1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011, 2012, 2014,
3 2020 Free Software Foundation, Inc.
5 This program is free software: you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published by
7 the Free Software Foundation, either version 3 of the License, or
8 (at your option) any later version.
10 This program is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program. If not, see <http://www.gnu.org/licenses/>. */
28 #include "libpspp/cast.h"
29 #include "libpspp/message.h"
30 #include "libpspp/pool.h"
32 #include "gl/c-ctype.h"
33 #include "gl/c-vasnprintf.h"
34 #include "gl/relocatable.h"
35 #include "gl/minmax.h"
36 #include "gl/xalloc.h"
37 #include "gl/xmemdup0.h"
40 /* Reverses the order of NBYTES bytes at address P, thus converting
41 between little- and big-endian byte orders. */
43 buf_reverse (char *p, size_t nbytes)
45 char *h = p, *t = &h[nbytes - 1];
57 /* Compares the SIZE bytes in A to those in B, disregarding case,
58 and returns a strcmp()-type result. */
60 buf_compare_case (const char *a_, const char *b_, size_t size)
62 const unsigned char *a = (unsigned char *) a_;
63 const unsigned char *b = (unsigned char *) b_;
67 unsigned char ac = toupper (*a++);
68 unsigned char bc = toupper (*b++);
71 return ac > bc ? 1 : -1;
77 /* Compares A of length A_LEN to B of length B_LEN. The shorter
78 string is considered to be padded with spaces to the length of
81 buf_compare_rpad (const char *a, size_t a_len, const char *b, size_t b_len)
86 min_len = a_len < b_len ? a_len : b_len;
87 result = memcmp (a, b, min_len);
96 for (idx = min_len; idx < b_len; idx++)
98 return ' ' > b[idx] ? 1 : -1;
102 for (idx = min_len; idx < a_len; idx++)
104 return a[idx] > ' ' ? 1 : -1;
110 /* Compares strin A to string B. The shorter string is
111 considered to be padded with spaces to the length of the
114 str_compare_rpad (const char *a, const char *b)
116 return buf_compare_rpad (a, strlen (a), b, strlen (b));
119 /* Copies string SRC to buffer DST, of size DST_SIZE bytes.
120 DST is truncated to DST_SIZE bytes or padded on the right with
121 copies of PAD as needed. */
123 buf_copy_str_rpad (char *dst, size_t dst_size, const char *src, char pad)
125 size_t src_len = strlen (src);
126 if (src_len >= dst_size)
127 memcpy (dst, src, dst_size);
130 memcpy (dst, src, src_len);
131 memset (&dst[src_len], pad, dst_size - src_len);
135 /* Copies string SRC to buffer DST, of size DST_SIZE bytes.
136 DST is truncated to DST_SIZE bytes or padded on the left with
137 copies of PAD as needed. */
139 buf_copy_str_lpad (char *dst, size_t dst_size, const char *src, char pad)
141 size_t src_len = strlen (src);
142 if (src_len >= dst_size)
143 memcpy (dst, src, dst_size);
146 size_t pad_cnt = dst_size - src_len;
147 memset (&dst[0], pad, pad_cnt);
148 memcpy (dst + pad_cnt, src, src_len);
152 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
153 DST is truncated to DST_SIZE bytes or padded on the left with
154 copies of PAD as needed. */
156 buf_copy_lpad (char *dst, size_t dst_size,
157 const char *src, size_t src_size,
160 if (src_size >= dst_size)
161 memmove (dst, src, dst_size);
164 memset (dst, pad, dst_size - src_size);
165 memmove (&dst[dst_size - src_size], src, src_size);
169 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
170 DST is truncated to DST_SIZE bytes or padded on the right with
171 copies of PAD as needed. */
173 buf_copy_rpad (char *dst, size_t dst_size,
174 const char *src, size_t src_size,
177 if (src_size >= dst_size)
178 memmove (dst, src, dst_size);
181 memmove (dst, src, src_size);
182 memset (&dst[src_size], pad, dst_size - src_size);
186 /* Copies string SRC to string DST, which is in a buffer DST_SIZE
188 Truncates DST to DST_SIZE - 1 bytes or right-pads with
189 spaces to DST_SIZE - 1 bytes if necessary. */
191 str_copy_rpad (char *dst, size_t dst_size, const char *src)
195 size_t src_len = strlen (src);
196 if (src_len < dst_size - 1)
198 memcpy (dst, src, src_len);
199 memset (&dst[src_len], ' ', dst_size - 1 - src_len);
202 memcpy (dst, src, dst_size - 1);
203 dst[dst_size - 1] = 0;
207 /* Copies SRC to DST, which is in a buffer DST_SIZE bytes long.
208 Truncates DST to DST_SIZE - 1 bytes, if necessary. */
210 str_copy_trunc (char *dst, size_t dst_size, const char *src)
212 size_t src_len = strlen (src);
213 assert (dst_size > 0);
214 if (src_len + 1 < dst_size)
215 memcpy (dst, src, src_len + 1);
218 memcpy (dst, src, dst_size - 1);
219 dst[dst_size - 1] = '\0';
223 /* Copies buffer SRC, of SRC_LEN bytes,
224 to DST, which is in a buffer DST_SIZE bytes long.
225 Truncates DST to DST_SIZE - 1 bytes, if necessary. */
227 str_copy_buf_trunc (char *dst, size_t dst_size,
228 const char *src, size_t src_size)
231 assert (dst_size > 0);
233 dst_len = src_size < dst_size ? src_size : dst_size - 1;
234 memcpy (dst, src, dst_len);
238 /* Converts each byte in S to uppercase.
240 This is suitable only for ASCII strings. Use utf8_to_upper() for UTF-8
243 str_uppercase (char *s)
245 for (; *s != '\0'; s++)
246 *s = c_toupper ((unsigned char) *s);
249 /* Converts each byte in S to lowercase.
251 This is suitable only for ASCII strings. Use utf8_to_lower() for UTF-8
254 str_lowercase (char *s)
256 for (; *s != '\0'; s++)
257 *s = c_tolower ((unsigned char) *s);
260 /* Converts NUMBER into a string in 26-adic notation in BUFFER,
261 which has room for SIZE bytes. Uses uppercase if UPPERCASE is
262 true, otherwise lowercase, Returns true if successful, false
263 if NUMBER, plus a trailing null, is too large to fit in the
266 26-adic notation is "spreadsheet column numbering": 1 = A, 2 =
267 B, 3 = C, ... 26 = Z, 27 = AA, 28 = AB, 29 = AC, ...
269 26-adic notation is the special case of a k-adic numeration
270 system (aka bijective base-k numeration) with k=26. In k-adic
271 numeration, the digits are {1, 2, 3, ..., k} (there is no
272 digit 0), and integer 0 is represented by the empty string.
273 For more information, see
274 http://en.wikipedia.org/wiki/Bijective_numeration. */
276 str_format_26adic (unsigned long int number, bool uppercase,
277 char buffer[], size_t size)
280 = uppercase ? "ABCDEFGHIJKLMNOPQRSTUVWXYZ" : "abcdefghijklmnopqrstuvwxyz";
287 buffer[length++] = alphabet[number % 26];
293 buffer[length] = '\0';
295 buf_reverse (buffer, length);
304 /* Sets the SIZE bytes starting at BLOCK to C,
305 and returns the byte following BLOCK. */
307 mempset (void *block, int c, size_t size)
309 memset (block, c, size);
310 return (char *) block + size;
315 /* Returns a substring whose contents are the CNT bytes
316 starting at the (0-based) position START in SS. */
318 ss_substr (struct substring ss, size_t start, size_t cnt)
320 if (start < ss.length)
321 return ss_buffer (ss.string + start, MIN (cnt, ss.length - start));
323 return ss_buffer (ss.string + ss.length, 0);
326 /* Returns a substring whose contents are the first CNT
329 ss_head (struct substring ss, size_t cnt)
331 return ss_buffer (ss.string, MIN (cnt, ss.length));
334 /* Returns a substring whose contents are the last CNT bytes
337 ss_tail (struct substring ss, size_t cnt)
340 return ss_buffer (ss.string + (ss.length - cnt), cnt);
345 /* Makes a malloc()'d, null-terminated copy of the contents of OLD
346 and stores it in NEW. */
348 ss_alloc_substring (struct substring *new, struct substring old)
350 new->string = xmemdup0 (old.string, old.length);
351 new->length = old.length;
354 /* Allocates room for a CNT-byte string in NEW. */
356 ss_alloc_uninit (struct substring *new, size_t cnt)
358 new->string = xmalloc (cnt);
363 ss_realloc (struct substring *ss, size_t size)
365 ss->string = xrealloc (ss->string, size);
368 /* Makes a pool_alloc_unaligned()'d, null-terminated copy of the contents of
369 OLD in POOL, and stores it in NEW. */
371 ss_alloc_substring_pool (struct substring *new, struct substring old,
374 new->string = pool_alloc_unaligned (pool, old.length + 1);
375 new->length = old.length;
376 memcpy (new->string, old.string, old.length);
377 new->string[old.length] = '\0';
380 /* Allocates room for a CNT-byte string in NEW in POOL. */
382 ss_alloc_uninit_pool (struct substring *new, size_t cnt, struct pool *pool)
384 new->string = pool_alloc_unaligned (pool, cnt);
388 /* Frees the string that SS points to. */
390 ss_dealloc (struct substring *ss)
395 /* Truncates SS to at most CNT bytes in length. */
397 ss_truncate (struct substring *ss, size_t cnt)
399 if (ss->length > cnt)
403 /* Removes trailing bytes in TRIM_SET from SS.
404 Returns number of bytes removed. */
406 ss_rtrim (struct substring *ss, struct substring trim_set)
409 while (cnt < ss->length
410 && ss_find_byte (trim_set,
411 ss->string[ss->length - cnt - 1]) != SIZE_MAX)
417 /* Removes leading bytes in TRIM_SET from SS.
418 Returns number of bytes removed. */
420 ss_ltrim (struct substring *ss, struct substring trim_set)
422 size_t cnt = ss_span (*ss, trim_set);
423 ss_advance (ss, cnt);
427 /* Trims leading and trailing bytes in TRIM_SET from SS. */
429 ss_trim (struct substring *ss, struct substring trim_set)
431 ss_ltrim (ss, trim_set);
432 ss_rtrim (ss, trim_set);
435 /* If the last byte in SS is C, removes it and returns true.
436 Otherwise, returns false without changing the string. */
438 ss_chomp_byte (struct substring *ss, char c)
440 if (ss_last (*ss) == c)
449 /* If SS ends with SUFFIX, removes it and returns true.
450 Otherwise, returns false without changing the string. */
452 ss_chomp (struct substring *ss, struct substring suffix)
454 if (ss_ends_with (*ss, suffix))
456 ss->length -= suffix.length;
463 /* Divides SS into tokens separated by any of the DELIMITERS.
464 Each call replaces TOKEN by the next token in SS, or by an
465 empty string if no tokens remain. Returns true if a token was
466 obtained, false otherwise.
468 Before the first call, initialize *SAVE_IDX to 0. Do not
469 modify *SAVE_IDX between calls.
471 SS divides into exactly one more tokens than it contains
472 delimiters. That is, a delimiter at the start or end of SS or
473 a pair of adjacent delimiters yields an empty token, and the
474 empty string contains a single token. */
476 ss_separate (struct substring ss, struct substring delimiters,
477 size_t *save_idx, struct substring *token)
479 if (*save_idx <= ss_length (ss))
481 struct substring tmp = ss_substr (ss, *save_idx, SIZE_MAX);
482 size_t length = ss_cspan (tmp, delimiters);
483 *token = ss_head (tmp, length);
484 *save_idx += length + 1;
489 *token = ss_empty ();
494 /* Divides SS into tokens separated by any of the DELIMITERS,
495 merging adjacent delimiters so that the empty string is never
496 produced as a token. Each call replaces TOKEN by the next
497 token in SS, or by an empty string if no tokens remain, and
498 then skips past the first delimiter following the token.
499 Returns true if a token was obtained, false otherwise.
501 Before the first call, initialize *SAVE_IDX to 0. Do not
502 modify *SAVE_IDX between calls. */
504 ss_tokenize (struct substring ss, struct substring delimiters,
505 size_t *save_idx, struct substring *token)
509 ss_advance (&ss, *save_idx);
510 *save_idx += ss_ltrim (&ss, delimiters);
511 ss_get_bytes (&ss, ss_cspan (ss, delimiters), token);
513 found_token = ss_length (*token) > 0;
514 *save_idx += ss_length (*token) + found_token;
518 /* Removes the first CNT bytes from SS. */
520 ss_advance (struct substring *ss, size_t cnt)
522 if (cnt > ss->length)
528 /* If the first byte in SS is C, removes it and returns true.
529 Otherwise, returns false without changing the string. */
531 ss_match_byte (struct substring *ss, char c)
533 if (ss_first (*ss) == c)
543 /* If the first byte in SS is in MATCH, removes it and
544 returns the byte that was removed.
545 Otherwise, returns EOF without changing the string. */
547 ss_match_byte_in (struct substring *ss, struct substring match)
551 && memchr (match.string, ss->string[0], match.length) != NULL)
560 /* If SS begins with TARGET, removes it and returns true.
561 Otherwise, returns false without changing SS. */
563 ss_match_string (struct substring *ss, const struct substring target)
565 size_t length = ss_length (target);
566 if (ss_equals (ss_head (*ss, length), target))
568 ss_advance (ss, length);
575 /* Removes the first byte from SS and returns it.
576 If SS is empty, returns EOF without modifying SS. */
578 ss_get_byte (struct substring *ss)
580 int c = ss_first (*ss);
589 /* Stores the prefix of SS up to the first DELIMITER in OUT (if
590 any). Trims those same bytes from SS. DELIMITER is
591 removed from SS but not made part of OUT. Returns true if
592 DELIMITER was found (and removed), false otherwise. */
594 ss_get_until (struct substring *ss, char delimiter, struct substring *out)
596 ss_get_bytes (ss, ss_cspan (*ss, ss_buffer (&delimiter, 1)), out);
597 return ss_match_byte (ss, delimiter);
600 /* Stores the first CNT bytes in SS in OUT (or fewer, if SS
601 is shorter than CNT bytes). Trims the same bytes
602 from the beginning of SS. Returns CNT. */
604 ss_get_bytes (struct substring *ss, size_t cnt, struct substring *out)
606 *out = ss_head (*ss, cnt);
607 ss_advance (ss, cnt);
611 /* Parses and removes an optionally signed decimal integer from
612 the beginning of SS. Returns 0 if an error occurred,
613 otherwise the number of bytes removed from SS. Stores
614 the integer's value into *VALUE. */
616 ss_get_long (struct substring *ss, long *value)
621 length = ss_span (*ss, ss_cstr ("+-"));
622 length += ss_span (ss_substr (*ss, length, SIZE_MAX), ss_cstr (CC_DIGITS));
623 if (length > 0 && length < sizeof tmp)
627 memcpy (tmp, ss_data (*ss), length);
630 *value = strtol (tmp, &tail, 10);
631 if (tail - tmp == length)
633 ss_advance (ss, length);
641 /* Returns true if SS is empty (has length 0 bytes),
644 ss_is_empty (struct substring ss)
646 return ss.length == 0;
649 /* Returns the number of bytes in SS. */
651 ss_length (struct substring ss)
656 /* Returns a pointer to the bytes in SS. */
658 ss_data (struct substring ss)
663 /* Returns a pointer just past the last byte in SS. */
665 ss_end (struct substring ss)
667 return ss.string + ss.length;
670 /* Returns the byte in position IDX in SS, as a value in the
671 range of unsigned char. Returns EOF if IDX is out of the
672 range of indexes for SS. */
674 ss_at (struct substring ss, size_t idx)
676 return idx < ss.length ? (unsigned char) ss.string[idx] : EOF;
679 /* Returns the first byte in SS as a value in the range of
680 unsigned char. Returns EOF if SS is the empty string. */
682 ss_first (struct substring ss)
684 return ss_at (ss, 0);
687 /* Returns the last byte in SS as a value in the range of
688 unsigned char. Returns EOF if SS is the empty string. */
690 ss_last (struct substring ss)
692 return ss.length > 0 ? (unsigned char) ss.string[ss.length - 1] : EOF;
695 /* Returns true if SS starts with PREFIX, false otherwise. */
697 ss_starts_with (struct substring ss, struct substring prefix)
699 return (ss.length >= prefix.length
700 && !memcmp (ss.string, prefix.string, prefix.length));
703 /* Returns true if SS ends with SUFFIX, false otherwise. */
705 ss_ends_with (struct substring ss, struct substring suffix)
707 return (ss.length >= suffix.length
708 && !memcmp (&ss.string[ss.length - suffix.length], suffix.string,
712 /* Returns the number of contiguous bytes at the beginning
713 of SS that are in SKIP_SET. */
715 ss_span (struct substring ss, struct substring skip_set)
718 for (i = 0; i < ss.length; i++)
719 if (ss_find_byte (skip_set, ss.string[i]) == SIZE_MAX)
724 /* Returns the number of contiguous bytes at the beginning
725 of SS that are not in SKIP_SET. */
727 ss_cspan (struct substring ss, struct substring stop_set)
730 for (i = 0; i < ss.length; i++)
731 if (ss_find_byte (stop_set, ss.string[i]) != SIZE_MAX)
736 /* Returns the offset in SS of the first instance of C,
737 or SIZE_MAX if C does not occur in SS. */
739 ss_find_byte (struct substring ss, char c)
741 const char *p = memchr (ss.string, c, ss.length);
742 return p != NULL ? p - ss.string : SIZE_MAX;
745 /* Returns the offset in HAYSTACK of the first instance of NEEDLE,
746 or SIZE_MAX if NEEDLE does not occur in HAYSTACK. */
748 ss_find_substring (struct substring haystack, struct substring needle)
750 const char *p = memmem (haystack.string, haystack.length,
751 needle.string, needle.length);
752 return p != NULL ? p - haystack.string : SIZE_MAX;
755 /* Compares A and B and returns a strcmp()-type comparison
758 ss_compare (struct substring a, struct substring b)
760 int retval = memcmp (a.string, b.string, MIN (a.length, b.length));
762 retval = a.length < b.length ? -1 : a.length > b.length;
766 /* Compares A and B case-insensitively and returns a
767 strcmp()-type comparison result. */
769 ss_compare_case (struct substring a, struct substring b)
771 int retval = memcasecmp (a.string, b.string, MIN (a.length, b.length));
773 retval = a.length < b.length ? -1 : a.length > b.length;
777 /* Compares A and B and returns true if their contents are
778 identical, false otherwise. */
780 ss_equals (struct substring a, struct substring b)
782 return a.length == b.length && !memcmp (a.string, b.string, a.length);
785 /* Compares A and B and returns true if their contents are
786 identical except possibly for case differences, false
789 ss_equals_case (struct substring a, struct substring b)
791 return a.length == b.length && !memcasecmp (a.string, b.string, a.length);
794 /* Returns the position in SS that the byte at P occupies.
795 P must point within SS or one past its end. */
797 ss_pointer_to_position (struct substring ss, const char *p)
799 size_t pos = p - ss.string;
800 assert (pos <= ss.length);
804 /* Allocates and returns a null-terminated string that contains
807 ss_xstrdup (struct substring ss)
809 char *s = xmalloc (ss.length + 1);
810 memcpy (s, ss.string, ss.length);
816 /* Returns the character represented by the UTF-8 sequence at the start of S.
817 The return value is either a Unicode code point in the range 0 to 0x10ffff,
818 or UINT32_MAX if S is empty. */
820 ss_first_mb (struct substring s)
822 return ss_at_mb (s, 0);
825 /* Returns the number of bytes in the UTF-8 character at the beginning of S.
827 The return value is 0 if S is empty, otherwise between 1 and 4. */
829 ss_first_mblen (struct substring s)
831 return ss_at_mblen (s, 0);
834 /* Advances S past the UTF-8 character at its beginning. Returns the Unicode
835 code point that was skipped (in the range 0 to 0x10ffff), or UINT32_MAX if S
836 was not modified because it was initially empty. */
838 ss_get_mb (struct substring *s)
845 n = u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s->string), s->length);
854 /* Returns the character represented by the UTF-8 sequence starting OFS bytes
855 into S. The return value is either a Unicode code point in the range 0 to
856 0x10ffff, or UINT32_MAX if OFS is past the last byte in S.
858 (Returns 0xfffd if OFS points into the middle, not the beginning, of a UTF-8
861 ss_at_mb (struct substring s, size_t ofs)
866 u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
874 /* Returns the number of bytes represented by the UTF-8 sequence starting OFS
875 bytes into S. The return value is 0 if OFS is past the last byte in S,
876 otherwise between 1 and 4. */
878 ss_at_mblen (struct substring s, size_t ofs)
883 return u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
890 /* Initializes ST as an empty string. */
892 ds_init_empty (struct string *st)
894 st->ss = ss_empty ();
898 /* Initializes ST with initial contents S. */
900 ds_init_string (struct string *st, const struct string *s)
902 ds_init_substring (st, ds_ss (s));
905 /* Initializes ST with initial contents SS. */
907 ds_init_substring (struct string *st, struct substring ss)
909 st->capacity = MAX (8, ss.length * 2);
910 st->ss.string = xmalloc (st->capacity + 1);
911 memcpy (st->ss.string, ss.string, ss.length);
912 st->ss.length = ss.length;
915 /* Initializes ST with initial contents S. */
917 ds_init_cstr (struct string *st, const char *s)
919 ds_init_substring (st, ss_cstr (s));
924 ds_destroy (struct string *st)
928 ss_dealloc (&st->ss);
929 st->ss.string = NULL;
935 /* Swaps the contents of strings A and B. */
937 ds_swap (struct string *a, struct string *b)
939 struct string tmp = *a;
944 /* Helper function for ds_register_pool. */
946 free_string (void *st_)
948 struct string *st = st_;
952 /* Arranges for ST to be destroyed automatically as part of
955 ds_register_pool (struct string *st, struct pool *pool)
957 pool_register (pool, free_string, st);
960 /* Cancels the arrangement for ST to be destroyed automatically
963 ds_unregister_pool (struct string *st, struct pool *pool)
965 pool_unregister (pool, st);
968 /* Copies SRC into DST.
969 DST and SRC may be the same string. */
971 ds_assign_string (struct string *dst, const struct string *src)
973 ds_assign_substring (dst, ds_ss (src));
976 /* Replaces DST by SS.
977 SS may be a substring of DST. */
979 ds_assign_substring (struct string *dst, struct substring ss)
981 dst->ss.length = ss.length;
982 ds_extend (dst, ss.length);
983 memmove (dst->ss.string, ss.string, ss.length);
986 /* Replaces DST by null-terminated string SRC. SRC may overlap
989 ds_assign_cstr (struct string *dst, const char *src)
991 ds_assign_substring (dst, ss_cstr (src));
994 /* Truncates ST to zero length. */
996 ds_clear (struct string *st)
1001 /* Returns a substring that contains ST. */
1003 ds_ss (const struct string *st)
1008 /* Returns a substring that contains CNT bytes from ST
1009 starting at position START.
1011 If START is greater than or equal to the length of ST, then
1012 the substring will be the empty string. If START + CNT
1013 exceeds the length of ST, then the substring will only be
1014 ds_length(ST) - START bytes long. */
1016 ds_substr (const struct string *st, size_t start, size_t cnt)
1018 return ss_substr (ds_ss (st), start, cnt);
1021 /* Returns a substring that contains the first CNT bytes in
1022 ST. If CNT exceeds the length of ST, then the substring will
1023 contain all of ST. */
1025 ds_head (const struct string *st, size_t cnt)
1027 return ss_head (ds_ss (st), cnt);
1030 /* Returns a substring that contains the last CNT bytes in
1031 ST. If CNT exceeds the length of ST, then the substring will
1032 contain all of ST. */
1034 ds_tail (const struct string *st, size_t cnt)
1036 return ss_tail (ds_ss (st), cnt);
1039 /* Ensures that ST can hold at least MIN_CAPACITY bytes plus a null
1042 ds_extend (struct string *st, size_t min_capacity)
1044 if (min_capacity > st->capacity)
1047 if (st->capacity < min_capacity)
1048 st->capacity = 2 * min_capacity;
1050 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
1054 /* Shrink ST to the minimum capacity need to contain its content. */
1056 ds_shrink (struct string *st)
1058 if (st->capacity != st->ss.length)
1060 st->capacity = st->ss.length;
1061 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
1065 /* Truncates ST to at most LENGTH bytes long. */
1067 ds_truncate (struct string *st, size_t length)
1069 ss_truncate (&st->ss, length);
1072 /* Removes trailing bytes in TRIM_SET from ST.
1073 Returns number of bytes removed. */
1075 ds_rtrim (struct string *st, struct substring trim_set)
1077 return ss_rtrim (&st->ss, trim_set);
1080 /* Removes leading bytes in TRIM_SET from ST.
1081 Returns number of bytes removed. */
1083 ds_ltrim (struct string *st, struct substring trim_set)
1085 size_t cnt = ds_span (st, trim_set);
1087 ds_assign_substring (st, ds_substr (st, cnt, SIZE_MAX));
1091 /* Trims leading and trailing bytes in TRIM_SET from ST.
1092 Returns number of bytes removed. */
1094 ds_trim (struct string *st, struct substring trim_set)
1096 size_t cnt = ds_rtrim (st, trim_set);
1097 return cnt + ds_ltrim (st, trim_set);
1100 /* If the last byte in ST is C, removes it and returns true.
1101 Otherwise, returns false without modifying ST. */
1103 ds_chomp_byte (struct string *st, char c)
1105 return ss_chomp_byte (&st->ss, c);
1108 /* If ST ends with SUFFIX, removes it and returns true.
1109 Otherwise, returns false without modifying ST. */
1111 ds_chomp (struct string *st, struct substring suffix)
1113 return ss_chomp (&st->ss, suffix);
1116 /* Divides ST into tokens separated by any of the DELIMITERS.
1117 Each call replaces TOKEN by the next token in ST, or by an
1118 empty string if no tokens remain. Returns true if a token was
1119 obtained, false otherwise.
1121 Before the first call, initialize *SAVE_IDX to 0. Do not
1122 modify *SAVE_IDX between calls.
1124 ST divides into exactly one more tokens than it contains
1125 delimiters. That is, a delimiter at the start or end of ST or
1126 a pair of adjacent delimiters yields an empty token, and the
1127 empty string contains a single token. */
1129 ds_separate (const struct string *st, struct substring delimiters,
1130 size_t *save_idx, struct substring *token)
1132 return ss_separate (ds_ss (st), delimiters, save_idx, token);
1135 /* Divides ST into tokens separated by any of the DELIMITERS,
1136 merging adjacent delimiters so that the empty string is never
1137 produced as a token. Each call replaces TOKEN by the next
1138 token in ST, or by an empty string if no tokens remain.
1139 Returns true if a token was obtained, false otherwise.
1141 Before the first call, initialize *SAVE_IDX to 0. Do not
1142 modify *SAVE_IDX between calls. */
1144 ds_tokenize (const struct string *st, struct substring delimiters,
1145 size_t *save_idx, struct substring *token)
1147 return ss_tokenize (ds_ss (st), delimiters, save_idx, token);
1150 /* Pad ST on the right with copies of PAD until ST is at least
1151 LENGTH bytes in size. If ST is initially LENGTH
1152 bytes or longer, this is a no-op. */
1154 ds_rpad (struct string *st, size_t length, char pad)
1156 if (length > st->ss.length)
1157 ds_put_byte_multiple (st, pad, length - st->ss.length);
1160 /* Sets the length of ST to exactly NEW_LENGTH,
1161 either by truncating bytes from the end,
1162 or by padding on the right with PAD. */
1164 ds_set_length (struct string *st, size_t new_length, char pad)
1166 if (st->ss.length < new_length)
1167 ds_rpad (st, new_length, pad);
1169 st->ss.length = new_length;
1172 /* Removes N bytes from ST starting at offset START. */
1174 ds_remove (struct string *st, size_t start, size_t n)
1176 if (n > 0 && start < st->ss.length)
1178 if (st->ss.length - start <= n)
1180 /* All bytes at or beyond START are deleted. */
1181 st->ss.length = start;
1185 /* Some bytes remain and must be shifted into
1187 memmove (st->ss.string + st->ss.length,
1188 st->ss.string + st->ss.length + n,
1189 st->ss.length - start - n);
1195 /* There are no bytes to delete or no bytes at or
1196 beyond START, hence deletion is a no-op. */
1200 /* Returns true if ST is empty, false otherwise. */
1202 ds_is_empty (const struct string *st)
1204 return ss_is_empty (st->ss);
1207 /* Returns the length of ST. */
1209 ds_length (const struct string *st)
1211 return ss_length (ds_ss (st));
1214 /* Returns the string data inside ST. */
1216 ds_data (const struct string *st)
1218 return ss_data (ds_ss (st));
1221 /* Returns a pointer to the null terminator ST.
1222 This might not be an actual null byte unless ds_c_str() has
1223 been called since the last modification to ST. */
1225 ds_end (const struct string *st)
1227 return ss_end (ds_ss (st));
1230 /* Returns the byte in position IDX in ST, as a value in the
1231 range of unsigned char. Returns EOF if IDX is out of the
1232 range of indexes for ST. */
1234 ds_at (const struct string *st, size_t idx)
1236 return ss_at (ds_ss (st), idx);
1239 /* Returns the first byte in ST as a value in the range of
1240 unsigned char. Returns EOF if ST is the empty string. */
1242 ds_first (const struct string *st)
1244 return ss_first (ds_ss (st));
1247 /* Returns the last byte in ST as a value in the range of
1248 unsigned char. Returns EOF if ST is the empty string. */
1250 ds_last (const struct string *st)
1252 return ss_last (ds_ss (st));
1255 /* Returns true if ST ends with SUFFIX, false otherwise. */
1257 ds_ends_with (const struct string *st, struct substring suffix)
1259 return ss_ends_with (st->ss, suffix);
1262 /* Returns the number of consecutive bytes at the beginning
1263 of ST that are in SKIP_SET. */
1265 ds_span (const struct string *st, struct substring skip_set)
1267 return ss_span (ds_ss (st), skip_set);
1270 /* Returns the number of consecutive bytes at the beginning
1271 of ST that are not in STOP_SET. */
1273 ds_cspan (const struct string *st, struct substring stop_set)
1275 return ss_cspan (ds_ss (st), stop_set);
1278 /* Returns the position of the first occurrence of byte C in
1279 ST at or after position OFS, or SIZE_MAX if there is no such
1282 ds_find_byte (const struct string *st, char c)
1284 return ss_find_byte (ds_ss (st), c);
1287 /* Compares A and B and returns a strcmp()-type comparison
1290 ds_compare (const struct string *a, const struct string *b)
1292 return ss_compare (ds_ss (a), ds_ss (b));
1295 /* Returns the position in ST that the byte at P occupies.
1296 P must point within ST or one past its end. */
1298 ds_pointer_to_position (const struct string *st, const char *p)
1300 return ss_pointer_to_position (ds_ss (st), p);
1303 /* Allocates and returns a null-terminated string that contains
1306 ds_xstrdup (const struct string *st)
1308 return ss_xstrdup (ds_ss (st));
1311 /* Returns the allocation size of ST. */
1313 ds_capacity (const struct string *st)
1315 return st->capacity;
1318 /* Returns the value of ST as a null-terminated string. */
1320 ds_cstr (const struct string *st_)
1322 struct string *st = CONST_CAST (struct string *, st_);
1323 if (st->ss.string == NULL)
1325 st->ss.string[st->ss.length] = '\0';
1326 return st->ss.string;
1329 /* Returns the value of ST as a null-terminated string and then
1330 reinitialized ST as an empty string. The caller must free the
1331 returned string with free(). */
1333 ds_steal_cstr (struct string *st)
1335 char *s = ds_cstr (st);
1340 /* Reads bytes from STREAM and appends them to ST, stopping
1341 after MAX_LENGTH bytes, after appending a newline, or
1342 after an I/O error or end of file was encountered, whichever
1343 comes first. Returns true if at least one byte was added
1344 to ST, false if no bytes were read before an I/O error or
1345 end of file (or if MAX_LENGTH was 0).
1347 This function treats LF and CR LF sequences as new-line,
1348 translating each of them to a single '\n' in ST. */
1350 ds_read_line (struct string *st, FILE *stream, size_t max_length)
1354 for (length = 0; length < max_length; length++)
1356 int c = getc (stream);
1363 ds_put_byte (st, c);
1370 /* CR followed by LF is special: translate to \n. */
1371 ds_put_byte (st, '\n');
1376 /* CR followed by anything else is just CR. */
1377 ds_put_byte (st, '\r');
1385 ds_put_byte (st, c);
1392 /* Removes a comment introduced by `#' from ST,
1393 ignoring occurrences inside quoted strings. */
1395 remove_comment (struct string *st)
1400 for (cp = ds_data (st); cp < ds_end (st); cp++)
1405 else if (*cp == '\\')
1408 else if (*cp == '\'' || *cp == '"')
1410 else if (*cp == '#')
1412 ds_truncate (st, cp - ds_cstr (st));
1417 /* Reads a line from STREAM into ST, then preprocesses as follows:
1419 - Splices lines terminated with `\'.
1421 - Deletes comments introduced by `#' outside of single or double
1424 - Deletes trailing white space.
1426 Returns true if a line was successfully read, false on
1427 failure. If LINE_NUMBER is non-null, then *LINE_NUMBER is
1428 incremented by the number of lines read. */
1430 ds_read_config_line (struct string *st, int *line_number, FILE *stream)
1435 if (!ds_read_line (st, stream, SIZE_MAX))
1438 ds_rtrim (st, ss_cstr (CC_SPACES));
1440 while (ds_chomp_byte (st, '\\'));
1442 remove_comment (st);
1446 /* Attempts to read SIZE * CNT bytes from STREAM and append them
1448 Returns true if all the requested data was read, false otherwise. */
1450 ds_read_stream (struct string *st, size_t size, size_t cnt, FILE *stream)
1454 size_t try_bytes = xtimes (cnt, size);
1455 if (size_in_bounds_p (xsum (ds_length (st), try_bytes)))
1457 char *buffer = ds_put_uninit (st, try_bytes);
1458 size_t got_bytes = fread (buffer, 1, try_bytes, stream);
1459 ds_truncate (st, ds_length (st) - (try_bytes - got_bytes));
1460 return got_bytes == try_bytes;
1472 /* Concatenates S onto ST. */
1474 ds_put_cstr (struct string *st, const char *s)
1477 ds_put_substring (st, ss_cstr (s));
1480 /* Concatenates SS to ST. */
1482 ds_put_substring (struct string *st, struct substring ss)
1484 memcpy (ds_put_uninit (st, ss_length (ss)), ss_data (ss), ss_length (ss));
1487 /* Returns ds_end(ST) and THEN increases the length by INCR. */
1489 ds_put_uninit (struct string *st, size_t incr)
1492 ds_extend (st, ds_length (st) + incr);
1494 st->ss.length += incr;
1498 /* Moves the bytes in ST following offset OFS + OLD_LEN in ST to offset OFS +
1499 NEW_LEN and returns the byte at offset OFS. The first min(OLD_LEN, NEW_LEN)
1500 bytes at the returned position are unchanged; if NEW_LEN > OLD_LEN then the
1501 following NEW_LEN - OLD_LEN bytes are initially indeterminate.
1503 The intention is that the caller should write NEW_LEN bytes at the returned
1504 position, to effectively replace the OLD_LEN bytes previously at that
1507 ds_splice_uninit (struct string *st,
1508 size_t ofs, size_t old_len, size_t new_len)
1510 if (new_len != old_len)
1512 if (new_len > old_len)
1513 ds_extend (st, ds_length (st) + (new_len - old_len));
1515 assert (ds_length (st) >= ofs + old_len);
1517 memmove (ds_data (st) + (ofs + new_len),
1518 ds_data (st) + (ofs + old_len),
1519 ds_length (st) - (ofs + old_len));
1520 st->ss.length += new_len - old_len;
1522 return ds_data (st) + ofs;
1525 /* Formats FORMAT as a printf string and appends the result to ST. */
1527 ds_put_format (struct string *st, const char *format, ...)
1531 va_start (args, format);
1532 ds_put_vformat (st, format, args);
1536 /* Formats FORMAT as a printf string as if in the C locale and appends the result to ST. */
1538 ds_put_c_format (struct string *st, const char *format, ...)
1542 va_start (args, format);
1543 ds_put_c_vformat (st, format, args);
1547 /* Formats FORMAT as a printf string and appends the result to ST. */
1549 ds_put_vformat (struct string *st, const char *format, va_list args_)
1554 va_copy (args, args_);
1555 avail = st->ss.string != NULL ? st->capacity - st->ss.length + 1 : 0;
1556 needed = vsnprintf (st->ss.string + st->ss.length, avail, format, args);
1559 if (needed >= avail)
1561 va_copy (args, args_);
1562 vsnprintf (ds_put_uninit (st, needed), needed + 1, format, args);
1567 /* Some old libc's returned -1 when the destination string
1569 while (needed == -1)
1571 ds_extend (st, (st->capacity + 1) * 2);
1572 avail = st->capacity - st->ss.length + 1;
1574 va_copy (args, args_);
1575 needed = vsnprintf (ds_end (st), avail, format, args);
1578 st->ss.length += needed;
1582 /* Formats FORMAT as a printf string, as if in the C locale,
1583 and appends the result to ST. */
1585 ds_put_c_vformat (struct string *st, const char *format, va_list args)
1588 size_t len = sizeof buf;
1589 char *output = c_vasnprintf (buf, &len, format, args);
1592 ds_put_cstr (st, output);
1598 /* Appends byte CH to ST. */
1600 ds_put_byte (struct string *st, int ch)
1602 ds_put_uninit (st, 1)[0] = ch;
1605 /* Appends CNT copies of byte CH to ST. */
1607 ds_put_byte_multiple (struct string *st, int ch, size_t cnt)
1609 memset (ds_put_uninit (st, cnt), ch, cnt);
1612 /* Appends Unicode code point UC to ST in UTF-8 encoding. */
1614 ds_put_unichar (struct string *st, ucs4_t uc)
1616 ds_extend (st, ds_length (st) + 6);
1617 st->ss.length += u8_uctomb (CHAR_CAST (uint8_t *, ds_end (st)), uc, 6);
1620 /* If relocation has been enabled, replace ST,
1621 with its relocated version */
1623 ds_relocate (struct string *st)
1625 const char *orig = ds_cstr (st);
1626 const char *rel = relocate (orig);
1631 ds_put_cstr (st, rel);
1632 /* The documentation for relocate says that casting away const
1633 and then freeing is appropriate ... */
1634 free (CONST_CAST (char *, rel));
1641 /* Operations on uint8_t "strings" */
1643 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
1644 DST is truncated to DST_SIZE bytes or padded on the right with
1645 copies of PAD as needed. */
1647 u8_buf_copy_rpad (uint8_t *dst, size_t dst_size,
1648 const uint8_t *src, size_t src_size,
1651 if (src_size >= dst_size)
1652 memmove (dst, src, dst_size);
1655 memmove (dst, src, src_size);
1656 memset (&dst[src_size], pad, dst_size - src_size);