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/i18n.h"
30 #include "libpspp/message.h"
31 #include "libpspp/pool.h"
33 #include "gl/c-ctype.h"
34 #include "gl/c-vasnprintf.h"
35 #include "gl/relocatable.h"
36 #include "gl/minmax.h"
37 #include "gl/xalloc.h"
38 #include "gl/xmemdup0.h"
41 /* Reverses the order of NBYTES bytes at address P, thus converting
42 between little- and big-endian byte orders. */
44 buf_reverse (char *p, size_t nbytes)
46 char *h = p, *t = &h[nbytes - 1];
58 /* Compares the SIZE bytes in A to those in B, disregarding case,
59 and returns a strcmp()-type result. */
61 buf_compare_case (const char *a_, const char *b_, size_t size)
63 const unsigned char *a = (unsigned char *) a_;
64 const unsigned char *b = (unsigned char *) b_;
68 unsigned char ac = toupper (*a++);
69 unsigned char bc = toupper (*b++);
72 return ac > bc ? 1 : -1;
78 /* Compares A of length A_LEN to B of length B_LEN. The shorter
79 string is considered to be padded with spaces to the length of
82 buf_compare_rpad (const char *a, size_t a_len, const char *b, size_t b_len)
87 min_len = a_len < b_len ? a_len : b_len;
88 result = memcmp (a, b, min_len);
97 for (idx = min_len; idx < b_len; idx++)
99 return ' ' > b[idx] ? 1 : -1;
103 for (idx = min_len; idx < a_len; idx++)
105 return a[idx] > ' ' ? 1 : -1;
111 /* Compares strin A to string B. The shorter string is
112 considered to be padded with spaces to the length of the
115 str_compare_rpad (const char *a, const char *b)
117 return buf_compare_rpad (a, strlen (a), b, strlen (b));
120 /* Copies string SRC to buffer DST, of size DST_SIZE bytes.
121 DST is truncated to DST_SIZE bytes or padded on the right with
122 copies of PAD as needed. */
124 buf_copy_str_rpad (char *dst, size_t dst_size, const char *src, char pad)
126 size_t src_len = strlen (src);
127 if (src_len >= dst_size)
128 memcpy (dst, src, dst_size);
131 memcpy (dst, src, src_len);
132 memset (&dst[src_len], pad, dst_size - src_len);
136 /* Copies string SRC to buffer DST, of size DST_SIZE bytes.
137 DST is truncated to DST_SIZE bytes or padded on the left with
138 copies of PAD as needed. */
140 buf_copy_str_lpad (char *dst, size_t dst_size, const char *src, char pad)
142 size_t src_len = strlen (src);
143 if (src_len >= dst_size)
144 memcpy (dst, src, dst_size);
147 size_t n_pad = dst_size - src_len;
148 memset (&dst[0], pad, n_pad);
149 memcpy (dst + n_pad, src, src_len);
153 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
154 DST is truncated to DST_SIZE bytes or padded on the left with
155 copies of PAD as needed. */
157 buf_copy_lpad (char *dst, size_t dst_size,
158 const char *src, size_t src_size,
161 if (src_size >= dst_size)
162 memmove (dst, src, dst_size);
165 memset (dst, pad, dst_size - src_size);
166 memmove (&dst[dst_size - src_size], src, src_size);
170 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
171 DST is truncated to DST_SIZE bytes or padded on the right with
172 copies of PAD as needed. */
174 buf_copy_rpad (char *dst, size_t dst_size,
175 const char *src, size_t src_size,
178 if (src_size >= dst_size)
179 memmove (dst, src, dst_size);
182 memmove (dst, src, src_size);
183 memset (&dst[src_size], pad, dst_size - src_size);
187 /* Copies string SRC to string DST, which is in a buffer DST_SIZE
189 Truncates DST to DST_SIZE - 1 bytes or right-pads with
190 spaces to DST_SIZE - 1 bytes if necessary. */
192 str_copy_rpad (char *dst, size_t dst_size, const char *src)
196 size_t src_len = strlen (src);
197 if (src_len < dst_size - 1)
199 memcpy (dst, src, src_len);
200 memset (&dst[src_len], ' ', dst_size - 1 - src_len);
203 memcpy (dst, src, dst_size - 1);
204 dst[dst_size - 1] = 0;
208 /* Copies SRC to DST, which is in a buffer DST_SIZE bytes long.
209 Truncates DST to DST_SIZE - 1 bytes, if necessary. */
211 str_copy_trunc (char *dst, size_t dst_size, const char *src)
213 size_t src_len = strlen (src);
214 assert (dst_size > 0);
215 if (src_len + 1 < dst_size)
216 memcpy (dst, src, src_len + 1);
219 memcpy (dst, src, dst_size - 1);
220 dst[dst_size - 1] = '\0';
224 /* Copies buffer SRC, of SRC_LEN bytes,
225 to DST, which is in a buffer DST_SIZE bytes long.
226 Truncates DST to DST_SIZE - 1 bytes, if necessary. */
228 str_copy_buf_trunc (char *dst, size_t dst_size,
229 const char *src, size_t src_size)
232 assert (dst_size > 0);
234 dst_len = src_size < dst_size ? src_size : dst_size - 1;
235 memcpy (dst, src, dst_len);
239 /* Converts each byte in S to uppercase.
241 This is suitable only for ASCII strings. Use utf8_to_upper() for UTF-8
244 str_uppercase (char *s)
246 for (; *s != '\0'; s++)
247 *s = c_toupper ((unsigned char) *s);
250 /* Converts each byte in S to lowercase.
252 This is suitable only for ASCII strings. Use utf8_to_lower() for UTF-8
255 str_lowercase (char *s)
257 for (; *s != '\0'; s++)
258 *s = c_tolower ((unsigned char) *s);
261 /* Converts NUMBER into a string in 26-adic notation in BUFFER,
262 which has room for SIZE bytes. Uses uppercase if UPPERCASE is
263 true, otherwise lowercase, Returns true if successful, false
264 if NUMBER, plus a trailing null, is too large to fit in the
267 26-adic notation is "spreadsheet column numbering": 1 = A, 2 =
268 B, 3 = C, ... 26 = Z, 27 = AA, 28 = AB, 29 = AC, ...
270 26-adic notation is the special case of a k-adic numeration
271 system (aka bijective base-k numeration) with k=26. In k-adic
272 numeration, the digits are {1, 2, 3, ..., k} (there is no
273 digit 0), and integer 0 is represented by the empty string.
274 For more information, see
275 http://en.wikipedia.org/wiki/Bijective_numeration. */
277 str_format_26adic (unsigned long int number, bool uppercase,
278 char buffer[], size_t size)
281 = uppercase ? "ABCDEFGHIJKLMNOPQRSTUVWXYZ" : "abcdefghijklmnopqrstuvwxyz";
288 buffer[length++] = alphabet[number % 26];
294 buffer[length] = '\0';
296 buf_reverse (buffer, length);
305 /* Copies IN to buffer OUT with size OUT_SIZE, appending a null terminator. If
306 IN is too long for OUT, or if IN contains a new-line, replaces the tail with
309 OUT_SIZE must be at least 16. */
311 str_ellipsize (struct substring in, char *out, size_t out_size)
313 assert (out_size >= 16);
315 size_t out_maxlen = out_size - 1;
316 if (in.length > out_maxlen - 3)
320 while (out_len < in.length
321 && in.string[out_len] != '\n'
322 && in.string[out_len] != '\0'
323 && (in.string[out_len] != '\r'
324 || out_len + 1 >= in.length
325 || in.string[out_len + 1] != '\n'))
327 int mblen = u8_mblen (CHAR_CAST (const uint8_t *, in.string + out_len),
328 in.length - out_len);
329 if (mblen < 0 || out_len + mblen > out_maxlen)
334 memcpy (out, in.string, out_len);
335 strcpy (&out[out_len], out_len < in.length ? "..." : "");
338 /* Sets the SIZE bytes starting at BLOCK to C,
339 and returns the byte following BLOCK. */
341 mempset (void *block, int c, size_t size)
343 memset (block, c, size);
344 return (char *) block + size;
349 /* Returns a substring whose contents are the N bytes
350 starting at the (0-based) position START in SS. */
352 ss_substr (struct substring ss, size_t start, size_t n)
354 if (start < ss.length)
355 return ss_buffer (ss.string + start, MIN (n, ss.length - start));
357 return ss_buffer (ss.string + ss.length, 0);
360 /* Returns a substring whose contents are the first N
363 ss_head (struct substring ss, size_t n)
365 return ss_buffer (ss.string, MIN (n, ss.length));
368 /* Returns a substring whose contents are the last N bytes
371 ss_tail (struct substring ss, size_t n)
374 return ss_buffer (ss.string + (ss.length - n), n);
379 /* Makes a malloc()'d, null-terminated copy of the contents of OLD
380 and stores it in NEW. */
382 ss_alloc_substring (struct substring *new, struct substring old)
384 new->string = xmemdup0 (old.string, old.length);
385 new->length = old.length;
388 /* Allocates room for a N-byte string in NEW. */
390 ss_alloc_uninit (struct substring *new, size_t n)
392 new->string = xmalloc (n);
397 ss_realloc (struct substring *ss, size_t size)
399 ss->string = xrealloc (ss->string, size);
402 /* Makes a pool_alloc_unaligned()'d, null-terminated copy of the contents of
403 OLD in POOL, and stores it in NEW. */
405 ss_alloc_substring_pool (struct substring *new, struct substring old,
408 new->string = pool_alloc_unaligned (pool, old.length + 1);
409 new->length = old.length;
410 memcpy (new->string, old.string, old.length);
411 new->string[old.length] = '\0';
414 /* Allocates room for a N-byte string in NEW in POOL. */
416 ss_alloc_uninit_pool (struct substring *new, size_t n, struct pool *pool)
418 new->string = pool_alloc_unaligned (pool, n);
422 /* Frees the string that SS points to. */
424 ss_dealloc (struct substring *ss)
429 /* Exchanges the contents of A and B. */
431 ss_swap (struct substring *a, struct substring *b)
433 struct substring tmp = *a;
438 /* Truncates SS to at most N bytes in length. */
440 ss_truncate (struct substring *ss, size_t n)
446 /* Removes trailing bytes in TRIM_SET from SS.
447 Returns number of bytes removed. */
449 ss_rtrim (struct substring *ss, struct substring trim_set)
452 while (n < ss->length
453 && ss_find_byte (trim_set,
454 ss->string[ss->length - n - 1]) != SIZE_MAX)
460 /* Removes leading bytes in TRIM_SET from SS.
461 Returns number of bytes removed. */
463 ss_ltrim (struct substring *ss, struct substring trim_set)
465 size_t n = ss_span (*ss, trim_set);
470 /* Trims leading and trailing bytes in TRIM_SET from SS. */
472 ss_trim (struct substring *ss, struct substring trim_set)
474 ss_ltrim (ss, trim_set);
475 ss_rtrim (ss, trim_set);
478 /* If the last byte in SS is C, removes it and returns true.
479 Otherwise, returns false without changing the string. */
481 ss_chomp_byte (struct substring *ss, char c)
483 if (ss_last (*ss) == c)
492 /* If SS ends with SUFFIX, removes it and returns true.
493 Otherwise, returns false without changing the string. */
495 ss_chomp (struct substring *ss, struct substring suffix)
497 if (ss_ends_with (*ss, suffix))
499 ss->length -= suffix.length;
506 /* Divides SS into tokens separated by any of the DELIMITERS.
507 Each call replaces TOKEN by the next token in SS, or by an
508 empty string if no tokens remain. Returns true if a token was
509 obtained, false otherwise.
511 Before the first call, initialize *SAVE_IDX to 0. Do not
512 modify *SAVE_IDX between calls.
514 SS divides into exactly one more tokens than it contains
515 delimiters. That is, a delimiter at the start or end of SS or
516 a pair of adjacent delimiters yields an empty token, and the
517 empty string contains a single token. */
519 ss_separate (struct substring ss, struct substring delimiters,
520 size_t *save_idx, struct substring *token)
522 if (*save_idx <= ss_length (ss))
524 struct substring tmp = ss_substr (ss, *save_idx, SIZE_MAX);
525 size_t length = ss_cspan (tmp, delimiters);
526 *token = ss_head (tmp, length);
527 *save_idx += length + 1;
532 *token = ss_empty ();
537 /* Divides SS into tokens separated by any of the DELIMITERS,
538 merging adjacent delimiters so that the empty string is never
539 produced as a token. Each call replaces TOKEN by the next
540 token in SS, or by an empty string if no tokens remain, and
541 then skips past the first delimiter following the token.
542 Returns true if a token was obtained, false otherwise.
544 Before the first call, initialize *SAVE_IDX to 0. Do not
545 modify *SAVE_IDX between calls. */
547 ss_tokenize (struct substring ss, struct substring delimiters,
548 size_t *save_idx, struct substring *token)
552 ss_advance (&ss, *save_idx);
553 *save_idx += ss_ltrim (&ss, delimiters);
554 ss_get_bytes (&ss, ss_cspan (ss, delimiters), token);
556 found_token = ss_length (*token) > 0;
557 *save_idx += ss_length (*token) + (found_token?1:0);
561 /* Removes the first N bytes from SS. */
563 ss_advance (struct substring *ss, size_t n)
571 /* If the first byte in SS is C, removes it and returns true.
572 Otherwise, returns false without changing the string. */
574 ss_match_byte (struct substring *ss, char c)
576 if (ss_first (*ss) == c)
586 /* If the first byte in SS is in MATCH, removes it and
587 returns the byte that was removed.
588 Otherwise, returns EOF without changing the string. */
590 ss_match_byte_in (struct substring *ss, struct substring match)
594 && memchr (match.string, ss->string[0], match.length) != NULL)
603 /* If SS begins with TARGET, removes it and returns true.
604 Otherwise, returns false without changing SS. */
606 ss_match_string (struct substring *ss, const struct substring target)
608 size_t length = ss_length (target);
609 if (ss_equals (ss_head (*ss, length), target))
611 ss_advance (ss, length);
618 /* If SS begins with TARGET, except possibly for case differences, removes it
619 and returns true. Otherwise, returns false without changing SS. */
621 ss_match_string_case (struct substring *ss, const struct substring target)
623 size_t length = ss_length (target);
624 if (ss_equals_case (ss_head (*ss, length), target))
626 ss_advance (ss, length);
633 /* Removes the first byte from SS and returns it.
634 If SS is empty, returns EOF without modifying SS. */
636 ss_get_byte (struct substring *ss)
638 int c = ss_first (*ss);
647 /* Stores the prefix of SS up to the first DELIMITER in OUT (if
648 any). Trims those same bytes from SS. DELIMITER is
649 removed from SS but not made part of OUT. Returns true if
650 DELIMITER was found (and removed), false otherwise. */
652 ss_get_until (struct substring *ss, char delimiter, struct substring *out)
654 ss_get_bytes (ss, ss_cspan (*ss, ss_buffer (&delimiter, 1)), out);
655 return ss_match_byte (ss, delimiter);
658 /* Stores the first N bytes in SS in OUT (or fewer, if SS
659 is shorter than N bytes). Trims the same bytes
660 from the beginning of SS. Returns N. */
662 ss_get_bytes (struct substring *ss, size_t n, struct substring *out)
664 *out = ss_head (*ss, n);
669 /* Parses and removes an optionally signed decimal integer from
670 the beginning of SS. Returns 0 if an error occurred,
671 otherwise the number of bytes removed from SS. Stores
672 the integer's value into *VALUE. */
674 ss_get_long (struct substring *ss, long *value)
679 length = ss_span (*ss, ss_cstr ("+-"));
680 length += ss_span (ss_substr (*ss, length, SIZE_MAX), ss_cstr (CC_DIGITS));
681 if (length > 0 && length < sizeof tmp)
685 memcpy (tmp, ss_data (*ss), length);
688 *value = strtol (tmp, &tail, 10);
689 if (tail - tmp == length)
691 ss_advance (ss, length);
699 /* Returns true if SS is empty (has length 0 bytes),
702 ss_is_empty (struct substring ss)
704 return ss.length == 0;
707 /* Returns the number of bytes in SS. */
709 ss_length (struct substring ss)
714 /* Returns a pointer to the bytes in SS. */
716 ss_data (struct substring ss)
721 /* Returns a pointer just past the last byte in SS. */
723 ss_end (struct substring ss)
725 return ss.string + ss.length;
728 /* Returns the byte in position IDX in SS, as a value in the
729 range of unsigned char. Returns EOF if IDX is out of the
730 range of indexes for SS. */
732 ss_at (struct substring ss, size_t idx)
734 return idx < ss.length ? (unsigned char) ss.string[idx] : EOF;
737 /* Returns the first byte in SS as a value in the range of
738 unsigned char. Returns EOF if SS is the empty string. */
740 ss_first (struct substring ss)
742 return ss_at (ss, 0);
745 /* Returns the last byte in SS as a value in the range of
746 unsigned char. Returns EOF if SS is the empty string. */
748 ss_last (struct substring ss)
750 return ss.length > 0 ? (unsigned char) ss.string[ss.length - 1] : EOF;
753 /* Returns true if SS starts with PREFIX, false otherwise. */
755 ss_starts_with (struct substring ss, struct substring prefix)
757 return (ss.length >= prefix.length
758 && !memcmp (ss.string, prefix.string, prefix.length));
761 /* Returns true if SS ends with SUFFIX, false otherwise. */
763 ss_ends_with (struct substring ss, struct substring suffix)
765 return (ss.length >= suffix.length
766 && !memcmp (&ss.string[ss.length - suffix.length], suffix.string,
770 /* Returns the number of contiguous bytes at the beginning
771 of SS that are in SKIP_SET. */
773 ss_span (struct substring ss, struct substring skip_set)
776 for (i = 0; i < ss.length; i++)
777 if (ss_find_byte (skip_set, ss.string[i]) == SIZE_MAX)
782 /* Returns the number of contiguous bytes at the beginning
783 of SS that are not in SKIP_SET. */
785 ss_cspan (struct substring ss, struct substring stop_set)
788 for (i = 0; i < ss.length; i++)
789 if (ss_find_byte (stop_set, ss.string[i]) != SIZE_MAX)
794 /* Returns the offset in SS of the first instance of C,
795 or SIZE_MAX if C does not occur in SS. */
797 ss_find_byte (struct substring ss, char c)
799 const char *p = memchr (ss.string, (int) c, ss.length);
800 return p != NULL ? p - ss.string : SIZE_MAX;
803 /* Returns the offset in HAYSTACK of the first instance of NEEDLE,
804 or SIZE_MAX if NEEDLE does not occur in HAYSTACK. */
806 ss_find_substring (struct substring haystack, struct substring needle)
808 const char *p = memmem (haystack.string, haystack.length,
809 needle.string, needle.length);
810 return p != NULL ? p - haystack.string : SIZE_MAX;
813 /* Compares A and B and returns a strcmp()-type comparison
816 ss_compare (struct substring a, struct substring b)
818 int retval = memcmp (a.string, b.string, MIN (a.length, b.length));
820 retval = a.length < b.length ? -1 : a.length > b.length;
824 /* Compares A and B case-insensitively and returns a
825 strcmp()-type comparison result. */
827 ss_compare_case (struct substring a, struct substring b)
829 int retval = memcasecmp (a.string, b.string, MIN (a.length, b.length));
831 retval = a.length < b.length ? -1 : a.length > b.length;
835 /* Compares A and B and returns true if their contents are
836 identical, false otherwise. */
838 ss_equals (struct substring a, struct substring b)
840 return a.length == b.length && !memcmp (a.string, b.string, a.length);
843 /* Compares A and B and returns true if their contents are
844 identical except possibly for case differences, false
847 ss_equals_case (struct substring a, struct substring b)
849 return a.length == b.length && !memcasecmp (a.string, b.string, a.length);
852 /* Returns the position in SS that the byte at P occupies.
853 P must point within SS or one past its end. */
855 ss_pointer_to_position (struct substring ss, const char *p)
857 size_t pos = p - ss.string;
858 assert (pos <= ss.length);
862 /* Allocates and returns a null-terminated string that contains
865 ss_xstrdup (struct substring ss)
867 char *s = xmalloc (ss.length + 1);
868 memcpy (s, ss.string, ss.length);
874 /* Returns the character represented by the UTF-8 sequence at the start of S.
875 The return value is either a Unicode code point in the range 0 to 0x10ffff,
876 or UINT32_MAX if S is empty. */
878 ss_first_mb (struct substring s)
880 return ss_at_mb (s, 0);
883 /* Returns the number of bytes in the UTF-8 character at the beginning of S.
885 The return value is 0 if S is empty, otherwise between 1 and 4. */
887 ss_first_mblen (struct substring s)
889 return ss_at_mblen (s, 0);
892 /* Advances S past the UTF-8 character at its beginning. Returns the Unicode
893 code point that was skipped (in the range 0 to 0x10ffff), or UINT32_MAX if S
894 was not modified because it was initially empty. */
896 ss_get_mb (struct substring *s)
903 n = u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s->string), s->length);
912 /* Returns the character represented by the UTF-8 sequence starting OFS bytes
913 into S. The return value is either a Unicode code point in the range 0 to
914 0x10ffff, or UINT32_MAX if OFS is past the last byte in S.
916 (Returns 0xfffd if OFS points into the middle, not the beginning, of a UTF-8
919 ss_at_mb (struct substring s, size_t ofs)
924 u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
932 /* Returns the number of bytes represented by the UTF-8 sequence starting OFS
933 bytes into S. The return value is 0 if OFS is past the last byte in S,
934 otherwise between 1 and 4. */
936 ss_at_mblen (struct substring s, size_t ofs)
941 return u8_mbtouc (&uc, CHAR_CAST (const uint8_t *, s.string + ofs),
949 ss_utf8_count_columns (struct substring s)
951 return utf8_count_columns (s.string, s.length);
954 /* Returns a substring of S starting at 0-based display column START and
955 running for N display columns. */
957 ss_utf8_columns (struct substring s, size_t start, size_t n)
959 ss_advance (&s, utf8_columns_to_bytes (s.string, s.length, start));
960 s.length = utf8_columns_to_bytes (s.string, s.length, n);
964 /* Initializes ST as an empty string. */
966 ds_init_empty (struct string *st)
968 st->ss = ss_empty ();
972 /* Initializes ST with initial contents S. */
974 ds_init_string (struct string *st, const struct string *s)
976 ds_init_substring (st, ds_ss (s));
979 /* Initializes ST with initial contents SS. */
981 ds_init_substring (struct string *st, struct substring ss)
983 st->capacity = MAX (8, ss.length * 2);
984 st->ss.string = xmalloc (st->capacity + 1);
985 memcpy (st->ss.string, ss.string, ss.length);
986 st->ss.length = ss.length;
989 /* Initializes ST with initial contents S. */
991 ds_init_cstr (struct string *st, const char *s)
993 ds_init_substring (st, ss_cstr (s));
998 ds_destroy (struct string *st)
1002 ss_dealloc (&st->ss);
1003 st->ss.string = NULL;
1009 /* Swaps the contents of strings A and B. */
1011 ds_swap (struct string *a, struct string *b)
1013 struct string tmp = *a;
1018 /* Helper function for ds_register_pool. */
1020 free_string (void *st_)
1022 struct string *st = st_;
1026 /* Arranges for ST to be destroyed automatically as part of
1029 ds_register_pool (struct string *st, struct pool *pool)
1031 pool_register (pool, free_string, st);
1034 /* Cancels the arrangement for ST to be destroyed automatically
1037 ds_unregister_pool (struct string *st, struct pool *pool)
1039 pool_unregister (pool, st);
1042 /* Copies SRC into DST.
1043 DST and SRC may be the same string. */
1045 ds_assign_string (struct string *dst, const struct string *src)
1047 ds_assign_substring (dst, ds_ss (src));
1050 /* Replaces DST by SS.
1051 SS may be a substring of DST. */
1053 ds_assign_substring (struct string *dst, struct substring ss)
1055 dst->ss.length = ss.length;
1056 ds_extend (dst, ss.length);
1057 memmove (dst->ss.string, ss.string, ss.length);
1060 /* Replaces DST by null-terminated string SRC. SRC may overlap
1063 ds_assign_cstr (struct string *dst, const char *src)
1065 ds_assign_substring (dst, ss_cstr (src));
1068 /* Truncates ST to zero length. */
1070 ds_clear (struct string *st)
1075 /* Returns a substring that contains ST. */
1077 ds_ss (const struct string *st)
1082 /* Returns a substring that contains N bytes from ST
1083 starting at position START.
1085 If START is greater than or equal to the length of ST, then
1086 the substring will be the empty string. If START + N
1087 exceeds the length of ST, then the substring will only be
1088 ds_length(ST) - START bytes long. */
1090 ds_substr (const struct string *st, size_t start, size_t n)
1092 return ss_substr (ds_ss (st), start, n);
1095 /* Returns a substring that contains the first N bytes in
1096 ST. If N exceeds the length of ST, then the substring will
1097 contain all of ST. */
1099 ds_head (const struct string *st, size_t n)
1101 return ss_head (ds_ss (st), n);
1104 /* Returns a substring that contains the last N bytes in
1105 ST. If N exceeds the length of ST, then the substring will
1106 contain all of ST. */
1108 ds_tail (const struct string *st, size_t n)
1110 return ss_tail (ds_ss (st), n);
1113 /* Ensures that ST can hold at least MIN_CAPACITY bytes plus a null
1116 ds_extend (struct string *st, size_t min_capacity)
1118 if (min_capacity > st->capacity)
1121 if (st->capacity < min_capacity)
1122 st->capacity = 2 * min_capacity;
1124 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
1128 /* Shrink ST to the minimum capacity need to contain its content. */
1130 ds_shrink (struct string *st)
1132 if (st->capacity != st->ss.length)
1134 st->capacity = st->ss.length;
1135 st->ss.string = xrealloc (st->ss.string, st->capacity + 1);
1139 /* Truncates ST to at most LENGTH bytes long. */
1141 ds_truncate (struct string *st, size_t length)
1143 ss_truncate (&st->ss, length);
1146 /* Removes trailing bytes in TRIM_SET from ST.
1147 Returns number of bytes removed. */
1149 ds_rtrim (struct string *st, struct substring trim_set)
1151 return ss_rtrim (&st->ss, trim_set);
1154 /* Removes leading bytes in TRIM_SET from ST.
1155 Returns number of bytes removed. */
1157 ds_ltrim (struct string *st, struct substring trim_set)
1159 size_t n = ds_span (st, trim_set);
1161 ds_assign_substring (st, ds_substr (st, n, SIZE_MAX));
1165 /* Trims leading and trailing bytes in TRIM_SET from ST.
1166 Returns number of bytes removed. */
1168 ds_trim (struct string *st, struct substring trim_set)
1170 size_t n = ds_rtrim (st, trim_set);
1171 return n + ds_ltrim (st, trim_set);
1174 /* If the last byte in ST is C, removes it and returns true.
1175 Otherwise, returns false without modifying ST. */
1177 ds_chomp_byte (struct string *st, char c)
1179 return ss_chomp_byte (&st->ss, c);
1182 /* If ST ends with SUFFIX, removes it and returns true.
1183 Otherwise, returns false without modifying ST. */
1185 ds_chomp (struct string *st, struct substring suffix)
1187 return ss_chomp (&st->ss, suffix);
1190 /* Divides ST into tokens separated by any of the DELIMITERS.
1191 Each call replaces TOKEN by the next token in ST, or by an
1192 empty string if no tokens remain. Returns true if a token was
1193 obtained, false otherwise.
1195 Before the first call, initialize *SAVE_IDX to 0. Do not
1196 modify *SAVE_IDX between calls.
1198 ST divides into exactly one more tokens than it contains
1199 delimiters. That is, a delimiter at the start or end of ST or
1200 a pair of adjacent delimiters yields an empty token, and the
1201 empty string contains a single token. */
1203 ds_separate (const struct string *st, struct substring delimiters,
1204 size_t *save_idx, struct substring *token)
1206 return ss_separate (ds_ss (st), delimiters, save_idx, token);
1209 /* Divides ST into tokens separated by any of the DELIMITERS,
1210 merging adjacent delimiters so that the empty string is never
1211 produced as a token. Each call replaces TOKEN by the next
1212 token in ST, or by an empty string if no tokens remain.
1213 Returns true if a token was obtained, false otherwise.
1215 Before the first call, initialize *SAVE_IDX to 0. Do not
1216 modify *SAVE_IDX between calls. */
1218 ds_tokenize (const struct string *st, struct substring delimiters,
1219 size_t *save_idx, struct substring *token)
1221 return ss_tokenize (ds_ss (st), delimiters, save_idx, token);
1224 /* Pad ST on the right with copies of PAD until ST is at least
1225 LENGTH bytes in size. If ST is initially LENGTH
1226 bytes or longer, this is a no-op. */
1228 ds_rpad (struct string *st, size_t length, char pad)
1230 if (length > st->ss.length)
1231 ds_put_byte_multiple (st, pad, length - st->ss.length);
1234 /* Sets the length of ST to exactly NEW_LENGTH,
1235 either by truncating bytes from the end,
1236 or by padding on the right with PAD. */
1238 ds_set_length (struct string *st, size_t new_length, char pad)
1240 if (st->ss.length < new_length)
1241 ds_rpad (st, new_length, pad);
1243 st->ss.length = new_length;
1246 /* Removes N bytes from ST starting at offset START. */
1248 ds_remove (struct string *st, size_t start, size_t n)
1250 if (n > 0 && start < st->ss.length)
1252 if (st->ss.length - start <= n)
1254 /* All bytes at or beyond START are deleted. */
1255 st->ss.length = start;
1259 /* Some bytes remain and must be shifted into
1261 memmove (st->ss.string + st->ss.length,
1262 st->ss.string + st->ss.length + n,
1263 st->ss.length - start - n);
1269 /* There are no bytes to delete or no bytes at or
1270 beyond START, hence deletion is a no-op. */
1274 /* Returns true if ST is empty, false otherwise. */
1276 ds_is_empty (const struct string *st)
1278 return ss_is_empty (st->ss);
1281 /* Returns the length of ST. */
1283 ds_length (const struct string *st)
1285 return ss_length (ds_ss (st));
1288 /* Returns the string data inside ST. */
1290 ds_data (const struct string *st)
1292 return ss_data (ds_ss (st));
1295 /* Returns a pointer to the null terminator ST.
1296 This might not be an actual null byte unless ds_c_str() has
1297 been called since the last modification to ST. */
1299 ds_end (const struct string *st)
1301 return ss_end (ds_ss (st));
1304 /* Returns the byte in position IDX in ST, as a value in the
1305 range of unsigned char. Returns EOF if IDX is out of the
1306 range of indexes for ST. */
1308 ds_at (const struct string *st, size_t idx)
1310 return ss_at (ds_ss (st), idx);
1313 /* Returns the first byte in ST as a value in the range of
1314 unsigned char. Returns EOF if ST is the empty string. */
1316 ds_first (const struct string *st)
1318 return ss_first (ds_ss (st));
1321 /* Returns the last byte in ST as a value in the range of
1322 unsigned char. Returns EOF if ST is the empty string. */
1324 ds_last (const struct string *st)
1326 return ss_last (ds_ss (st));
1329 /* Returns true if ST ends with SUFFIX, false otherwise. */
1331 ds_ends_with (const struct string *st, struct substring suffix)
1333 return ss_ends_with (st->ss, suffix);
1336 /* Returns the number of consecutive bytes at the beginning
1337 of ST that are in SKIP_SET. */
1339 ds_span (const struct string *st, struct substring skip_set)
1341 return ss_span (ds_ss (st), skip_set);
1344 /* Returns the number of consecutive bytes at the beginning
1345 of ST that are not in STOP_SET. */
1347 ds_cspan (const struct string *st, struct substring stop_set)
1349 return ss_cspan (ds_ss (st), stop_set);
1352 /* Returns the position of the first occurrence of byte C in
1353 ST at or after position OFS, or SIZE_MAX if there is no such
1356 ds_find_byte (const struct string *st, char c)
1358 return ss_find_byte (ds_ss (st), c);
1361 /* Compares A and B and returns a strcmp()-type comparison
1364 ds_compare (const struct string *a, const struct string *b)
1366 return ss_compare (ds_ss (a), ds_ss (b));
1369 /* Returns the position in ST that the byte at P occupies.
1370 P must point within ST or one past its end. */
1372 ds_pointer_to_position (const struct string *st, const char *p)
1374 return ss_pointer_to_position (ds_ss (st), p);
1377 /* Allocates and returns a null-terminated string that contains
1380 ds_xstrdup (const struct string *st)
1382 return ss_xstrdup (ds_ss (st));
1385 /* Returns the allocation size of ST. */
1387 ds_capacity (const struct string *st)
1389 return st->capacity;
1392 /* Returns the value of ST as a null-terminated string. */
1394 ds_cstr (const struct string *st_)
1396 struct string *st = CONST_CAST (struct string *, st_);
1397 if (st->ss.string == NULL)
1399 st->ss.string[st->ss.length] = '\0';
1400 return st->ss.string;
1403 /* Returns the value of ST as a null-terminated string and then
1404 reinitialized ST as an empty string. The caller must free the
1405 returned string with free(). */
1407 ds_steal_cstr (struct string *st)
1409 char *s = ds_cstr (st);
1414 /* Reads bytes from STREAM and appends them to ST, stopping
1415 after MAX_LENGTH bytes, after appending a newline, or
1416 after an I/O error or end of file was encountered, whichever
1417 comes first. Returns true if at least one byte was added
1418 to ST, false if no bytes were read before an I/O error or
1419 end of file (or if MAX_LENGTH was 0).
1421 This function treats LF and CR LF sequences as new-line,
1422 translating each of them to a single '\n' in ST. */
1424 ds_read_line (struct string *st, FILE *stream, size_t max_length)
1428 for (length = 0; length < max_length; length++)
1430 int c = getc (stream);
1437 ds_put_byte (st, c);
1444 /* CR followed by LF is special: translate to \n. */
1445 ds_put_byte (st, '\n');
1450 /* CR followed by anything else is just CR. */
1451 ds_put_byte (st, '\r');
1459 ds_put_byte (st, c);
1466 /* Removes a comment introduced by `#' from ST,
1467 ignoring occurrences inside quoted strings. */
1469 remove_comment (struct string *st)
1474 for (cp = ds_data (st); cp < ds_end (st); cp++)
1479 else if (*cp == '\\')
1482 else if (*cp == '\'' || *cp == '"')
1484 else if (*cp == '#')
1486 ds_truncate (st, cp - ds_cstr (st));
1491 /* Reads a line from STREAM into ST, then preprocesses as follows:
1493 - Splices lines terminated with `\'.
1495 - Deletes comments introduced by `#' outside of single or double
1498 - Deletes trailing white space.
1500 Returns true if a line was successfully read, false on
1501 failure. If LINE_NUMBER is non-null, then *LINE_NUMBER is
1502 incremented by the number of lines read. */
1504 ds_read_config_line (struct string *st, int *line_number, FILE *stream)
1509 if (!ds_read_line (st, stream, SIZE_MAX))
1512 ds_rtrim (st, ss_cstr (CC_SPACES));
1514 while (ds_chomp_byte (st, '\\'));
1516 remove_comment (st);
1520 /* Attempts to read SIZE * N bytes from STREAM and append them
1522 Returns true if all the requested data was read, false otherwise. */
1524 ds_read_stream (struct string *st, size_t size, size_t n, FILE *stream)
1528 size_t try_bytes = xtimes (n, size);
1529 if (size_in_bounds_p (xsum (ds_length (st), try_bytes)))
1531 char *buffer = ds_put_uninit (st, try_bytes);
1532 size_t got_bytes = fread (buffer, 1, try_bytes, stream);
1533 ds_truncate (st, ds_length (st) - (try_bytes - got_bytes));
1534 return got_bytes == try_bytes;
1546 /* Concatenates S onto ST. */
1548 ds_put_cstr (struct string *st, const char *s)
1551 ds_put_substring (st, ss_cstr (s));
1554 /* Concatenates SS to ST. */
1556 ds_put_substring (struct string *st, struct substring ss)
1559 memcpy (ds_put_uninit (st, ss_length (ss)), ss_data (ss), ss_length (ss));
1562 /* Returns ds_end(ST) and THEN increases the length by INCR. */
1564 ds_put_uninit (struct string *st, size_t incr)
1567 ds_extend (st, ds_length (st) + incr);
1569 st->ss.length += incr;
1573 /* Moves the bytes in ST following offset OFS + OLD_LEN in ST to offset OFS +
1574 NEW_LEN and returns the byte at offset OFS. The first min(OLD_LEN, NEW_LEN)
1575 bytes at the returned position are unchanged; if NEW_LEN > OLD_LEN then the
1576 following NEW_LEN - OLD_LEN bytes are initially indeterminate.
1578 The intention is that the caller should write NEW_LEN bytes at the returned
1579 position, to effectively replace the OLD_LEN bytes previously at that
1582 ds_splice_uninit (struct string *st,
1583 size_t ofs, size_t old_len, size_t new_len)
1585 if (new_len != old_len)
1587 if (new_len > old_len)
1588 ds_extend (st, ds_length (st) + (new_len - old_len));
1590 assert (ds_length (st) >= ofs + old_len);
1592 memmove (ds_data (st) + (ofs + new_len),
1593 ds_data (st) + (ofs + old_len),
1594 ds_length (st) - (ofs + old_len));
1595 st->ss.length += new_len - old_len;
1597 return ds_data (st) + ofs;
1600 /* Formats FORMAT as a printf string and appends the result to ST. */
1602 ds_put_format (struct string *st, const char *format, ...)
1606 va_start (args, format);
1607 ds_put_vformat (st, format, args);
1611 /* Formats FORMAT as a printf string as if in the C locale and appends the result to ST. */
1613 ds_put_c_format (struct string *st, const char *format, ...)
1617 va_start (args, format);
1618 ds_put_c_vformat (st, format, args);
1622 /* Formats FORMAT as a printf string and appends the result to ST. */
1624 ds_put_vformat (struct string *st, const char *format, va_list args_)
1629 va_copy (args, args_);
1630 avail = st->ss.string != NULL ? st->capacity - st->ss.length + 1 : 0;
1631 needed = vsnprintf (st->ss.string + st->ss.length, avail, format, args);
1634 if (needed >= avail)
1636 va_copy (args, args_);
1637 vsnprintf (ds_put_uninit (st, needed), needed + 1, format, args);
1642 /* Some old libc's returned -1 when the destination string
1644 while (needed == -1)
1646 ds_extend (st, (st->capacity + 1) * 2);
1647 avail = st->capacity - st->ss.length + 1;
1649 va_copy (args, args_);
1650 needed = vsnprintf (ds_end (st), avail, format, args);
1653 st->ss.length += needed;
1657 /* Formats FORMAT as a printf string, as if in the C locale,
1658 and appends the result to ST. */
1660 ds_put_c_vformat (struct string *st, const char *format, va_list args)
1663 size_t len = sizeof buf;
1664 char *output = c_vasnprintf (buf, &len, format, args);
1667 ds_put_cstr (st, output);
1673 /* Appends byte CH to ST. */
1675 ds_put_byte (struct string *st, int ch)
1677 ds_put_uninit (st, 1)[0] = ch;
1680 /* Appends N copies of byte CH to ST. */
1682 ds_put_byte_multiple (struct string *st, int ch, size_t n)
1684 memset (ds_put_uninit (st, n), ch, n);
1687 /* Appends Unicode code point UC to ST in UTF-8 encoding. */
1689 ds_put_unichar (struct string *st, ucs4_t uc)
1691 ds_extend (st, ds_length (st) + 6);
1692 st->ss.length += u8_uctomb (CHAR_CAST (uint8_t *, ds_end (st)), uc, 6);
1695 /* Appends N copies of S to ST. */
1697 ds_put_substring_multiple (struct string *dst, struct substring src, size_t n)
1699 char *p = ds_put_uninit (dst, n * src.length);
1700 for (size_t i = 0; i < n; i++)
1702 memcpy (p, src.string, src.length);
1707 /* If relocation has been enabled, replace ST,
1708 with its relocated version */
1710 ds_relocate (struct string *st)
1712 const char *orig = ds_cstr (st);
1713 const char *rel = relocate (orig);
1718 ds_put_cstr (st, rel);
1719 /* The documentation for relocate says that casting away const
1720 and then freeing is appropriate ... */
1721 free (CONST_CAST (char *, rel));
1728 /* Operations on uint8_t "strings" */
1730 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
1731 DST is truncated to DST_SIZE bytes or padded on the right with
1732 copies of PAD as needed. */
1734 u8_buf_copy_rpad (uint8_t *dst, size_t dst_size,
1735 const uint8_t *src, size_t src_size,
1738 if (src_size >= dst_size)
1739 memmove (dst, src, dst_size);
1742 memmove (dst, src, src_size);
1743 memset (&dst[src_size], pad, dst_size - src_size);