1 | /* $Id: string.cpp 67646 2017-06-27 16:37:41Z vboxsync $ */
|
---|
2 | /** @file
|
---|
3 | * MS COM / XPCOM Abstraction Layer - UTF-8 and UTF-16 string classes.
|
---|
4 | */
|
---|
5 |
|
---|
6 | /*
|
---|
7 | * Copyright (C) 2006-2016 Oracle Corporation
|
---|
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 (GPL) as published by the Free Software
|
---|
13 | * Foundation, in version 2 as it comes in the "COPYING" file of the
|
---|
14 | * VirtualBox OSE distribution. VirtualBox OSE is distributed in the
|
---|
15 | * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
|
---|
16 | */
|
---|
17 |
|
---|
18 | #include "VBox/com/string.h"
|
---|
19 |
|
---|
20 | #include <iprt/err.h>
|
---|
21 | #include <iprt/path.h>
|
---|
22 | #include <iprt/log.h>
|
---|
23 | #include <iprt/string.h>
|
---|
24 | #include <iprt/uni.h>
|
---|
25 |
|
---|
26 | namespace com
|
---|
27 | {
|
---|
28 |
|
---|
29 | // BSTR representing a null wide char with 32 bits of length prefix (0);
|
---|
30 | // this will work on Windows as well as other platforms where BSTR does
|
---|
31 | // not use length prefixes
|
---|
32 | const OLECHAR g_achEmptyBstr[3] = { 0, 0, 0 };
|
---|
33 | const BSTR g_bstrEmpty = (BSTR)&g_achEmptyBstr[2];
|
---|
34 |
|
---|
35 | /* static */
|
---|
36 | const Bstr Bstr::Empty; /* default ctor is OK */
|
---|
37 |
|
---|
38 | void Bstr::copyFromN(const char *a_pszSrc, size_t a_cchMax)
|
---|
39 | {
|
---|
40 | /*
|
---|
41 | * Initialize m_bstr first in case of throws further down in the code, then
|
---|
42 | * check for empty input (m_bstr == NULL means empty, there are no NULL
|
---|
43 | * strings).
|
---|
44 | */
|
---|
45 | m_bstr = NULL;
|
---|
46 | if (!a_cchMax || !a_pszSrc || !*a_pszSrc)
|
---|
47 | return;
|
---|
48 |
|
---|
49 | /*
|
---|
50 | * Calculate the length and allocate a BSTR string buffer of the right
|
---|
51 | * size, i.e. optimize heap usage.
|
---|
52 | */
|
---|
53 | size_t cwc;
|
---|
54 | int vrc = ::RTStrCalcUtf16LenEx(a_pszSrc, a_cchMax, &cwc);
|
---|
55 | if (RT_SUCCESS(vrc))
|
---|
56 | {
|
---|
57 | m_bstr = ::SysAllocStringByteLen(NULL, (unsigned)(cwc * sizeof(OLECHAR)));
|
---|
58 | if (RT_LIKELY(m_bstr))
|
---|
59 | {
|
---|
60 | PRTUTF16 pwsz = (PRTUTF16)m_bstr;
|
---|
61 | vrc = ::RTStrToUtf16Ex(a_pszSrc, a_cchMax, &pwsz, cwc + 1, NULL);
|
---|
62 | if (RT_SUCCESS(vrc))
|
---|
63 | return;
|
---|
64 |
|
---|
65 | /* This should not happen! */
|
---|
66 | AssertRC(vrc);
|
---|
67 | cleanup();
|
---|
68 | }
|
---|
69 | }
|
---|
70 | else /* ASSUME: input is valid Utf-8. Fake out of memory error. */
|
---|
71 | AssertLogRelMsgFailed(("%Rrc %.*Rhxs\n", vrc, RTStrNLen(a_pszSrc, a_cchMax), a_pszSrc));
|
---|
72 | throw std::bad_alloc();
|
---|
73 | }
|
---|
74 |
|
---|
75 | int Bstr::compareUtf8(const char *a_pszRight, CaseSensitivity a_enmCase /*= CaseSensitive*/) const
|
---|
76 | {
|
---|
77 | PCRTUTF16 pwszLeft = m_bstr;
|
---|
78 |
|
---|
79 | /*
|
---|
80 | * Special case for null/empty strings. Unlike RTUtf16Cmp we
|
---|
81 | * treat null and empty equally.
|
---|
82 | */
|
---|
83 | if (!pwszLeft)
|
---|
84 | return !a_pszRight || *a_pszRight == '\0' ? 0 : -1;
|
---|
85 | if (!a_pszRight)
|
---|
86 | return *pwszLeft == '\0' ? 0 : 1;
|
---|
87 |
|
---|
88 | /*
|
---|
89 | * Compare with a UTF-8 string by enumerating them char by char.
|
---|
90 | */
|
---|
91 | for (;;)
|
---|
92 | {
|
---|
93 | RTUNICP ucLeft;
|
---|
94 | int rc = RTUtf16GetCpEx(&pwszLeft, &ucLeft);
|
---|
95 | AssertRCReturn(rc, 1);
|
---|
96 |
|
---|
97 | RTUNICP ucRight;
|
---|
98 | rc = RTStrGetCpEx(&a_pszRight, &ucRight);
|
---|
99 | AssertRCReturn(rc, -1);
|
---|
100 | if (ucLeft == ucRight)
|
---|
101 | {
|
---|
102 | if (ucLeft)
|
---|
103 | continue;
|
---|
104 | return 0;
|
---|
105 | }
|
---|
106 |
|
---|
107 | if (a_enmCase == CaseInsensitive)
|
---|
108 | {
|
---|
109 | if (RTUniCpToUpper(ucLeft) == RTUniCpToUpper(ucRight))
|
---|
110 | continue;
|
---|
111 | if (RTUniCpToLower(ucLeft) == RTUniCpToLower(ucRight))
|
---|
112 | continue;
|
---|
113 | }
|
---|
114 |
|
---|
115 | return ucLeft < ucRight ? -1 : 1;
|
---|
116 | }
|
---|
117 | }
|
---|
118 |
|
---|
119 |
|
---|
120 | /* static */
|
---|
121 | const Utf8Str Utf8Str::Empty; /* default ctor is OK */
|
---|
122 |
|
---|
123 | #if defined(VBOX_WITH_XPCOM)
|
---|
124 | void Utf8Str::cloneTo(char **pstr) const
|
---|
125 | {
|
---|
126 | size_t cb = length() + 1;
|
---|
127 | *pstr = (char *)nsMemory::Alloc(cb);
|
---|
128 | if (RT_LIKELY(*pstr))
|
---|
129 | memcpy(*pstr, c_str(), cb);
|
---|
130 | else
|
---|
131 | throw std::bad_alloc();
|
---|
132 | }
|
---|
133 |
|
---|
134 | HRESULT Utf8Str::cloneToEx(char **pstr) const
|
---|
135 | {
|
---|
136 | size_t cb = length() + 1;
|
---|
137 | *pstr = (char *)nsMemory::Alloc(cb);
|
---|
138 | if (RT_LIKELY(*pstr))
|
---|
139 | {
|
---|
140 | memcpy(*pstr, c_str(), cb);
|
---|
141 | return S_OK;
|
---|
142 | }
|
---|
143 | return E_OUTOFMEMORY;
|
---|
144 | }
|
---|
145 | #endif
|
---|
146 |
|
---|
147 | Utf8Str& Utf8Str::stripTrailingSlash()
|
---|
148 | {
|
---|
149 | if (length())
|
---|
150 | {
|
---|
151 | ::RTPathStripTrailingSlash(m_psz);
|
---|
152 | jolt();
|
---|
153 | }
|
---|
154 | return *this;
|
---|
155 | }
|
---|
156 |
|
---|
157 | Utf8Str& Utf8Str::stripFilename()
|
---|
158 | {
|
---|
159 | if (length())
|
---|
160 | {
|
---|
161 | RTPathStripFilename(m_psz);
|
---|
162 | jolt();
|
---|
163 | }
|
---|
164 | return *this;
|
---|
165 | }
|
---|
166 |
|
---|
167 | Utf8Str& Utf8Str::stripPath()
|
---|
168 | {
|
---|
169 | if (length())
|
---|
170 | {
|
---|
171 | char *pszName = ::RTPathFilename(m_psz);
|
---|
172 | if (pszName)
|
---|
173 | {
|
---|
174 | size_t cchName = length() - (pszName - m_psz);
|
---|
175 | memmove(m_psz, pszName, cchName + 1);
|
---|
176 | jolt();
|
---|
177 | }
|
---|
178 | else
|
---|
179 | cleanup();
|
---|
180 | }
|
---|
181 | return *this;
|
---|
182 | }
|
---|
183 |
|
---|
184 | Utf8Str& Utf8Str::stripSuffix()
|
---|
185 | {
|
---|
186 | if (length())
|
---|
187 | {
|
---|
188 | RTPathStripSuffix(m_psz);
|
---|
189 | jolt();
|
---|
190 | }
|
---|
191 | return *this;
|
---|
192 | }
|
---|
193 |
|
---|
194 | size_t Utf8Str::parseKeyValue(Utf8Str &a_rKey, Utf8Str &a_rValue, size_t a_offStart /* = 0*/,
|
---|
195 | const Utf8Str &a_rPairSeparator /*= ","*/, const Utf8Str &a_rKeyValueSeparator /*= "="*/) const
|
---|
196 | {
|
---|
197 | /* Find the end of the next pair, skipping empty pairs.
|
---|
198 | Note! The skipping allows us to pass the return value of a parseKeyValue()
|
---|
199 | call as offStart to the next call. */
|
---|
200 | size_t offEnd;
|
---|
201 | while ( a_offStart == (offEnd = find(&a_rPairSeparator, a_offStart))
|
---|
202 | && offEnd != npos)
|
---|
203 | a_offStart++;
|
---|
204 |
|
---|
205 | /* Look for a key/value separator before the end of the pair.
|
---|
206 | ASSUMES npos value returned by find when the substring is not found is
|
---|
207 | really high. */
|
---|
208 | size_t offKeyValueSep = find(&a_rKeyValueSeparator, a_offStart);
|
---|
209 | if (offKeyValueSep < offEnd)
|
---|
210 | {
|
---|
211 | a_rKey = substr(a_offStart, offKeyValueSep - a_offStart);
|
---|
212 | if (offEnd == npos)
|
---|
213 | offEnd = m_cch; /* No confusing npos when returning strings. */
|
---|
214 | a_rValue = substr(offKeyValueSep + 1, offEnd - offKeyValueSep - 1);
|
---|
215 | }
|
---|
216 | else
|
---|
217 | {
|
---|
218 | a_rKey.setNull();
|
---|
219 | a_rValue.setNull();
|
---|
220 | }
|
---|
221 |
|
---|
222 | return offEnd;
|
---|
223 | }
|
---|
224 |
|
---|
225 | /**
|
---|
226 | * Internal function used in Utf8Str copy constructors and assignment when
|
---|
227 | * copying from a UTF-16 string.
|
---|
228 | *
|
---|
229 | * As with the RTCString::copyFrom() variants, this unconditionally sets the
|
---|
230 | * members to a copy of the given other strings and makes no assumptions about
|
---|
231 | * previous contents. This can therefore be used both in copy constructors,
|
---|
232 | * when member variables have no defined value, and in assignments after having
|
---|
233 | * called cleanup().
|
---|
234 | *
|
---|
235 | * This variant converts from a UTF-16 string, most probably from
|
---|
236 | * a Bstr assignment.
|
---|
237 | *
|
---|
238 | * @param a_pbstr The source string. The caller guarantees that this
|
---|
239 | * is valid UTF-16.
|
---|
240 | * @param a_cwcMax The number of characters to be copied. If set to RTSTR_MAX,
|
---|
241 | * the entire string will be copied.
|
---|
242 | *
|
---|
243 | * @sa RTCString::copyFromN
|
---|
244 | */
|
---|
245 | void Utf8Str::copyFrom(CBSTR a_pbstr, size_t a_cwcMax)
|
---|
246 | {
|
---|
247 | if (a_pbstr && *a_pbstr)
|
---|
248 | {
|
---|
249 | int vrc = RTUtf16ToUtf8Ex((PCRTUTF16)a_pbstr,
|
---|
250 | a_cwcMax, // size_t cwcString: translate entire string
|
---|
251 | &m_psz, // char **ppsz: output buffer
|
---|
252 | 0, // size_t cch: if 0, func allocates buffer in *ppsz
|
---|
253 | &m_cch); // size_t *pcch: receives the size of the output string, excluding the terminator.
|
---|
254 | if (RT_SUCCESS(vrc))
|
---|
255 | m_cbAllocated = m_cch + 1;
|
---|
256 | else
|
---|
257 | {
|
---|
258 | if ( vrc != VERR_NO_STR_MEMORY
|
---|
259 | && vrc != VERR_NO_MEMORY)
|
---|
260 | {
|
---|
261 | /* ASSUME: input is valid Utf-16. Fake out of memory error. */
|
---|
262 | AssertLogRelMsgFailed(("%Rrc %.*Rhxs\n", vrc, RTUtf16Len(a_pbstr) * sizeof(RTUTF16), a_pbstr));
|
---|
263 | }
|
---|
264 |
|
---|
265 | m_cch = 0;
|
---|
266 | m_cbAllocated = 0;
|
---|
267 | m_psz = NULL;
|
---|
268 |
|
---|
269 | throw std::bad_alloc();
|
---|
270 | }
|
---|
271 | }
|
---|
272 | else
|
---|
273 | {
|
---|
274 | m_cch = 0;
|
---|
275 | m_cbAllocated = 0;
|
---|
276 | m_psz = NULL;
|
---|
277 | }
|
---|
278 | }
|
---|
279 |
|
---|
280 | /**
|
---|
281 | * A variant of Utf8Str::copyFrom that does not throw any exceptions but returns
|
---|
282 | * E_OUTOFMEMORY instead.
|
---|
283 | *
|
---|
284 | * @param a_pbstr The source string.
|
---|
285 | * @returns S_OK or E_OUTOFMEMORY.
|
---|
286 | */
|
---|
287 | HRESULT Utf8Str::copyFromEx(CBSTR a_pbstr)
|
---|
288 | {
|
---|
289 | if (a_pbstr && *a_pbstr)
|
---|
290 | {
|
---|
291 | int vrc = RTUtf16ToUtf8Ex((PCRTUTF16)a_pbstr,
|
---|
292 | RTSTR_MAX, // size_t cwcString: translate entire string
|
---|
293 | &m_psz, // char **ppsz: output buffer
|
---|
294 | 0, // size_t cch: if 0, func allocates buffer in *ppsz
|
---|
295 | &m_cch); // size_t *pcch: receives the size of the output string, excluding the terminator.
|
---|
296 | if (RT_SUCCESS(vrc))
|
---|
297 | m_cbAllocated = m_cch + 1;
|
---|
298 | else
|
---|
299 | {
|
---|
300 | if ( vrc != VERR_NO_STR_MEMORY
|
---|
301 | && vrc != VERR_NO_MEMORY)
|
---|
302 | {
|
---|
303 | /* ASSUME: input is valid Utf-16. Fake out of memory error. */
|
---|
304 | AssertLogRelMsgFailed(("%Rrc %.*Rhxs\n", vrc, RTUtf16Len(a_pbstr) * sizeof(RTUTF16), a_pbstr));
|
---|
305 | }
|
---|
306 |
|
---|
307 | m_cch = 0;
|
---|
308 | m_cbAllocated = 0;
|
---|
309 | m_psz = NULL;
|
---|
310 |
|
---|
311 | return E_OUTOFMEMORY;
|
---|
312 | }
|
---|
313 | }
|
---|
314 | else
|
---|
315 | {
|
---|
316 | m_cch = 0;
|
---|
317 | m_cbAllocated = 0;
|
---|
318 | m_psz = NULL;
|
---|
319 | }
|
---|
320 | return S_OK;
|
---|
321 | }
|
---|
322 |
|
---|
323 |
|
---|
324 | /**
|
---|
325 | * A variant of Utf8Str::copyFromN that does not throw any exceptions but
|
---|
326 | * returns E_OUTOFMEMORY instead.
|
---|
327 | *
|
---|
328 | * @param a_pcszSrc The source string.
|
---|
329 | * @param a_offSrc Start offset to copy from.
|
---|
330 | * @param a_cchSrc The source string.
|
---|
331 | * @returns S_OK or E_OUTOFMEMORY.
|
---|
332 | *
|
---|
333 | * @remarks This calls cleanup() first, so the caller doesn't have to. (Saves
|
---|
334 | * code space.)
|
---|
335 | */
|
---|
336 | HRESULT Utf8Str::copyFromExNComRC(const char *a_pcszSrc, size_t a_offSrc, size_t a_cchSrc)
|
---|
337 | {
|
---|
338 | cleanup();
|
---|
339 | if (a_cchSrc)
|
---|
340 | {
|
---|
341 | m_psz = RTStrAlloc(a_cchSrc + 1);
|
---|
342 | if (RT_LIKELY(m_psz))
|
---|
343 | {
|
---|
344 | m_cch = a_cchSrc;
|
---|
345 | m_cbAllocated = a_cchSrc + 1;
|
---|
346 | memcpy(m_psz, a_pcszSrc + a_offSrc, a_cchSrc);
|
---|
347 | m_psz[a_cchSrc] = '\0';
|
---|
348 | }
|
---|
349 | else
|
---|
350 | {
|
---|
351 | m_cch = 0;
|
---|
352 | m_cbAllocated = 0;
|
---|
353 | return E_OUTOFMEMORY;
|
---|
354 | }
|
---|
355 | }
|
---|
356 | else
|
---|
357 | {
|
---|
358 | m_cch = 0;
|
---|
359 | m_cbAllocated = 0;
|
---|
360 | m_psz = NULL;
|
---|
361 | }
|
---|
362 | return S_OK;
|
---|
363 | }
|
---|
364 |
|
---|
365 | } /* namespace com */
|
---|