1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
27 #include "libpspp/cast.h"
28 #include "libpspp/message.h"
29 #include "libpspp/pool.h"
31 #include "gl/relocatable.h"
32 #include "gl/minmax.h"
33 #include "gl/xalloc.h"
34 #include "gl/xmemdup0.h"
37 /* Reverses the order of NBYTES bytes at address P, thus converting
38 between little- and big-endian byte orders. */
40 buf_reverse (char *p, size_t nbytes)
42 char *h = p, *t = &h[nbytes - 1];
54 /* Compares the SIZE bytes in A to those in B, disregarding case,
55 and returns a strcmp()-type result. */
57 buf_compare_case (const char *a_, const char *b_, size_t size)
59 const unsigned char *a = (unsigned char *) a_;
60 const unsigned char *b = (unsigned char *) b_;
64 unsigned char ac = toupper (*a++);
65 unsigned char bc = toupper (*b++);
68 return ac > bc ? 1 : -1;
74 /* Compares A of length A_LEN to B of length B_LEN. The shorter
75 string is considered to be padded with spaces to the length of
78 buf_compare_rpad (const char *a, size_t a_len, const char *b, size_t b_len)
83 min_len = a_len < b_len ? a_len : b_len;
84 result = memcmp (a, b, min_len);
93 for (idx = min_len; idx < b_len; idx++)
95 return ' ' > b[idx] ? 1 : -1;
99 for (idx = min_len; idx < a_len; idx++)
101 return a[idx] > ' ' ? 1 : -1;
107 /* Compares strin A to string B. The shorter string is
108 considered to be padded with spaces to the length of the
111 str_compare_rpad (const char *a, const char *b)
113 return buf_compare_rpad (a, strlen (a), b, strlen (b));
116 /* Copies string SRC to buffer DST, of size DST_SIZE bytes.
117 DST is truncated to DST_SIZE bytes or padded on the right with
118 copies of PAD as needed. */
120 buf_copy_str_rpad (char *dst, size_t dst_size, const char *src, char pad)
122 size_t src_len = strlen (src);
123 if (src_len >= dst_size)
124 memcpy (dst, src, dst_size);
127 memcpy (dst, src, src_len);
128 memset (&dst[src_len], pad, dst_size - src_len);
132 /* Copies string SRC to buffer DST, of size DST_SIZE bytes.
133 DST is truncated to DST_SIZE bytes or padded on the left with
134 copies of PAD as needed. */
136 buf_copy_str_lpad (char *dst, size_t dst_size, const char *src, char pad)
138 size_t src_len = strlen (src);
139 if (src_len >= dst_size)
140 memcpy (dst, src, dst_size);
143 size_t pad_cnt = dst_size - src_len;
144 memset (&dst[0], pad, pad_cnt);
145 memcpy (dst + pad_cnt, src, src_len);
149 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
150 DST is truncated to DST_SIZE bytes or padded on the left with
151 copies of PAD as needed. */
153 buf_copy_lpad (char *dst, size_t dst_size,
154 const char *src, size_t src_size,
157 if (src_size >= dst_size)
158 memmove (dst, src, dst_size);
161 memset (dst, pad, dst_size - src_size);
162 memmove (&dst[dst_size - src_size], src, src_size);
166 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
167 DST is truncated to DST_SIZE bytes or padded on the right with
168 copies of PAD as needed. */
170 buf_copy_rpad (char *dst, size_t dst_size,
171 const char *src, size_t src_size,
174 if (src_size >= dst_size)
175 memmove (dst, src, dst_size);
178 memmove (dst, src, src_size);
179 memset (&dst[src_size], pad, dst_size - src_size);
183 /* Copies string SRC to string DST, which is in a buffer DST_SIZE
185 Truncates DST to DST_SIZE - 1 bytes or right-pads with
186 spaces to DST_SIZE - 1 bytes if necessary. */
188 str_copy_rpad (char *dst, size_t dst_size, const char *src)
192 size_t src_len = strlen (src);
193 if (src_len < dst_size - 1)
195 memcpy (dst, src, src_len);
196 memset (&dst[src_len], ' ', dst_size - 1 - src_len);
199 memcpy (dst, src, dst_size - 1);
200 dst[dst_size - 1] = 0;
204 /* Copies SRC to DST, which is in a buffer DST_SIZE bytes long.
205 Truncates DST to DST_SIZE - 1 bytes, if necessary. */
207 str_copy_trunc (char *dst, size_t dst_size, const char *src)
209 size_t src_len = strlen (src);
210 assert (dst_size > 0);
211 if (src_len + 1 < dst_size)
212 memcpy (dst, src, src_len + 1);
215 memcpy (dst, src, dst_size - 1);
216 dst[dst_size - 1] = '\0';
220 /* Copies buffer SRC, of SRC_LEN bytes,
221 to DST, which is in a buffer DST_SIZE bytes long.
222 Truncates DST to DST_SIZE - 1 bytes, if necessary. */
224 str_copy_buf_trunc (char *dst, size_t dst_size,
225 const char *src, size_t src_size)
228 assert (dst_size > 0);
230 dst_len = src_size < dst_size ? src_size : dst_size - 1;
231 memcpy (dst, src, dst_len);
235 /* Converts each byte in S to uppercase. */
237 str_uppercase (char *s)
239 for (; *s != '\0'; s++)
240 *s = toupper ((unsigned char) *s);
243 /* Converts each byte in S to lowercase. */
245 str_lowercase (char *s)
247 for (; *s != '\0'; s++)
248 *s = tolower ((unsigned char) *s);
251 /* Converts NUMBER into a string in 26-adic notation in BUFFER,
252 which has room for SIZE bytes. Returns true if successful,
253 false if NUMBER, plus a trailing null, is too large to fit in
256 26-adic notation is "spreadsheet column numbering": 1 = A, 2 =
257 B, 3 = C, ... 26 = Z, 27 = AA, 28 = AB, 29 = AC, ...
259 26-adic notation is the special case of a k-adic numeration
260 system (aka bijective base-k numeration) with k=26. In k-adic
261 numeration, the digits are {1, 2, 3, ..., k} (there is no
262 digit 0), and integer 0 is represented by the empty string.
263 For more information, see
264 http://en.wikipedia.org/wiki/Bijective_numeration. */
266 str_format_26adic (unsigned long int number, char buffer[], size_t size)
274 buffer[length++] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"[number % 26];
280 buffer[length] = '\0';
282 buf_reverse (buffer, length);
286 /* Sets the SIZE bytes starting at BLOCK to C,
287 and returns the byte following BLOCK. */
289 mempset (void *block, int c, size_t size)
291 memset (block, c, size);
292 return (char *) block + size;
297 /* Returns a substring whose contents are the CNT bytes
298 starting at the (0-based) position START in SS. */
300 ss_substr (struct substring ss, size_t start, size_t cnt)
302 if (start < ss.length)
303 return ss_buffer (ss.string + start, MIN (cnt, ss.length - start));
305 return ss_buffer (ss.string + ss.length, 0);
308 /* Returns a substring whose contents are the first CNT
311 ss_head (struct substring ss, size_t cnt)
313 return ss_buffer (ss.string, MIN (cnt, ss.length));
316 /* Returns a substring whose contents are the last CNT bytes
319 ss_tail (struct substring ss, size_t cnt)
322 return ss_buffer (ss.string + (ss.length - cnt), cnt);
327 /* Makes a malloc()'d, null-terminated copy of the contents of OLD
328 and stores it in NEW. */
330 ss_alloc_substring (struct substring *new, struct substring old)
332 new->string = xmemdup0 (old.string, old.length);
333 new->length = old.length;
336 /* Allocates room for a CNT-byte string in NEW. */
338 ss_alloc_uninit (struct substring *new, size_t cnt)
340 new->string = xmalloc (cnt);
345 ss_realloc (struct substring *ss, size_t size)
347 ss->string = xrealloc (ss->string, size);
350 /* Makes a pool_alloc_unaligned()'d copy of the contents of OLD
351 in POOL, and stores it in NEW. */
353 ss_alloc_substring_pool (struct substring *new, struct substring old,
356 new->string = pool_alloc_unaligned (pool, old.length);
357 new->length = old.length;
358 memcpy (new->string, old.string, old.length);
361 /* Allocates room for a CNT-byte string in NEW in POOL. */
363 ss_alloc_uninit_pool (struct substring *new, size_t cnt, struct pool *pool)
365 new->string = pool_alloc_unaligned (pool, cnt);
369 /* Frees the string that SS points to. */
371 ss_dealloc (struct substring *ss)
376 /* Truncates SS to at most CNT bytes in length. */
378 ss_truncate (struct substring *ss, size_t cnt)
380 if (ss->length > cnt)
384 /* Removes trailing bytes in TRIM_SET from SS.
385 Returns number of bytes removed. */
387 ss_rtrim (struct substring *ss, struct substring trim_set)
390 while (cnt < ss->length
391 && ss_find_byte (trim_set,
392 ss->string[ss->length - cnt - 1]) != SIZE_MAX)
398 /* Removes leading bytes in TRIM_SET from SS.
399 Returns number of bytes removed. */
401 ss_ltrim (struct substring *ss, struct substring trim_set)
403 size_t cnt = ss_span (*ss, trim_set);
404 ss_advance (ss, cnt);
408 /* Trims leading and trailing bytes in TRIM_SET from SS. */
410 ss_trim (struct substring *ss, struct substring trim_set)
412 ss_ltrim (ss, trim_set);
413 ss_rtrim (ss, trim_set);
416 /* If the last byte in SS is C, removes it and returns true.
417 Otherwise, returns false without changing the string. */
419 ss_chomp_byte (struct substring *ss, char c)
421 if (ss_last (*ss) == c)
430 /* If SS ends with SUFFIX, removes it and returns true.
431 Otherwise, returns false without changing the string. */
433 ss_chomp (struct substring *ss, struct substring suffix)
435 if (ss_ends_with (*ss, suffix))
437 ss->length -= suffix.length;
444 /* Divides SS into tokens separated by any of the DELIMITERS.
445 Each call replaces TOKEN by the next token in SS, or by an
446 empty string if no tokens remain. Returns true if a token was
447 obtained, false otherwise.
449 Before the first call, initialize *SAVE_IDX to 0. Do not
450 modify *SAVE_IDX between calls.
452 SS divides into exactly one more tokens than it contains
453 delimiters. That is, a delimiter at the start or end of SS or
454 a pair of adjacent delimiters yields an empty token, and the
455 empty string contains a single token. */
457 ss_separate (struct substring ss, struct substring delimiters,
458 size_t *save_idx, struct substring *token)
460 if (*save_idx <= ss_length (ss))
462 struct substring tmp = ss_substr (ss, *save_idx, SIZE_MAX);
463 size_t length = ss_cspan (tmp, delimiters);
464 *token = ss_head (tmp, length);
465 *save_idx += length + 1;
470 *token = ss_empty ();
475 /* Divides SS into tokens separated by any of the DELIMITERS,
476 merging adjacent delimiters so that the empty string is never
477 produced as a token. Each call replaces TOKEN by the next
478 token in SS, or by an empty string if no tokens remain, and
479 then skips past the first delimiter following the token.
480 Returns true if a token was obtained, false otherwise.
482 Before the first call, initialize *SAVE_IDX to 0. Do not
483 modify *SAVE_IDX between calls. */
485 ss_tokenize (struct substring ss, struct substring delimiters,
486 size_t *save_idx, struct substring *token)
488 ss_advance (&ss, *save_idx);
489 *save_idx += ss_ltrim (&ss, delimiters);
490 ss_get_bytes (&ss, ss_cspan (ss, delimiters), token);
491 *save_idx += ss_length (*token) + 1;
492 return ss_length (*token) > 0;
495 /* Removes the first CNT bytes from SS. */
497 ss_advance (struct substring *ss, size_t cnt)
499 if (cnt > ss->length)
505 /* If the first byte in SS is C, removes it and returns true.
506 Otherwise, returns false without changing the string. */
508 ss_match_byte (struct substring *ss, char c)
510 if (ss_first (*ss) == c)
520 /* If the first byte in SS is in MATCH, removes it and
521 returns the byte that was removed.
522 Otherwise, returns EOF without changing the string. */
524 ss_match_byte_in (struct substring *ss, struct substring match)
528 && memchr (match.string, ss->string[0], match.length) != NULL)
537 /* If SS begins with TARGET, removes it and returns true.
538 Otherwise, returns false without changing SS. */
540 ss_match_string (struct substring *ss, const struct substring target)
542 size_t length = ss_length (target);
543 if (ss_equals (ss_head (*ss, length), target))
545 ss_advance (ss, length);
552 /* Removes the first byte from SS and returns it.
553 If SS is empty, returns EOF without modifying SS. */
555 ss_get_byte (struct substring *ss)
557 int c = ss_first (*ss);
566 /* Stores the prefix of SS up to the first DELIMITER in OUT (if
567 any). Trims those same bytes from SS. DELIMITER is
568 removed from SS but not made part of OUT. Returns true if
569 DELIMITER was found (and removed), false otherwise. */
571 ss_get_until (struct substring *ss, char delimiter, struct substring *out)
573 ss_get_bytes (ss, ss_cspan (*ss, ss_buffer (&delimiter, 1)), out);
574 return ss_match_byte (ss, delimiter);
577 /* Stores the first CNT bytes in SS in OUT (or fewer, if SS
578 is shorter than CNT bytes). Trims the same bytes
579 from the beginning of SS. Returns CNT. */
581 ss_get_bytes (struct substring *ss, size_t cnt, struct substring *out)
583 *out = ss_head (*ss, cnt);
584 ss_advance (ss, cnt);
588 /* Parses and removes an optionally signed decimal integer from
589 the beginning of SS. Returns 0 if an error occurred,
590 otherwise the number of bytes removed from SS. Stores
591 the integer's value into *VALUE. */
593 ss_get_long (struct substring *ss, long *value)
598 length = ss_span (*ss, ss_cstr ("+-"));
599 length += ss_span (ss_substr (*ss, length, SIZE_MAX), ss_cstr (CC_DIGITS));
600 if (length > 0 && length < sizeof tmp)
604 memcpy (tmp, ss_data (*ss), length);
607 *value = strtol (tmp, &tail, 10);
608 if (tail - tmp == length)
610 ss_advance (ss, length);
618 /* Returns true if SS is empty (has length 0 bytes),
621 ss_is_empty (struct substring ss)
623 return ss.length == 0;
626 /* Returns the number of bytes in SS. */
628 ss_length (struct substring ss)
633 /* Returns a pointer to the bytes in SS. */
635 ss_data (struct substring ss)
640 /* Returns a pointer just past the last byte in SS. */
642 ss_end (struct substring ss)
644 return ss.string + ss.length;
647 /* Returns the byte in position IDX in SS, as a value in the
648 range of unsigned char. Returns EOF if IDX is out of the
649 range of indexes for SS. */
651 ss_at (struct substring ss, size_t idx)
653 return idx < ss.length ? (unsigned char) ss.string[idx] : EOF;
656 /* Returns the first byte in SS as a value in the range of
657 unsigned char. Returns EOF if SS is the empty string. */
659 ss_first (struct substring ss)
661 return ss_at (ss, 0);
664 /* Returns the last byte in SS as a value in the range of
665 unsigned char. Returns EOF if SS is the empty string. */
667 ss_last (struct substring ss)
669 return ss.length > 0 ? (unsigned char) ss.string[ss.length - 1] : EOF;
672 /* Returns true if SS ends with SUFFIX, false otherwise. */
674 ss_ends_with (struct substring ss, struct substring suffix)
676 return (ss.length >= suffix.length
677 && !memcmp (&ss.string[ss.length - suffix.length], suffix.string,
681 /* Returns the number of contiguous bytes at the beginning
682 of SS that are in SKIP_SET. */
684 ss_span (struct substring ss, struct substring skip_set)
687 for (i = 0; i < ss.length; i++)
688 if (ss_find_byte (skip_set, ss.string[i]) == SIZE_MAX)
693 /* Returns the number of contiguous bytes at the beginning
694 of SS that are not in SKIP_SET. */
696 ss_cspan (struct substring ss, struct substring stop_set)
699 for (i = 0; i < ss.length; i++)
700 if (ss_find_byte (stop_set, ss.string[i]) != SIZE_MAX)
705 /* Returns the offset in SS of the first instance of C,
706 or SIZE_MAX if C does not occur in SS. */
708 ss_find_byte (struct substring ss, char c)
710 const char *p = memchr (ss.string, c, ss.length);
711 return p != NULL ? p - ss.string : SIZE_MAX;
714 /* Compares A and B and returns a strcmp()-type comparison
717 ss_compare (struct substring a, struct substring b)
719 int retval = memcmp (a.string, b.string, MIN (a.length, b.length));
721 retval = a.length < b.length ? -1 : a.length > b.length;
725 /* Compares A and B case-insensitively and returns a
726 strcmp()-type comparison result. */
728 ss_compare_case (struct substring a, struct substring b)
730 int retval = memcasecmp (a.string, b.string, MIN (a.length, b.length));
732 retval = a.length < b.length ? -1 : a.length > b.length;
736 /* Compares A and B and returns true if their contents are
737 identical, false otherwise. */
739 ss_equals (struct substring a, struct substring b)
741 return a.length == b.length && !memcmp (a.string, b.string, a.length);
744 /* Compares A and B and returns true if their contents are
745 identical except possibly for case differences, false
748 ss_equals_case (struct substring a, struct substring b)
750 return a.length == b.length && !memcasecmp (a.string, b.string, a.length);
753 /* Returns the position in SS that the byte at P occupies.
754 P must point within SS or one past its end. */
756 ss_pointer_to_position (struct substring ss, const char *p)
758 size_t pos = p - ss.string;
759 assert (pos <= ss.length);
763 /* Allocates and returns a null-terminated string that contains
766 ss_xstrdup (struct substring ss)
768 char *s = xmalloc (ss.length + 1);
769 memcpy (s, ss.string, ss.length);
775 /* Returns the character represented by the UTF-8 sequence at the start of S.
776 The return value is either a Unicode code point in the range 0 to 0x10ffff,
777 or UINT32_MAX if S is empty. */
779 ss_first_mb (struct substring s)
781 return ss_at_mb (s, 0);
784 /* Returns the number of bytes in the UTF-8 character at the beginning of S.
786 The return value is 0 if S is empty, otherwise between 1 and 4. */
788 ss_first_mblen (struct substring s)
790 return ss_at_mblen (s, 0);
793 /* Advances S past the UTF-8 character at its beginning. Returns the Unicode
794 code point that was skipped (in the range 0 to 0x10ffff), or UINT32_MAX if S
795 was not modified because it was initially empty. */
797 ss_get_mb (struct substring *s)
804 n = u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s->string), s->length);
813 /* Returns the character represented by the UTF-8 sequence starting OFS bytes
814 into S. The return value is either a Unicode code point in the range 0 to
815 0x10ffff, or UINT32_MAX if OFS is past the last byte in S.
817 (Returns 0xfffd if OFS points into the middle, not the beginning, of a UTF-8
820 ss_at_mb (struct substring s, size_t ofs)
825 u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
833 /* Returns the number of bytes represented by the UTF-8 sequence starting OFS
834 bytes into S. The return value is 0 if OFS is past the last byte in S,
835 otherwise between 1 and 4. */
837 ss_at_mblen (struct substring s, size_t ofs)
842 return u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
849 /* Initializes ST as an empty string. */
851 ds_init_empty (struct string *st)
853 st->ss = ss_empty ();
857 /* Initializes ST with initial contents S. */
859 ds_init_string (struct string *st, const struct string *s)
861 ds_init_substring (st, ds_ss (s));
864 /* Initializes ST with initial contents SS. */
866 ds_init_substring (struct string *st, struct substring ss)
868 st->capacity = MAX (8, ss.length * 2);
869 st->ss.string = xmalloc (st->capacity + 1);
870 memcpy (st->ss.string, ss.string, ss.length);
871 st->ss.length = ss.length;
874 /* Initializes ST with initial contents S. */
876 ds_init_cstr (struct string *st, const char *s)
878 ds_init_substring (st, ss_cstr (s));
883 ds_destroy (struct string *st)
887 ss_dealloc (&st->ss);
888 st->ss.string = NULL;
894 /* Swaps the contents of strings A and B. */
896 ds_swap (struct string *a, struct string *b)
898 struct string tmp = *a;
903 /* Helper function for ds_register_pool. */
905 free_string (void *st_)
907 struct string *st = st_;
911 /* Arranges for ST to be destroyed automatically as part of
914 ds_register_pool (struct string *st, struct pool *pool)
916 pool_register (pool, free_string, st);
919 /* Cancels the arrangement for ST to be destroyed automatically
922 ds_unregister_pool (struct string *st, struct pool *pool)
924 pool_unregister (pool, st);
927 /* Copies SRC into DST.
928 DST and SRC may be the same string. */
930 ds_assign_string (struct string *dst, const struct string *src)
932 ds_assign_substring (dst, ds_ss (src));
935 /* Replaces DST by SS.
936 SS may be a substring of DST. */
938 ds_assign_substring (struct string *dst, struct substring ss)
940 dst->ss.length = ss.length;
941 ds_extend (dst, ss.length);
942 memmove (dst->ss.string, ss.string, ss.length);
945 /* Replaces DST by null-terminated string SRC. SRC may overlap
948 ds_assign_cstr (struct string *dst, const char *src)
950 ds_assign_substring (dst, ss_cstr (src));
953 /* Truncates ST to zero length. */
955 ds_clear (struct string *st)
960 /* Returns a substring that contains ST. */
962 ds_ss (const struct string *st)
967 /* Returns a substring that contains CNT bytes from ST
968 starting at position START.
970 If START is greater than or equal to the length of ST, then
971 the substring will be the empty string. If START + CNT
972 exceeds the length of ST, then the substring will only be
973 ds_length(ST) - START bytes long. */
975 ds_substr (const struct string *st, size_t start, size_t cnt)
977 return ss_substr (ds_ss (st), start, cnt);
980 /* Returns a substring that contains the first CNT bytes in
981 ST. If CNT exceeds the length of ST, then the substring will
982 contain all of ST. */
984 ds_head (const struct string *st, size_t cnt)
986 return ss_head (ds_ss (st), cnt);
989 /* Returns a substring that contains the last CNT bytes in
990 ST. If CNT exceeds the length of ST, then the substring will
991 contain all of ST. */
993 ds_tail (const struct string *st, size_t cnt)
995 return ss_tail (ds_ss (st), cnt);
998 /* Ensures that ST can hold at least MIN_CAPACITY bytes plus a null
1001 ds_extend (struct string *st, size_t min_capacity)
1003 if (min_capacity > st->capacity)
1006 if (st->capacity < min_capacity)
1007 st->capacity = 2 * min_capacity;
1009 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
1013 /* Shrink ST to the minimum capacity need to contain its content. */
1015 ds_shrink (struct string *st)
1017 if (st->capacity != st->ss.length)
1019 st->capacity = st->ss.length;
1020 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
1024 /* Truncates ST to at most LENGTH bytes long. */
1026 ds_truncate (struct string *st, size_t length)
1028 ss_truncate (&st->ss, length);
1031 /* Removes trailing bytes in TRIM_SET from ST.
1032 Returns number of bytes removed. */
1034 ds_rtrim (struct string *st, struct substring trim_set)
1036 return ss_rtrim (&st->ss, trim_set);
1039 /* Removes leading bytes in TRIM_SET from ST.
1040 Returns number of bytes removed. */
1042 ds_ltrim (struct string *st, struct substring trim_set)
1044 size_t cnt = ds_span (st, trim_set);
1046 ds_assign_substring (st, ds_substr (st, cnt, SIZE_MAX));
1050 /* Trims leading and trailing bytes in TRIM_SET from ST.
1051 Returns number of bytes removed. */
1053 ds_trim (struct string *st, struct substring trim_set)
1055 size_t cnt = ds_rtrim (st, trim_set);
1056 return cnt + ds_ltrim (st, trim_set);
1059 /* If the last byte in ST is C, removes it and returns true.
1060 Otherwise, returns false without modifying ST. */
1062 ds_chomp_byte (struct string *st, char c)
1064 return ss_chomp_byte (&st->ss, c);
1067 /* If ST ends with SUFFIX, removes it and returns true.
1068 Otherwise, returns false without modifying ST. */
1070 ds_chomp (struct string *st, struct substring suffix)
1072 return ss_chomp (&st->ss, suffix);
1075 /* Divides ST into tokens separated by any of the DELIMITERS.
1076 Each call replaces TOKEN by the next token in ST, or by an
1077 empty string if no tokens remain. Returns true if a token was
1078 obtained, false otherwise.
1080 Before the first call, initialize *SAVE_IDX to 0. Do not
1081 modify *SAVE_IDX between calls.
1083 ST divides into exactly one more tokens than it contains
1084 delimiters. That is, a delimiter at the start or end of ST or
1085 a pair of adjacent delimiters yields an empty token, and the
1086 empty string contains a single token. */
1088 ds_separate (const struct string *st, struct substring delimiters,
1089 size_t *save_idx, struct substring *token)
1091 return ss_separate (ds_ss (st), delimiters, save_idx, token);
1094 /* Divides ST into tokens separated by any of the DELIMITERS,
1095 merging adjacent delimiters so that the empty string is never
1096 produced as a token. Each call replaces TOKEN by the next
1097 token in ST, or by an empty string if no tokens remain.
1098 Returns true if a token was obtained, false otherwise.
1100 Before the first call, initialize *SAVE_IDX to 0. Do not
1101 modify *SAVE_IDX between calls. */
1103 ds_tokenize (const struct string *st, struct substring delimiters,
1104 size_t *save_idx, struct substring *token)
1106 return ss_tokenize (ds_ss (st), delimiters, save_idx, token);
1109 /* Pad ST on the right with copies of PAD until ST is at least
1110 LENGTH bytes in size. If ST is initially LENGTH
1111 bytes or longer, this is a no-op. */
1113 ds_rpad (struct string *st, size_t length, char pad)
1115 if (length > st->ss.length)
1116 ds_put_byte_multiple (st, pad, length - st->ss.length);
1119 /* Sets the length of ST to exactly NEW_LENGTH,
1120 either by truncating bytes from the end,
1121 or by padding on the right with PAD. */
1123 ds_set_length (struct string *st, size_t new_length, char pad)
1125 if (st->ss.length < new_length)
1126 ds_rpad (st, new_length, pad);
1128 st->ss.length = new_length;
1131 /* Removes N bytes from ST starting at offset START. */
1133 ds_remove (struct string *st, size_t start, size_t n)
1135 if (n > 0 && start < st->ss.length)
1137 if (st->ss.length - start <= n)
1139 /* All bytes at or beyond START are deleted. */
1140 st->ss.length = start;
1144 /* Some bytes remain and must be shifted into
1146 memmove (st->ss.string + st->ss.length,
1147 st->ss.string + st->ss.length + n,
1148 st->ss.length - start - n);
1154 /* There are no bytes to delete or no bytes at or
1155 beyond START, hence deletion is a no-op. */
1159 /* Returns true if ST is empty, false otherwise. */
1161 ds_is_empty (const struct string *st)
1163 return ss_is_empty (st->ss);
1166 /* Returns the length of ST. */
1168 ds_length (const struct string *st)
1170 return ss_length (ds_ss (st));
1173 /* Returns the string data inside ST. */
1175 ds_data (const struct string *st)
1177 return ss_data (ds_ss (st));
1180 /* Returns a pointer to the null terminator ST.
1181 This might not be an actual null byte unless ds_c_str() has
1182 been called since the last modification to ST. */
1184 ds_end (const struct string *st)
1186 return ss_end (ds_ss (st));
1189 /* Returns the byte in position IDX in ST, as a value in the
1190 range of unsigned char. Returns EOF if IDX is out of the
1191 range of indexes for ST. */
1193 ds_at (const struct string *st, size_t idx)
1195 return ss_at (ds_ss (st), idx);
1198 /* Returns the first byte in ST as a value in the range of
1199 unsigned char. Returns EOF if ST is the empty string. */
1201 ds_first (const struct string *st)
1203 return ss_first (ds_ss (st));
1206 /* Returns the last byte in ST as a value in the range of
1207 unsigned char. Returns EOF if ST is the empty string. */
1209 ds_last (const struct string *st)
1211 return ss_last (ds_ss (st));
1214 /* Returns true if ST ends with SUFFIX, false otherwise. */
1216 ds_ends_with (const struct string *st, struct substring suffix)
1218 return ss_ends_with (st->ss, suffix);
1221 /* Returns the number of consecutive bytes at the beginning
1222 of ST that are in SKIP_SET. */
1224 ds_span (const struct string *st, struct substring skip_set)
1226 return ss_span (ds_ss (st), skip_set);
1229 /* Returns the number of consecutive bytes at the beginning
1230 of ST that are not in STOP_SET. */
1232 ds_cspan (const struct string *st, struct substring stop_set)
1234 return ss_cspan (ds_ss (st), stop_set);
1237 /* Returns the position of the first occurrence of byte C in
1238 ST at or after position OFS, or SIZE_MAX if there is no such
1241 ds_find_byte (const struct string *st, char c)
1243 return ss_find_byte (ds_ss (st), c);
1246 /* Compares A and B and returns a strcmp()-type comparison
1249 ds_compare (const struct string *a, const struct string *b)
1251 return ss_compare (ds_ss (a), ds_ss (b));
1254 /* Returns the position in ST that the byte at P occupies.
1255 P must point within ST or one past its end. */
1257 ds_pointer_to_position (const struct string *st, const char *p)
1259 return ss_pointer_to_position (ds_ss (st), p);
1262 /* Allocates and returns a null-terminated string that contains
1265 ds_xstrdup (const struct string *st)
1267 return ss_xstrdup (ds_ss (st));
1270 /* Returns the allocation size of ST. */
1272 ds_capacity (const struct string *st)
1274 return st->capacity;
1277 /* Returns the value of ST as a null-terminated string. */
1279 ds_cstr (const struct string *st_)
1281 struct string *st = CONST_CAST (struct string *, st_);
1282 if (st->ss.string == NULL)
1284 st->ss.string[st->ss.length] = '\0';
1285 return st->ss.string;
1288 /* Returns the value of ST as a null-terminated string and then
1289 reinitialized ST as an empty string. The caller must free the
1290 returned string with free(). */
1292 ds_steal_cstr (struct string *st)
1294 char *s = ds_cstr (st);
1299 /* Reads bytes from STREAM and appends them to ST, stopping
1300 after MAX_LENGTH bytes, after appending a newline, or
1301 after an I/O error or end of file was encountered, whichever
1302 comes first. Returns true if at least one byte was added
1303 to ST, false if no bytes were read before an I/O error or
1304 end of file (or if MAX_LENGTH was 0).
1306 This function treats LF and CR LF sequences as new-line,
1307 translating each of them to a single '\n' in ST. */
1309 ds_read_line (struct string *st, FILE *stream, size_t max_length)
1313 for (length = 0; length < max_length; length++)
1315 int c = getc (stream);
1322 ds_put_byte (st, c);
1329 /* CR followed by LF is special: translate to \n. */
1330 ds_put_byte (st, '\n');
1335 /* CR followed by anything else is just CR. */
1336 ds_put_byte (st, '\r');
1344 ds_put_byte (st, c);
1351 /* Removes a comment introduced by `#' from ST,
1352 ignoring occurrences inside quoted strings. */
1354 remove_comment (struct string *st)
1359 for (cp = ds_data (st); cp < ds_end (st); cp++)
1364 else if (*cp == '\\')
1367 else if (*cp == '\'' || *cp == '"')
1369 else if (*cp == '#')
1371 ds_truncate (st, cp - ds_cstr (st));
1376 /* Reads a line from STREAM into ST, then preprocesses as follows:
1378 - Splices lines terminated with `\'.
1380 - Deletes comments introduced by `#' outside of single or double
1383 - Deletes trailing white space.
1385 Returns true if a line was successfully read, false on
1386 failure. If LINE_NUMBER is non-null, then *LINE_NUMBER is
1387 incremented by the number of lines read. */
1389 ds_read_config_line (struct string *st, int *line_number, FILE *stream)
1394 if (!ds_read_line (st, stream, SIZE_MAX))
1397 ds_rtrim (st, ss_cstr (CC_SPACES));
1399 while (ds_chomp_byte (st, '\\'));
1401 remove_comment (st);
1405 /* Attempts to read SIZE * CNT bytes from STREAM and append them
1407 Returns true if all the requested data was read, false otherwise. */
1409 ds_read_stream (struct string *st, size_t size, size_t cnt, FILE *stream)
1413 size_t try_bytes = xtimes (cnt, size);
1414 if (size_in_bounds_p (xsum (ds_length (st), try_bytes)))
1416 char *buffer = ds_put_uninit (st, try_bytes);
1417 size_t got_bytes = fread (buffer, 1, try_bytes, stream);
1418 ds_truncate (st, ds_length (st) - (try_bytes - got_bytes));
1419 return got_bytes == try_bytes;
1431 /* Concatenates S onto ST. */
1433 ds_put_cstr (struct string *st, const char *s)
1436 ds_put_substring (st, ss_cstr (s));
1439 /* Concatenates SS to ST. */
1441 ds_put_substring (struct string *st, struct substring ss)
1443 memcpy (ds_put_uninit (st, ss_length (ss)), ss_data (ss), ss_length (ss));
1446 /* Returns ds_end(ST) and THEN increases the length by INCR. */
1448 ds_put_uninit (struct string *st, size_t incr)
1451 ds_extend (st, ds_length (st) + incr);
1453 st->ss.length += incr;
1457 /* Formats FORMAT as a printf string and appends the result to ST. */
1459 ds_put_format (struct string *st, const char *format, ...)
1463 va_start (args, format);
1464 ds_put_vformat (st, format, args);
1468 /* Formats FORMAT as a printf string and appends the result to ST. */
1470 ds_put_vformat (struct string *st, const char *format, va_list args_)
1475 va_copy (args, args_);
1476 avail = st->ss.string != NULL ? st->capacity - st->ss.length + 1 : 0;
1477 needed = vsnprintf (st->ss.string + st->ss.length, avail, format, args);
1480 if (needed >= avail)
1482 va_copy (args, args_);
1483 vsprintf (ds_put_uninit (st, needed), format, args);
1488 /* Some old libc's returned -1 when the destination string
1490 while (needed == -1)
1492 ds_extend (st, (st->capacity + 1) * 2);
1493 avail = st->capacity - st->ss.length + 1;
1495 va_copy (args, args_);
1496 needed = vsnprintf (ds_end (st), avail, format, args);
1499 st->ss.length += needed;
1503 /* Appends byte CH to ST. */
1505 ds_put_byte (struct string *st, int ch)
1507 ds_put_uninit (st, 1)[0] = ch;
1510 /* Appends CNT copies of byte CH to ST. */
1512 ds_put_byte_multiple (struct string *st, int ch, size_t cnt)
1514 memset (ds_put_uninit (st, cnt), ch, cnt);
1518 /* If relocation has been enabled, replace ST,
1519 with its relocated version */
1521 ds_relocate (struct string *st)
1523 const char *orig = ds_cstr (st);
1524 const char *rel = relocate (orig);
1529 ds_put_cstr (st, rel);
1530 /* The documentation for relocate says that casting away const
1531 and then freeing is appropriate ... */
1532 free (CONST_CAST (char *, rel));
1539 /* Operations on uint8_t "strings" */
1541 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
1542 DST is truncated to DST_SIZE bytes or padded on the right with
1543 copies of PAD as needed. */
1545 u8_buf_copy_rpad (uint8_t *dst, size_t dst_size,
1546 const uint8_t *src, size_t src_size,
1549 if (src_size >= dst_size)
1550 memmove (dst, src, dst_size);
1553 memmove (dst, src, src_size);
1554 memset (&dst[src_size], pad, dst_size - src_size);