utf-16.cpp@ 4071

Last change on this file since 4071 was 4071, checked in by vboxsync, 17 years ago
Biggest check-in ever. New source code headers for all (C) innotek files.
Property svn:eol-style set to `native` Property svn:keywords set to `Id`
File size: 16.4 KB

Line
1	/* $Id: utf-16.cpp 4071 2007-08-07 17:07:59Z vboxsync $ */
2	/** @file
3	* innotek Portable Runtime - UTF-16
4	*/
5
6	/*
7	* Copyright (C) 2006-2007 innotek GmbH
8	*
9	* This file is part of VirtualBox Open Source Edition (OSE), as
10	* available from http://www.virtualbox.org. This file is free software;
11	* you can redistribute it and/or modify it under the terms of the GNU
12	* General Public License as published by the Free Software Foundation,
13	* in version 2 as it comes in the "COPYING" file of the VirtualBox OSE
14	* distribution. VirtualBox OSE is distributed in the hope that it will
15	* be useful, but WITHOUT ANY WARRANTY of any kind.
16	*/
17
18
19	/*******************************************************************************
20	* Header Files *
21	*******************************************************************************/
22	#include <iprt/string.h>
23	#include <iprt/uni.h>
24	#include <iprt/alloc.h>
25	#include <iprt/assert.h>
26	#include <iprt/err.h>
27	#include "internal/string.h"
28
29
30
31	RTDECL(void) RTUtf16Free(PRTUTF16 pwszString)
32	{
33	if (pwszString)
34	RTMemTmpFree(pwszString);
35	}
36
37
38	RTDECL(PRTUTF16) RTUtf16Dup(PCRTUTF16 pwszString)
39	{
40	Assert(pwszString);
41	size_t cb = (RTUtf16Len(pwszString) + 1) * sizeof(RTUTF16);
42	PRTUTF16 pwsz = (PRTUTF16)RTMemAlloc(cb);
43	if (pwsz)
44	memcpy(pwsz, pwszString, cb);
45	return pwsz;
46	}
47
48
49	RTDECL(int) RTUtf16DupEx(PRTUTF16 *ppwszString, PCRTUTF16 pwszString, size_t cwcExtra)
50	{
51	Assert(pwszString);
52	size_t cb = (RTUtf16Len(pwszString) + 1) * sizeof(RTUTF16);
53	PRTUTF16 pwsz = (PRTUTF16)RTMemAlloc(cb + cwcExtra * sizeof(RTUTF16));
54	if (pwsz)
55	{
56	memcpy(pwsz, pwszString, cb);
57	*ppwszString = pwsz;
58	return VINF_SUCCESS;
59	}
60	return VERR_NO_MEMORY;
61	}
62
63
64	RTDECL(size_t) RTUtf16Len(PCRTUTF16 pwszString)
65	{
66	if (!pwszString)
67	return 0;
68
69	PCRTUTF16 pwsz = pwszString;
70	while (*pwsz)
71	pwsz++;
72	return pwsz - pwszString;
73	}
74
75
76	RTDECL(int) RTUtf16Cmp(register PCRTUTF16 pwsz1, register PCRTUTF16 pwsz2)
77	{
78	if (pwsz1 == pwsz2)
79	return 0;
80	if (!pwsz1)
81	return -1;
82	if (!pwsz2)
83	return 1;
84
85	for (;;)
86	{
87	register RTUTF16 wcs = *pwsz1;
88	register int iDiff = wcs - *pwsz2;
89	if (iDiff \|\| !wcs)
90	return iDiff;
91	pwsz1++;
92	pwsz2++;
93	}
94	}
95
96
97	RTDECL(int) RTUtf16ICmp(register PCRTUTF16 pwsz1, register PCRTUTF16 pwsz2)
98	{
99	if (pwsz1 == pwsz2)
100	return 0;
101	if (!pwsz1)
102	return -1;
103	if (!pwsz2)
104	return 1;
105
106	PCRTUTF16 pwsz1Start = pwsz1; /* keep it around in case we have to backtrack on a surrogate pair */
107	for (;;)
108	{
109	register RTUTF16 wc1 = *pwsz1;
110	register RTUTF16 wc2 = *pwsz2;
111	register int iDiff = wc1 - wc2;
112	if (iDiff)
113	{
114	/* unless they are both surrogate pairs, there is no chance they'll be identical. */
115	if ( wc1 < 0xd800
116	\|\| wc2 < 0xd800
117	\|\| wc1 > 0xdfff
118	\|\| wc2 > 0xdfff)
119	{
120	/* simple UCS-2 char */
121	iDiff = RTUniCpToUpper(wc1) - RTUniCpToUpper(wc2);
122	if (iDiff)
123	iDiff = RTUniCpToLower(wc1) - RTUniCpToLower(wc2);
124	}
125	else
126	{
127	/* a damned pair */
128	RTUNICP uc1;
129	RTUNICP uc2;
130	if (wc1 >= 0xdc00)
131	{
132	if (pwsz1Start == pwsz1)
133	return iDiff;
134	uc1 = pwsz1[-1];
135	if (uc1 < 0xd800 \|\| uc1 >= 0xdc00)
136	return iDiff;
137	uc1 = 0x10000 + (((uc1 & 0x3ff) << 10) \| (wc1 & 0x3ff));
138	uc2 = 0x10000 + (((pwsz2[-1] & 0x3ff) << 10) \| (wc2 & 0x3ff));
139	}
140	else
141	{
142	uc1 = *++pwsz1;
143	if (uc1 < 0xdc00 \|\| uc1 >= 0xe000)
144	return iDiff;
145	uc1 = 0x10000 + (((wc1 & 0x3ff) << 10) \| (uc1 & 0x3ff));
146	uc2 = 0x10000 + (((wc2 & 0x3ff) << 10) \| (*++pwsz2 & 0x3ff));
147	}
148	iDiff = RTUniCpToUpper(uc1) - RTUniCpToUpper(uc2);
149	if (iDiff)
150	iDiff = RTUniCpToLower(uc1) - RTUniCpToLower(uc2); /* serious paranoia! */
151	}
152	if (iDiff)
153	return iDiff;
154	}
155	if (!wc1)
156	return 0;
157	pwsz1++;
158	pwsz2++;
159	}
160	}
161
162
163	RTDECL(PRTUTF16) RTUtf16ToLower(PRTUTF16 pwsz)
164	{
165	PRTUTF16 pwc = pwsz;
166	for (;;)
167	{
168	RTUTF16 wc = *pwc;
169	if (!wc)
170	break;
171	if (wc < 0xd800 \|\| wc >= 0xdc00)
172	{
173	RTUNICP ucFolded = RTUniCpToLower(wc);
174	if (ucFolded < 0x10000)
175	*pwc++ = RTUniCpToLower(wc);
176	}
177	else
178	{
179	/* surrogate */
180	RTUTF16 wc2 = pwc[1];
181	if (wc2 >= 0xdc00 && wc2 <= 0xdfff)
182	{
183	RTUNICP uc = 0x10000 + (((wc & 0x3ff) << 10) \| (wc2 & 0x3ff));
184	RTUNICP ucFolded = RTUniCpToLower(uc);
185	if (uc != ucFolded && ucFolded >= 0x10000) /* we don't support shrinking the string */
186	{
187	uc -= 0x10000;
188	*pwc++ = 0xd800 \| (uc >> 10);
189	*pwc++ = 0xdc00 \| (uc & 0x3ff);
190	}
191	}
192	else /* invalid encoding. */
193	pwc++;
194	}
195	}
196	return pwsz;
197	}
198
199
200	RTDECL(PRTUTF16) RTUtf16ToUpper(PRTUTF16 pwsz)
201	{
202	PRTUTF16 pwc = pwsz;
203	for (;;)
204	{
205	RTUTF16 wc = *pwc;
206	if (!wc)
207	break;
208	if (wc < 0xd800 \|\| wc >= 0xdc00)
209	*pwc++ = RTUniCpToUpper(wc);
210	else
211	{
212	/* surrogate */
213	RTUTF16 wc2 = pwc[1];
214	if (wc2 >= 0xdc00 && wc2 <= 0xdfff)
215	{
216	RTUNICP uc = 0x10000 + (((wc & 0x3ff) << 10) \| (wc2 & 0x3ff));
217	RTUNICP ucFolded = RTUniCpToUpper(uc);
218	if (uc != ucFolded && ucFolded >= 0x10000) /* we don't support shrinking the string */
219	{
220	uc -= 0x10000;
221	*pwc++ = 0xd800 \| (uc >> 10);
222	*pwc++ = 0xdc00 \| (uc & 0x3ff);
223	}
224	}
225	else /* invalid encoding. */
226	pwc++;
227	}
228	}
229	return pwsz;
230	}
231
232
233	/**
234	* Validate the UTF-16 encoding and calculates the length of an UTF-8 encoding.
235	*
236	* @returns iprt status code.
237	* @param pwsz The UTF-16 string.
238	* @param cwc The max length of the UTF-16 string to consider.
239	* @param pcch Where to store the length (excluding '\\0') of the UTF-8 string. (cch == cb, btw)
240	*/
241	static int rtUtf16CalcUtf8Length(PCRTUTF16 pwsz, size_t cwc, size_t *pcch)
242	{
243	int rc = VINF_SUCCESS;
244	size_t cch = 0;
245	while (cwc > 0)
246	{
247	RTUTF16 wc = *pwsz++; cwc--;
248	if (!wc)
249	break;
250	else if (wc < 0xd800 \|\| wc > 0xdfff)
251	{
252	if (wc < 0x80)
253	cch++;
254	else if (wc < 0x800)
255	cch += 2;
256	else if (wc < 0xfffe)
257	cch += 3;
258	else
259	{
260	RTStrAssertMsgFailed(("endian indicator! wc=%#x\n", wc));
261	rc = VERR_CODE_POINT_ENDIAN_INDICATOR;
262	break;
263	}
264	}
265	else
266	{
267	if (wc >= 0xdc00)
268	{
269	RTStrAssertMsgFailed(("Wrong 1st char in surrogate! wc=%#x\n", wc));
270	rc = VERR_INVALID_UTF16_ENCODING;
271	break;
272	}
273	if (cwc <= 0)
274	{
275	RTStrAssertMsgFailed(("Invalid length! wc=%#x\n", wc));
276	rc = VERR_INVALID_UTF16_ENCODING;
277	break;
278	}
279	wc = *pwsz++; cwc--;
280	if (wc < 0xdc00 \|\| wc > 0xdfff)
281	{
282	RTStrAssertMsgFailed(("Wrong 2nd char in surrogate! wc=%#x\n", wc));
283	rc = VERR_INVALID_UTF16_ENCODING;
284	break;
285	}
286	cch += 4;
287	}
288	}
289
290
291	/* done */
292	*pcch = cch;
293	return rc;
294	}
295
296
297	/**
298	* Recodes an valid UTF-16 string as UTF-8.
299	*
300	* @returns iprt status code.
301	* @param pwsz The UTF-16 string.
302	* @param cwc The number of RTUTF16 characters to process from pwsz. The recoding
303	* will stop when cwc or '\\0' is reached.
304	* @param psz Where to store the UTF-8 string.
305	* @param cch The size of the UTF-8 buffer, excluding the terminator.
306	* @param pcch Where to store the number of octets actually encoded.
307	*/
308	static int rtUtf16RecodeAsUtf8(PCRTUTF16 pwsz, size_t cwc, char psz, size_t cch, size_t pcch)
309	{
310	unsigned char pwch = (unsigned char )psz;
311	int rc = VINF_SUCCESS;
312	while (cwc > 0)
313	{
314	RTUTF16 wc = *pwsz++; cwc--;
315	if (!wc)
316	break;
317	else if (wc < 0xd800 \|\| wc > 0xdfff)
318	{
319	if (wc < 0x80)
320	{
321	if (cch < 1)
322	{
323	RTStrAssertMsgFailed(("Buffer overflow! 1\n"));
324	rc = VERR_BUFFER_OVERFLOW;
325	break;
326	}
327	cch--;
328	*pwch++ = (unsigned char)wc;
329	}
330	else if (wc < 0x800)
331	{
332	if (cch < 2)
333	{
334	RTStrAssertMsgFailed(("Buffer overflow! 2\n"));
335	rc = VERR_BUFFER_OVERFLOW;
336	break;
337	}
338	cch -= 2;
339	*pwch++ = 0xc0 \| (wc >> 6);
340	*pwch++ = 0x80 \| (wc & 0x3f);
341	}
342	else if (wc < 0xfffe)
343	{
344	if (cch < 3)
345	{
346	RTStrAssertMsgFailed(("Buffer overflow! 3\n"));
347	rc = VERR_BUFFER_OVERFLOW;
348	break;
349	}
350	cch -= 3;
351	*pwch++ = 0xe0 \| (wc >> 12);
352	*pwch++ = 0x80 \| ((wc >> 6) & 0x3f);
353	*pwch++ = 0x80 \| (wc & 0x3f);
354	}
355	else
356	{
357	RTStrAssertMsgFailed(("endian indicator! wc=%#x\n", wc));
358	rc = VERR_CODE_POINT_ENDIAN_INDICATOR;
359	break;
360	}
361	}
362	else
363	{
364	if (wc >= 0xdc00)
365	{
366	RTStrAssertMsgFailed(("Wrong 1st char in surrogate! wc=%#x\n", wc));
367	rc = VERR_INVALID_UTF16_ENCODING;
368	break;
369	}
370	if (cwc <= 0)
371	{
372	RTStrAssertMsgFailed(("Invalid length! wc=%#x\n", wc));
373	rc = VERR_INVALID_UTF16_ENCODING;
374	break;
375	}
376	RTUTF16 wc2 = *pwsz++; cwc--;
377	if (wc2 < 0xdc00 \|\| wc2 > 0xdfff)
378	{
379	RTStrAssertMsgFailed(("Wrong 2nd char in surrogate! wc=%#x\n", wc));
380	rc = VERR_INVALID_UTF16_ENCODING;
381	break;
382	}
383	uint32_t CodePoint = 0x10000
384	+ ( ((wc & 0x3ff) << 10)
385	\| (wc2 & 0x3ff));
386	if (cch < 4)
387	{
388	RTStrAssertMsgFailed(("Buffer overflow! 4\n"));
389	rc = VERR_BUFFER_OVERFLOW;
390	break;
391	}
392	cch -= 4;
393	*pwch++ = 0xf0 \| (CodePoint >> 18);
394	*pwch++ = 0x80 \| ((CodePoint >> 12) & 0x3f);
395	*pwch++ = 0x80 \| ((CodePoint >> 6) & 0x3f);
396	*pwch++ = 0x80 \| (CodePoint & 0x3f);
397	}
398	}
399
400	/* done */
401	*pwch = '\0';
402	pcch = (char )pwch - psz;
403	return rc;
404	}
405
406
407
408	RTDECL(int) RTUtf16ToUtf8(PCRTUTF16 pwszString, char **ppszString)
409	{
410	/*
411	* Validate input.
412	*/
413	Assert(VALID_PTR(ppszString));
414	Assert(VALID_PTR(pwszString));
415	*ppszString = NULL;
416
417	/*
418	* Validate the UTF-16 string and calculate the length of the UTF-8 encoding of it.
419	*/
420	size_t cch;
421	int rc = rtUtf16CalcUtf8Length(pwszString, RTSTR_MAX, &cch);
422	if (RT_SUCCESS(rc))
423	{
424	/*
425	* Allocate buffer and recode it.
426	*/
427	char pszResult = (char )RTMemAlloc(cch + 1);
428	if (pszResult)
429	{
430	rc = rtUtf16RecodeAsUtf8(pwszString, RTSTR_MAX, pszResult, cch, &cch);
431	if (RT_SUCCESS(rc))
432	{
433	*ppszString = pszResult;
434	return rc;
435	}
436
437	RTMemFree(pszResult);
438	}
439	else
440	rc = VERR_NO_STR_MEMORY;
441	}
442	return rc;
443	}
444
445
446	RTDECL(int) RTUtf16ToUtf8Ex(PCRTUTF16 pwszString, size_t cwcString, char *ppsz, size_t cch, size_t pcch)
447	{
448	/*
449	* Validate input.
450	*/
451	Assert(VALID_PTR(pwszString));
452	Assert(VALID_PTR(ppsz));
453	Assert(!pcch \|\| VALID_PTR(pcch));
454
455	/*
456	* Validate the UTF-16 string and calculate the length of the UTF-8 encoding of it.
457	*/
458	size_t cchResult;
459	int rc = rtUtf16CalcUtf8Length(pwszString, cwcString, &cchResult);
460	if (RT_SUCCESS(rc))
461	{
462	if (pcch)
463	*pcch = cchResult;
464
465	/*
466	* Check buffer size / Allocate buffer and recode it.
467	*/
468	bool fShouldFree;
469	char *pszResult;
470	if (cch > 0 && *ppsz)
471	{
472	fShouldFree = false;
473	if (cch <= cchResult)
474	return VERR_BUFFER_OVERFLOW;
475	pszResult = *ppsz;
476	}
477	else
478	{
479	*ppsz = NULL;
480	fShouldFree = true;
481	cch = RT_MAX(cch, cchResult + 1);
482	pszResult = (char *)RTMemAlloc(cch);
483	}
484	if (pszResult)
485	{
486	rc = rtUtf16RecodeAsUtf8(pwszString, cwcString, pszResult, cch - 1, &cch);
487	if (RT_SUCCESS(rc))
488	{
489	*ppsz = pszResult;
490	return rc;
491	}
492
493	if (fShouldFree)
494	RTMemFree(pszResult);
495	}
496	else
497	rc = VERR_NO_STR_MEMORY;
498	}
499	return rc;
500	}
501
502
503	RTDECL(RTUNICP) RTUtf16GetCpInternal(PCRTUTF16 pwsz)
504	{
505	const RTUTF16 wc = *pwsz;
506
507	/* simple */
508	if (wc < 0xd800 \|\| (wc > 0xdfff && wc < 0xfffe))
509	return wc;
510	if (wc < 0xfffe)
511	{
512	/* surrogate pair */
513	if (wc < 0xdc00)
514	{
515	const RTUTF16 wc2 = pwsz[1];
516	if (wc2 >= 0xdc00 && wc2 <= 0xdfff)
517	{
518	RTUNICP uc = 0x10000 + (((wc & 0x3ff) << 10) \| (wc2 & 0x3ff));
519	return uc;
520	}
521
522	RTStrAssertMsgFailed(("wc=%#08x wc2=%#08x - invalid 2nd char in surrogate pair\n", wc, wc2));
523	}
524	else
525	RTStrAssertMsgFailed(("wc=%#08x - invalid surrogate pair order\n", wc));
526	}
527	else
528	RTStrAssertMsgFailed(("wc=%#08x - endian indicator\n", wc));
529	return RTUNICP_INVALID;
530	}
531
532
533	RTDECL(int) RTUtf16GetCpExInternal(PCRTUTF16 *ppwsz, PRTUNICP pCp)
534	{
535	const RTUTF16 wc = **ppwsz;
536
537	/* simple */
538	if (wc < 0xd800 \|\| (wc > 0xdfff && wc < 0xfffe))
539	{
540	(*ppwsz)++;
541	*pCp = wc;
542	return VINF_SUCCESS;
543	}
544
545	int rc;
546	if (wc < 0xfffe)
547	{
548	/* surrogate pair */
549	if (wc < 0xdc00)
550	{
551	const RTUTF16 wc2 = (*ppwsz)[1];
552	if (wc2 >= 0xdc00 && wc2 <= 0xdfff)
553	{
554	RTUNICP uc = 0x10000 + (((wc & 0x3ff) << 10) \| (wc2 & 0x3ff));
555	*pCp = uc;
556	(*ppwsz) += 2;
557	return VINF_SUCCESS;
558	}
559
560	RTStrAssertMsgFailed(("wc=%#08x wc2=%#08x - invalid 2nd char in surrogate pair\n", wc, wc2));
561	}
562	else
563	RTStrAssertMsgFailed(("wc=%#08x - invalid surrogate pair order\n", wc));
564	rc = VERR_INVALID_UTF16_ENCODING;
565	}
566	else
567	{
568	RTStrAssertMsgFailed(("wc=%#08x - endian indicator\n", wc));
569	rc = VERR_CODE_POINT_ENDIAN_INDICATOR;
570	}
571	*pCp = RTUNICP_INVALID;
572	(*ppwsz)++;
573	return rc;
574	}
575
576
577	RTDECL(PRTUTF16) RTUtf16PutCpInternal(PRTUTF16 pwsz, RTUNICP CodePoint)
578	{
579	/* simple */
580	if ( CodePoint < 0xd800
581	\|\| ( CodePoint > 0xdfff
582	&& CodePoint < 0xfffe))
583	{
584	*pwsz++ = (RTUTF16)CodePoint;
585	return pwsz;
586	}
587
588	/* surrogate pair */
589	if (CodePoint >= 0x10000 && CodePoint <= 0x0010ffff)
590	{
591	CodePoint -= 0x10000;
592	*pwsz++ = 0xd800 \| (CodePoint >> 10);
593	*pwsz++ = 0xdc00 \| (CodePoint & 0x3ff);
594	return pwsz;
595	}
596
597	/* invalid code point. */
598	RTStrAssertMsgFailed(("Invalid codepoint %#x\n", CodePoint));
599	*pwsz++ = 0x7f;
600	return pwsz;
601	}
602
603

Note: See TracBrowser for help on using the repository browser.

source: vbox/trunk/src/VBox/Runtime/utf-16.cpp@ 4071

Download in other formats: