/src/icu/icu4c/source/common/charstr.cpp

Source
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*   Copyright (C) 2010-2015, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*   file name:  charstr.cpp
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2010may19
*   created by: Markus W. Scherer
*/

#include <cstdlib>

#include "unicode/utypes.h"
#include "unicode/putil.h"
#include "charstr.h"
#include "cmemory.h"
#include "cstring.h"
#include "uinvchar.h"
#include "ustr_imp.h"

U_NAMESPACE_BEGIN

CharString::CharString(CharString&& src) noexcept
        : buffer(std::move(src.buffer)), len(src.len) {
    src.len = 0;  // not strictly necessary because we make no guarantees on the source string
}

CharString& CharString::operator=(CharString&& src) noexcept {
    buffer = std::move(src.buffer);
    len = src.len;
    src.len = 0;  // not strictly necessary because we make no guarantees on the source string
    return *this;
}

char *CharString::cloneData(UErrorCode &errorCode) const {
    if (U_FAILURE(errorCode)) { return nullptr; }
    char *p = static_cast<char *>(uprv_malloc(len + 1));
    if (p == nullptr) {
        errorCode = U_MEMORY_ALLOCATION_ERROR;
        return nullptr;
    }
    uprv_memcpy(p, buffer.getAlias(), len + 1);
    return p;
}

int32_t CharString::extract(char *dest, int32_t capacity, UErrorCode &errorCode) const {
    if (U_FAILURE(errorCode)) { return len; }
    if (capacity < 0 || (capacity > 0 && dest == nullptr)) {
        errorCode = U_ILLEGAL_ARGUMENT_ERROR;
        return len;
    }
    const char *src = buffer.getAlias();
    if (0 < len && len <= capacity && src != dest) {
        uprv_memcpy(dest, src, len);
    }
    return u_terminateChars(dest, capacity, len, &errorCode);
}

CharString &CharString::copyFrom(const CharString &s, UErrorCode &errorCode) {
    if(U_SUCCESS(errorCode) && this!=&s && ensureCapacity(s.len+1, 0, errorCode)) {
        len=s.len;
        uprv_memcpy(buffer.getAlias(), s.buffer.getAlias(), len+1);
    }
    return *this;
}

CharString &CharString::copyFrom(StringPiece s, UErrorCode &errorCode) {
    if (U_FAILURE(errorCode)) {
        return *this;
    }
    len = 0;
    append(s, errorCode);
    return *this;
}

int32_t CharString::lastIndexOf(char c) const {
    for(int32_t i=len; i>0;) {
        if(buffer[--i]==c) {
            return i;
        }
    }
    return -1;
}

bool CharString::contains(StringPiece s) const {
    if (s.empty()) { return false; }
    const char *p = buffer.getAlias();
    int32_t lastStart = len - s.length();
    for (int32_t i = 0; i <= lastStart; ++i) {
        if (uprv_memcmp(p + i, s.data(), s.length()) == 0) {
            return true;
        }
    }
    return false;
}

CharString &CharString::truncate(int32_t newLength) {
    if(newLength<0) {
        newLength=0;
    }
    if(newLength<len) {
        buffer[len=newLength]=0;
    }
    return *this;
}

CharString &CharString::append(char c, UErrorCode &errorCode) {
    if(ensureCapacity(len+2, 0, errorCode)) {
        buffer[len++]=c;
        buffer[len]=0;
    }
    return *this;
}

CharString &CharString::append(const char *s, int32_t sLength, UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) {
        return *this;
    }
    if(sLength<-1 || (s==nullptr && sLength!=0)) {
        errorCode=U_ILLEGAL_ARGUMENT_ERROR;
        return *this;
    }
    if(sLength<0) {
        sLength= static_cast<int32_t>(uprv_strlen(s));
    }
    if(sLength>0) {
        if(s==(buffer.getAlias()+len)) {
            // The caller wrote into the getAppendBuffer().
            if(sLength>=(buffer.getCapacity()-len)) {
                // The caller wrote too much.
                errorCode=U_INTERNAL_PROGRAM_ERROR;
            } else {
                buffer[len+=sLength]=0;
            }
        } else if(buffer.getAlias()<=s && s<(buffer.getAlias()+len) &&
                  sLength>=(buffer.getCapacity()-len)
        ) {
            // (Part of) this string is appended to itself which requires reallocation,
            // so we have to make a copy of the substring and append that.
            return append(CharString(s, sLength, errorCode), errorCode);
        } else if(ensureCapacity(len+sLength+1, 0, errorCode)) {
            uprv_memcpy(buffer.getAlias()+len, s, sLength);
            buffer[len+=sLength]=0;
        }
    }
    return *this;
}

CharString &CharString::appendNumber(int64_t number, UErrorCode &status) {
    if (number < 0) {
        this->append('-', status);
        if (U_FAILURE(status)) {
            return *this;
        }
    }

    if (number == 0) {
        this->append('0', status);
        return *this;
    }

    int32_t numLen = 0;
    while (number != 0) {
        int32_t residue = number % 10;
        number /= 10;
        this->append(std::abs(residue) + '0', status);
        numLen++;
        if (U_FAILURE(status)) {
            return *this;
        }
    }

    int32_t start = this->length() - numLen, end = this->length() - 1;
    while(start < end) {
        std::swap(this->data()[start++], this->data()[end--]);
    }

    return *this;
}

char *CharString::getAppendBuffer(int32_t minCapacity,
                                  int32_t desiredCapacityHint,
                                  int32_t &resultCapacity,
                                  UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) {
        resultCapacity=0;
        return nullptr;
    }
    int32_t appendCapacity=buffer.getCapacity()-len-1;  // -1 for NUL
    if(appendCapacity>=minCapacity) {
        resultCapacity=appendCapacity;
        return buffer.getAlias()+len;
    }
    if(ensureCapacity(len+minCapacity+1, len+desiredCapacityHint+1, errorCode)) {
        resultCapacity=buffer.getCapacity()-len-1;
        return buffer.getAlias()+len;
    }
    resultCapacity=0;
    return nullptr;
}

CharString &CharString::appendInvariantChars(const UnicodeString &s, UErrorCode &errorCode) {
    return appendInvariantChars(s.getBuffer(), s.length(), errorCode);
}

CharString &CharString::appendInvariantChars(const char16_t* uchars, int32_t ucharsLen, UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) {
        return *this;
    }
    if (!uprv_isInvariantUString(uchars, ucharsLen)) {
        errorCode = U_INVARIANT_CONVERSION_ERROR;
        return *this;
    }
    if(ensureCapacity(len+ucharsLen+1, 0, errorCode)) {
        u_UCharsToChars(uchars, buffer.getAlias()+len, ucharsLen);
        len += ucharsLen;
        buffer[len] = 0;
    }
    return *this;
}

UBool CharString::ensureCapacity(int32_t capacity,
                                 int32_t desiredCapacityHint,
                                 UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) {
        return false;
    }
    if(capacity>buffer.getCapacity()) {
        if(desiredCapacityHint==0) {
            desiredCapacityHint=capacity+buffer.getCapacity();
        }
        if( (desiredCapacityHint<=capacity || buffer.resize(desiredCapacityHint, len+1)==nullptr) &&
            buffer.resize(capacity, len+1)==nullptr
        ) {
            errorCode=U_MEMORY_ALLOCATION_ERROR;
            return false;
        }
    }
    return true;
}

CharString &CharString::appendPathPart(StringPiece s, UErrorCode &errorCode) {
    if(U_FAILURE(errorCode)) {
        return *this;
    }
    if(s.length()==0) {
        return *this;
    }
    char c;
    if(len>0 && (c=buffer[len-1])!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) {
        append(getDirSepChar(), errorCode);
    }
    append(s, errorCode);
    return *this;
}

CharString &CharString::ensureEndsWithFileSeparator(UErrorCode &errorCode) {
    char c;
    if(U_SUCCESS(errorCode) && len>0 &&
            (c=buffer[len-1])!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) {
        append(getDirSepChar(), errorCode);
    }
    return *this;
}

char CharString::getDirSepChar() const {
    char dirSepChar = U_FILE_SEP_CHAR;
#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR)
    // We may need to return a different directory separator when building for Cygwin or MSYS2.
    if(len>0 && !uprv_strchr(data(), U_FILE_SEP_CHAR) && uprv_strchr(data(), U_FILE_ALT_SEP_CHAR))
        dirSepChar = U_FILE_ALT_SEP_CHAR;
#endif
    return dirSepChar;
}

U_NAMESPACE_END

Coverage Report

Created: 2025-12-07 06:36

Line	Count	Source
1		// © 2016 and later: Unicode, Inc. and others.
2		// License & terms of use: http://www.unicode.org/copyright.html
3		/*
4		*******************************************************************************
5		* Copyright (C) 2010-2015, International Business Machines
6		* Corporation and others. All Rights Reserved.
7		*******************************************************************************
8		* file name: charstr.cpp
9		* encoding: UTF-8
10		* tab size: 8 (not used)
11		* indentation:4
12		*
13		* created on: 2010may19
14		* created by: Markus W. Scherer
15		*/
16
17		#include <cstdlib>
18
19		#include "unicode/utypes.h"
20		#include "unicode/putil.h"
21		#include "charstr.h"
22		#include "cmemory.h"
23		#include "cstring.h"
24		#include "uinvchar.h"
25		#include "ustr_imp.h"
26
27		U_NAMESPACE_BEGIN
28
29		CharString::CharString(CharString&& src) noexcept
30	1.87M	: buffer(std::move(src.buffer)), len(src.len) {
31	1.87M	src.len = 0; // not strictly necessary because we make no guarantees on the source string
32	1.87M	}
33
34	1.62M	CharString& CharString::operator=(CharString&& src) noexcept {
35	1.62M	buffer = std::move(src.buffer);
36	1.62M	len = src.len;
37	1.62M	src.len = 0; // not strictly necessary because we make no guarantees on the source string
38	1.62M	return *this;
39	1.62M	}
40
41	101k	char *CharString::cloneData(UErrorCode &errorCode) const {
42	101k	if (U_FAILURE(errorCode)) { return nullptr; }
43	101k	char p = static_cast<char >(uprv_malloc(len + 1));
44	101k	if (p == nullptr) {
45	0	errorCode = U_MEMORY_ALLOCATION_ERROR;
46	0	return nullptr;
47	0	}
48	101k	uprv_memcpy(p, buffer.getAlias(), len + 1);
49	101k	return p;
50	101k	}
51
52	49.4k	int32_t CharString::extract(char *dest, int32_t capacity, UErrorCode &errorCode) const {
53	49.4k	if (U_FAILURE(errorCode)) { return len; }
54	49.4k	if (capacity < 0 \|\| (capacity > 0 && dest == nullptr)) {
55	0	errorCode = U_ILLEGAL_ARGUMENT_ERROR;
56	0	return len;
57	0	}
58	49.4k	const char *src = buffer.getAlias();
59	49.4k	if (0 < len && len <= capacity && src != dest) {
60	49.4k	uprv_memcpy(dest, src, len);
61	49.4k	}
62	49.4k	return u_terminateChars(dest, capacity, len, &errorCode);
63	49.4k	}
64
65	2.15M	CharString &CharString::copyFrom(const CharString &s, UErrorCode &errorCode) {
66	2.15M	if(U_SUCCESS(errorCode) && this!=&s && ensureCapacity(s.len+1, 0, errorCode)) {
67	2.15M	len=s.len;
68	2.15M	uprv_memcpy(buffer.getAlias(), s.buffer.getAlias(), len+1);
69	2.15M	}
70	2.15M	return *this;
71	2.15M	}
72
73	166	CharString &CharString::copyFrom(StringPiece s, UErrorCode &errorCode) {
74	166	if (U_FAILURE(errorCode)) {
75	0	return *this;
76	0	}
77	166	len = 0;
78	166	append(s, errorCode);
79	166	return *this;
80	166	}
81
82	6	int32_t CharString::lastIndexOf(char c) const {
83	24	for(int32_t i=len; i>0;) {
84	24	if(buffer[--i]==c) {
85	6	return i;
86	6	}
87	24	}
88	0	return -1;
89	6	}
90
91	0	bool CharString::contains(StringPiece s) const {
92	0	if (s.empty()) { return false; }
93	0	const char *p = buffer.getAlias();
94	0	int32_t lastStart = len - s.length();
95	0	for (int32_t i = 0; i <= lastStart; ++i) {
96	0	if (uprv_memcmp(p + i, s.data(), s.length()) == 0) {
97	0	return true;
98	0	}
99	0	}
100	0	return false;
101	0	}
102
103	4.44k	CharString &CharString::truncate(int32_t newLength) {
104	4.44k	if(newLength<0) {
105	0	newLength=0;
106	0	}
107	4.44k	if(newLength<len) {
108	4.44k	buffer[len=newLength]=0;
109	4.44k	}
110	4.44k	return *this;
111	4.44k	}
112
113	87.6M	CharString &CharString::append(char c, UErrorCode &errorCode) {
114	87.6M	if(ensureCapacity(len+2, 0, errorCode)) {
115	87.5M	buffer[len++]=c;
116	87.5M	buffer[len]=0;
117	87.5M	}
118	87.6M	return *this;
119	87.6M	}
120
121	272M	CharString &CharString::append(const char *s, int32_t sLength, UErrorCode &errorCode) {
122	272M	if(U_FAILURE(errorCode)) {
123	228k	return *this;
124	228k	}
125	272M	if(sLength<-1 \|\| (s==nullptr && sLength!=0)) {
126	0	errorCode=U_ILLEGAL_ARGUMENT_ERROR;
127	0	return *this;
128	0	}
129	272M	if(sLength<0) {
130	1.24M	sLength= static_cast<int32_t>(uprv_strlen(s));
131	1.24M	}
132	272M	if(sLength>0) {
133	266M	if(s==(buffer.getAlias()+len)) {
134		// The caller wrote into the getAppendBuffer().
135	21.8M	if(sLength>=(buffer.getCapacity()-len)) {
136		// The caller wrote too much.
137	0	errorCode=U_INTERNAL_PROGRAM_ERROR;
138	21.8M	} else {
139	21.8M	buffer[len+=sLength]=0;
140	21.8M	}
141	244M	} else if(buffer.getAlias()<=s && s<(buffer.getAlias()+len) &&
142	0	sLength>=(buffer.getCapacity()-len)
143	244M	) {
144		// (Part of) this string is appended to itself which requires reallocation,
145		// so we have to make a copy of the substring and append that.
146	0	return append(CharString(s, sLength, errorCode), errorCode);
147	244M	} else if(ensureCapacity(len+sLength+1, 0, errorCode)) {
148	244M	uprv_memcpy(buffer.getAlias()+len, s, sLength);
149	244M	buffer[len+=sLength]=0;
150	244M	}
151	266M	}
152	272M	return *this;
153	272M	}
154
155	0	CharString &CharString::appendNumber(int64_t number, UErrorCode &status) {
156	0	if (number < 0) {
157	0	this->append('-', status);
158	0	if (U_FAILURE(status)) {
159	0	return *this;
160	0	}
161	0	}
162
163	0	if (number == 0) {
164	0	this->append('0', status);
165	0	return *this;
166	0	}
167
168	0	int32_t numLen = 0;
169	0	while (number != 0) {
170	0	int32_t residue = number % 10;
171	0	number /= 10;
172	0	this->append(std::abs(residue) + '0', status);
173	0	numLen++;
174	0	if (U_FAILURE(status)) {
175	0	return *this;
176	0	}
177	0	}
178
179	0	int32_t start = this->length() - numLen, end = this->length() - 1;
180	0	while(start < end) {
181	0	std::swap(this->data()[start++], this->data()[end--]);
182	0	}
183
184	0	return *this;
185	0	}
186
187		char *CharString::getAppendBuffer(int32_t minCapacity,
188		int32_t desiredCapacityHint,
189		int32_t &resultCapacity,
190	23.3M	UErrorCode &errorCode) {
191	23.3M	if(U_FAILURE(errorCode)) {
192	0	resultCapacity=0;
193	0	return nullptr;
194	0	}
195	23.3M	int32_t appendCapacity=buffer.getCapacity()-len-1; // -1 for NUL
196	23.3M	if(appendCapacity>=minCapacity) {
197	23.3M	resultCapacity=appendCapacity;
198	23.3M	return buffer.getAlias()+len;
199	23.3M	}
200	22.4k	if(ensureCapacity(len+minCapacity+1, len+desiredCapacityHint+1, errorCode)) {
201	22.4k	resultCapacity=buffer.getCapacity()-len-1;
202	22.4k	return buffer.getAlias()+len;
203	22.4k	}
204	0	resultCapacity=0;
205	0	return nullptr;
206	22.4k	}
207
208	630k	CharString &CharString::appendInvariantChars(const UnicodeString &s, UErrorCode &errorCode) {
209	630k	return appendInvariantChars(s.getBuffer(), s.length(), errorCode);
210	630k	}
211
212	2.66M	CharString &CharString::appendInvariantChars(const char16_t* uchars, int32_t ucharsLen, UErrorCode &errorCode) {
213	2.66M	if(U_FAILURE(errorCode)) {
214	0	return *this;
215	0	}
216	2.66M	if (!uprv_isInvariantUString(uchars, ucharsLen)) {
217	32	errorCode = U_INVARIANT_CONVERSION_ERROR;
218	32	return *this;
219	32	}
220	2.66M	if(ensureCapacity(len+ucharsLen+1, 0, errorCode)) {
221	2.66M	u_UCharsToChars(uchars, buffer.getAlias()+len, ucharsLen);
222	2.66M	len += ucharsLen;
223	2.66M	buffer[len] = 0;
224	2.66M	}
225	2.66M	return *this;
226	2.66M	}
227
228		UBool CharString::ensureCapacity(int32_t capacity,
229		int32_t desiredCapacityHint,
230	337M	UErrorCode &errorCode) {
231	337M	if(U_FAILURE(errorCode)) {
232	155k	return false;
233	155k	}
234	337M	if(capacity>buffer.getCapacity()) {
235	862k	if(desiredCapacityHint==0) {
236	840k	desiredCapacityHint=capacity+buffer.getCapacity();
237	840k	}
238	862k	if( (desiredCapacityHint<=capacity \|\| buffer.resize(desiredCapacityHint, len+1)==nullptr) &&
239	22.4k	buffer.resize(capacity, len+1)==nullptr
240	862k	) {
241	0	errorCode=U_MEMORY_ALLOCATION_ERROR;
242	0	return false;
243	0	}
244	862k	}
245	337M	return true;
246	337M	}
247
248	0	CharString &CharString::appendPathPart(StringPiece s, UErrorCode &errorCode) {
249	0	if(U_FAILURE(errorCode)) {
250	0	return *this;
251	0	}
252	0	if(s.length()==0) {
253	0	return *this;
254	0	}
255	0	char c;
256	0	if(len>0 && (c=buffer[len-1])!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) {
257	0	append(getDirSepChar(), errorCode);
258	0	}
259	0	append(s, errorCode);
260	0	return *this;
261	0	}
262
263	0	CharString &CharString::ensureEndsWithFileSeparator(UErrorCode &errorCode) {
264	0	char c;
265	0	if(U_SUCCESS(errorCode) && len>0 &&
266	0	(c=buffer[len-1])!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) {
267	0	append(getDirSepChar(), errorCode);
268	0	}
269	0	return *this;
270	0	}
271
272	0	char CharString::getDirSepChar() const {
273	0	char dirSepChar = U_FILE_SEP_CHAR;
274		#if (U_FILE_SEP_CHAR != U_FILE_ALT_SEP_CHAR)
275		// We may need to return a different directory separator when building for Cygwin or MSYS2.
276		if(len>0 && !uprv_strchr(data(), U_FILE_SEP_CHAR) && uprv_strchr(data(), U_FILE_ALT_SEP_CHAR))
277		dirSepChar = U_FILE_ALT_SEP_CHAR;
278		#endif
279	0	return dirSepChar;
280	0	}
281
282		U_NAMESPACE_END