doc: use 'symmetric difference' consistently
[git/debian.git] / strbuf.c
blob958387562184fe27c7e476ebb24d5547a55ae4ad
1 #include "cache.h"
2 #include "refs.h"
3 #include "utf8.h"
5 int starts_with(const char *str, const char *prefix)
7 for (; ; str++, prefix++)
8 if (!*prefix)
9 return 1;
10 else if (*str != *prefix)
11 return 0;
15 * Used as the default ->buf value, so that people can always assume
16 * buf is non NULL and ->buf is NUL terminated even for a freshly
17 * initialized strbuf.
19 char strbuf_slopbuf[1];
21 void strbuf_init(struct strbuf *sb, size_t hint)
23 sb->alloc = sb->len = 0;
24 sb->buf = strbuf_slopbuf;
25 if (hint)
26 strbuf_grow(sb, hint);
29 void strbuf_release(struct strbuf *sb)
31 if (sb->alloc) {
32 free(sb->buf);
33 strbuf_init(sb, 0);
37 char *strbuf_detach(struct strbuf *sb, size_t *sz)
39 char *res;
40 strbuf_grow(sb, 0);
41 res = sb->buf;
42 if (sz)
43 *sz = sb->len;
44 strbuf_init(sb, 0);
45 return res;
48 void strbuf_attach(struct strbuf *sb, void *buf, size_t len, size_t alloc)
50 strbuf_release(sb);
51 sb->buf = buf;
52 sb->len = len;
53 sb->alloc = alloc;
54 strbuf_grow(sb, 0);
55 sb->buf[sb->len] = '\0';
58 void strbuf_grow(struct strbuf *sb, size_t extra)
60 int new_buf = !sb->alloc;
61 if (unsigned_add_overflows(extra, 1) ||
62 unsigned_add_overflows(sb->len, extra + 1))
63 die("you want to use way too much memory");
64 if (new_buf)
65 sb->buf = NULL;
66 ALLOC_GROW(sb->buf, sb->len + extra + 1, sb->alloc);
67 if (new_buf)
68 sb->buf[0] = '\0';
71 void strbuf_trim(struct strbuf *sb)
73 strbuf_rtrim(sb);
74 strbuf_ltrim(sb);
76 void strbuf_rtrim(struct strbuf *sb)
78 while (sb->len > 0 && isspace((unsigned char)sb->buf[sb->len - 1]))
79 sb->len--;
80 sb->buf[sb->len] = '\0';
83 void strbuf_ltrim(struct strbuf *sb)
85 char *b = sb->buf;
86 while (sb->len > 0 && isspace(*b)) {
87 b++;
88 sb->len--;
90 memmove(sb->buf, b, sb->len);
91 sb->buf[sb->len] = '\0';
94 int strbuf_reencode(struct strbuf *sb, const char *from, const char *to)
96 char *out;
97 int len;
99 if (same_encoding(from, to))
100 return 0;
102 out = reencode_string_len(sb->buf, sb->len, to, from, &len);
103 if (!out)
104 return -1;
106 strbuf_attach(sb, out, len, len);
107 return 0;
110 void strbuf_tolower(struct strbuf *sb)
112 char *p = sb->buf, *end = sb->buf + sb->len;
113 for (; p < end; p++)
114 *p = tolower(*p);
117 struct strbuf **strbuf_split_buf(const char *str, size_t slen,
118 int terminator, int max)
120 struct strbuf **ret = NULL;
121 size_t nr = 0, alloc = 0;
122 struct strbuf *t;
124 while (slen) {
125 int len = slen;
126 if (max <= 0 || nr + 1 < max) {
127 const char *end = memchr(str, terminator, slen);
128 if (end)
129 len = end - str + 1;
131 t = xmalloc(sizeof(struct strbuf));
132 strbuf_init(t, len);
133 strbuf_add(t, str, len);
134 ALLOC_GROW(ret, nr + 2, alloc);
135 ret[nr++] = t;
136 str += len;
137 slen -= len;
139 ALLOC_GROW(ret, nr + 1, alloc); /* In case string was empty */
140 ret[nr] = NULL;
141 return ret;
144 void strbuf_list_free(struct strbuf **sbs)
146 struct strbuf **s = sbs;
148 while (*s) {
149 strbuf_release(*s);
150 free(*s++);
152 free(sbs);
155 int strbuf_cmp(const struct strbuf *a, const struct strbuf *b)
157 int len = a->len < b->len ? a->len: b->len;
158 int cmp = memcmp(a->buf, b->buf, len);
159 if (cmp)
160 return cmp;
161 return a->len < b->len ? -1: a->len != b->len;
164 void strbuf_splice(struct strbuf *sb, size_t pos, size_t len,
165 const void *data, size_t dlen)
167 if (unsigned_add_overflows(pos, len))
168 die("you want to use way too much memory");
169 if (pos > sb->len)
170 die("`pos' is too far after the end of the buffer");
171 if (pos + len > sb->len)
172 die("`pos + len' is too far after the end of the buffer");
174 if (dlen >= len)
175 strbuf_grow(sb, dlen - len);
176 memmove(sb->buf + pos + dlen,
177 sb->buf + pos + len,
178 sb->len - pos - len);
179 memcpy(sb->buf + pos, data, dlen);
180 strbuf_setlen(sb, sb->len + dlen - len);
183 void strbuf_insert(struct strbuf *sb, size_t pos, const void *data, size_t len)
185 strbuf_splice(sb, pos, 0, data, len);
188 void strbuf_remove(struct strbuf *sb, size_t pos, size_t len)
190 strbuf_splice(sb, pos, len, NULL, 0);
193 void strbuf_add(struct strbuf *sb, const void *data, size_t len)
195 strbuf_grow(sb, len);
196 memcpy(sb->buf + sb->len, data, len);
197 strbuf_setlen(sb, sb->len + len);
200 void strbuf_adddup(struct strbuf *sb, size_t pos, size_t len)
202 strbuf_grow(sb, len);
203 memcpy(sb->buf + sb->len, sb->buf + pos, len);
204 strbuf_setlen(sb, sb->len + len);
207 void strbuf_addchars(struct strbuf *sb, int c, size_t n)
209 strbuf_grow(sb, n);
210 memset(sb->buf + sb->len, c, n);
211 strbuf_setlen(sb, sb->len + n);
214 void strbuf_addf(struct strbuf *sb, const char *fmt, ...)
216 va_list ap;
217 va_start(ap, fmt);
218 strbuf_vaddf(sb, fmt, ap);
219 va_end(ap);
222 static void add_lines(struct strbuf *out,
223 const char *prefix1,
224 const char *prefix2,
225 const char *buf, size_t size)
227 while (size) {
228 const char *prefix;
229 const char *next = memchr(buf, '\n', size);
230 next = next ? (next + 1) : (buf + size);
232 prefix = ((prefix2 && (buf[0] == '\n' || buf[0] == '\t'))
233 ? prefix2 : prefix1);
234 strbuf_addstr(out, prefix);
235 strbuf_add(out, buf, next - buf);
236 size -= next - buf;
237 buf = next;
239 strbuf_complete_line(out);
242 void strbuf_add_commented_lines(struct strbuf *out, const char *buf, size_t size)
244 static char prefix1[3];
245 static char prefix2[2];
247 if (prefix1[0] != comment_line_char) {
248 sprintf(prefix1, "%c ", comment_line_char);
249 sprintf(prefix2, "%c", comment_line_char);
251 add_lines(out, prefix1, prefix2, buf, size);
254 void strbuf_commented_addf(struct strbuf *sb, const char *fmt, ...)
256 va_list params;
257 struct strbuf buf = STRBUF_INIT;
258 int incomplete_line = sb->len && sb->buf[sb->len - 1] != '\n';
260 va_start(params, fmt);
261 strbuf_vaddf(&buf, fmt, params);
262 va_end(params);
264 strbuf_add_commented_lines(sb, buf.buf, buf.len);
265 if (incomplete_line)
266 sb->buf[--sb->len] = '\0';
268 strbuf_release(&buf);
271 void strbuf_vaddf(struct strbuf *sb, const char *fmt, va_list ap)
273 int len;
274 va_list cp;
276 if (!strbuf_avail(sb))
277 strbuf_grow(sb, 64);
278 va_copy(cp, ap);
279 len = vsnprintf(sb->buf + sb->len, sb->alloc - sb->len, fmt, cp);
280 va_end(cp);
281 if (len < 0)
282 die("BUG: your vsnprintf is broken (returned %d)", len);
283 if (len > strbuf_avail(sb)) {
284 strbuf_grow(sb, len);
285 len = vsnprintf(sb->buf + sb->len, sb->alloc - sb->len, fmt, ap);
286 if (len > strbuf_avail(sb))
287 die("BUG: your vsnprintf is broken (insatiable)");
289 strbuf_setlen(sb, sb->len + len);
292 void strbuf_expand(struct strbuf *sb, const char *format, expand_fn_t fn,
293 void *context)
295 for (;;) {
296 const char *percent;
297 size_t consumed;
299 percent = strchrnul(format, '%');
300 strbuf_add(sb, format, percent - format);
301 if (!*percent)
302 break;
303 format = percent + 1;
305 if (*format == '%') {
306 strbuf_addch(sb, '%');
307 format++;
308 continue;
311 consumed = fn(sb, format, context);
312 if (consumed)
313 format += consumed;
314 else
315 strbuf_addch(sb, '%');
319 size_t strbuf_expand_dict_cb(struct strbuf *sb, const char *placeholder,
320 void *context)
322 struct strbuf_expand_dict_entry *e = context;
323 size_t len;
325 for (; e->placeholder && (len = strlen(e->placeholder)); e++) {
326 if (!strncmp(placeholder, e->placeholder, len)) {
327 if (e->value)
328 strbuf_addstr(sb, e->value);
329 return len;
332 return 0;
335 void strbuf_addbuf_percentquote(struct strbuf *dst, const struct strbuf *src)
337 int i, len = src->len;
339 for (i = 0; i < len; i++) {
340 if (src->buf[i] == '%')
341 strbuf_addch(dst, '%');
342 strbuf_addch(dst, src->buf[i]);
346 size_t strbuf_fread(struct strbuf *sb, size_t size, FILE *f)
348 size_t res;
349 size_t oldalloc = sb->alloc;
351 strbuf_grow(sb, size);
352 res = fread(sb->buf + sb->len, 1, size, f);
353 if (res > 0)
354 strbuf_setlen(sb, sb->len + res);
355 else if (oldalloc == 0)
356 strbuf_release(sb);
357 return res;
360 ssize_t strbuf_read(struct strbuf *sb, int fd, size_t hint)
362 size_t oldlen = sb->len;
363 size_t oldalloc = sb->alloc;
365 strbuf_grow(sb, hint ? hint : 8192);
366 for (;;) {
367 ssize_t want = sb->alloc - sb->len - 1;
368 ssize_t got = read_in_full(fd, sb->buf + sb->len, want);
370 if (got < 0) {
371 if (oldalloc == 0)
372 strbuf_release(sb);
373 else
374 strbuf_setlen(sb, oldlen);
375 return -1;
377 sb->len += got;
378 if (got < want)
379 break;
380 strbuf_grow(sb, 8192);
383 sb->buf[sb->len] = '\0';
384 return sb->len - oldlen;
387 #define STRBUF_MAXLINK (2*PATH_MAX)
389 int strbuf_readlink(struct strbuf *sb, const char *path, size_t hint)
391 size_t oldalloc = sb->alloc;
393 if (hint < 32)
394 hint = 32;
396 while (hint < STRBUF_MAXLINK) {
397 int len;
399 strbuf_grow(sb, hint);
400 len = readlink(path, sb->buf, hint);
401 if (len < 0) {
402 if (errno != ERANGE)
403 break;
404 } else if (len < hint) {
405 strbuf_setlen(sb, len);
406 return 0;
409 /* .. the buffer was too small - try again */
410 hint *= 2;
412 if (oldalloc == 0)
413 strbuf_release(sb);
414 return -1;
417 int strbuf_getcwd(struct strbuf *sb)
419 size_t oldalloc = sb->alloc;
420 size_t guessed_len = 128;
422 for (;; guessed_len *= 2) {
423 strbuf_grow(sb, guessed_len);
424 if (getcwd(sb->buf, sb->alloc)) {
425 strbuf_setlen(sb, strlen(sb->buf));
426 return 0;
428 if (errno != ERANGE)
429 break;
431 if (oldalloc == 0)
432 strbuf_release(sb);
433 else
434 strbuf_reset(sb);
435 return -1;
438 #ifdef HAVE_GETDELIM
439 int strbuf_getwholeline(struct strbuf *sb, FILE *fp, int term)
441 ssize_t r;
443 if (feof(fp))
444 return EOF;
446 strbuf_reset(sb);
448 /* Translate slopbuf to NULL, as we cannot call realloc on it */
449 if (!sb->alloc)
450 sb->buf = NULL;
451 r = getdelim(&sb->buf, &sb->alloc, term, fp);
453 if (r > 0) {
454 sb->len = r;
455 return 0;
457 assert(r == -1);
460 * Normally we would have called xrealloc, which will try to free
461 * memory and recover. But we have no way to tell getdelim() to do so.
462 * Worse, we cannot try to recover ENOMEM ourselves, because we have
463 * no idea how many bytes were read by getdelim.
465 * Dying here is reasonable. It mirrors what xrealloc would do on
466 * catastrophic memory failure. We skip the opportunity to free pack
467 * memory and retry, but that's unlikely to help for a malloc small
468 * enough to hold a single line of input, anyway.
470 if (errno == ENOMEM)
471 die("Out of memory, getdelim failed");
473 /* Restore slopbuf that we moved out of the way before */
474 if (!sb->buf)
475 strbuf_init(sb, 0);
476 return EOF;
478 #else
479 int strbuf_getwholeline(struct strbuf *sb, FILE *fp, int term)
481 int ch;
483 if (feof(fp))
484 return EOF;
486 strbuf_reset(sb);
487 flockfile(fp);
488 while ((ch = getc_unlocked(fp)) != EOF) {
489 if (!strbuf_avail(sb))
490 strbuf_grow(sb, 1);
491 sb->buf[sb->len++] = ch;
492 if (ch == term)
493 break;
495 funlockfile(fp);
496 if (ch == EOF && sb->len == 0)
497 return EOF;
499 sb->buf[sb->len] = '\0';
500 return 0;
502 #endif
504 int strbuf_getline(struct strbuf *sb, FILE *fp, int term)
506 if (strbuf_getwholeline(sb, fp, term))
507 return EOF;
508 if (sb->buf[sb->len-1] == term)
509 strbuf_setlen(sb, sb->len-1);
510 return 0;
513 int strbuf_getwholeline_fd(struct strbuf *sb, int fd, int term)
515 strbuf_reset(sb);
517 while (1) {
518 char ch;
519 ssize_t len = xread(fd, &ch, 1);
520 if (len <= 0)
521 return EOF;
522 strbuf_addch(sb, ch);
523 if (ch == term)
524 break;
526 return 0;
529 ssize_t strbuf_read_file(struct strbuf *sb, const char *path, size_t hint)
531 int fd;
532 ssize_t len;
534 fd = open(path, O_RDONLY);
535 if (fd < 0)
536 return -1;
537 len = strbuf_read(sb, fd, hint);
538 close(fd);
539 if (len < 0)
540 return -1;
542 return len;
545 void strbuf_add_lines(struct strbuf *out, const char *prefix,
546 const char *buf, size_t size)
548 add_lines(out, prefix, NULL, buf, size);
551 void strbuf_addstr_xml_quoted(struct strbuf *buf, const char *s)
553 while (*s) {
554 size_t len = strcspn(s, "\"<>&");
555 strbuf_add(buf, s, len);
556 s += len;
557 switch (*s) {
558 case '"':
559 strbuf_addstr(buf, "&quot;");
560 break;
561 case '<':
562 strbuf_addstr(buf, "&lt;");
563 break;
564 case '>':
565 strbuf_addstr(buf, "&gt;");
566 break;
567 case '&':
568 strbuf_addstr(buf, "&amp;");
569 break;
570 case 0:
571 return;
573 s++;
577 static int is_rfc3986_reserved(char ch)
579 switch (ch) {
580 case '!': case '*': case '\'': case '(': case ')': case ';':
581 case ':': case '@': case '&': case '=': case '+': case '$':
582 case ',': case '/': case '?': case '#': case '[': case ']':
583 return 1;
585 return 0;
588 static int is_rfc3986_unreserved(char ch)
590 return isalnum(ch) ||
591 ch == '-' || ch == '_' || ch == '.' || ch == '~';
594 static void strbuf_add_urlencode(struct strbuf *sb, const char *s, size_t len,
595 int reserved)
597 strbuf_grow(sb, len);
598 while (len--) {
599 char ch = *s++;
600 if (is_rfc3986_unreserved(ch) ||
601 (!reserved && is_rfc3986_reserved(ch)))
602 strbuf_addch(sb, ch);
603 else
604 strbuf_addf(sb, "%%%02x", ch);
608 void strbuf_addstr_urlencode(struct strbuf *sb, const char *s,
609 int reserved)
611 strbuf_add_urlencode(sb, s, strlen(s), reserved);
614 void strbuf_humanise_bytes(struct strbuf *buf, off_t bytes)
616 if (bytes > 1 << 30) {
617 strbuf_addf(buf, "%u.%2.2u GiB",
618 (int)(bytes >> 30),
619 (int)(bytes & ((1 << 30) - 1)) / 10737419);
620 } else if (bytes > 1 << 20) {
621 int x = bytes + 5243; /* for rounding */
622 strbuf_addf(buf, "%u.%2.2u MiB",
623 x >> 20, ((x & ((1 << 20) - 1)) * 100) >> 20);
624 } else if (bytes > 1 << 10) {
625 int x = bytes + 5; /* for rounding */
626 strbuf_addf(buf, "%u.%2.2u KiB",
627 x >> 10, ((x & ((1 << 10) - 1)) * 100) >> 10);
628 } else {
629 strbuf_addf(buf, "%u bytes", (int)bytes);
633 void strbuf_add_absolute_path(struct strbuf *sb, const char *path)
635 if (!*path)
636 die("The empty string is not a valid path");
637 if (!is_absolute_path(path)) {
638 struct stat cwd_stat, pwd_stat;
639 size_t orig_len = sb->len;
640 char *cwd = xgetcwd();
641 char *pwd = getenv("PWD");
642 if (pwd && strcmp(pwd, cwd) &&
643 !stat(cwd, &cwd_stat) &&
644 (cwd_stat.st_dev || cwd_stat.st_ino) &&
645 !stat(pwd, &pwd_stat) &&
646 pwd_stat.st_dev == cwd_stat.st_dev &&
647 pwd_stat.st_ino == cwd_stat.st_ino)
648 strbuf_addstr(sb, pwd);
649 else
650 strbuf_addstr(sb, cwd);
651 if (sb->len > orig_len && !is_dir_sep(sb->buf[sb->len - 1]))
652 strbuf_addch(sb, '/');
653 free(cwd);
655 strbuf_addstr(sb, path);
658 int printf_ln(const char *fmt, ...)
660 int ret;
661 va_list ap;
662 va_start(ap, fmt);
663 ret = vprintf(fmt, ap);
664 va_end(ap);
665 if (ret < 0 || putchar('\n') == EOF)
666 return -1;
667 return ret + 1;
670 int fprintf_ln(FILE *fp, const char *fmt, ...)
672 int ret;
673 va_list ap;
674 va_start(ap, fmt);
675 ret = vfprintf(fp, fmt, ap);
676 va_end(ap);
677 if (ret < 0 || putc('\n', fp) == EOF)
678 return -1;
679 return ret + 1;
682 char *xstrdup_tolower(const char *string)
684 char *result;
685 size_t len, i;
687 len = strlen(string);
688 result = xmalloc(len + 1);
689 for (i = 0; i < len; i++)
690 result[i] = tolower(string[i]);
691 result[i] = '\0';
692 return result;
695 char *xstrvfmt(const char *fmt, va_list ap)
697 struct strbuf buf = STRBUF_INIT;
698 strbuf_vaddf(&buf, fmt, ap);
699 return strbuf_detach(&buf, NULL);
702 char *xstrfmt(const char *fmt, ...)
704 va_list ap;
705 char *ret;
707 va_start(ap, fmt);
708 ret = xstrvfmt(fmt, ap);
709 va_end(ap);
711 return ret;
714 void strbuf_addftime(struct strbuf *sb, const char *fmt, const struct tm *tm)
716 size_t hint = 128;
717 size_t len;
719 if (!*fmt)
720 return;
722 strbuf_grow(sb, hint);
723 len = strftime(sb->buf + sb->len, sb->alloc - sb->len, fmt, tm);
725 if (!len) {
727 * strftime reports "0" if it could not fit the result in the buffer.
728 * Unfortunately, it also reports "0" if the requested time string
729 * takes 0 bytes. So our strategy is to munge the format so that the
730 * output contains at least one character, and then drop the extra
731 * character before returning.
733 struct strbuf munged_fmt = STRBUF_INIT;
734 strbuf_addf(&munged_fmt, "%s ", fmt);
735 while (!len) {
736 hint *= 2;
737 strbuf_grow(sb, hint);
738 len = strftime(sb->buf + sb->len, sb->alloc - sb->len,
739 munged_fmt.buf, tm);
741 strbuf_release(&munged_fmt);
742 len--; /* drop munged space */
744 strbuf_setlen(sb, sb->len + len);
748 * Returns the length of a line, without trailing spaces.
750 * If the line ends with newline, it will be removed too.
752 static size_t cleanup(char *line, size_t len)
754 while (len) {
755 unsigned char c = line[len - 1];
756 if (!isspace(c))
757 break;
758 len--;
761 return len;
765 * Remove empty lines from the beginning and end
766 * and also trailing spaces from every line.
768 * Turn multiple consecutive empty lines between paragraphs
769 * into just one empty line.
771 * If the input has only empty lines and spaces,
772 * no output will be produced.
774 * If last line does not have a newline at the end, one is added.
776 * Enable skip_comments to skip every line starting with comment
777 * character.
779 void strbuf_stripspace(struct strbuf *sb, int skip_comments)
781 int empties = 0;
782 size_t i, j, len, newlen;
783 char *eol;
785 /* We may have to add a newline. */
786 strbuf_grow(sb, 1);
788 for (i = j = 0; i < sb->len; i += len, j += newlen) {
789 eol = memchr(sb->buf + i, '\n', sb->len - i);
790 len = eol ? eol - (sb->buf + i) + 1 : sb->len - i;
792 if (skip_comments && len && sb->buf[i] == comment_line_char) {
793 newlen = 0;
794 continue;
796 newlen = cleanup(sb->buf + i, len);
798 /* Not just an empty line? */
799 if (newlen) {
800 if (empties > 0 && j > 0)
801 sb->buf[j++] = '\n';
802 empties = 0;
803 memmove(sb->buf + j, sb->buf + i, newlen);
804 sb->buf[newlen + j++] = '\n';
805 } else {
806 empties++;
810 strbuf_setlen(sb, j);