while (len) {
298*ae771770SStanislav Sedov 	if (olen < 1)
299*ae771770SStanislav Sedov 	    return WIND_ERR_OVERRUN;
300*ae771770SStanislav Sedov 	if (little)
301*ae771770SStanislav Sedov 	    *out = (p[1] << 8) + p[0];
302*ae771770SStanislav Sedov 	else
303*ae771770SStanislav Sedov 	    *out = (p[0] << 8) + p[1];
304*ae771770SStanislav Sedov 	out++; p += 2; len -= 2; olen--;
305*ae771770SStanislav Sedov     }
306*ae771770SStanislav Sedov     *out_len -= olen;
307*ae771770SStanislav Sedov     return 0;
308*ae771770SStanislav Sedov }
309*ae771770SStanislav Sedov 
310*ae771770SStanislav Sedov /**
311*ae771770SStanislav Sedov  * Write an UCS2 string to a buffer.
312*ae771770SStanislav Sedov  *
313*ae771770SStanislav Sedov  * @param in The input UCS2 string.
314*ae771770SStanislav Sedov  * @param in_len the length of the input buffer.
315*ae771770SStanislav Sedov  * @param flags Flags to control the behavior of the function.
316*ae771770SStanislav Sedov  * @param ptr The input buffer to write to, the array must be at least
317*ae771770SStanislav Sedov  * (in + 1) * 2 bytes long.
318*ae771770SStanislav Sedov  * @param out_len the output length
319*ae771770SStanislav Sedov  *
320*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise.
321*ae771770SStanislav Sedov  * @ingroup wind
322*ae771770SStanislav Sedov  */
323*ae771770SStanislav Sedov 
324*ae771770SStanislav Sedov int
325*ae771770SStanislav Sedov wind_ucs2write(const uint16_t *in, size_t in_len, unsigned int *flags,
326*ae771770SStanislav Sedov 	       void *ptr, size_t *out_len)
327*ae771770SStanislav Sedov {
328*ae771770SStanislav Sedov     unsigned char *p = ptr;
329*ae771770SStanislav Sedov     size_t len = *out_len;
330*ae771770SStanislav Sedov 
331*ae771770SStanislav Sedov     /** If in buffer is not of length be mod 2, WIND_ERR_LENGTH_NOT_MOD2 is returned*/
332*ae771770SStanislav Sedov     if (len & 1)
333*ae771770SStanislav Sedov 	return WIND_ERR_LENGTH_NOT_MOD2;
334*ae771770SStanislav Sedov 
335*ae771770SStanislav Sedov     /** On zero input length, flags are preserved */
336*ae771770SStanislav Sedov     if (in_len == 0) {
337*ae771770SStanislav Sedov 	*out_len = 0;
338*ae771770SStanislav Sedov 	return 0;
339*ae771770SStanislav Sedov     }
340*ae771770SStanislav Sedov     /** If flags have WIND_RW_BOM set, the byte order mark is written
341*ae771770SStanislav Sedov      * first to the output data */
342*ae771770SStanislav Sedov     if ((*flags) & WIND_RW_BOM) {
343*ae771770SStanislav Sedov 	uint16_t bom = 0xfffe;
344*ae771770SStanislav Sedov 
345*ae771770SStanislav Sedov 	if (len < 2)
346*ae771770SStanislav Sedov 	    return WIND_ERR_OVERRUN;
347*ae771770SStanislav Sedov 
348*ae771770SStanislav Sedov 	if ((*flags) & WIND_RW_LE) {
349*ae771770SStanislav Sedov 	    p[0] = (bom >> 8) & 0xff;
350*ae771770SStanislav Sedov 	    p[1] = (bom     ) & 0xff;
351*ae771770SStanislav Sedov 	} else {
352*ae771770SStanislav Sedov 	    p[1] = (bom     ) & 0xff;
353*ae771770SStanislav Sedov 	    p[0] = (bom >> 8) & 0xff;
354*ae771770SStanislav Sedov 	}
355*ae771770SStanislav Sedov 	len -= 2;
356*ae771770SStanislav Sedov     }
357*ae771770SStanislav Sedov 
358*ae771770SStanislav Sedov     while (in_len) {
359*ae771770SStanislav Sedov 	/** If the output wont fit into out_len, WIND_ERR_OVERRUN is returned */
360*ae771770SStanislav Sedov 	if (len < 2)
361*ae771770SStanislav Sedov 	    return WIND_ERR_OVERRUN;
362*ae771770SStanislav Sedov 	if ((*flags) & WIND_RW_LE) {
363*ae771770SStanislav Sedov 	    p[0] = (in[0] >> 8) & 0xff;
364*ae771770SStanislav Sedov 	    p[1] = (in[0]     ) & 0xff;
365*ae771770SStanislav Sedov 	} else {
366*ae771770SStanislav Sedov 	    p[1] = (in[0]     ) & 0xff;
367*ae771770SStanislav Sedov 	    p[0] = (in[0] >> 8) & 0xff;
368*ae771770SStanislav Sedov 	}
369*ae771770SStanislav Sedov 	len -= 2;
370*ae771770SStanislav Sedov 	in_len--;
371*ae771770SStanislav Sedov 	p += 2;
372*ae771770SStanislav Sedov 	in++;
373*ae771770SStanislav Sedov     }
374*ae771770SStanislav Sedov     *out_len -= len;
375*ae771770SStanislav Sedov     return 0;
376*ae771770SStanislav Sedov }
377*ae771770SStanislav Sedov 
378*ae771770SStanislav Sedov 
379*ae771770SStanislav Sedov /**
380*ae771770SStanislav Sedov  * Convert an UTF-8 string to an UCS2 string.
381*ae771770SStanislav Sedov  *
382*ae771770SStanislav Sedov  * @param in an UTF-8 string to convert.
383*ae771770SStanislav Sedov  * @param out the resulting UCS2 strint, must be at least
384*ae771770SStanislav Sedov  * wind_utf8ucs2_length() long.  If out is NULL, the function will
385*ae771770SStanislav Sedov  * calculate the needed space for the out variable (just like
386*ae771770SStanislav Sedov  * wind_utf8ucs2_length()).
387*ae771770SStanislav Sedov  * @param out_len before processing out_len should be the length of
388*ae771770SStanislav Sedov  * the out variable, after processing it will be the length of the out
389*ae771770SStanislav Sedov  * string.
390*ae771770SStanislav Sedov  *
391*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
392*ae771770SStanislav Sedov  * @ingroup wind
393*ae771770SStanislav Sedov  */
394*ae771770SStanislav Sedov 
395*ae771770SStanislav Sedov int
396*ae771770SStanislav Sedov wind_utf8ucs2(const char *in, uint16_t *out, size_t *out_len)
397*ae771770SStanislav Sedov {
398*ae771770SStanislav Sedov     const unsigned char *p;
399*ae771770SStanislav Sedov     size_t o = 0;
400*ae771770SStanislav Sedov     int ret;
401*ae771770SStanislav Sedov 
402*ae771770SStanislav Sedov     for (p = (const unsigned char *)in; *p != '\0'; ++p) {
403*ae771770SStanislav Sedov 	uint32_t u;
404*ae771770SStanislav Sedov 
405*ae771770SStanislav Sedov 	ret = utf8toutf32(&p, &u);
406*ae771770SStanislav Sedov 	if (ret)
407*ae771770SStanislav Sedov 	    return ret;
408*ae771770SStanislav Sedov 
409*ae771770SStanislav Sedov 	if (u & 0xffff0000)
410*ae771770SStanislav Sedov 	    return WIND_ERR_NOT_UTF16;
411*ae771770SStanislav Sedov 
412*ae771770SStanislav Sedov 	if (out) {
413*ae771770SStanislav Sedov 	    if (o >= *out_len)
414*ae771770SStanislav Sedov 		return WIND_ERR_OVERRUN;
415*ae771770SStanislav Sedov 	    out[o] = u;
416*ae771770SStanislav Sedov 	}
417*ae771770SStanislav Sedov 	o++;
418*ae771770SStanislav Sedov     }
419*ae771770SStanislav Sedov     *out_len = o;
420*ae771770SStanislav Sedov     return 0;
421*ae771770SStanislav Sedov }
422*ae771770SStanislav Sedov 
423*ae771770SStanislav Sedov /**
424*ae771770SStanislav Sedov  * Calculate the length of from converting a UTF-8 string to a UCS2
425*ae771770SStanislav Sedov  * string.
426*ae771770SStanislav Sedov  *
427*ae771770SStanislav Sedov  * @param in an UTF-8 string to convert.
428*ae771770SStanislav Sedov  * @param out_len the length of the resulting UCS4 string.
429*ae771770SStanislav Sedov  *
430*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
431*ae771770SStanislav Sedov  * @ingroup wind
432*ae771770SStanislav Sedov  */
433*ae771770SStanislav Sedov 
434*ae771770SStanislav Sedov int
435*ae771770SStanislav Sedov wind_utf8ucs2_length(const char *in, size_t *out_len)
436*ae771770SStanislav Sedov {
437*ae771770SStanislav Sedov     return wind_utf8ucs2(in, NULL, out_len);
438*ae771770SStanislav Sedov }
439*ae771770SStanislav Sedov 
440*ae771770SStanislav Sedov /**
441*ae771770SStanislav Sedov  * Convert an UCS2 string to a UTF-8 string.
442*ae771770SStanislav Sedov  *
443*ae771770SStanislav Sedov  * @param in an UCS2 string to convert.
444*ae771770SStanislav Sedov  * @param in_len the length of the in UCS2 string.
445*ae771770SStanislav Sedov  * @param out the resulting UTF-8 strint, must be at least
446*ae771770SStanislav Sedov  * wind_ucs2utf8_length() long.  If out is NULL, the function will
447*ae771770SStanislav Sedov  * calculate the needed space for the out variable (just like
448*ae771770SStanislav Sedov  * wind_ucs2utf8_length()).
449*ae771770SStanislav Sedov  * @param out_len before processing out_len should be the length of
450*ae771770SStanislav Sedov  * the out variable, after processing it will be the length of the out
451*ae771770SStanislav Sedov  * string.
452*ae771770SStanislav Sedov  *
453*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
454*ae771770SStanislav Sedov  * @ingroup wind
455*ae771770SStanislav Sedov  */
456*ae771770SStanislav Sedov 
457*ae771770SStanislav Sedov int
458*ae771770SStanislav Sedov wind_ucs2utf8(const uint16_t *in, size_t in_len, char *out, size_t *out_len)
459*ae771770SStanislav Sedov {
460*ae771770SStanislav Sedov     uint16_t ch;
461*ae771770SStanislav Sedov     size_t i, len, o;
462*ae771770SStanislav Sedov 
463*ae771770SStanislav Sedov     for (o = 0, i = 0; i < in_len; i++) {
464*ae771770SStanislav Sedov 	ch = in[i];
465*ae771770SStanislav Sedov 
466*ae771770SStanislav Sedov 	if (ch < 0x80) {
467*ae771770SStanislav Sedov 	    len = 1;
468*ae771770SStanislav Sedov 	} else if (ch < 0x800) {
469*ae771770SStanislav Sedov 	    len = 2;
470*ae771770SStanislav Sedov 	} else
471*ae771770SStanislav Sedov 	    len = 3;
472*ae771770SStanislav Sedov 
473*ae771770SStanislav Sedov 	o += len;
474*ae771770SStanislav Sedov 
475*ae771770SStanislav Sedov 	if (out) {
476*ae771770SStanislav Sedov 	    if (o >= *out_len)
477*ae771770SStanislav Sedov 		return WIND_ERR_OVERRUN;
478*ae771770SStanislav Sedov 
479*ae771770SStanislav Sedov 	    switch(len) {
480*ae771770SStanislav Sedov 	    case 3:
481*ae771770SStanislav Sedov 		out[2] = (ch | 0x80) & 0xbf;
482*ae771770SStanislav Sedov 		ch = ch << 6;
483*ae771770SStanislav Sedov 	    case 2:
484*ae771770SStanislav Sedov 		out[1] = (ch | 0x80) & 0xbf;
485*ae771770SStanislav Sedov 		ch = ch << 6;
486*ae771770SStanislav Sedov 	    case 1:
487*ae771770SStanislav Sedov 		out[0] = ch | first_char[len - 1];
488*ae771770SStanislav Sedov 	    }
489*ae771770SStanislav Sedov 	    out += len;
490*ae771770SStanislav Sedov 	}
491*ae771770SStanislav Sedov     }
492*ae771770SStanislav Sedov     if (out) {
493*ae771770SStanislav Sedov 	if (o >= *out_len)
494*ae771770SStanislav Sedov 	    return WIND_ERR_OVERRUN;
495*ae771770SStanislav Sedov 	*out = '\0';
496*ae771770SStanislav Sedov     }
497*ae771770SStanislav Sedov     *out_len = o;
498*ae771770SStanislav Sedov     return 0;
499*ae771770SStanislav Sedov }
500*ae771770SStanislav Sedov 
501*ae771770SStanislav Sedov /**
502*ae771770SStanislav Sedov  * Calculate the length of from converting a UCS2 string to an UTF-8 string.
503*ae771770SStanislav Sedov  *
504*ae771770SStanislav Sedov  * @param in an UCS2 string to convert.
505*ae771770SStanislav Sedov  * @param in_len an UCS2 string length to convert.
506*ae771770SStanislav Sedov  * @param out_len the length of the resulting UTF-8 string.
507*ae771770SStanislav Sedov  *
508*ae771770SStanislav Sedov  * @return returns 0 on success, an wind error code otherwise
509*ae771770SStanislav Sedov  * @ingroup wind
510*ae771770SStanislav Sedov  */
511*ae771770SStanislav Sedov 
512*ae771770SStanislav Sedov int
513*ae771770SStanislav Sedov wind_ucs2utf8_length(const uint16_t *in, size_t in_len, size_t *out_len)
514*ae771770SStanislav Sedov {
515*ae771770SStanislav Sedov     return wind_ucs2utf8(in, in_len, NULL, out_len);
516*ae771770SStanislav Sedov }
517