1 /* SPDX-License-Identifier: LGPL-2.1-or-later */
2 
3 #include <errno.h>
4 #include <stdarg.h>
5 #include <stdint.h>
6 #include <stdio.h>
7 #include <stdlib.h>
8 
9 #include "alloc-util.h"
10 #include "escape.h"
11 #include "extract-word.h"
12 #include "fileio.h"
13 #include "gunicode.h"
14 #include "locale-util.h"
15 #include "macro.h"
16 #include "memory-util.h"
17 #include "string-util.h"
18 #include "strv.h"
19 #include "terminal-util.h"
20 #include "utf8.h"
21 #include "util.h"
22 
first_word(const char * s,const char * word)23 char* first_word(const char *s, const char *word) {
24         size_t sl, wl;
25         const char *p;
26 
27         assert(s);
28         assert(word);
29 
30         /* Checks if the string starts with the specified word, either
31          * followed by NUL or by whitespace. Returns a pointer to the
32          * NUL or the first character after the whitespace. */
33 
34         sl = strlen(s);
35         wl = strlen(word);
36 
37         if (sl < wl)
38                 return NULL;
39 
40         if (wl == 0)
41                 return (char*) s;
42 
43         if (memcmp(s, word, wl) != 0)
44                 return NULL;
45 
46         p = s + wl;
47         if (*p == 0)
48                 return (char*) p;
49 
50         if (!strchr(WHITESPACE, *p))
51                 return NULL;
52 
53         p += strspn(p, WHITESPACE);
54         return (char*) p;
55 }
56 
strnappend(const char * s,const char * suffix,size_t b)57 char *strnappend(const char *s, const char *suffix, size_t b) {
58         size_t a;
59         char *r;
60 
61         if (!s && !suffix)
62                 return strdup("");
63 
64         if (!s)
65                 return strndup(suffix, b);
66 
67         if (!suffix)
68                 return strdup(s);
69 
70         assert(s);
71         assert(suffix);
72 
73         a = strlen(s);
74         if (b > SIZE_MAX - a)
75                 return NULL;
76 
77         r = new(char, a+b+1);
78         if (!r)
79                 return NULL;
80 
81         memcpy(r, s, a);
82         memcpy(r+a, suffix, b);
83         r[a+b] = 0;
84 
85         return r;
86 }
87 
strjoin_real(const char * x,...)88 char *strjoin_real(const char *x, ...) {
89         va_list ap;
90         size_t l = 1;
91         char *r, *p;
92 
93         va_start(ap, x);
94         for (const char *t = x; t; t = va_arg(ap, const char *)) {
95                 size_t n;
96 
97                 n = strlen(t);
98                 if (n > SIZE_MAX - l) {
99                         va_end(ap);
100                         return NULL;
101                 }
102                 l += n;
103         }
104         va_end(ap);
105 
106         p = r = new(char, l);
107         if (!r)
108                 return NULL;
109 
110         va_start(ap, x);
111         for (const char *t = x; t; t = va_arg(ap, const char *))
112                 p = stpcpy(p, t);
113         va_end(ap);
114 
115         *p = 0;
116 
117         return r;
118 }
119 
strstrip(char * s)120 char *strstrip(char *s) {
121         if (!s)
122                 return NULL;
123 
124         /* Drops trailing whitespace. Modifies the string in place. Returns pointer to first non-space character */
125 
126         return delete_trailing_chars(skip_leading_chars(s, WHITESPACE), WHITESPACE);
127 }
128 
delete_chars(char * s,const char * bad)129 char *delete_chars(char *s, const char *bad) {
130         char *f, *t;
131 
132         /* Drops all specified bad characters, regardless where in the string */
133 
134         if (!s)
135                 return NULL;
136 
137         if (!bad)
138                 bad = WHITESPACE;
139 
140         for (f = s, t = s; *f; f++) {
141                 if (strchr(bad, *f))
142                         continue;
143 
144                 *(t++) = *f;
145         }
146 
147         *t = 0;
148 
149         return s;
150 }
151 
delete_trailing_chars(char * s,const char * bad)152 char *delete_trailing_chars(char *s, const char *bad) {
153         char *c = s;
154 
155         /* Drops all specified bad characters, at the end of the string */
156 
157         if (!s)
158                 return NULL;
159 
160         if (!bad)
161                 bad = WHITESPACE;
162 
163         for (char *p = s; *p; p++)
164                 if (!strchr(bad, *p))
165                         c = p + 1;
166 
167         *c = 0;
168 
169         return s;
170 }
171 
truncate_nl(char * s)172 char *truncate_nl(char *s) {
173         assert(s);
174 
175         s[strcspn(s, NEWLINE)] = 0;
176         return s;
177 }
178 
ascii_tolower(char x)179 char ascii_tolower(char x) {
180 
181         if (x >= 'A' && x <= 'Z')
182                 return x - 'A' + 'a';
183 
184         return x;
185 }
186 
ascii_toupper(char x)187 char ascii_toupper(char x) {
188 
189         if (x >= 'a' && x <= 'z')
190                 return x - 'a' + 'A';
191 
192         return x;
193 }
194 
ascii_strlower(char * t)195 char *ascii_strlower(char *t) {
196         assert(t);
197 
198         for (char *p = t; *p; p++)
199                 *p = ascii_tolower(*p);
200 
201         return t;
202 }
203 
ascii_strupper(char * t)204 char *ascii_strupper(char *t) {
205         assert(t);
206 
207         for (char *p = t; *p; p++)
208                 *p = ascii_toupper(*p);
209 
210         return t;
211 }
212 
ascii_strlower_n(char * t,size_t n)213 char *ascii_strlower_n(char *t, size_t n) {
214         if (n <= 0)
215                 return t;
216 
217         for (size_t i = 0; i < n; i++)
218                 t[i] = ascii_tolower(t[i]);
219 
220         return t;
221 }
222 
ascii_strcasecmp_n(const char * a,const char * b,size_t n)223 int ascii_strcasecmp_n(const char *a, const char *b, size_t n) {
224 
225         for (; n > 0; a++, b++, n--) {
226                 int x, y;
227 
228                 x = (int) (uint8_t) ascii_tolower(*a);
229                 y = (int) (uint8_t) ascii_tolower(*b);
230 
231                 if (x != y)
232                         return x - y;
233         }
234 
235         return 0;
236 }
237 
ascii_strcasecmp_nn(const char * a,size_t n,const char * b,size_t m)238 int ascii_strcasecmp_nn(const char *a, size_t n, const char *b, size_t m) {
239         int r;
240 
241         r = ascii_strcasecmp_n(a, b, MIN(n, m));
242         if (r != 0)
243                 return r;
244 
245         return CMP(n, m);
246 }
247 
chars_intersect(const char * a,const char * b)248 bool chars_intersect(const char *a, const char *b) {
249         /* Returns true if any of the chars in a are in b. */
250         for (const char *p = a; *p; p++)
251                 if (strchr(b, *p))
252                         return true;
253 
254         return false;
255 }
256 
string_has_cc(const char * p,const char * ok)257 bool string_has_cc(const char *p, const char *ok) {
258         assert(p);
259 
260         /*
261          * Check if a string contains control characters. If 'ok' is
262          * non-NULL it may be a string containing additional CCs to be
263          * considered OK.
264          */
265 
266         for (const char *t = p; *t; t++) {
267                 if (ok && strchr(ok, *t))
268                         continue;
269 
270                 if (char_is_cc(*t))
271                         return true;
272         }
273 
274         return false;
275 }
276 
write_ellipsis(char * buf,bool unicode)277 static int write_ellipsis(char *buf, bool unicode) {
278         if (unicode || is_locale_utf8()) {
279                 buf[0] = 0xe2; /* tri-dot ellipsis: … */
280                 buf[1] = 0x80;
281                 buf[2] = 0xa6;
282         } else {
283                 buf[0] = '.';
284                 buf[1] = '.';
285                 buf[2] = '.';
286         }
287 
288         return 3;
289 }
290 
ascii_ellipsize_mem(const char * s,size_t old_length,size_t new_length,unsigned percent)291 static char *ascii_ellipsize_mem(const char *s, size_t old_length, size_t new_length, unsigned percent) {
292         size_t x, need_space, suffix_len;
293         char *t;
294 
295         assert(s);
296         assert(percent <= 100);
297         assert(new_length != SIZE_MAX);
298 
299         if (old_length <= new_length)
300                 return strndup(s, old_length);
301 
302         /* Special case short ellipsations */
303         switch (new_length) {
304 
305         case 0:
306                 return strdup("");
307 
308         case 1:
309                 if (is_locale_utf8())
310                         return strdup("…");
311                 else
312                         return strdup(".");
313 
314         case 2:
315                 if (!is_locale_utf8())
316                         return strdup("..");
317 
318                 break;
319 
320         default:
321                 break;
322         }
323 
324         /* Calculate how much space the ellipsis will take up. If we are in UTF-8 mode we only need space for one
325          * character ("…"), otherwise for three characters ("..."). Note that in both cases we need 3 bytes of storage,
326          * either for the UTF-8 encoded character or for three ASCII characters. */
327         need_space = is_locale_utf8() ? 1 : 3;
328 
329         t = new(char, new_length+3);
330         if (!t)
331                 return NULL;
332 
333         assert(new_length >= need_space);
334 
335         x = ((new_length - need_space) * percent + 50) / 100;
336         assert(x <= new_length - need_space);
337 
338         memcpy(t, s, x);
339         write_ellipsis(t + x, false);
340         suffix_len = new_length - x - need_space;
341         memcpy(t + x + 3, s + old_length - suffix_len, suffix_len);
342         *(t + x + 3 + suffix_len) = '\0';
343 
344         return t;
345 }
346 
ellipsize_mem(const char * s,size_t old_length,size_t new_length,unsigned percent)347 char *ellipsize_mem(const char *s, size_t old_length, size_t new_length, unsigned percent) {
348         size_t x, k, len, len2;
349         const char *i, *j;
350         char *e;
351         int r;
352 
353         /* Note that 'old_length' refers to bytes in the string, while 'new_length' refers to character cells taken up
354          * on screen. This distinction doesn't matter for ASCII strings, but it does matter for non-ASCII UTF-8
355          * strings.
356          *
357          * Ellipsation is done in a locale-dependent way:
358          * 1. If the string passed in is fully ASCII and the current locale is not UTF-8, three dots are used ("...")
359          * 2. Otherwise, a unicode ellipsis is used ("…")
360          *
361          * In other words: you'll get a unicode ellipsis as soon as either the string contains non-ASCII characters or
362          * the current locale is UTF-8.
363          */
364 
365         assert(s);
366         assert(percent <= 100);
367 
368         if (new_length == SIZE_MAX)
369                 return strndup(s, old_length);
370 
371         if (new_length == 0)
372                 return strdup("");
373 
374         /* If no multibyte characters use ascii_ellipsize_mem for speed */
375         if (ascii_is_valid_n(s, old_length))
376                 return ascii_ellipsize_mem(s, old_length, new_length, percent);
377 
378         x = ((new_length - 1) * percent) / 100;
379         assert(x <= new_length - 1);
380 
381         k = 0;
382         for (i = s; i < s + old_length; i = utf8_next_char(i)) {
383                 char32_t c;
384                 int w;
385 
386                 r = utf8_encoded_to_unichar(i, &c);
387                 if (r < 0)
388                         return NULL;
389 
390                 w = unichar_iswide(c) ? 2 : 1;
391                 if (k + w <= x)
392                         k += w;
393                 else
394                         break;
395         }
396 
397         for (j = s + old_length; j > i; ) {
398                 char32_t c;
399                 int w;
400                 const char *jj;
401 
402                 jj = utf8_prev_char(j);
403                 r = utf8_encoded_to_unichar(jj, &c);
404                 if (r < 0)
405                         return NULL;
406 
407                 w = unichar_iswide(c) ? 2 : 1;
408                 if (k + w <= new_length) {
409                         k += w;
410                         j = jj;
411                 } else
412                         break;
413         }
414         assert(i <= j);
415 
416         /* we don't actually need to ellipsize */
417         if (i == j)
418                 return memdup_suffix0(s, old_length);
419 
420         /* make space for ellipsis, if possible */
421         if (j < s + old_length)
422                 j = utf8_next_char(j);
423         else if (i > s)
424                 i = utf8_prev_char(i);
425 
426         len = i - s;
427         len2 = s + old_length - j;
428         e = new(char, len + 3 + len2 + 1);
429         if (!e)
430                 return NULL;
431 
432         /*
433         printf("old_length=%zu new_length=%zu x=%zu len=%u len2=%u k=%u\n",
434                old_length, new_length, x, len, len2, k);
435         */
436 
437         memcpy(e, s, len);
438         write_ellipsis(e + len, true);
439         memcpy(e + len + 3, j, len2);
440         *(e + len + 3 + len2) = '\0';
441 
442         return e;
443 }
444 
cellescape(char * buf,size_t len,const char * s)445 char *cellescape(char *buf, size_t len, const char *s) {
446         /* Escape and ellipsize s into buffer buf of size len. Only non-control ASCII
447          * characters are copied as they are, everything else is escaped. The result
448          * is different then if escaping and ellipsization was performed in two
449          * separate steps, because each sequence is either stored in full or skipped.
450          *
451          * This function should be used for logging about strings which expected to
452          * be plain ASCII in a safe way.
453          *
454          * An ellipsis will be used if s is too long. It was always placed at the
455          * very end.
456          */
457 
458         size_t i = 0, last_char_width[4] = {}, k = 0;
459 
460         assert(len > 0); /* at least a terminating NUL */
461 
462         for (;;) {
463                 char four[4];
464                 int w;
465 
466                 if (*s == 0) /* terminating NUL detected? then we are done! */
467                         goto done;
468 
469                 w = cescape_char(*s, four);
470                 if (i + w + 1 > len) /* This character doesn't fit into the buffer anymore? In that case let's
471                                       * ellipsize at the previous location */
472                         break;
473 
474                 /* OK, there was space, let's add this escaped character to the buffer */
475                 memcpy(buf + i, four, w);
476                 i += w;
477 
478                 /* And remember its width in the ring buffer */
479                 last_char_width[k] = w;
480                 k = (k + 1) % 4;
481 
482                 s++;
483         }
484 
485         /* Ellipsation is necessary. This means we might need to truncate the string again to make space for 4
486          * characters ideally, but the buffer is shorter than that in the first place take what we can get */
487         for (size_t j = 0; j < ELEMENTSOF(last_char_width); j++) {
488 
489                 if (i + 4 <= len) /* nice, we reached our space goal */
490                         break;
491 
492                 k = k == 0 ? 3 : k - 1;
493                 if (last_char_width[k] == 0) /* bummer, we reached the beginning of the strings */
494                         break;
495 
496                 assert(i >= last_char_width[k]);
497                 i -= last_char_width[k];
498         }
499 
500         if (i + 4 <= len) /* yay, enough space */
501                 i += write_ellipsis(buf + i, false);
502         else if (i + 3 <= len) { /* only space for ".." */
503                 buf[i++] = '.';
504                 buf[i++] = '.';
505         } else if (i + 2 <= len) /* only space for a single "." */
506                 buf[i++] = '.';
507         else
508                 assert(i + 1 <= len);
509 
510  done:
511         buf[i] = '\0';
512         return buf;
513 }
514 
strshorten(char * s,size_t l)515 char* strshorten(char *s, size_t l) {
516         assert(s);
517 
518         if (strnlen(s, l+1) > l)
519                 s[l] = 0;
520 
521         return s;
522 }
523 
strreplace(const char * text,const char * old_string,const char * new_string)524 char *strreplace(const char *text, const char *old_string, const char *new_string) {
525         size_t l, old_len, new_len;
526         char *t, *ret = NULL;
527         const char *f;
528 
529         assert(old_string);
530         assert(new_string);
531 
532         if (!text)
533                 return NULL;
534 
535         old_len = strlen(old_string);
536         new_len = strlen(new_string);
537 
538         l = strlen(text);
539         if (!GREEDY_REALLOC(ret, l+1))
540                 return NULL;
541 
542         f = text;
543         t = ret;
544         while (*f) {
545                 size_t d, nl;
546 
547                 if (!startswith(f, old_string)) {
548                         *(t++) = *(f++);
549                         continue;
550                 }
551 
552                 d = t - ret;
553                 nl = l - old_len + new_len;
554 
555                 if (!GREEDY_REALLOC(ret, nl + 1))
556                         return mfree(ret);
557 
558                 l = nl;
559                 t = ret + d;
560 
561                 t = stpcpy(t, new_string);
562                 f += old_len;
563         }
564 
565         *t = 0;
566         return ret;
567 }
568 
advance_offsets(ssize_t diff,size_t offsets[2],size_t shift[static2],size_t size)569 static void advance_offsets(
570                 ssize_t diff,
571                 size_t offsets[2], /* note: we can't use [static 2] here, since this may be NULL */
572                 size_t shift[static 2],
573                 size_t size) {
574 
575         if (!offsets)
576                 return;
577 
578         assert(shift);
579 
580         if ((size_t) diff < offsets[0])
581                 shift[0] += size;
582         if ((size_t) diff < offsets[1])
583                 shift[1] += size;
584 }
585 
strip_tab_ansi(char ** ibuf,size_t * _isz,size_t highlight[2])586 char *strip_tab_ansi(char **ibuf, size_t *_isz, size_t highlight[2]) {
587         const char *begin = NULL;
588         enum {
589                 STATE_OTHER,
590                 STATE_ESCAPE,
591                 STATE_CSI,
592                 STATE_CSO,
593         } state = STATE_OTHER;
594         char *obuf = NULL;
595         size_t osz = 0, isz, shift[2] = {}, n_carriage_returns = 0;
596         FILE *f;
597 
598         assert(ibuf);
599         assert(*ibuf);
600 
601         /* This does three things:
602          *
603          * 1. Replaces TABs by 8 spaces
604          * 2. Strips ANSI color sequences (a subset of CSI), i.e. ESC '[' … 'm' sequences
605          * 3. Strips ANSI operating system sequences (CSO), i.e. ESC ']' … BEL sequences
606          * 4. Strip trailing \r characters (since they would "move the cursor", but have no
607          *    other effect).
608          *
609          * Everything else will be left as it is. In particular other ANSI sequences are left as they are, as
610          * are any other special characters. Truncated ANSI sequences are left-as is too. This call is
611          * supposed to suppress the most basic formatting noise, but nothing else.
612          *
613          * Why care for CSO sequences? Well, to undo what terminal_urlify() and friends generate. */
614 
615         isz = _isz ? *_isz : strlen(*ibuf);
616 
617         /* Note we turn off internal locking on f for performance reasons. It's safe to do so since we
618          * created f here and it doesn't leave our scope. */
619         f = open_memstream_unlocked(&obuf, &osz);
620         if (!f)
621                 return NULL;
622 
623         for (const char *i = *ibuf; i < *ibuf + isz + 1; i++) {
624 
625                 switch (state) {
626 
627                 case STATE_OTHER:
628                         if (i >= *ibuf + isz) /* EOT */
629                                 break;
630 
631                         if (*i == '\r') {
632                                 n_carriage_returns++;
633                                 break;
634                         } else if (*i == '\n')
635                                 /* Ignore carriage returns before new line */
636                                 n_carriage_returns = 0;
637                         for (; n_carriage_returns > 0; n_carriage_returns--)
638                                 fputc('\r', f);
639 
640                         if (*i == '\x1B')
641                                 state = STATE_ESCAPE;
642                         else if (*i == '\t') {
643                                 fputs("        ", f);
644                                 advance_offsets(i - *ibuf, highlight, shift, 7);
645                         } else
646                                 fputc(*i, f);
647 
648                         break;
649 
650                 case STATE_ESCAPE:
651                         assert(n_carriage_returns == 0);
652 
653                         if (i >= *ibuf + isz) { /* EOT */
654                                 fputc('\x1B', f);
655                                 advance_offsets(i - *ibuf, highlight, shift, 1);
656                                 break;
657                         } else if (*i == '[') { /* ANSI CSI */
658                                 state = STATE_CSI;
659                                 begin = i + 1;
660                         } else if (*i == ']') { /* ANSI CSO */
661                                 state = STATE_CSO;
662                                 begin = i + 1;
663                         } else {
664                                 fputc('\x1B', f);
665                                 fputc(*i, f);
666                                 advance_offsets(i - *ibuf, highlight, shift, 1);
667                                 state = STATE_OTHER;
668                         }
669 
670                         break;
671 
672                 case STATE_CSI:
673                         assert(n_carriage_returns == 0);
674 
675                         if (i >= *ibuf + isz || /* EOT … */
676                             !strchr("01234567890;m", *i)) { /* … or invalid chars in sequence */
677                                 fputc('\x1B', f);
678                                 fputc('[', f);
679                                 advance_offsets(i - *ibuf, highlight, shift, 2);
680                                 state = STATE_OTHER;
681                                 i = begin-1;
682                         } else if (*i == 'm')
683                                 state = STATE_OTHER;
684 
685                         break;
686 
687                 case STATE_CSO:
688                         assert(n_carriage_returns == 0);
689 
690                         if (i >= *ibuf + isz || /* EOT … */
691                             (*i != '\a' && (uint8_t) *i < 32U) || (uint8_t) *i > 126U) { /* … or invalid chars in sequence */
692                                 fputc('\x1B', f);
693                                 fputc(']', f);
694                                 advance_offsets(i - *ibuf, highlight, shift, 2);
695                                 state = STATE_OTHER;
696                                 i = begin-1;
697                         } else if (*i == '\a')
698                                 state = STATE_OTHER;
699 
700                         break;
701                 }
702         }
703 
704         if (fflush_and_check(f) < 0) {
705                 fclose(f);
706                 return mfree(obuf);
707         }
708         fclose(f);
709 
710         free_and_replace(*ibuf, obuf);
711 
712         if (_isz)
713                 *_isz = osz;
714 
715         if (highlight) {
716                 highlight[0] += shift[0];
717                 highlight[1] += shift[1];
718         }
719 
720         return *ibuf;
721 }
722 
strextend_with_separator_internal(char ** x,const char * separator,...)723 char *strextend_with_separator_internal(char **x, const char *separator, ...) {
724         size_t f, l, l_separator;
725         bool need_separator;
726         char *nr, *p;
727         va_list ap;
728 
729         assert(x);
730 
731         l = f = strlen_ptr(*x);
732 
733         need_separator = !isempty(*x);
734         l_separator = strlen_ptr(separator);
735 
736         va_start(ap, separator);
737         for (;;) {
738                 const char *t;
739                 size_t n;
740 
741                 t = va_arg(ap, const char *);
742                 if (!t)
743                         break;
744 
745                 n = strlen(t);
746 
747                 if (need_separator)
748                         n += l_separator;
749 
750                 if (n >= SIZE_MAX - l) {
751                         va_end(ap);
752                         return NULL;
753                 }
754 
755                 l += n;
756                 need_separator = true;
757         }
758         va_end(ap);
759 
760         need_separator = !isempty(*x);
761 
762         nr = realloc(*x, GREEDY_ALLOC_ROUND_UP(l+1));
763         if (!nr)
764                 return NULL;
765 
766         *x = nr;
767         p = nr + f;
768 
769         va_start(ap, separator);
770         for (;;) {
771                 const char *t;
772 
773                 t = va_arg(ap, const char *);
774                 if (!t)
775                         break;
776 
777                 if (need_separator && separator)
778                         p = stpcpy(p, separator);
779 
780                 p = stpcpy(p, t);
781 
782                 need_separator = true;
783         }
784         va_end(ap);
785 
786         assert(p == nr + l);
787 
788         *p = 0;
789 
790         return p;
791 }
792 
strextendf_with_separator(char ** x,const char * separator,const char * format,...)793 int strextendf_with_separator(char **x, const char *separator, const char *format, ...) {
794         size_t m, a, l_separator;
795         va_list ap;
796         int l;
797 
798         /* Appends a formatted string to the specified string. Don't use this in inner loops, since then
799          * we'll spend a tonload of time in determining the length of the string passed in, over and over
800          * again. */
801 
802         assert(x);
803         assert(format);
804 
805         l_separator = isempty(*x) ? 0 : strlen_ptr(separator);
806 
807         /* Let's try to use the allocated buffer, if there's room at the end still. Otherwise let's extend by 64 chars. */
808         if (*x) {
809                 m = strlen(*x);
810                 a = MALLOC_SIZEOF_SAFE(*x);
811                 assert(a >= m + 1);
812         } else
813                 m = a = 0;
814 
815         if (a - m < 17 + l_separator) { /* if there's less than 16 chars space, then enlarge the buffer first */
816                 char *n;
817 
818                 if (_unlikely_(l_separator > SIZE_MAX - 64)) /* overflow check #1 */
819                         return -ENOMEM;
820                 if (_unlikely_(m > SIZE_MAX - 64 - l_separator)) /* overflow check #2 */
821                         return -ENOMEM;
822 
823                 n = realloc(*x, m + 64 + l_separator);
824                 if (!n)
825                         return -ENOMEM;
826 
827                 *x = n;
828                 a = MALLOC_SIZEOF_SAFE(*x);
829         }
830 
831         /* Now, let's try to format the string into it */
832         memcpy_safe(*x + m, separator, l_separator);
833         va_start(ap, format);
834         l = vsnprintf(*x + m + l_separator, a - m - l_separator, format, ap);
835         va_end(ap);
836 
837         assert(l >= 0);
838 
839         if ((size_t) l < a - m - l_separator) {
840                 char *n;
841 
842                 /* Nice! This worked. We are done. But first, let's return the extra space we don't
843                  * need. This should be a cheap operation, since we only lower the allocation size here,
844                  * never increase. */
845                 n = realloc(*x, m + (size_t) l + l_separator + 1);
846                 if (n)
847                         *x = n;
848         } else {
849                 char *n;
850 
851                 /* Wasn't enough. Then let's allocate exactly what we need. */
852 
853                 if (_unlikely_((size_t) l > SIZE_MAX - (l_separator + 1))) /* overflow check #1 */
854                         goto oom;
855                 if (_unlikely_(m > SIZE_MAX - ((size_t) l + l_separator + 1))) /* overflow check #2 */
856                         goto oom;
857 
858                 a = m + (size_t) l + l_separator + 1;
859                 n = realloc(*x, a);
860                 if (!n)
861                         goto oom;
862                 *x = n;
863 
864                 va_start(ap, format);
865                 l = vsnprintf(*x + m + l_separator, a - m - l_separator, format, ap);
866                 va_end(ap);
867 
868                 assert((size_t) l < a - m - l_separator);
869         }
870 
871         return 0;
872 
873 oom:
874         /* truncate the bytes added after the first vsnprintf() attempt again */
875         (*x)[m] = 0;
876         return -ENOMEM;
877 }
878 
strrep(const char * s,unsigned n)879 char *strrep(const char *s, unsigned n) {
880         char *r, *p;
881         size_t l;
882 
883         assert(s);
884 
885         l = strlen(s);
886         p = r = malloc(l * n + 1);
887         if (!r)
888                 return NULL;
889 
890         for (unsigned i = 0; i < n; i++)
891                 p = stpcpy(p, s);
892 
893         *p = 0;
894         return r;
895 }
896 
split_pair(const char * s,const char * sep,char ** l,char ** r)897 int split_pair(const char *s, const char *sep, char **l, char **r) {
898         char *x, *a, *b;
899 
900         assert(s);
901         assert(sep);
902         assert(l);
903         assert(r);
904 
905         if (isempty(sep))
906                 return -EINVAL;
907 
908         x = strstr(s, sep);
909         if (!x)
910                 return -EINVAL;
911 
912         a = strndup(s, x - s);
913         if (!a)
914                 return -ENOMEM;
915 
916         b = strdup(x + strlen(sep));
917         if (!b) {
918                 free(a);
919                 return -ENOMEM;
920         }
921 
922         *l = a;
923         *r = b;
924 
925         return 0;
926 }
927 
free_and_strdup(char ** p,const char * s)928 int free_and_strdup(char **p, const char *s) {
929         char *t;
930 
931         assert(p);
932 
933         /* Replaces a string pointer with a strdup()ed new string,
934          * possibly freeing the old one. */
935 
936         if (streq_ptr(*p, s))
937                 return 0;
938 
939         if (s) {
940                 t = strdup(s);
941                 if (!t)
942                         return -ENOMEM;
943         } else
944                 t = NULL;
945 
946         free(*p);
947         *p = t;
948 
949         return 1;
950 }
951 
free_and_strndup(char ** p,const char * s,size_t l)952 int free_and_strndup(char **p, const char *s, size_t l) {
953         char *t;
954 
955         assert(p);
956         assert(s || l == 0);
957 
958         /* Replaces a string pointer with a strndup()ed new string,
959          * freeing the old one. */
960 
961         if (!*p && !s)
962                 return 0;
963 
964         if (*p && s && strneq(*p, s, l) && (l > strlen(*p) || (*p)[l] == '\0'))
965                 return 0;
966 
967         if (s) {
968                 t = strndup(s, l);
969                 if (!t)
970                         return -ENOMEM;
971         } else
972                 t = NULL;
973 
974         free_and_replace(*p, t);
975         return 1;
976 }
977 
string_is_safe(const char * p)978 bool string_is_safe(const char *p) {
979         if (!p)
980                 return false;
981 
982         /* Checks if the specified string contains no quotes or control characters */
983 
984         for (const char *t = p; *t; t++) {
985                 if (*t > 0 && *t < ' ') /* no control characters */
986                         return false;
987 
988                 if (strchr(QUOTES "\\\x7f", *t))
989                         return false;
990         }
991 
992         return true;
993 }
994 
string_erase(char * x)995 char* string_erase(char *x) {
996         if (!x)
997                 return NULL;
998 
999         /* A delicious drop of snake-oil! To be called on memory where we stored passphrases or so, after we
1000          * used them. */
1001         explicit_bzero_safe(x, strlen(x));
1002         return x;
1003 }
1004 
string_truncate_lines(const char * s,size_t n_lines,char ** ret)1005 int string_truncate_lines(const char *s, size_t n_lines, char **ret) {
1006         const char *p = s, *e = s;
1007         bool truncation_applied = false;
1008         char *copy;
1009         size_t n = 0;
1010 
1011         assert(s);
1012 
1013         /* Truncate after the specified number of lines. Returns > 0 if a truncation was applied or == 0 if
1014          * there were fewer lines in the string anyway. Trailing newlines on input are ignored, and not
1015          * generated either. */
1016 
1017         for (;;) {
1018                 size_t k;
1019 
1020                 k = strcspn(p, "\n");
1021 
1022                 if (p[k] == 0) {
1023                         if (k == 0) /* final empty line */
1024                                 break;
1025 
1026                         if (n >= n_lines) /* above threshold */
1027                                 break;
1028 
1029                         e = p + k; /* last line to include */
1030                         break;
1031                 }
1032 
1033                 assert(p[k] == '\n');
1034 
1035                 if (n >= n_lines)
1036                         break;
1037 
1038                 if (k > 0)
1039                         e = p + k;
1040 
1041                 p += k + 1;
1042                 n++;
1043         }
1044 
1045         /* e points after the last character we want to keep */
1046         if (isempty(e))
1047                 copy = strdup(s);
1048         else {
1049                 if (!in_charset(e, "\n")) /* We only consider things truncated if we remove something that
1050                                            * isn't a new-line or a series of them */
1051                         truncation_applied = true;
1052 
1053                 copy = strndup(s, e - s);
1054         }
1055         if (!copy)
1056                 return -ENOMEM;
1057 
1058         *ret = copy;
1059         return truncation_applied;
1060 }
1061 
string_extract_line(const char * s,size_t i,char ** ret)1062 int string_extract_line(const char *s, size_t i, char **ret) {
1063         const char *p = s;
1064         size_t c = 0;
1065 
1066         /* Extract the i'nth line from the specified string. Returns > 0 if there are more lines after that,
1067          * and == 0 if we are looking at the last line or already beyond the last line. As special
1068          * optimization, if the first line is requested and the string only consists of one line we return
1069          * NULL, indicating the input string should be used as is, and avoid a memory allocation for a very
1070          * common case. */
1071 
1072         for (;;) {
1073                 const char *q;
1074 
1075                 q = strchr(p, '\n');
1076                 if (i == c) {
1077                         /* The line we are looking for! */
1078 
1079                         if (q) {
1080                                 char *m;
1081 
1082                                 m = strndup(p, q - p);
1083                                 if (!m)
1084                                         return -ENOMEM;
1085 
1086                                 *ret = m;
1087                                 return !isempty(q + 1); /* more coming? */
1088                         } else {
1089                                 if (p == s)
1090                                         *ret = NULL; /* Just use the input string */
1091                                 else {
1092                                         char *m;
1093 
1094                                         m = strdup(p);
1095                                         if (!m)
1096                                                 return -ENOMEM;
1097 
1098                                         *ret = m;
1099                                 }
1100 
1101                                 return 0; /* The end */
1102                         }
1103                 }
1104 
1105                 if (!q) {
1106                         char *m;
1107 
1108                         /* No more lines, return empty line */
1109 
1110                         m = strdup("");
1111                         if (!m)
1112                                 return -ENOMEM;
1113 
1114                         *ret = m;
1115                         return 0; /* The end */
1116                 }
1117 
1118                 p = q + 1;
1119                 c++;
1120         }
1121 }
1122 
string_contains_word_strv(const char * string,const char * separators,char ** words,const char ** ret_word)1123 int string_contains_word_strv(const char *string, const char *separators, char **words, const char **ret_word) {
1124         /* In the default mode with no separators specified, we split on whitespace and
1125          * don't coalesce separators. */
1126         const ExtractFlags flags = separators ? EXTRACT_DONT_COALESCE_SEPARATORS : 0;
1127 
1128         const char *found = NULL;
1129 
1130         for (const char *p = string;;) {
1131                 _cleanup_free_ char *w = NULL;
1132                 int r;
1133 
1134                 r = extract_first_word(&p, &w, separators, flags);
1135                 if (r < 0)
1136                         return r;
1137                 if (r == 0)
1138                         break;
1139 
1140                 found = strv_find(words, w);
1141                 if (found)
1142                         break;
1143         }
1144 
1145         if (ret_word)
1146                 *ret_word = found;
1147         return !!found;
1148 }
1149 
streq_skip_trailing_chars(const char * s1,const char * s2,const char * ok)1150 bool streq_skip_trailing_chars(const char *s1, const char *s2, const char *ok) {
1151         if (!s1 && !s2)
1152                 return true;
1153         if (!s1 || !s2)
1154                 return false;
1155 
1156         if (!ok)
1157                 ok = WHITESPACE;
1158 
1159         for (; *s1 && *s2; s1++, s2++)
1160                 if (*s1 != *s2)
1161                         break;
1162 
1163         return in_charset(s1, ok) && in_charset(s2, ok);
1164 }
1165 
string_replace_char(char * str,char old_char,char new_char)1166 char *string_replace_char(char *str, char old_char, char new_char) {
1167         assert(str);
1168         assert(old_char != '\0');
1169         assert(new_char != '\0');
1170         assert(old_char != new_char);
1171 
1172         for (char *p = strchr(str, old_char); p; p = strchr(p + 1, old_char))
1173                 *p = new_char;
1174 
1175         return str;
1176 }
1177 
strspn_from_end(const char * str,const char * accept)1178 size_t strspn_from_end(const char *str, const char *accept) {
1179         size_t n = 0;
1180 
1181         if (isempty(str))
1182                 return 0;
1183 
1184         if (isempty(accept))
1185                 return 0;
1186 
1187         for (const char *p = str + strlen(str); p > str && strchr(accept, p[-1]); p--)
1188                 n++;
1189 
1190         return n;
1191 }
1192