1 | /* SDSLib 2.0 -- A C dynamic strings library |
2 | * |
3 | * Copyright (c) 2006-2015, Salvatore Sanfilippo <antirez at gmail dot com> |
4 | * Copyright (c) 2015, Oran Agra |
5 | * Copyright (c) 2015, Redis Labs, Inc |
6 | * All rights reserved. |
7 | * |
8 | * Redistribution and use in source and binary forms, with or without |
9 | * modification, are permitted provided that the following conditions are met: |
10 | * |
11 | * * Redistributions of source code must retain the above copyright notice, |
12 | * this list of conditions and the following disclaimer. |
13 | * * Redistributions in binary form must reproduce the above copyright |
14 | * notice, this list of conditions and the following disclaimer in the |
15 | * documentation and/or other materials provided with the distribution. |
16 | * * Neither the name of Redis nor the names of its contributors may be used |
17 | * to endorse or promote products derived from this software without |
18 | * specific prior written permission. |
19 | * |
20 | * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" |
21 | * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
22 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
23 | * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE |
24 | * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
25 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
26 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
27 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
28 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
29 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
30 | * POSSIBILITY OF SUCH DAMAGE. |
31 | */ |
32 | |
33 | #include <stdio.h> |
34 | #include <stdlib.h> |
35 | #include <string.h> |
36 | #include <ctype.h> |
37 | #include <assert.h> |
38 | #include <limits.h> |
39 | #include "sds.hpp" |
40 | |
41 | static inline int sdsHdrSize(char type) { |
42 | switch(type&SDS_TYPE_MASK) { |
43 | case SDS_TYPE_5: |
44 | return sizeof(struct sdshdr5); |
45 | case SDS_TYPE_8: |
46 | return sizeof(struct sdshdr8); |
47 | case SDS_TYPE_16: |
48 | return sizeof(struct sdshdr16); |
49 | case SDS_TYPE_32: |
50 | return sizeof(struct sdshdr32); |
51 | case SDS_TYPE_64: |
52 | return sizeof(struct sdshdr64); |
53 | } |
54 | return 0; |
55 | } |
56 | |
57 | static inline char sdsReqType(size_t string_size) { |
58 | if (string_size < 1<<5) |
59 | return SDS_TYPE_5; |
60 | if (string_size < 1<<8) |
61 | return SDS_TYPE_8; |
62 | if (string_size < 1<<16) |
63 | return SDS_TYPE_16; |
64 | #if (LONG_MAX == LLONG_MAX) |
65 | if (string_size < 1ll<<32) |
66 | return SDS_TYPE_32; |
67 | return SDS_TYPE_64; |
68 | #else |
69 | return SDS_TYPE_32; |
70 | #endif |
71 | } |
72 | |
73 | /* Create a new sds string with the content specified by the 'init' pointer |
74 | * and 'initlen'. |
75 | * If NULL is used for 'init' the string is initialized with zero bytes. |
76 | * If SDS_NOINIT is used, the buffer is left uninitialized; |
77 | * |
78 | * The string is always null-termined (all the sds strings are, always) so |
79 | * even if you create an sds string with: |
80 | * |
81 | * mystring = sdsnewlen("abc",3); |
82 | * |
83 | * You can print the string with printf() as there is an implicit \0 at the |
84 | * end of the string. However the string is binary safe and can contain |
85 | * \0 characters in the middle, as the length is stored in the sds header. */ |
86 | sds sdsnewlen(const void *init, size_t initlen) { |
87 | void *sh; |
88 | sds s; |
89 | char type = sdsReqType(initlen); |
90 | /* Empty strings are usually created in order to append. Use type 8 |
91 | * since type 5 is not good at this. */ |
92 | if (type == SDS_TYPE_5 && initlen == 0) type = SDS_TYPE_8; |
93 | int hdrlen = sdsHdrSize(type); |
94 | unsigned char *fp; /* flags pointer. */ |
95 | |
96 | sh = malloc(hdrlen+initlen+1); |
97 | if (!init) |
98 | memset(sh, 0, hdrlen+initlen+1); |
99 | if (sh == NULL) return NULL; |
100 | s = (char*)sh+hdrlen; |
101 | fp = ((unsigned char*)s)-1; |
102 | switch(type) { |
103 | case SDS_TYPE_5: { |
104 | *fp = type | (initlen << SDS_TYPE_BITS); |
105 | break; |
106 | } |
107 | case SDS_TYPE_8: { |
108 | SDS_HDR_VAR(8,s); |
109 | sh->len = initlen; |
110 | sh->alloc = initlen; |
111 | *fp = type; |
112 | break; |
113 | } |
114 | case SDS_TYPE_16: { |
115 | SDS_HDR_VAR(16,s); |
116 | sh->len = initlen; |
117 | sh->alloc = initlen; |
118 | *fp = type; |
119 | break; |
120 | } |
121 | case SDS_TYPE_32: { |
122 | SDS_HDR_VAR(32,s); |
123 | sh->len = initlen; |
124 | sh->alloc = initlen; |
125 | *fp = type; |
126 | break; |
127 | } |
128 | case SDS_TYPE_64: { |
129 | SDS_HDR_VAR(64,s); |
130 | sh->len = initlen; |
131 | sh->alloc = initlen; |
132 | *fp = type; |
133 | break; |
134 | } |
135 | } |
136 | if (initlen && init) |
137 | memcpy(s, init, initlen); |
138 | s[initlen] = '\0'; |
139 | return s; |
140 | } |
141 | |
142 | /* Create an empty (zero length) sds string. Even in this case the string |
143 | * always has an implicit null term. */ |
144 | sds sdsempty(void) { |
145 | return sdsnewlen("" ,0); |
146 | } |
147 | |
148 | /* Create a new sds string starting from a null terminated C string. */ |
149 | sds sdsnew(const char *init) { |
150 | size_t initlen = (init == NULL) ? 0 : strlen(init); |
151 | return sdsnewlen(init, initlen); |
152 | } |
153 | |
154 | /* Duplicate an sds string. */ |
155 | sds sdsdup(const sds s) { |
156 | return sdsnewlen(s, sdslen(s)); |
157 | } |
158 | |
159 | /* Free an sds string. No operation is performed if 's' is NULL. */ |
160 | void sdsfree(sds s) { |
161 | if (s == NULL) return; |
162 | free((char*)s-sdsHdrSize(s[-1])); |
163 | } |
164 | |
165 | /* Set the sds string length to the length as obtained with strlen(), so |
166 | * considering as content only up to the first null term character. |
167 | * |
168 | * This function is useful when the sds string is hacked manually in some |
169 | * way, like in the following example: |
170 | * |
171 | * s = sdsnew("foobar"); |
172 | * s[2] = '\0'; |
173 | * sdsupdatelen(s); |
174 | * printf("%d\n", sdslen(s)); |
175 | * |
176 | * The output will be "2", but if we comment out the call to sdsupdatelen() |
177 | * the output will be "6" as the string was modified but the logical length |
178 | * remains 6 bytes. */ |
179 | void sdsupdatelen(sds s) { |
180 | size_t reallen = strlen(s); |
181 | sdssetlen(s, reallen); |
182 | } |
183 | |
184 | /* Modify an sds string in-place to make it empty (zero length). |
185 | * However all the existing buffer is not discarded but set as free space |
186 | * so that next append operations will not require allocations up to the |
187 | * number of bytes previously available. */ |
188 | void sdsclear(sds s) { |
189 | sdssetlen(s, 0); |
190 | s[0] = '\0'; |
191 | } |
192 | |
193 | /* Enlarge the free space at the end of the sds string so that the caller |
194 | * is sure that after calling this function can overwrite up to addlen |
195 | * bytes after the end of the string, plus one more byte for nul term. |
196 | * |
197 | * Note: this does not change the *length* of the sds string as returned |
198 | * by sdslen(), but only the free buffer space we have. */ |
199 | sds sdsMakeRoomFor(sds s, size_t addlen) { |
200 | void *sh, *newsh; |
201 | size_t avail = sdsavail(s); |
202 | size_t len, newlen; |
203 | char type, oldtype = s[-1] & SDS_TYPE_MASK; |
204 | int hdrlen; |
205 | |
206 | /* Return ASAP if there is enough space left. */ |
207 | if (avail >= addlen) return s; |
208 | |
209 | len = sdslen(s); |
210 | sh = (char*)s-sdsHdrSize(oldtype); |
211 | newlen = (len+addlen); |
212 | if (newlen < SDS_MAX_PREALLOC) |
213 | newlen *= 2; |
214 | else |
215 | newlen += SDS_MAX_PREALLOC; |
216 | |
217 | type = sdsReqType(newlen); |
218 | |
219 | /* Don't use type 5: the user is appending to the string and type 5 is |
220 | * not able to remember empty space, so sdsMakeRoomFor() must be called |
221 | * at every appending operation. */ |
222 | if (type == SDS_TYPE_5) type = SDS_TYPE_8; |
223 | |
224 | hdrlen = sdsHdrSize(type); |
225 | if (oldtype==type) { |
226 | newsh = realloc(sh, hdrlen+newlen+1); |
227 | if (newsh == NULL) return NULL; |
228 | s = (char*)newsh+hdrlen; |
229 | } else { |
230 | /* Since the header size changes, need to move the string forward, |
231 | * and can't use realloc */ |
232 | newsh = malloc(hdrlen+newlen+1); |
233 | if (newsh == NULL) return NULL; |
234 | memcpy((char*)newsh+hdrlen, s, len+1); |
235 | free(sh); |
236 | s = (char*)newsh+hdrlen; |
237 | s[-1] = type; |
238 | sdssetlen(s, len); |
239 | } |
240 | sdssetalloc(s, newlen); |
241 | return s; |
242 | } |
243 | |
244 | /* Reallocate the sds string so that it has no free space at the end. The |
245 | * contained string remains not altered, but next concatenation operations |
246 | * will require a reallocation. |
247 | * |
248 | * After the call, the passed sds string is no longer valid and all the |
249 | * references must be substituted with the new pointer returned by the call. */ |
250 | sds sdsRemoveFreeSpace(sds s) { |
251 | void *sh, *newsh; |
252 | char type, oldtype = s[-1] & SDS_TYPE_MASK; |
253 | int hdrlen, oldhdrlen = sdsHdrSize(oldtype); |
254 | size_t len = sdslen(s); |
255 | sh = (char*)s-oldhdrlen; |
256 | |
257 | /* Check what would be the minimum SDS header that is just good enough to |
258 | * fit this string. */ |
259 | type = sdsReqType(len); |
260 | hdrlen = sdsHdrSize(type); |
261 | |
262 | /* If the type is the same, or at least a large enough type is still |
263 | * required, we just realloc(), letting the allocator to do the copy |
264 | * only if really needed. Otherwise if the change is huge, we manually |
265 | * reallocate the string to use the different header type. */ |
266 | if (oldtype==type || type > SDS_TYPE_8) { |
267 | newsh = realloc(sh, oldhdrlen+len+1); |
268 | if (newsh == NULL) return NULL; |
269 | s = (char*)newsh+oldhdrlen; |
270 | } else { |
271 | newsh = malloc(hdrlen+len+1); |
272 | if (newsh == NULL) return NULL; |
273 | memcpy((char*)newsh+hdrlen, s, len+1); |
274 | free(sh); |
275 | s = (char*)newsh+hdrlen; |
276 | s[-1] = type; |
277 | sdssetlen(s, len); |
278 | } |
279 | sdssetalloc(s, len); |
280 | return s; |
281 | } |
282 | |
283 | /* Return the total size of the allocation of the specified sds string, |
284 | * including: |
285 | * 1) The sds header before the pointer. |
286 | * 2) The string. |
287 | * 3) The free buffer at the end if any. |
288 | * 4) The implicit null term. |
289 | */ |
290 | size_t sdsAllocSize(sds s) { |
291 | size_t alloc = sdsalloc(s); |
292 | return sdsHdrSize(s[-1])+alloc+1; |
293 | } |
294 | |
295 | /* Return the pointer of the actual SDS allocation (normally SDS strings |
296 | * are referenced by the start of the string buffer). */ |
297 | void *sdsAllocPtr(sds s) { |
298 | return (void*) (s-sdsHdrSize(s[-1])); |
299 | } |
300 | |
301 | /* Increment the sds length and decrements the left free space at the |
302 | * end of the string according to 'incr'. Also set the null term |
303 | * in the new end of the string. |
304 | * |
305 | * This function is used in order to fix the string length after the |
306 | * user calls sdsMakeRoomFor(), writes something after the end of |
307 | * the current string, and finally needs to set the new length. |
308 | * |
309 | * Note: it is possible to use a negative increment in order to |
310 | * right-trim the string. |
311 | * |
312 | * Usage example: |
313 | * |
314 | * Using sdsIncrLen() and sdsMakeRoomFor() it is possible to mount the |
315 | * following schema, to cat bytes coming from the kernel to the end of an |
316 | * sds string without copying into an intermediate buffer: |
317 | * |
318 | * oldlen = sdslen(s); |
319 | * s = sdsMakeRoomFor(s, BUFFER_SIZE); |
320 | * nread = read(fd, s+oldlen, BUFFER_SIZE); |
321 | * ... check for nread <= 0 and handle it ... |
322 | * sdsIncrLen(s, nread); |
323 | */ |
324 | void sdsIncrLen(sds s, ssize_t incr) { |
325 | unsigned char flags = s[-1]; |
326 | size_t len; |
327 | switch(flags&SDS_TYPE_MASK) { |
328 | case SDS_TYPE_5: { |
329 | unsigned char *fp = ((unsigned char*)s)-1; |
330 | unsigned char oldlen = SDS_TYPE_5_LEN(flags); |
331 | assert((incr > 0 && oldlen+incr < 32) || (incr < 0 && oldlen >= (unsigned int)(-incr))); |
332 | *fp = SDS_TYPE_5 | ((oldlen+incr) << SDS_TYPE_BITS); |
333 | len = oldlen+incr; |
334 | break; |
335 | } |
336 | case SDS_TYPE_8: { |
337 | SDS_HDR_VAR(8,s); |
338 | assert((incr >= 0 && sh->alloc-sh->len >= incr) || (incr < 0 && sh->len >= (unsigned int)(-incr))); |
339 | len = (sh->len += incr); |
340 | break; |
341 | } |
342 | case SDS_TYPE_16: { |
343 | SDS_HDR_VAR(16,s); |
344 | assert((incr >= 0 && sh->alloc-sh->len >= incr) || (incr < 0 && sh->len >= (unsigned int)(-incr))); |
345 | len = (sh->len += incr); |
346 | break; |
347 | } |
348 | case SDS_TYPE_32: { |
349 | SDS_HDR_VAR(32,s); |
350 | assert((incr >= 0 && sh->alloc-sh->len >= (unsigned int)incr) || (incr < 0 && sh->len >= (unsigned int)(-incr))); |
351 | len = (sh->len += incr); |
352 | break; |
353 | } |
354 | case SDS_TYPE_64: { |
355 | SDS_HDR_VAR(64,s); |
356 | assert((incr >= 0 && sh->alloc-sh->len >= (uint64_t)incr) || (incr < 0 && sh->len >= (uint64_t)(-incr))); |
357 | len = (sh->len += incr); |
358 | break; |
359 | } |
360 | default: len = 0; /* Just to avoid compilation warnings. */ |
361 | } |
362 | s[len] = '\0'; |
363 | } |
364 | |
365 | /* Grow the sds to have the specified length. Bytes that were not part of |
366 | * the original length of the sds will be set to zero. |
367 | * |
368 | * if the specified length is smaller than the current length, no operation |
369 | * is performed. */ |
370 | sds sdsgrowzero(sds s, size_t len) { |
371 | size_t curlen = sdslen(s); |
372 | |
373 | if (len <= curlen) return s; |
374 | s = sdsMakeRoomFor(s,len-curlen); |
375 | if (s == NULL) return NULL; |
376 | |
377 | /* Make sure added region doesn't contain garbage */ |
378 | memset(s+curlen,0,(len-curlen+1)); /* also set trailing \0 byte */ |
379 | sdssetlen(s, len); |
380 | return s; |
381 | } |
382 | |
383 | /* Append the specified binary-safe string pointed by 't' of 'len' bytes to the |
384 | * end of the specified sds string 's'. |
385 | * |
386 | * After the call, the passed sds string is no longer valid and all the |
387 | * references must be substituted with the new pointer returned by the call. */ |
388 | sds sdscatlen(sds s, const void *t, size_t len) { |
389 | size_t curlen = sdslen(s); |
390 | |
391 | s = sdsMakeRoomFor(s,len); |
392 | if (s == NULL) return NULL; |
393 | memcpy(s+curlen, t, len); |
394 | sdssetlen(s, curlen+len); |
395 | s[curlen+len] = '\0'; |
396 | return s; |
397 | } |
398 | |
399 | /* Append the specified null termianted C string to the sds string 's'. |
400 | * |
401 | * After the call, the passed sds string is no longer valid and all the |
402 | * references must be substituted with the new pointer returned by the call. */ |
403 | sds sdscat(sds s, const char *t) { |
404 | return sdscatlen(s, t, strlen(t)); |
405 | } |
406 | |
407 | /* Append the specified sds 't' to the existing sds 's'. |
408 | * |
409 | * After the call, the modified sds string is no longer valid and all the |
410 | * references must be substituted with the new pointer returned by the call. */ |
411 | sds sdscatsds(sds s, const sds t) { |
412 | return sdscatlen(s, t, sdslen(t)); |
413 | } |
414 | |
415 | /* Destructively modify the sds string 's' to hold the specified binary |
416 | * safe string pointed by 't' of length 'len' bytes. */ |
417 | sds sdscpylen(sds s, const char *t, size_t len) { |
418 | if (sdsalloc(s) < len) { |
419 | s = sdsMakeRoomFor(s,len-sdslen(s)); |
420 | if (s == NULL) return NULL; |
421 | } |
422 | memcpy(s, t, len); |
423 | s[len] = '\0'; |
424 | sdssetlen(s, len); |
425 | return s; |
426 | } |
427 | |
428 | /* Like sdscpylen() but 't' must be a null-termined string so that the length |
429 | * of the string is obtained with strlen(). */ |
430 | sds sdscpy(sds s, const char *t) { |
431 | return sdscpylen(s, t, strlen(t)); |
432 | } |
433 | |
434 | /* Helper for sdscatlonglong() doing the actual number -> string |
435 | * conversion. 's' must point to a string with room for at least |
436 | * SDS_LLSTR_SIZE bytes. |
437 | * |
438 | * The function returns the length of the null-terminated string |
439 | * representation stored at 's'. */ |
440 | #define SDS_LLSTR_SIZE 21 |
441 | int sdsll2str(char *s, long long value) { |
442 | char *p, aux; |
443 | unsigned long long v; |
444 | size_t l; |
445 | |
446 | /* Generate the string representation, this method produces |
447 | * an reversed string. */ |
448 | v = (value < 0) ? -value : value; |
449 | p = s; |
450 | do { |
451 | *p++ = '0'+(v%10); |
452 | v /= 10; |
453 | } while(v); |
454 | if (value < 0) *p++ = '-'; |
455 | |
456 | /* Compute length and add null term. */ |
457 | l = p-s; |
458 | *p = '\0'; |
459 | |
460 | /* Reverse the string. */ |
461 | p--; |
462 | while(s < p) { |
463 | aux = *s; |
464 | *s = *p; |
465 | *p = aux; |
466 | s++; |
467 | p--; |
468 | } |
469 | return l; |
470 | } |
471 | |
472 | /* Identical sdsll2str(), but for unsigned long long type. */ |
473 | int sdsull2str(char *s, unsigned long long v) { |
474 | char *p, aux; |
475 | size_t l; |
476 | |
477 | /* Generate the string representation, this method produces |
478 | * an reversed string. */ |
479 | p = s; |
480 | do { |
481 | *p++ = '0'+(v%10); |
482 | v /= 10; |
483 | } while(v); |
484 | |
485 | /* Compute length and add null term. */ |
486 | l = p-s; |
487 | *p = '\0'; |
488 | |
489 | /* Reverse the string. */ |
490 | p--; |
491 | while(s < p) { |
492 | aux = *s; |
493 | *s = *p; |
494 | *p = aux; |
495 | s++; |
496 | p--; |
497 | } |
498 | return l; |
499 | } |
500 | |
501 | /* Create an sds string from a long long value. It is much faster than: |
502 | * |
503 | * sdscatprintf(sdsempty(),"%lld\n", value); |
504 | */ |
505 | sds sdsfromlonglong(long long value) { |
506 | char buf[SDS_LLSTR_SIZE]; |
507 | int len = sdsll2str(buf,value); |
508 | |
509 | return sdsnewlen(buf,len); |
510 | } |
511 | |
512 | /* Like sdscatprintf() but gets va_list instead of being variadic. */ |
513 | sds sdscatvprintf(sds s, const char *fmt, va_list ap) { |
514 | va_list cpy; |
515 | char staticbuf[1024], *buf = staticbuf, *t; |
516 | size_t buflen = strlen(fmt)*2; |
517 | |
518 | /* We try to start using a static buffer for speed. |
519 | * If not possible we revert to heap allocation. */ |
520 | if (buflen > sizeof(staticbuf)) { |
521 | buf = (char*) malloc(buflen); |
522 | if (buf == NULL) return NULL; |
523 | } else { |
524 | buflen = sizeof(staticbuf); |
525 | } |
526 | |
527 | /* Try with buffers two times bigger every time we fail to |
528 | * fit the string in the current buffer size. */ |
529 | while(1) { |
530 | buf[buflen-2] = '\0'; |
531 | va_copy(cpy,ap); |
532 | vsnprintf(buf, buflen, fmt, cpy); |
533 | va_end(cpy); |
534 | if (buf[buflen-2] != '\0') { |
535 | if (buf != staticbuf) free(buf); |
536 | buflen *= 2; |
537 | buf = (char*) malloc(buflen); |
538 | if (buf == NULL) return NULL; |
539 | continue; |
540 | } |
541 | break; |
542 | } |
543 | |
544 | /* Finally concat the obtained string to the SDS string and return it. */ |
545 | t = sdscat(s, buf); |
546 | if (buf != staticbuf) free(buf); |
547 | return t; |
548 | } |
549 | |
550 | /* Append to the sds string 's' a string obtained using printf-alike format |
551 | * specifier. |
552 | * |
553 | * After the call, the modified sds string is no longer valid and all the |
554 | * references must be substituted with the new pointer returned by the call. |
555 | * |
556 | * Example: |
557 | * |
558 | * s = sdsnew("Sum is: "); |
559 | * s = sdscatprintf(s,"%d+%d = %d",a,b,a+b). |
560 | * |
561 | * Often you need to create a string from scratch with the printf-alike |
562 | * format. When this is the need, just use sdsempty() as the target string: |
563 | * |
564 | * s = sdscatprintf(sdsempty(), "... your format ...", args); |
565 | */ |
566 | sds sdscatprintf(sds s, const char *fmt, ...) { |
567 | va_list ap; |
568 | char *t; |
569 | va_start(ap, fmt); |
570 | t = sdscatvprintf(s,fmt,ap); |
571 | va_end(ap); |
572 | return t; |
573 | } |
574 | |
575 | /* This function is similar to sdscatprintf, but much faster as it does |
576 | * not rely on sprintf() family functions implemented by the libc that |
577 | * are often very slow. Moreover directly handling the sds string as |
578 | * new data is concatenated provides a performance improvement. |
579 | * |
580 | * However this function only handles an incompatible subset of printf-alike |
581 | * format specifiers: |
582 | * |
583 | * %s - C String |
584 | * %S - SDS string |
585 | * %i - signed int |
586 | * %I - 64 bit signed integer (long long, int64_t) |
587 | * %u - unsigned int |
588 | * %U - 64 bit unsigned integer (unsigned long long, uint64_t) |
589 | * %% - Verbatim "%" character. |
590 | */ |
591 | sds sdscatfmt(sds s, char const *fmt, ...) { |
592 | size_t initlen = sdslen(s); |
593 | const char *f = fmt; |
594 | long i; |
595 | va_list ap; |
596 | |
597 | va_start(ap,fmt); |
598 | f = fmt; /* Next format specifier byte to process. */ |
599 | i = initlen; /* Position of the next byte to write to dest str. */ |
600 | while(*f) { |
601 | char next, *str; |
602 | size_t l; |
603 | long long num; |
604 | unsigned long long unum; |
605 | |
606 | /* Make sure there is always space for at least 1 char. */ |
607 | if (sdsavail(s)==0) { |
608 | s = sdsMakeRoomFor(s,1); |
609 | } |
610 | |
611 | switch(*f) { |
612 | case '%': |
613 | next = *(f+1); |
614 | f++; |
615 | switch(next) { |
616 | case 's': |
617 | case 'S': |
618 | str = va_arg(ap,char*); |
619 | l = (next == 's') ? strlen(str) : sdslen(str); |
620 | if (sdsavail(s) < l) { |
621 | s = sdsMakeRoomFor(s,l); |
622 | } |
623 | memcpy(s+i,str,l); |
624 | sdsinclen(s,l); |
625 | i += l; |
626 | break; |
627 | case 'i': |
628 | case 'I': |
629 | if (next == 'i') |
630 | num = va_arg(ap,int); |
631 | else |
632 | num = va_arg(ap,long long); |
633 | { |
634 | char buf[SDS_LLSTR_SIZE]; |
635 | l = sdsll2str(buf,num); |
636 | if (sdsavail(s) < l) { |
637 | s = sdsMakeRoomFor(s,l); |
638 | } |
639 | memcpy(s+i,buf,l); |
640 | sdsinclen(s,l); |
641 | i += l; |
642 | } |
643 | break; |
644 | case 'u': |
645 | case 'U': |
646 | if (next == 'u') |
647 | unum = va_arg(ap,unsigned int); |
648 | else |
649 | unum = va_arg(ap,unsigned long long); |
650 | { |
651 | char buf[SDS_LLSTR_SIZE]; |
652 | l = sdsull2str(buf,unum); |
653 | if (sdsavail(s) < l) { |
654 | s = sdsMakeRoomFor(s,l); |
655 | } |
656 | memcpy(s+i,buf,l); |
657 | sdsinclen(s,l); |
658 | i += l; |
659 | } |
660 | break; |
661 | default: /* Handle %% and generally %<unknown>. */ |
662 | s[i++] = next; |
663 | sdsinclen(s,1); |
664 | break; |
665 | } |
666 | break; |
667 | default: |
668 | s[i++] = *f; |
669 | sdsinclen(s,1); |
670 | break; |
671 | } |
672 | f++; |
673 | } |
674 | va_end(ap); |
675 | |
676 | /* Add null-term */ |
677 | s[i] = '\0'; |
678 | return s; |
679 | } |
680 | |
681 | /* Remove the part of the string from left and from right composed just of |
682 | * contiguous characters found in 'cset', that is a null terminted C string. |
683 | * |
684 | * After the call, the modified sds string is no longer valid and all the |
685 | * references must be substituted with the new pointer returned by the call. |
686 | * |
687 | * Example: |
688 | * |
689 | * s = sdsnew("AA...AA.a.aa.aHelloWorld :::"); |
690 | * s = sdstrim(s,"Aa. :"); |
691 | * printf("%s\n", s); |
692 | * |
693 | * Output will be just "Hello World". |
694 | */ |
695 | sds sdstrim(sds s, const char *cset) { |
696 | char *start, *end, *sp, *ep; |
697 | size_t len; |
698 | |
699 | sp = start = s; |
700 | ep = end = s+sdslen(s)-1; |
701 | while(sp <= end && strchr(cset, *sp)) sp++; |
702 | while(ep > sp && strchr(cset, *ep)) ep--; |
703 | len = (sp > ep) ? 0 : ((ep-sp)+1); |
704 | if (s != sp) memmove(s, sp, len); |
705 | s[len] = '\0'; |
706 | sdssetlen(s,len); |
707 | return s; |
708 | } |
709 | |
710 | /* Turn the string into a smaller (or equal) string containing only the |
711 | * substring specified by the 'start' and 'end' indexes. |
712 | * |
713 | * start and end can be negative, where -1 means the last character of the |
714 | * string, -2 the penultimate character, and so forth. |
715 | * |
716 | * The interval is inclusive, so the start and end characters will be part |
717 | * of the resulting string. |
718 | * |
719 | * The string is modified in-place. |
720 | * |
721 | * Example: |
722 | * |
723 | * s = sdsnew("Hello World"); |
724 | * sdsrange(s,1,-1); => "ello World" |
725 | */ |
726 | void sdsrange(sds s, ssize_t start, ssize_t end) { |
727 | size_t newlen, len = sdslen(s); |
728 | |
729 | if (len == 0) return; |
730 | if (start < 0) { |
731 | start = len+start; |
732 | if (start < 0) start = 0; |
733 | } |
734 | if (end < 0) { |
735 | end = len+end; |
736 | if (end < 0) end = 0; |
737 | } |
738 | newlen = (start > end) ? 0 : (end-start)+1; |
739 | if (newlen != 0) { |
740 | if (start >= (ssize_t)len) { |
741 | newlen = 0; |
742 | } else if (end >= (ssize_t)len) { |
743 | end = len-1; |
744 | newlen = (start > end) ? 0 : (end-start)+1; |
745 | } |
746 | } else { |
747 | start = 0; |
748 | } |
749 | if (start && newlen) memmove(s, s+start, newlen); |
750 | s[newlen] = 0; |
751 | sdssetlen(s,newlen); |
752 | } |
753 | |
754 | /* Apply tolower() to every character of the sds string 's'. */ |
755 | void sdstolower(sds s) { |
756 | size_t len = sdslen(s), j; |
757 | |
758 | for (j = 0; j < len; j++) s[j] = tolower(s[j]); |
759 | } |
760 | |
761 | /* Apply toupper() to every character of the sds string 's'. */ |
762 | void sdstoupper(sds s) { |
763 | size_t len = sdslen(s), j; |
764 | |
765 | for (j = 0; j < len; j++) s[j] = toupper(s[j]); |
766 | } |
767 | |
768 | /* Compare two sds strings s1 and s2 with memcmp(). |
769 | * |
770 | * Return value: |
771 | * |
772 | * positive if s1 > s2. |
773 | * negative if s1 < s2. |
774 | * 0 if s1 and s2 are exactly the same binary string. |
775 | * |
776 | * If two strings share exactly the same prefix, but one of the two has |
777 | * additional characters, the longer string is considered to be greater than |
778 | * the smaller one. */ |
779 | int sdscmp(const sds s1, const sds s2) { |
780 | size_t l1, l2, minlen; |
781 | int cmp; |
782 | |
783 | l1 = sdslen(s1); |
784 | l2 = sdslen(s2); |
785 | minlen = (l1 < l2) ? l1 : l2; |
786 | cmp = memcmp(s1,s2,minlen); |
787 | if (cmp == 0) return l1>l2? 1: (l1<l2? -1: 0); |
788 | return cmp; |
789 | } |
790 | |
791 | /* Split 's' with separator in 'sep'. An array |
792 | * of sds strings is returned. *count will be set |
793 | * by reference to the number of tokens returned. |
794 | * |
795 | * On out of memory, zero length string, zero length |
796 | * separator, NULL is returned. |
797 | * |
798 | * Note that 'sep' is able to split a string using |
799 | * a multi-character separator. For example |
800 | * sdssplit("foo_-_bar","_-_"); will return two |
801 | * elements "foo" and "bar". |
802 | * |
803 | * This version of the function is binary-safe but |
804 | * requires length arguments. sdssplit() is just the |
805 | * same function but for zero-terminated strings. |
806 | */ |
807 | sds *sdssplitlen(const char *s, ssize_t len, const char *sep, int seplen, int *count) { |
808 | int elements = 0, slots = 5; |
809 | long start = 0, j; |
810 | sds *tokens; |
811 | |
812 | if (seplen < 1 || len < 0) return NULL; |
813 | |
814 | tokens = (sds*) malloc(sizeof(sds)*slots); |
815 | if (tokens == NULL) return NULL; |
816 | |
817 | if (len == 0) { |
818 | *count = 0; |
819 | return tokens; |
820 | } |
821 | for (j = 0; j < (len-(seplen-1)); j++) { |
822 | /* make sure there is room for the next element and the final one */ |
823 | if (slots < elements+2) { |
824 | sds *newtokens; |
825 | |
826 | slots *= 2; |
827 | newtokens = (sds*) realloc(tokens,sizeof(sds)*slots); |
828 | if (newtokens == NULL) goto cleanup; |
829 | tokens = newtokens; |
830 | } |
831 | /* search the separator */ |
832 | if ((seplen == 1 && *(s+j) == sep[0]) || (memcmp(s+j,sep,seplen) == 0)) { |
833 | tokens[elements] = sdsnewlen(s+start,j-start); |
834 | if (tokens[elements] == NULL) goto cleanup; |
835 | elements++; |
836 | start = j+seplen; |
837 | j = j+seplen-1; /* skip the separator */ |
838 | } |
839 | } |
840 | /* Add the final element. We are sure there is room in the tokens array. */ |
841 | tokens[elements] = sdsnewlen(s+start,len-start); |
842 | if (tokens[elements] == NULL) goto cleanup; |
843 | elements++; |
844 | *count = elements; |
845 | return tokens; |
846 | |
847 | cleanup: |
848 | { |
849 | int i; |
850 | for (i = 0; i < elements; i++) sdsfree(tokens[i]); |
851 | free(tokens); |
852 | *count = 0; |
853 | return NULL; |
854 | } |
855 | } |
856 | |
857 | /* Free the result returned by sdssplitlen(), or do nothing if 'tokens' is NULL. */ |
858 | void sdsfreesplitres(sds *tokens, int count) { |
859 | if (!tokens) return; |
860 | while(count--) |
861 | sdsfree(tokens[count]); |
862 | free(tokens); |
863 | } |
864 | |
865 | /* Append to the sds string "s" an escaped string representation where |
866 | * all the non-printable characters (tested with isprint()) are turned into |
867 | * escapes in the form "\n\r\a...." or "\x<hex-number>". |
868 | * |
869 | * After the call, the modified sds string is no longer valid and all the |
870 | * references must be substituted with the new pointer returned by the call. */ |
871 | sds sdscatrepr(sds s, const char *p, size_t len) { |
872 | s = sdscatlen(s,"\"" ,1); |
873 | while(len--) { |
874 | switch(*p) { |
875 | case '\\': |
876 | case '"': |
877 | s = sdscatprintf(s,"\\%c" ,*p); |
878 | break; |
879 | case '\n': s = sdscatlen(s,"\\n" ,2); break; |
880 | case '\r': s = sdscatlen(s,"\\r" ,2); break; |
881 | case '\t': s = sdscatlen(s,"\\t" ,2); break; |
882 | case '\a': s = sdscatlen(s,"\\a" ,2); break; |
883 | case '\b': s = sdscatlen(s,"\\b" ,2); break; |
884 | default: |
885 | if (isprint(*p)) |
886 | s = sdscatprintf(s,"%c" ,*p); |
887 | else |
888 | s = sdscatprintf(s,"\\x%02x" ,(unsigned char)*p); |
889 | break; |
890 | } |
891 | p++; |
892 | } |
893 | return sdscatlen(s,"\"" ,1); |
894 | } |
895 | |
896 | /* Helper function for sdssplitargs() that returns non zero if 'c' |
897 | * is a valid hex digit. */ |
898 | int is_hex_digit(char c) { |
899 | return (c >= '0' && c <= '9') || (c >= 'a' && c <= 'f') || |
900 | (c >= 'A' && c <= 'F'); |
901 | } |
902 | |
903 | /* Helper function for sdssplitargs() that converts a hex digit into an |
904 | * integer from 0 to 15 */ |
905 | int hex_digit_to_int(char c) { |
906 | switch(c) { |
907 | case '0': return 0; |
908 | case '1': return 1; |
909 | case '2': return 2; |
910 | case '3': return 3; |
911 | case '4': return 4; |
912 | case '5': return 5; |
913 | case '6': return 6; |
914 | case '7': return 7; |
915 | case '8': return 8; |
916 | case '9': return 9; |
917 | case 'a': case 'A': return 10; |
918 | case 'b': case 'B': return 11; |
919 | case 'c': case 'C': return 12; |
920 | case 'd': case 'D': return 13; |
921 | case 'e': case 'E': return 14; |
922 | case 'f': case 'F': return 15; |
923 | default: return 0; |
924 | } |
925 | } |
926 | |
927 | /* Split a line into arguments, where every argument can be in the |
928 | * following programming-language REPL-alike form: |
929 | * |
930 | * foo bar "newline are supported\n" and "\xff\x00otherstuff" |
931 | * |
932 | * The number of arguments is stored into *argc, and an array |
933 | * of sds is returned. |
934 | * |
935 | * The caller should free the resulting array of sds strings with |
936 | * sdsfreesplitres(). |
937 | * |
938 | * Note that sdscatrepr() is able to convert back a string into |
939 | * a quoted string in the same format sdssplitargs() is able to parse. |
940 | * |
941 | * The function returns the allocated tokens on success, even when the |
942 | * input string is empty, or NULL if the input contains unbalanced |
943 | * quotes or closed quotes followed by non space characters |
944 | * as in: "foo"bar or "foo' |
945 | */ |
946 | sds *sdssplitargs(const char *line, int *argc) { |
947 | const char *p = line; |
948 | char *current = NULL; |
949 | char **vector = NULL; |
950 | |
951 | *argc = 0; |
952 | while(1) { |
953 | /* skip blanks */ |
954 | while(*p && isspace(*p)) p++; |
955 | if (*p) { |
956 | /* get a token */ |
957 | int inq=0; /* set to 1 if we are in "quotes" */ |
958 | int insq=0; /* set to 1 if we are in 'single quotes' */ |
959 | int done=0; |
960 | |
961 | if (current == NULL) current = sdsempty(); |
962 | while(!done) { |
963 | if (inq) { |
964 | if (*p == '\\' && *(p+1) == 'x' && |
965 | is_hex_digit(*(p+2)) && |
966 | is_hex_digit(*(p+3))) |
967 | { |
968 | unsigned char byte; |
969 | |
970 | byte = (hex_digit_to_int(*(p+2))*16)+ |
971 | hex_digit_to_int(*(p+3)); |
972 | current = sdscatlen(current,(char*)&byte,1); |
973 | p += 3; |
974 | } else if (*p == '\\' && *(p+1)) { |
975 | char c; |
976 | |
977 | p++; |
978 | switch(*p) { |
979 | case 'n': c = '\n'; break; |
980 | case 'r': c = '\r'; break; |
981 | case 't': c = '\t'; break; |
982 | case 'b': c = '\b'; break; |
983 | case 'a': c = '\a'; break; |
984 | default: c = *p; break; |
985 | } |
986 | current = sdscatlen(current,&c,1); |
987 | } else if (*p == '"') { |
988 | /* closing quote must be followed by a space or |
989 | * nothing at all. */ |
990 | if (*(p+1) && !isspace(*(p+1))) goto err; |
991 | done=1; |
992 | } else if (!*p) { |
993 | /* unterminated quotes */ |
994 | goto err; |
995 | } else { |
996 | current = sdscatlen(current,p,1); |
997 | } |
998 | } else if (insq) { |
999 | if (*p == '\\' && *(p+1) == '\'') { |
1000 | p++; |
1001 | current = sdscatlen(current,"'" ,1); |
1002 | } else if (*p == '\'') { |
1003 | /* closing quote must be followed by a space or |
1004 | * nothing at all. */ |
1005 | if (*(p+1) && !isspace(*(p+1))) goto err; |
1006 | done=1; |
1007 | } else if (!*p) { |
1008 | /* unterminated quotes */ |
1009 | goto err; |
1010 | } else { |
1011 | current = sdscatlen(current,p,1); |
1012 | } |
1013 | } else { |
1014 | switch(*p) { |
1015 | case ' ': |
1016 | case '\n': |
1017 | case '\r': |
1018 | case '\t': |
1019 | case '\0': |
1020 | done=1; |
1021 | break; |
1022 | case '"': |
1023 | inq=1; |
1024 | break; |
1025 | case '\'': |
1026 | insq=1; |
1027 | break; |
1028 | default: |
1029 | current = sdscatlen(current,p,1); |
1030 | break; |
1031 | } |
1032 | } |
1033 | if (*p) p++; |
1034 | } |
1035 | /* add the token to the vector */ |
1036 | vector = (char**) realloc(vector,((*argc)+1)*sizeof(char*)); |
1037 | vector[*argc] = current; |
1038 | (*argc)++; |
1039 | current = NULL; |
1040 | } else { |
1041 | /* Even on empty input string return something not NULL. */ |
1042 | if (vector == NULL) vector = (char**) malloc(sizeof(void*)); |
1043 | return vector; |
1044 | } |
1045 | } |
1046 | |
1047 | err: |
1048 | while((*argc)--) |
1049 | sdsfree(vector[*argc]); |
1050 | free(vector); |
1051 | if (current) sdsfree(current); |
1052 | *argc = 0; |
1053 | return NULL; |
1054 | } |
1055 | |
1056 | /* Modify the string substituting all the occurrences of the set of |
1057 | * characters specified in the 'from' string to the corresponding character |
1058 | * in the 'to' array. |
1059 | * |
1060 | * For instance: sdsmapchars(mystring, "ho", "01", 2) |
1061 | * will have the effect of turning the string "hello" into "0ell1". |
1062 | * |
1063 | * The function returns the sds string pointer, that is always the same |
1064 | * as the input pointer since no resize is needed. */ |
1065 | sds sdsmapchars(sds s, const char *from, const char *to, size_t setlen) { |
1066 | size_t j, i, l = sdslen(s); |
1067 | |
1068 | for (j = 0; j < l; j++) { |
1069 | for (i = 0; i < setlen; i++) { |
1070 | if (s[j] == from[i]) { |
1071 | s[j] = to[i]; |
1072 | break; |
1073 | } |
1074 | } |
1075 | } |
1076 | return s; |
1077 | } |
1078 | |
1079 | /* Join an array of C strings using the specified separator (also a C string). |
1080 | * Returns the result as an sds string. */ |
1081 | sds sdsjoin(char **argv, int argc, char *sep) { |
1082 | sds join = sdsempty(); |
1083 | int j; |
1084 | |
1085 | for (j = 0; j < argc; j++) { |
1086 | join = sdscat(join, argv[j]); |
1087 | if (j != argc-1) join = sdscat(join,sep); |
1088 | } |
1089 | return join; |
1090 | } |
1091 | |
1092 | /* Like sdsjoin, but joins an array of SDS strings. */ |
1093 | sds sdsjoinsds(sds *argv, int argc, const char *sep, size_t seplen) { |
1094 | sds join = sdsempty(); |
1095 | int j; |
1096 | |
1097 | for (j = 0; j < argc; j++) { |
1098 | join = sdscatsds(join, argv[j]); |
1099 | if (j != argc-1) join = sdscatlen(join,sep,seplen); |
1100 | } |
1101 | return join; |
1102 | } |
1103 | |
1104 | /* Wrappers to the allocators used by SDS. Note that SDS will actually |
1105 | * just use the macros defined into sdsalloc.h in order to avoid to pay |
1106 | * the overhead of function calls. Here we define these wrappers only for |
1107 | * the programs SDS is linked to, if they want to touch the SDS internals |
1108 | * even if they use a different allocator. */ |
1109 | void *sdmalloc(size_t size) { return malloc(size); } |
1110 | void *sdrealloc(void *ptr, size_t size) { return realloc(ptr,size); } |
1111 | void sdfree(void *ptr) { free(ptr); } |
1112 | |