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 /* Divides SS into tokens separated by any of the DELIMITERS.
431 Each call replaces TOKEN by the next token in SS, or by an
432 empty string if no tokens remain. Returns true if a token was
433 obtained, false otherwise.
435 Before the first call, initialize *SAVE_IDX to 0. Do not
436 modify *SAVE_IDX between calls.
438 SS divides into exactly one more tokens than it contains
439 delimiters. That is, a delimiter at the start or end of SS or
440 a pair of adjacent delimiters yields an empty token, and the
441 empty string contains a single token. */
443 ss_separate (struct substring ss, struct substring delimiters,
444 size_t *save_idx, struct substring *token)
446 if (*save_idx <= ss_length (ss))
448 struct substring tmp = ss_substr (ss, *save_idx, SIZE_MAX);
449 size_t length = ss_cspan (tmp, delimiters);
450 *token = ss_head (tmp, length);
451 *save_idx += length + 1;
456 *token = ss_empty ();
461 /* Divides SS into tokens separated by any of the DELIMITERS,
462 merging adjacent delimiters so that the empty string is never
463 produced as a token. Each call replaces TOKEN by the next
464 token in SS, or by an empty string if no tokens remain, and
465 then skips past the first delimiter following the token.
466 Returns true if a token was obtained, false otherwise.
468 Before the first call, initialize *SAVE_IDX to 0. Do not
469 modify *SAVE_IDX between calls. */
471 ss_tokenize (struct substring ss, struct substring delimiters,
472 size_t *save_idx, struct substring *token)
474 ss_advance (&ss, *save_idx);
475 *save_idx += ss_ltrim (&ss, delimiters);
476 ss_get_bytes (&ss, ss_cspan (ss, delimiters), token);
477 *save_idx += ss_length (*token) + 1;
478 return ss_length (*token) > 0;
481 /* Removes the first CNT bytes from SS. */
483 ss_advance (struct substring *ss, size_t cnt)
485 if (cnt > ss->length)
491 /* If the first byte in SS is C, removes it and returns true.
492 Otherwise, returns false without changing the string. */
494 ss_match_byte (struct substring *ss, char c)
496 if (ss_first (*ss) == c)
506 /* If the first byte in SS is in MATCH, removes it and
507 returns the byte that was removed.
508 Otherwise, returns EOF without changing the string. */
510 ss_match_byte_in (struct substring *ss, struct substring match)
514 && memchr (match.string, ss->string[0], match.length) != NULL)
523 /* If SS begins with TARGET, removes it and returns true.
524 Otherwise, returns false without changing SS. */
526 ss_match_string (struct substring *ss, const struct substring target)
528 size_t length = ss_length (target);
529 if (ss_equals (ss_head (*ss, length), target))
531 ss_advance (ss, length);
538 /* Removes the first byte from SS and returns it.
539 If SS is empty, returns EOF without modifying SS. */
541 ss_get_byte (struct substring *ss)
543 int c = ss_first (*ss);
552 /* Stores the prefix of SS up to the first DELIMITER in OUT (if
553 any). Trims those same bytes from SS. DELIMITER is
554 removed from SS but not made part of OUT. Returns true if
555 DELIMITER was found (and removed), false otherwise. */
557 ss_get_until (struct substring *ss, char delimiter, struct substring *out)
559 ss_get_bytes (ss, ss_cspan (*ss, ss_buffer (&delimiter, 1)), out);
560 return ss_match_byte (ss, delimiter);
563 /* Stores the first CNT bytes in SS in OUT (or fewer, if SS
564 is shorter than CNT bytes). Trims the same bytes
565 from the beginning of SS. Returns CNT. */
567 ss_get_bytes (struct substring *ss, size_t cnt, struct substring *out)
569 *out = ss_head (*ss, cnt);
570 ss_advance (ss, cnt);
574 /* Parses and removes an optionally signed decimal integer from
575 the beginning of SS. Returns 0 if an error occurred,
576 otherwise the number of bytes removed from SS. Stores
577 the integer's value into *VALUE. */
579 ss_get_long (struct substring *ss, long *value)
584 length = ss_span (*ss, ss_cstr ("+-"));
585 length += ss_span (ss_substr (*ss, length, SIZE_MAX), ss_cstr (CC_DIGITS));
586 if (length > 0 && length < sizeof tmp)
590 memcpy (tmp, ss_data (*ss), length);
593 *value = strtol (tmp, &tail, 10);
594 if (tail - tmp == length)
596 ss_advance (ss, length);
604 /* Returns true if SS is empty (has length 0 bytes),
607 ss_is_empty (struct substring ss)
609 return ss.length == 0;
612 /* Returns the number of bytes in SS. */
614 ss_length (struct substring ss)
619 /* Returns a pointer to the bytes in SS. */
621 ss_data (struct substring ss)
626 /* Returns a pointer just past the last byte in SS. */
628 ss_end (struct substring ss)
630 return ss.string + ss.length;
633 /* Returns the byte in position IDX in SS, as a value in the
634 range of unsigned char. Returns EOF if IDX is out of the
635 range of indexes for SS. */
637 ss_at (struct substring ss, size_t idx)
639 return idx < ss.length ? (unsigned char) ss.string[idx] : EOF;
642 /* Returns the first byte in SS as a value in the range of
643 unsigned char. Returns EOF if SS is the empty string. */
645 ss_first (struct substring ss)
647 return ss_at (ss, 0);
650 /* Returns the last byte in SS as a value in the range of
651 unsigned char. Returns EOF if SS is the empty string. */
653 ss_last (struct substring ss)
655 return ss.length > 0 ? (unsigned char) ss.string[ss.length - 1] : EOF;
658 /* Returns the number of contiguous bytes at the beginning
659 of SS that are in SKIP_SET. */
661 ss_span (struct substring ss, struct substring skip_set)
664 for (i = 0; i < ss.length; i++)
665 if (ss_find_byte (skip_set, ss.string[i]) == SIZE_MAX)
670 /* Returns the number of contiguous bytes at the beginning
671 of SS that are not in SKIP_SET. */
673 ss_cspan (struct substring ss, struct substring stop_set)
676 for (i = 0; i < ss.length; i++)
677 if (ss_find_byte (stop_set, ss.string[i]) != SIZE_MAX)
682 /* Returns the offset in SS of the first instance of C,
683 or SIZE_MAX if C does not occur in SS. */
685 ss_find_byte (struct substring ss, char c)
687 const char *p = memchr (ss.string, c, ss.length);
688 return p != NULL ? p - ss.string : SIZE_MAX;
691 /* Compares A and B and returns a strcmp()-type comparison
694 ss_compare (struct substring a, struct substring b)
696 int retval = memcmp (a.string, b.string, MIN (a.length, b.length));
698 retval = a.length < b.length ? -1 : a.length > b.length;
702 /* Compares A and B case-insensitively and returns a
703 strcmp()-type comparison result. */
705 ss_compare_case (struct substring a, struct substring b)
707 int retval = memcasecmp (a.string, b.string, MIN (a.length, b.length));
709 retval = a.length < b.length ? -1 : a.length > b.length;
713 /* Compares A and B and returns true if their contents are
714 identical, false otherwise. */
716 ss_equals (struct substring a, struct substring b)
718 return a.length == b.length && !memcmp (a.string, b.string, a.length);
721 /* Compares A and B and returns true if their contents are
722 identical except possibly for case differences, false
725 ss_equals_case (struct substring a, struct substring b)
727 return a.length == b.length && !memcasecmp (a.string, b.string, a.length);
730 /* Returns the position in SS that the byte at P occupies.
731 P must point within SS or one past its end. */
733 ss_pointer_to_position (struct substring ss, const char *p)
735 size_t pos = p - ss.string;
736 assert (pos <= ss.length);
740 /* Allocates and returns a null-terminated string that contains
743 ss_xstrdup (struct substring ss)
745 char *s = xmalloc (ss.length + 1);
746 memcpy (s, ss.string, ss.length);
752 /* Returns the character represented by the UTF-8 sequence at the start of S.
753 The return value is either a Unicode code point in the range 0 to 0x10ffff,
754 or UINT32_MAX if S is empty. */
756 ss_first_mb (struct substring s)
758 return ss_at_mb (s, 0);
761 /* Returns the number of bytes in the UTF-8 character at the beginning of S.
763 The return value is 0 if S is empty, otherwise between 1 and 4. */
765 ss_first_mblen (struct substring s)
767 return ss_at_mblen (s, 0);
770 /* Advances S past the UTF-8 character at its beginning. Returns the Unicode
771 code point that was skipped (in the range 0 to 0x10ffff), or UINT32_MAX if S
772 was not modified because it was initially empty. */
774 ss_get_mb (struct substring *s)
781 n = u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s->string), s->length);
790 /* Returns the character represented by the UTF-8 sequence starting OFS bytes
791 into S. The return value is either a Unicode code point in the range 0 to
792 0x10ffff, or UINT32_MAX if OFS is past the last byte in S.
794 (Returns 0xfffd if OFS points into the middle, not the beginning, of a UTF-8
797 ss_at_mb (struct substring s, size_t ofs)
802 u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
810 /* Returns the number of bytes represented by the UTF-8 sequence starting OFS
811 bytes into S. The return value is 0 if OFS is past the last byte in S,
812 otherwise between 1 and 4. */
814 ss_at_mblen (struct substring s, size_t ofs)
819 return u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
826 /* Initializes ST as an empty string. */
828 ds_init_empty (struct string *st)
830 st->ss = ss_empty ();
834 /* Initializes ST with initial contents S. */
836 ds_init_string (struct string *st, const struct string *s)
838 ds_init_substring (st, ds_ss (s));
841 /* Initializes ST with initial contents SS. */
843 ds_init_substring (struct string *st, struct substring ss)
845 st->capacity = MAX (8, ss.length * 2);
846 st->ss.string = xmalloc (st->capacity + 1);
847 memcpy (st->ss.string, ss.string, ss.length);
848 st->ss.length = ss.length;
851 /* Initializes ST with initial contents S. */
853 ds_init_cstr (struct string *st, const char *s)
855 ds_init_substring (st, ss_cstr (s));
860 ds_destroy (struct string *st)
864 ss_dealloc (&st->ss);
865 st->ss.string = NULL;
871 /* Swaps the contents of strings A and B. */
873 ds_swap (struct string *a, struct string *b)
875 struct string tmp = *a;
880 /* Helper function for ds_register_pool. */
882 free_string (void *st_)
884 struct string *st = st_;
888 /* Arranges for ST to be destroyed automatically as part of
891 ds_register_pool (struct string *st, struct pool *pool)
893 pool_register (pool, free_string, st);
896 /* Cancels the arrangement for ST to be destroyed automatically
899 ds_unregister_pool (struct string *st, struct pool *pool)
901 pool_unregister (pool, st);
904 /* Copies SRC into DST.
905 DST and SRC may be the same string. */
907 ds_assign_string (struct string *dst, const struct string *src)
909 ds_assign_substring (dst, ds_ss (src));
912 /* Replaces DST by SS.
913 SS may be a substring of DST. */
915 ds_assign_substring (struct string *dst, struct substring ss)
917 dst->ss.length = ss.length;
918 ds_extend (dst, ss.length);
919 memmove (dst->ss.string, ss.string, ss.length);
922 /* Replaces DST by null-terminated string SRC. SRC may overlap
925 ds_assign_cstr (struct string *dst, const char *src)
927 ds_assign_substring (dst, ss_cstr (src));
930 /* Truncates ST to zero length. */
932 ds_clear (struct string *st)
937 /* Returns a substring that contains ST. */
939 ds_ss (const struct string *st)
944 /* Returns a substring that contains CNT bytes from ST
945 starting at position START.
947 If START is greater than or equal to the length of ST, then
948 the substring will be the empty string. If START + CNT
949 exceeds the length of ST, then the substring will only be
950 ds_length(ST) - START bytes long. */
952 ds_substr (const struct string *st, size_t start, size_t cnt)
954 return ss_substr (ds_ss (st), start, cnt);
957 /* Returns a substring that contains the first CNT bytes in
958 ST. If CNT exceeds the length of ST, then the substring will
959 contain all of ST. */
961 ds_head (const struct string *st, size_t cnt)
963 return ss_head (ds_ss (st), cnt);
966 /* Returns a substring that contains the last CNT bytes in
967 ST. If CNT exceeds the length of ST, then the substring will
968 contain all of ST. */
970 ds_tail (const struct string *st, size_t cnt)
972 return ss_tail (ds_ss (st), cnt);
975 /* Ensures that ST can hold at least MIN_CAPACITY bytes plus a null
978 ds_extend (struct string *st, size_t min_capacity)
980 if (min_capacity > st->capacity)
983 if (st->capacity < min_capacity)
984 st->capacity = 2 * min_capacity;
986 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
990 /* Shrink ST to the minimum capacity need to contain its content. */
992 ds_shrink (struct string *st)
994 if (st->capacity != st->ss.length)
996 st->capacity = st->ss.length;
997 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
1001 /* Truncates ST to at most LENGTH bytes long. */
1003 ds_truncate (struct string *st, size_t length)
1005 ss_truncate (&st->ss, length);
1008 /* Removes trailing bytes in TRIM_SET from ST.
1009 Returns number of bytes removed. */
1011 ds_rtrim (struct string *st, struct substring trim_set)
1013 return ss_rtrim (&st->ss, trim_set);
1016 /* Removes leading bytes in TRIM_SET from ST.
1017 Returns number of bytes removed. */
1019 ds_ltrim (struct string *st, struct substring trim_set)
1021 size_t cnt = ds_span (st, trim_set);
1023 ds_assign_substring (st, ds_substr (st, cnt, SIZE_MAX));
1027 /* Trims leading and trailing bytes in TRIM_SET from ST.
1028 Returns number of bytes removed. */
1030 ds_trim (struct string *st, struct substring trim_set)
1032 size_t cnt = ds_rtrim (st, trim_set);
1033 return cnt + ds_ltrim (st, trim_set);
1036 /* If the last byte in ST is C, removes it and returns true.
1037 Otherwise, returns false without modifying ST. */
1039 ds_chomp_byte (struct string *st, char c)
1041 return ss_chomp_byte (&st->ss, c);
1044 /* Divides ST into tokens separated by any of the DELIMITERS.
1045 Each call replaces TOKEN by the next token in ST, or by an
1046 empty string if no tokens remain. Returns true if a token was
1047 obtained, false otherwise.
1049 Before the first call, initialize *SAVE_IDX to 0. Do not
1050 modify *SAVE_IDX between calls.
1052 ST divides into exactly one more tokens than it contains
1053 delimiters. That is, a delimiter at the start or end of ST or
1054 a pair of adjacent delimiters yields an empty token, and the
1055 empty string contains a single token. */
1057 ds_separate (const struct string *st, struct substring delimiters,
1058 size_t *save_idx, struct substring *token)
1060 return ss_separate (ds_ss (st), delimiters, save_idx, token);
1063 /* Divides ST into tokens separated by any of the DELIMITERS,
1064 merging adjacent delimiters so that the empty string is never
1065 produced as a token. Each call replaces TOKEN by the next
1066 token in ST, or by an empty string if no tokens remain.
1067 Returns true if a token was obtained, false otherwise.
1069 Before the first call, initialize *SAVE_IDX to 0. Do not
1070 modify *SAVE_IDX between calls. */
1072 ds_tokenize (const struct string *st, struct substring delimiters,
1073 size_t *save_idx, struct substring *token)
1075 return ss_tokenize (ds_ss (st), delimiters, save_idx, token);
1078 /* Pad ST on the right with copies of PAD until ST is at least
1079 LENGTH bytes in size. If ST is initially LENGTH
1080 bytes or longer, this is a no-op. */
1082 ds_rpad (struct string *st, size_t length, char pad)
1084 if (length > st->ss.length)
1085 ds_put_byte_multiple (st, pad, length - st->ss.length);
1088 /* Sets the length of ST to exactly NEW_LENGTH,
1089 either by truncating bytes from the end,
1090 or by padding on the right with PAD. */
1092 ds_set_length (struct string *st, size_t new_length, char pad)
1094 if (st->ss.length < new_length)
1095 ds_rpad (st, new_length, pad);
1097 st->ss.length = new_length;
1100 /* Removes N bytes from ST starting at offset START. */
1102 ds_remove (struct string *st, size_t start, size_t n)
1104 if (n > 0 && start < st->ss.length)
1106 if (st->ss.length - start <= n)
1108 /* All bytes at or beyond START are deleted. */
1109 st->ss.length = start;
1113 /* Some bytes remain and must be shifted into
1115 memmove (st->ss.string + st->ss.length,
1116 st->ss.string + st->ss.length + n,
1117 st->ss.length - start - n);
1123 /* There are no bytes to delete or no bytes at or
1124 beyond START, hence deletion is a no-op. */
1128 /* Returns true if ST is empty, false otherwise. */
1130 ds_is_empty (const struct string *st)
1132 return ss_is_empty (st->ss);
1135 /* Returns the length of ST. */
1137 ds_length (const struct string *st)
1139 return ss_length (ds_ss (st));
1142 /* Returns the string data inside ST. */
1144 ds_data (const struct string *st)
1146 return ss_data (ds_ss (st));
1149 /* Returns a pointer to the null terminator ST.
1150 This might not be an actual null byte unless ds_c_str() has
1151 been called since the last modification to ST. */
1153 ds_end (const struct string *st)
1155 return ss_end (ds_ss (st));
1158 /* Returns the byte in position IDX in ST, as a value in the
1159 range of unsigned char. Returns EOF if IDX is out of the
1160 range of indexes for ST. */
1162 ds_at (const struct string *st, size_t idx)
1164 return ss_at (ds_ss (st), idx);
1167 /* Returns the first byte in ST as a value in the range of
1168 unsigned char. Returns EOF if ST is the empty string. */
1170 ds_first (const struct string *st)
1172 return ss_first (ds_ss (st));
1175 /* Returns the last byte in ST as a value in the range of
1176 unsigned char. Returns EOF if ST is the empty string. */
1178 ds_last (const struct string *st)
1180 return ss_last (ds_ss (st));
1183 /* Returns the number of consecutive bytes at the beginning
1184 of ST that are in SKIP_SET. */
1186 ds_span (const struct string *st, struct substring skip_set)
1188 return ss_span (ds_ss (st), skip_set);
1191 /* Returns the number of consecutive bytes at the beginning
1192 of ST that are not in STOP_SET. */
1194 ds_cspan (const struct string *st, struct substring stop_set)
1196 return ss_cspan (ds_ss (st), stop_set);
1199 /* Returns the position of the first occurrence of byte C in
1200 ST at or after position OFS, or SIZE_MAX if there is no such
1203 ds_find_byte (const struct string *st, char c)
1205 return ss_find_byte (ds_ss (st), c);
1208 /* Compares A and B and returns a strcmp()-type comparison
1211 ds_compare (const struct string *a, const struct string *b)
1213 return ss_compare (ds_ss (a), ds_ss (b));
1216 /* Returns the position in ST that the byte at P occupies.
1217 P must point within ST or one past its end. */
1219 ds_pointer_to_position (const struct string *st, const char *p)
1221 return ss_pointer_to_position (ds_ss (st), p);
1224 /* Allocates and returns a null-terminated string that contains
1227 ds_xstrdup (const struct string *st)
1229 return ss_xstrdup (ds_ss (st));
1232 /* Returns the allocation size of ST. */
1234 ds_capacity (const struct string *st)
1236 return st->capacity;
1239 /* Returns the value of ST as a null-terminated string. */
1241 ds_cstr (const struct string *st_)
1243 struct string *st = CONST_CAST (struct string *, st_);
1244 if (st->ss.string == NULL)
1246 st->ss.string[st->ss.length] = '\0';
1247 return st->ss.string;
1250 /* Returns the value of ST as a null-terminated string and then
1251 reinitialized ST as an empty string. The caller must free the
1252 returned string with free(). */
1254 ds_steal_cstr (struct string *st)
1256 char *s = ds_cstr (st);
1261 /* Reads bytes from STREAM and appends them to ST, stopping
1262 after MAX_LENGTH bytes, after appending a newline, or
1263 after an I/O error or end of file was encountered, whichever
1264 comes first. Returns true if at least one byte was added
1265 to ST, false if no bytes were read before an I/O error or
1266 end of file (or if MAX_LENGTH was 0).
1268 This function treats LF and CR LF sequences as new-line,
1269 translating each of them to a single '\n' in ST. */
1271 ds_read_line (struct string *st, FILE *stream, size_t max_length)
1275 for (length = 0; length < max_length; length++)
1277 int c = getc (stream);
1284 ds_put_byte (st, c);
1291 /* CR followed by LF is special: translate to \n. */
1292 ds_put_byte (st, '\n');
1297 /* CR followed by anything else is just CR. */
1298 ds_put_byte (st, '\r');
1306 ds_put_byte (st, c);
1313 /* Removes a comment introduced by `#' from ST,
1314 ignoring occurrences inside quoted strings. */
1316 remove_comment (struct string *st)
1321 for (cp = ds_data (st); cp < ds_end (st); cp++)
1326 else if (*cp == '\\')
1329 else if (*cp == '\'' || *cp == '"')
1331 else if (*cp == '#')
1333 ds_truncate (st, cp - ds_cstr (st));
1338 /* Reads a line from STREAM into ST, then preprocesses as follows:
1340 - Splices lines terminated with `\'.
1342 - Deletes comments introduced by `#' outside of single or double
1345 - Deletes trailing white space.
1347 Returns true if a line was successfully read, false on
1348 failure. If LINE_NUMBER is non-null, then *LINE_NUMBER is
1349 incremented by the number of lines read. */
1351 ds_read_config_line (struct string *st, int *line_number, FILE *stream)
1356 if (!ds_read_line (st, stream, SIZE_MAX))
1359 ds_rtrim (st, ss_cstr (CC_SPACES));
1361 while (ds_chomp_byte (st, '\\'));
1363 remove_comment (st);
1367 /* Attempts to read SIZE * CNT bytes from STREAM and append them
1369 Returns true if all the requested data was read, false otherwise. */
1371 ds_read_stream (struct string *st, size_t size, size_t cnt, FILE *stream)
1375 size_t try_bytes = xtimes (cnt, size);
1376 if (size_in_bounds_p (xsum (ds_length (st), try_bytes)))
1378 char *buffer = ds_put_uninit (st, try_bytes);
1379 size_t got_bytes = fread (buffer, 1, try_bytes, stream);
1380 ds_truncate (st, ds_length (st) - (try_bytes - got_bytes));
1381 return got_bytes == try_bytes;
1393 /* Concatenates S onto ST. */
1395 ds_put_cstr (struct string *st, const char *s)
1398 ds_put_substring (st, ss_cstr (s));
1401 /* Concatenates SS to ST. */
1403 ds_put_substring (struct string *st, struct substring ss)
1405 memcpy (ds_put_uninit (st, ss_length (ss)), ss_data (ss), ss_length (ss));
1408 /* Returns ds_end(ST) and THEN increases the length by INCR. */
1410 ds_put_uninit (struct string *st, size_t incr)
1413 ds_extend (st, ds_length (st) + incr);
1415 st->ss.length += incr;
1419 /* Formats FORMAT as a printf string and appends the result to ST. */
1421 ds_put_format (struct string *st, const char *format, ...)
1425 va_start (args, format);
1426 ds_put_vformat (st, format, args);
1430 /* Formats FORMAT as a printf string and appends the result to ST. */
1432 ds_put_vformat (struct string *st, const char *format, va_list args_)
1437 va_copy (args, args_);
1438 avail = st->ss.string != NULL ? st->capacity - st->ss.length + 1 : 0;
1439 needed = vsnprintf (st->ss.string + st->ss.length, avail, format, args);
1442 if (needed >= avail)
1444 va_copy (args, args_);
1445 vsprintf (ds_put_uninit (st, needed), format, args);
1450 /* Some old libc's returned -1 when the destination string
1452 while (needed == -1)
1454 ds_extend (st, (st->capacity + 1) * 2);
1455 avail = st->capacity - st->ss.length + 1;
1457 va_copy (args, args_);
1458 needed = vsnprintf (ds_end (st), avail, format, args);
1461 st->ss.length += needed;
1465 /* Appends byte CH to ST. */
1467 ds_put_byte (struct string *st, int ch)
1469 ds_put_uninit (st, 1)[0] = ch;
1472 /* Appends CNT copies of byte CH to ST. */
1474 ds_put_byte_multiple (struct string *st, int ch, size_t cnt)
1476 memset (ds_put_uninit (st, cnt), ch, cnt);
1480 /* If relocation has been enabled, replace ST,
1481 with its relocated version */
1483 ds_relocate (struct string *st)
1485 const char *orig = ds_cstr (st);
1486 const char *rel = relocate (orig);
1491 ds_put_cstr (st, rel);
1492 /* The documentation for relocate says that casting away const
1493 and then freeing is appropriate ... */
1494 free (CONST_CAST (char *, rel));
1501 /* Operations on uint8_t "strings" */
1503 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
1504 DST is truncated to DST_SIZE bytes or padded on the right with
1505 copies of PAD as needed. */
1507 u8_buf_copy_rpad (uint8_t *dst, size_t dst_size,
1508 const uint8_t *src, size_t src_size,
1511 if (src_size >= dst_size)
1512 memmove (dst, src, dst_size);
1515 memmove (dst, src, src_size);
1516 memset (&dst[src_size], pad, dst_size - src_size);