1 /* |
|
2 * Copyright (c) 2007-2008 Nokia Corporation and/or its subsidiary(-ies). |
|
3 * All rights reserved. |
|
4 * This component and the accompanying materials are made available |
|
5 * under the terms of "Eclipse Public License v1.0" |
|
6 * which accompanies this distribution, and is available |
|
7 * at the URL "http://www.eclipse.org/legal/epl-v10.html". |
|
8 * |
|
9 * Initial Contributors: |
|
10 * Nokia Corporation - initial contribution. |
|
11 * |
|
12 * Contributors: |
|
13 * |
|
14 * Description: String class implementation |
|
15 * |
|
16 */ |
|
17 |
|
18 |
|
19 #include "ustringimpl.h" |
|
20 |
|
21 #include <gstring.h> |
|
22 #include <stdexcept> |
|
23 #if defined(__GNUC__) |
|
24 #include <string.h> |
|
25 #else |
|
26 #include <string.h> |
|
27 #endif |
|
28 #include <osn/osnnew.h> |
|
29 using namespace std; |
|
30 |
|
31 |
|
32 namespace osncore |
|
33 { |
|
34 |
|
35 const int KMaxUtf8Length = 6; |
|
36 |
|
37 |
|
38 // ------------------------------------------------------------------------- |
|
39 // Returns the byte count for aStr with aCharCount length |
|
40 // ------------------------------------------------------------------------- |
|
41 // |
|
42 static long bytesFromCharacters(const Utf8* aStr, long aCharCount) |
|
43 { |
|
44 int byteCount = 0; |
|
45 |
|
46 if (aStr) |
|
47 { |
|
48 const char* srcEndOffset = g_utf8_offset_to_pointer(aStr,aCharCount); |
|
49 byteCount = srcEndOffset - aStr; |
|
50 } |
|
51 |
|
52 return byteCount; |
|
53 } |
|
54 |
|
55 // ------------------------------------------------------------------------- |
|
56 // Initializes the string from stack or heap depending on its length |
|
57 // ------------------------------------------------------------------------- |
|
58 // |
|
59 void UStringImpl::initialize(const Utf8* aStr) |
|
60 { |
|
61 if (len < MAX_STR_LEN) |
|
62 { |
|
63 str = internalStr; |
|
64 memcpy(internalStr, aStr, len); |
|
65 } |
|
66 else |
|
67 { |
|
68 str = (gchar*)calloc(len+1, sizeof (gchar)); |
|
69 if(!str) |
|
70 { |
|
71 throw bad_alloc(); |
|
72 } |
|
73 memcpy(str,aStr,len); |
|
74 isInHeap = true; |
|
75 } |
|
76 |
|
77 str[len] = 0; |
|
78 allocated_len = len+1; |
|
79 } |
|
80 |
|
81 // ------------------------------------------------------------------------- |
|
82 // Allocates memory from heap if needed, throws if allocation fails |
|
83 // ------------------------------------------------------------------------- |
|
84 // |
|
85 void UStringImpl::checkBuffer(unsigned int aLength) |
|
86 { |
|
87 if (aLength < allocated_len) |
|
88 { |
|
89 return; |
|
90 } |
|
91 if (aLength >= MAX_STR_LEN) |
|
92 { |
|
93 if (!isInHeap) //str is in stack, move it to heap |
|
94 { |
|
95 gchar* temp = (gchar*)calloc(aLength + 1, sizeof(gchar)); |
|
96 if (!temp) |
|
97 { |
|
98 throw bad_alloc(); |
|
99 } |
|
100 if(str) |
|
101 { |
|
102 memcpy(temp, str, len); |
|
103 } |
|
104 str = temp; |
|
105 isInHeap = true; |
|
106 } |
|
107 else |
|
108 { |
|
109 gchar* tmp(0); |
|
110 tmp = (gchar*)realloc(str,aLength+1); |
|
111 if(!tmp) |
|
112 { |
|
113 throw bad_alloc(); |
|
114 } |
|
115 str = tmp; |
|
116 } |
|
117 } |
|
118 else |
|
119 { |
|
120 // if ustring is constructed with null or "" set internal buffer to |
|
121 // point to str |
|
122 if (!str) |
|
123 str = internalStr; |
|
124 } |
|
125 allocated_len = aLength+1; |
|
126 |
|
127 |
|
128 // reset allocated space |
|
129 memset(str+len, 0, allocated_len-len); |
|
130 } |
|
131 // ------------------------------------------------------------------------- |
|
132 // Implements the actual insertion functionality |
|
133 // ------------------------------------------------------------------------- |
|
134 // |
|
135 void UStringImpl::insertImpl(unsigned int aIndex, const char* aInsertChars, unsigned int aInsertLength) |
|
136 { |
|
137 // Ensure that we have enough space. |
|
138 checkBuffer(len + aInsertLength); |
|
139 // Move tail to make space for inserted characters. |
|
140 memmove(str+aIndex+aInsertLength, str+aIndex, len-aIndex); |
|
141 // Insert characters. |
|
142 memcpy(str+aIndex, aInsertChars, aInsertLength); |
|
143 // Update len |
|
144 len += aInsertLength; |
|
145 } |
|
146 // ------------------------------------------------------------------------- |
|
147 // Implements the actual append functionality |
|
148 // ------------------------------------------------------------------------- |
|
149 // |
|
150 void UStringImpl::appendImpl(const char* aStr, unsigned int aLength) |
|
151 { |
|
152 insertImpl(len, aStr, aLength); |
|
153 } |
|
154 // ------------------------------------------------------------------------- |
|
155 // Implements the actual erase functionality |
|
156 // ------------------------------------------------------------------------- |
|
157 // |
|
158 void UStringImpl::eraseImpl(unsigned int aIndex, unsigned int aLength) |
|
159 { |
|
160 if(aIndex > len) |
|
161 { |
|
162 throw out_of_range("out_of_range"); |
|
163 } |
|
164 |
|
165 if(aIndex+aLength >= len) |
|
166 { |
|
167 // erase from given index to end of the string |
|
168 if(aIndex<len) |
|
169 { |
|
170 len = aIndex; |
|
171 } |
|
172 memset(str+aIndex,0,len-aIndex); |
|
173 } |
|
174 else |
|
175 { |
|
176 // Move tail (remove begin or between) to given index. |
|
177 memmove(str+aIndex, str+aIndex+aLength, len-(aIndex+aLength)); |
|
178 len -= aLength; |
|
179 memset(str+len,0,allocated_len-len); |
|
180 } |
|
181 } |
|
182 |
|
183 // ------------------------------------------------------------------------- |
|
184 // Constructor |
|
185 // ------------------------------------------------------------------------- |
|
186 // |
|
187 UStringImpl::UStringImpl():str(0),len(0),allocated_len(0), isInHeap(0) |
|
188 { |
|
189 #ifdef _DEBUG |
|
190 g_setenv("G_SLICE","always-malloc",1); |
|
191 #endif |
|
192 } |
|
193 // ------------------------------------------------------------------------- |
|
194 // Constructs the string from aStr |
|
195 // ------------------------------------------------------------------------- |
|
196 // |
|
197 UStringImpl::UStringImpl(const Utf8* aStr) |
|
198 :str(0),len(0),allocated_len(0), isInHeap(0) |
|
199 { |
|
200 #ifdef _DEBUG |
|
201 g_setenv("G_SLICE","always-malloc",1); |
|
202 #endif |
|
203 //NULL byte not accepted |
|
204 if (aStr) |
|
205 { |
|
206 len = strlen(aStr); |
|
207 initialize(aStr); |
|
208 } |
|
209 } |
|
210 |
|
211 // ------------------------------------------------------------------------- |
|
212 // Constructs the string from aStr with aByteCount length |
|
213 // ------------------------------------------------------------------------- |
|
214 // |
|
215 UStringImpl::UStringImpl(const Utf8* aStr, int aByteCount) |
|
216 :str(0),len(0),allocated_len(0), isInHeap(0) |
|
217 { |
|
218 #ifdef _DEBUG |
|
219 g_setenv("G_SLICE","always-malloc",1); |
|
220 #endif |
|
221 //NULL byte not accepted |
|
222 if (aStr) |
|
223 { |
|
224 len = aByteCount; |
|
225 initialize(aStr); |
|
226 } |
|
227 } |
|
228 |
|
229 // ------------------------------------------------------------------------- |
|
230 // Constructs the string from aCodePoint |
|
231 // ------------------------------------------------------------------------- |
|
232 // |
|
233 UStringImpl::UStringImpl(Unicode aCodePoint) |
|
234 :str(0),len(0),allocated_len(0), isInHeap(0) |
|
235 { |
|
236 #ifdef _DEBUG |
|
237 g_setenv("G_SLICE","always-malloc",1); |
|
238 #endif |
|
239 |
|
240 char string[KMaxUtf8Length]; |
|
241 long usedBytes = g_unichar_to_utf8((unsigned int)aCodePoint,&string[0]); |
|
242 if(usedBytes > 0) |
|
243 { |
|
244 len = usedBytes; |
|
245 initialize(&string[0]); |
|
246 } |
|
247 } |
|
248 |
|
249 // ------------------------------------------------------------------------- |
|
250 // Copy constructor |
|
251 // ------------------------------------------------------------------------- |
|
252 // |
|
253 UStringImpl::UStringImpl(const UStringImpl& aUStringImpl) |
|
254 :str(0),len(aUStringImpl.len),allocated_len(0), isInHeap(0) |
|
255 { |
|
256 #ifdef _DEBUG |
|
257 g_setenv("G_SLICE","always-malloc",1); |
|
258 #endif |
|
259 if(aUStringImpl.str) |
|
260 { |
|
261 initialize(aUStringImpl.str); |
|
262 } |
|
263 } |
|
264 |
|
265 |
|
266 // ------------------------------------------------------------------------- |
|
267 // Destructor |
|
268 // ------------------------------------------------------------------------- |
|
269 // |
|
270 UStringImpl::~UStringImpl() |
|
271 { |
|
272 if (str && isInHeap) |
|
273 { |
|
274 delete str; |
|
275 } |
|
276 } |
|
277 // ------------------------------------------------------------------------- |
|
278 // Returns true if the string is empty |
|
279 // ------------------------------------------------------------------------- |
|
280 // |
|
281 bool UStringImpl::empty()const |
|
282 { |
|
283 return (len > 0) ? false : true; |
|
284 } |
|
285 |
|
286 // ------------------------------------------------------------------------- |
|
287 // Returns true if the string is a null string |
|
288 // ------------------------------------------------------------------------- |
|
289 // |
|
290 bool UStringImpl::null()const |
|
291 { |
|
292 return (str) ? false : true; |
|
293 } |
|
294 |
|
295 // ------------------------------------------------------------------------- |
|
296 // Returns the length of the string in characters |
|
297 // ------------------------------------------------------------------------- |
|
298 // |
|
299 long UStringImpl::length()const |
|
300 { |
|
301 long ret(0); |
|
302 if(str && (len > 0)) |
|
303 { |
|
304 ret = g_utf8_strlen(str, len); |
|
305 } |
|
306 return ret; |
|
307 } |
|
308 |
|
309 // ------------------------------------------------------------------------- |
|
310 // Returns the length of the string in bytes |
|
311 // ------------------------------------------------------------------------- |
|
312 // |
|
313 long UStringImpl::bytes()const |
|
314 { |
|
315 return (long)len; |
|
316 } |
|
317 |
|
318 |
|
319 // ------------------------------------------------------------------------- |
|
320 // Returns the utf8 type string |
|
321 // ------------------------------------------------------------------------- |
|
322 // |
|
323 const Utf8* UStringImpl::utf8()const |
|
324 { |
|
325 return str; |
|
326 } |
|
327 |
|
328 // ------------------------------------------------------------------------- |
|
329 // Compares the string to another UString with strcmp |
|
330 // ------------------------------------------------------------------------- |
|
331 // |
|
332 int UStringImpl::compare(const UStringImpl& aUStringImpl)const |
|
333 { |
|
334 int ret(1); |
|
335 if(!str && !aUStringImpl.str) |
|
336 { |
|
337 ret = 0; |
|
338 } |
|
339 else if(str && aUStringImpl.str) |
|
340 { |
|
341 ret = strcmp(str, aUStringImpl.str); |
|
342 } |
|
343 else if(!str) |
|
344 { |
|
345 ret = -1; |
|
346 } |
|
347 |
|
348 return ret; |
|
349 } |
|
350 // ------------------------------------------------------------------------- |
|
351 // Compares the string to Utf8 string with strcmp |
|
352 // ------------------------------------------------------------------------- |
|
353 // |
|
354 int UStringImpl::compare(const Utf8* aStr)const |
|
355 { |
|
356 int ret(1); |
|
357 if(str && aStr) |
|
358 { |
|
359 ret = strcmp(str, aStr); |
|
360 } |
|
361 else if(!str && !strlen(aStr)) |
|
362 { |
|
363 ret = 0; |
|
364 } |
|
365 else if(!str) |
|
366 { |
|
367 ret = -1; |
|
368 } |
|
369 |
|
370 return ret; |
|
371 } |
|
372 |
|
373 // ------------------------------------------------------------------------- |
|
374 // Compares the string to another UString using the linguistically |
|
375 // correct rules for the current locale |
|
376 // ------------------------------------------------------------------------- |
|
377 // |
|
378 int UStringImpl::compareC(const UStringImpl& aUStringImpl)const |
|
379 { |
|
380 int ret(1); |
|
381 if(!str && !aUStringImpl.str) |
|
382 { |
|
383 ret = 0; |
|
384 } |
|
385 else if(str && aUStringImpl.str) |
|
386 { |
|
387 ret = g_utf8_collate(str, aUStringImpl.str); |
|
388 } |
|
389 else if(!str) |
|
390 { |
|
391 ret = -1; |
|
392 } |
|
393 |
|
394 return ret; |
|
395 } |
|
396 // ------------------------------------------------------------------------- |
|
397 // Compares the string to Utf8 string using the linguistically |
|
398 // correct rules for the current locale |
|
399 // ------------------------------------------------------------------------- |
|
400 // |
|
401 int UStringImpl::compareC(const Utf8* aStr)const |
|
402 { |
|
403 int ret(1); |
|
404 if(str && aStr) |
|
405 { |
|
406 ret = g_utf8_collate(str, aStr); |
|
407 } |
|
408 else if(!str && !strlen(aStr)) |
|
409 { |
|
410 ret = 0; |
|
411 } |
|
412 else if(!str) |
|
413 { |
|
414 ret = -1; |
|
415 } |
|
416 |
|
417 return ret; |
|
418 } |
|
419 |
|
420 // ------------------------------------------------------------------------- |
|
421 // Appends another UString to this string |
|
422 // ------------------------------------------------------------------------- |
|
423 // |
|
424 void UStringImpl::append(const UStringImpl& aUStringImpl) |
|
425 { |
|
426 if(str && aUStringImpl.str) |
|
427 { |
|
428 appendImpl(aUStringImpl.str, aUStringImpl.len); |
|
429 } |
|
430 } |
|
431 // ------------------------------------------------------------------------- |
|
432 // Appends a Utf8 string to this string |
|
433 // ------------------------------------------------------------------------- |
|
434 // |
|
435 void UStringImpl::append(const Utf8* aStr) |
|
436 { |
|
437 if(aStr && (strlen(aStr) !=0)) |
|
438 { |
|
439 appendImpl(aStr, strlen(aStr)); |
|
440 } |
|
441 } |
|
442 |
|
443 // ------------------------------------------------------------------------- |
|
444 // Inserts Utf8 string to this string |
|
445 // ------------------------------------------------------------------------- |
|
446 // |
|
447 void UStringImpl::insert( |
|
448 long aPos, |
|
449 const Utf8* aStr) |
|
450 { |
|
451 |
|
452 if(aStr) |
|
453 { |
|
454 if (!str) |
|
455 checkBuffer(0); |
|
456 long dstCharCount = g_utf8_strlen(str, -1); |
|
457 if(aPos < -1 || aPos > dstCharCount) |
|
458 { |
|
459 throw out_of_range("out_of_range"); |
|
460 } |
|
461 // find out byte position for given character position |
|
462 long startBytePos(-1); |
|
463 if(aPos >= 0) |
|
464 { |
|
465 if(str) |
|
466 { |
|
467 const char* startOffset = g_utf8_offset_to_pointer(str, aPos); |
|
468 startBytePos = startOffset - str; |
|
469 } |
|
470 } |
|
471 if(aPos == -1) |
|
472 { |
|
473 appendImpl(aStr, strlen(aStr)); |
|
474 } |
|
475 else |
|
476 { |
|
477 insertImpl(startBytePos, aStr, strlen(aStr)); |
|
478 } |
|
479 } |
|
480 } |
|
481 |
|
482 // ------------------------------------------------------------------------- |
|
483 // Inserts Utf8 string to this string with length of the Utf8 specified |
|
484 // ------------------------------------------------------------------------- |
|
485 // |
|
486 void UStringImpl::insert( |
|
487 long aPos, |
|
488 const Utf8* aStr, |
|
489 long aCharCount) |
|
490 { |
|
491 if (aCharCount < 1) |
|
492 { |
|
493 throw out_of_range("out_of_range"); |
|
494 } |
|
495 |
|
496 if (aStr) |
|
497 { |
|
498 long srcByteLen = bytesFromCharacters(aStr, aCharCount); |
|
499 if(!str) |
|
500 checkBuffer(0); |
|
501 |
|
502 // calculate how many bytes are needed for given characters |
|
503 long srcCharCount = g_utf8_strlen(aStr, srcByteLen); |
|
504 long dstCharCount = g_utf8_strlen(str, -1); |
|
505 long startBytePos(-1); // appends end of the string |
|
506 |
|
507 if(aPos < -1 || aPos > dstCharCount-1 || aCharCount > srcCharCount) |
|
508 { |
|
509 throw out_of_range("out_of_range"); |
|
510 } |
|
511 |
|
512 if(aPos != -1) |
|
513 { |
|
514 const char* startOffset = g_utf8_offset_to_pointer(str, aPos); |
|
515 startBytePos = startOffset - str; |
|
516 } |
|
517 const char* endOffset = g_utf8_offset_to_pointer(aStr, aCharCount); |
|
518 long insertByteLength = endOffset - aStr; |
|
519 |
|
520 if(aPos == -1) |
|
521 { |
|
522 appendImpl(aStr, insertByteLength); |
|
523 } |
|
524 else |
|
525 { |
|
526 insertImpl(startBytePos, aStr, srcByteLen); |
|
527 } |
|
528 } |
|
529 } |
|
530 |
|
531 // ------------------------------------------------------------------------- |
|
532 // Replaces Utf8 string to this string |
|
533 // ------------------------------------------------------------------------- |
|
534 // |
|
535 void UStringImpl::replace( |
|
536 long aPos, |
|
537 const Utf8* aStr) |
|
538 { |
|
539 if (aStr) |
|
540 { |
|
541 if (!str) |
|
542 checkBuffer(0); |
|
543 |
|
544 long srcByteLen = strlen(aStr); |
|
545 long srcCharCount = g_utf8_strlen(aStr,-1); |
|
546 long dstCharCount = g_utf8_strlen(str, -1); |
|
547 if(aPos > dstCharCount || aPos < 0) |
|
548 { |
|
549 throw out_of_range("out_of_range"); |
|
550 } |
|
551 |
|
552 // calculate start byte position |
|
553 if(str) |
|
554 { |
|
555 const char* startOffset = g_utf8_offset_to_pointer(str, aPos); |
|
556 long startBytePos = startOffset - str; |
|
557 |
|
558 // calculate erasable character count in bytes |
|
559 if(startOffset) |
|
560 { |
|
561 const char* endOffset = g_utf8_offset_to_pointer(startOffset, srcCharCount); |
|
562 long eraseByteLength = endOffset - startOffset; |
|
563 // erase existing characters |
|
564 eraseImpl(startBytePos, eraseByteLength); |
|
565 } |
|
566 insertImpl(startBytePos, aStr, srcByteLen); |
|
567 } |
|
568 } |
|
569 } |
|
570 |
|
571 |
|
572 // ------------------------------------------------------------------------- |
|
573 // Replaces a Utf8 string to this string with length of the Utf8 specified |
|
574 // ------------------------------------------------------------------------- |
|
575 // |
|
576 void UStringImpl::replace( |
|
577 long aPosOffset, |
|
578 const Utf8* aStr, |
|
579 long aCharCount) |
|
580 { |
|
581 if (aCharCount < 1) |
|
582 { |
|
583 throw out_of_range("out_of_range"); |
|
584 } |
|
585 |
|
586 if (aStr) |
|
587 { |
|
588 long srcByteLen = bytesFromCharacters(aStr, aCharCount); |
|
589 if(!str) |
|
590 checkBuffer(0); |
|
591 |
|
592 long srcCharCount = g_utf8_strlen(aStr, srcByteLen); |
|
593 long dstCharCount = g_utf8_strlen(str, -1); |
|
594 if(aPosOffset > dstCharCount || aPosOffset < 0 || aCharCount > srcCharCount) |
|
595 { |
|
596 throw out_of_range("out_of_range"); |
|
597 } |
|
598 // calculate start byte position |
|
599 if(str) |
|
600 { |
|
601 const char* startOffset = g_utf8_offset_to_pointer(str, aPosOffset); |
|
602 long startBytePos = startOffset - str; |
|
603 |
|
604 // calculate erasable character count in bytes |
|
605 if(startOffset) |
|
606 { |
|
607 const char* endOffset = g_utf8_offset_to_pointer(startOffset, srcCharCount); |
|
608 long eraseByteLength = endOffset - startOffset; |
|
609 // erase existing characters |
|
610 eraseImpl(startBytePos, eraseByteLength); |
|
611 } |
|
612 insertImpl(startBytePos, aStr, srcByteLen); |
|
613 } |
|
614 } |
|
615 } |
|
616 |
|
617 // ------------------------------------------------------------------------- |
|
618 // Erases aCharCount characters from the string at position specified by |
|
619 // aPosOffset |
|
620 // ------------------------------------------------------------------------- |
|
621 // |
|
622 void UStringImpl::erase( |
|
623 long aPosOffset, |
|
624 long aCharCount) |
|
625 { |
|
626 |
|
627 if(str) |
|
628 { |
|
629 long srcCharCount = g_utf8_strlen(str, -1); |
|
630 |
|
631 if (aPosOffset < 0 || aCharCount < 1 || aCharCount > srcCharCount) |
|
632 { |
|
633 throw out_of_range("out_of_range"); |
|
634 } |
|
635 |
|
636 // calc startpos |
|
637 const char* startOffset = g_utf8_offset_to_pointer(str, aPosOffset); |
|
638 long startBytePos = startOffset - str; |
|
639 |
|
640 // calc bytes to erase |
|
641 const char* endOffset = g_utf8_offset_to_pointer(startOffset, aCharCount); |
|
642 long eraseByteLength = endOffset - startOffset; |
|
643 |
|
644 eraseImpl(startBytePos, eraseByteLength); |
|
645 } |
|
646 } |
|
647 |
|
648 // ------------------------------------------------------------------------- |
|
649 // Returns unicode at aIndex position in the string |
|
650 // ------------------------------------------------------------------------- |
|
651 // |
|
652 Unicode UStringImpl::operator[](long aIndex) |
|
653 { |
|
654 Unicode ret(0xF8FF); |
|
655 |
|
656 if(str) |
|
657 { |
|
658 long charCount = g_utf8_strlen(str, -1); |
|
659 if(aIndex < 0 || aIndex > charCount-1) |
|
660 { |
|
661 throw out_of_range("out_of_range"); |
|
662 } |
|
663 // calc startpos |
|
664 const char* uniChar = g_utf8_offset_to_pointer(str, aIndex); |
|
665 |
|
666 ret = g_utf8_get_char(uniChar); |
|
667 } |
|
668 return ret; |
|
669 } |
|
670 |
|
671 } |
|
672 |
|
673 |
|