1// This is an open source non-commercial project. Dear PVS-Studio, please check
2// it. PVS-Studio Static Code Analyzer for C, C++ and C#: http://www.viva64.com
3
4/// @file charset.c
5///
6/// Code related to character sets.
7
8#include <assert.h>
9#include <string.h>
10#include <wctype.h>
11#include <wchar.h> // for towupper() and towlower()
12#include <inttypes.h>
13
14#include "nvim/vim.h"
15#include "nvim/ascii.h"
16#include "nvim/charset.h"
17#include "nvim/func_attr.h"
18#include "nvim/indent.h"
19#include "nvim/main.h"
20#include "nvim/mark.h"
21#include "nvim/mbyte.h"
22#include "nvim/memline.h"
23#include "nvim/memory.h"
24#include "nvim/misc1.h"
25#include "nvim/garray.h"
26#include "nvim/move.h"
27#include "nvim/option.h"
28#include "nvim/os_unix.h"
29#include "nvim/state.h"
30#include "nvim/strings.h"
31#include "nvim/path.h"
32#include "nvim/cursor.h"
33
34#ifdef INCLUDE_GENERATED_DECLARATIONS
35# include "charset.c.generated.h"
36#endif
37
38
39static bool chartab_initialized = false;
40
41// b_chartab[] is an array with 256 bits, each bit representing one of the
42// characters 0-255.
43#define SET_CHARTAB(buf, c) \
44 (buf)->b_chartab[(unsigned)(c) >> 6] |= (1ull << ((c) & 0x3f))
45#define RESET_CHARTAB(buf, c) \
46 (buf)->b_chartab[(unsigned)(c) >> 6] &= ~(1ull << ((c) & 0x3f))
47#define GET_CHARTAB_TAB(chartab, c) \
48 ((chartab)[(unsigned)(c) >> 6] & (1ull << ((c) & 0x3f)))
49#define GET_CHARTAB(buf, c) \
50 GET_CHARTAB_TAB((buf)->b_chartab, c)
51
52// Table used below, see init_chartab() for an explanation
53static char_u g_chartab[256];
54
55// Flags for g_chartab[].
56#define CT_CELL_MASK 0x07 ///< mask: nr of display cells (1, 2 or 4)
57#define CT_PRINT_CHAR 0x10 ///< flag: set for printable chars
58#define CT_ID_CHAR 0x20 ///< flag: set for ID chars
59#define CT_FNAME_CHAR 0x40 ///< flag: set for file name chars
60
61/// Fill g_chartab[]. Also fills curbuf->b_chartab[] with flags for keyword
62/// characters for current buffer.
63///
64/// Depends on the option settings 'iskeyword', 'isident', 'isfname',
65/// 'isprint' and 'encoding'.
66///
67/// The index in g_chartab[] is the character when first byte is up to 0x80,
68/// if the first byte is 0x80 and above it depends on further bytes.
69///
70/// The contents of g_chartab[]:
71/// - The lower two bits, masked by CT_CELL_MASK, give the number of display
72/// cells the character occupies (1 or 2). Not valid for UTF-8 above 0x80.
73/// - CT_PRINT_CHAR bit is set when the character is printable (no need to
74/// translate the character before displaying it). Note that only DBCS
75/// characters can have 2 display cells and still be printable.
76/// - CT_FNAME_CHAR bit is set when the character can be in a file name.
77/// - CT_ID_CHAR bit is set when the character can be in an identifier.
78///
79/// @return FAIL if 'iskeyword', 'isident', 'isfname' or 'isprint' option has
80/// an error, OK otherwise.
81int init_chartab(void)
82{
83 return buf_init_chartab(curbuf, true);
84}
85
86/// Helper for init_chartab
87///
88/// @param global false: only set buf->b_chartab[]
89///
90/// @return FAIL if 'iskeyword', 'isident', 'isfname' or 'isprint' option has
91/// an error, OK otherwise.
92int buf_init_chartab(buf_T *buf, int global)
93{
94 int c;
95 int c2;
96 int i;
97 bool tilde;
98 bool do_isalpha;
99
100 if (global) {
101 // Set the default size for printable characters:
102 // From <Space> to '~' is 1 (printable), others are 2 (not printable).
103 // This also inits all 'isident' and 'isfname' flags to false.
104 c = 0;
105
106 while (c < ' ') {
107 g_chartab[c++] = (dy_flags & DY_UHEX) ? 4 : 2;
108 }
109
110 while (c <= '~') {
111 g_chartab[c++] = 1 + CT_PRINT_CHAR;
112 }
113
114 while (c < 256) {
115 if (c >= 0xa0) {
116 // UTF-8: bytes 0xa0 - 0xff are printable (latin1)
117 g_chartab[c++] = CT_PRINT_CHAR + 1;
118 } else {
119 // the rest is unprintable by default
120 g_chartab[c++] = (dy_flags & DY_UHEX) ? 4 : 2;
121 }
122 }
123
124 // Assume that every multi-byte char is a filename character.
125 for (c = 1; c < 256; c++) {
126 if (c >= 0xa0) {
127 g_chartab[c] |= CT_FNAME_CHAR;
128 }
129 }
130 }
131
132 // Init word char flags all to false
133 memset(buf->b_chartab, 0, (size_t)32);
134
135 // In lisp mode the '-' character is included in keywords.
136 if (buf->b_p_lisp) {
137 SET_CHARTAB(buf, '-');
138 }
139
140 // Walk through the 'isident', 'iskeyword', 'isfname' and 'isprint'
141 // options Each option is a list of characters, character numbers or
142 // ranges, separated by commas, e.g.: "200-210,x,#-178,-"
143 for (i = global ? 0 : 3; i <= 3; i++) {
144 const char_u *p;
145 if (i == 0) {
146 // first round: 'isident'
147 p = p_isi;
148 } else if (i == 1) {
149 // second round: 'isprint'
150 p = p_isp;
151 } else if (i == 2) {
152 // third round: 'isfname'
153 p = p_isf;
154 } else { // i == 3
155 // fourth round: 'iskeyword'
156 p = buf->b_p_isk;
157 }
158
159 while (*p) {
160 tilde = false;
161 do_isalpha = false;
162
163 if ((*p == '^') && (p[1] != NUL)) {
164 tilde = true;
165 ++p;
166 }
167
168 if (ascii_isdigit(*p)) {
169 c = getdigits_int((char_u **)&p, true, 0);
170 } else {
171 c = mb_ptr2char_adv(&p);
172 }
173 c2 = -1;
174
175 if ((*p == '-') && (p[1] != NUL)) {
176 ++p;
177
178 if (ascii_isdigit(*p)) {
179 c2 = getdigits_int((char_u **)&p, true, 0);
180 } else {
181 c2 = mb_ptr2char_adv(&p);
182 }
183 }
184
185 if ((c <= 0)
186 || (c >= 256)
187 || ((c2 < c) && (c2 != -1))
188 || (c2 >= 256)
189 || !((*p == NUL) || (*p == ','))) {
190 return FAIL;
191 }
192
193 if (c2 == -1) { // not a range
194 // A single '@' (not "@-@"):
195 // Decide on letters being ID/printable/keyword chars with
196 // standard function isalpha(). This takes care of locale for
197 // single-byte characters).
198 if (c == '@') {
199 do_isalpha = true;
200 c = 1;
201 c2 = 255;
202 } else {
203 c2 = c;
204 }
205 }
206
207 while (c <= c2) {
208 // Use the MB_ functions here, because isalpha() doesn't
209 // work properly when 'encoding' is "latin1" and the locale is
210 // "C".
211 if (!do_isalpha
212 || mb_islower(c)
213 || mb_isupper(c)) {
214 if (i == 0) {
215 // (re)set ID flag
216 if (tilde) {
217 g_chartab[c] &= (uint8_t)~CT_ID_CHAR;
218 } else {
219 g_chartab[c] |= CT_ID_CHAR;
220 }
221 } else if (i == 1) {
222 // (re)set printable
223 // For double-byte we keep the cell width, so
224 // that we can detect it from the first byte.
225 if (((c < ' ') || (c > '~'))) {
226 if (tilde) {
227 g_chartab[c] = (uint8_t)((g_chartab[c] & ~CT_CELL_MASK)
228 + ((dy_flags & DY_UHEX) ? 4 : 2));
229 g_chartab[c] &= (uint8_t)~CT_PRINT_CHAR;
230 } else {
231 g_chartab[c] = (uint8_t)((g_chartab[c] & ~CT_CELL_MASK) + 1);
232 g_chartab[c] |= CT_PRINT_CHAR;
233 }
234 }
235 } else if (i == 2) {
236 // (re)set fname flag
237 if (tilde) {
238 g_chartab[c] &= (uint8_t)~CT_FNAME_CHAR;
239 } else {
240 g_chartab[c] |= CT_FNAME_CHAR;
241 }
242 } else { // i == 3
243 // (re)set keyword flag
244 if (tilde) {
245 RESET_CHARTAB(buf, c);
246 } else {
247 SET_CHARTAB(buf, c);
248 }
249 }
250 }
251 ++c;
252 }
253
254 c = *p;
255 p = skip_to_option_part(p);
256
257 if ((c == ',') && (*p == NUL)) {
258 // Trailing comma is not allowed.
259 return FAIL;
260 }
261 }
262 }
263 chartab_initialized = true;
264 return OK;
265}
266
267/// Translate any special characters in buf[bufsize] in-place.
268///
269/// The result is a string with only printable characters, but if there is not
270/// enough room, not all characters will be translated.
271///
272/// @param buf
273/// @param bufsize
274void trans_characters(char_u *buf, int bufsize)
275{
276 int len; // length of string needing translation
277 int room; // room in buffer after string
278 char_u *trs; // translated character
279 int trs_len; // length of trs[]
280
281 len = (int)STRLEN(buf);
282 room = bufsize - len;
283
284 while (*buf != 0) {
285 // Assume a multi-byte character doesn't need translation.
286 if ((trs_len = (*mb_ptr2len)(buf)) > 1) {
287 len -= trs_len;
288 } else {
289 trs = transchar_byte(*buf);
290 trs_len = (int)STRLEN(trs);
291
292 if (trs_len > 1) {
293 room -= trs_len - 1;
294 if (room <= 0) {
295 return;
296 }
297 memmove(buf + trs_len, buf + 1, (size_t)len);
298 }
299 memmove(buf, trs, (size_t)trs_len);
300 --len;
301 }
302 buf += trs_len;
303 }
304}
305
306/// Find length of a string capable of holding s with all specials replaced
307///
308/// Assumes replacing special characters with printable ones just like
309/// strtrans() does.
310///
311/// @param[in] s String to check.
312///
313/// @return number of bytes needed to hold a translation of `s`, NUL byte not
314/// included.
315size_t transstr_len(const char *const s)
316 FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_PURE
317{
318 const char *p = s;
319 size_t len = 0;
320
321 while (*p) {
322 const size_t l = (size_t)utfc_ptr2len((const char_u *)p);
323 if (l > 1) {
324 int pcc[MAX_MCO + 1];
325 pcc[0] = utfc_ptr2char((const char_u *)p, &pcc[1]);
326
327 if (vim_isprintc(pcc[0])) {
328 len += l;
329 } else {
330 for (size_t i = 0; i < ARRAY_SIZE(pcc) && pcc[i]; i++) {
331 char hexbuf[9];
332 len += transchar_hex(hexbuf, pcc[i]);
333 }
334 }
335 p += l;
336 } else {
337 const int b2c_l = byte2cells((uint8_t)(*p++));
338 // Illegal byte sequence may occupy up to 4 characters.
339 len += (size_t)(b2c_l > 0 ? b2c_l : 4);
340 }
341 }
342 return len;
343}
344
345/// Replace special characters with printable ones
346///
347/// @param[in] s String to replace characters from.
348/// @param[out] buf Buffer to which result should be saved.
349/// @param[in] len Buffer length. Resulting string may not occupy more then
350/// len - 1 bytes (one for trailing NUL byte).
351///
352/// @return length of the resulting string, without the NUL byte.
353size_t transstr_buf(const char *const s, char *const buf, const size_t len)
354 FUNC_ATTR_NONNULL_ALL
355{
356 const char *p = s;
357 char *buf_p = buf;
358 char *const buf_e = buf_p + len - 1;
359
360 while (*p != NUL && buf_p < buf_e) {
361 const size_t l = (size_t)utfc_ptr2len((const char_u *)p);
362 if (l > 1) {
363 if (buf_p + l > buf_e) {
364 break; // Exceeded `buf` size.
365 }
366 int pcc[MAX_MCO + 1];
367 pcc[0] = utfc_ptr2char((const char_u *)p, &pcc[1]);
368
369 if (vim_isprintc(pcc[0])) {
370 memmove(buf_p, p, l);
371 buf_p += l;
372 } else {
373 for (size_t i = 0; i < ARRAY_SIZE(pcc) && pcc[i]; i++) {
374 char hexbuf[9]; // <up to 6 bytes>NUL
375 const size_t hexlen = transchar_hex(hexbuf, pcc[i]);
376 if (buf_p + hexlen > buf_e) {
377 break;
378 }
379 memmove(buf_p, hexbuf, hexlen);
380 buf_p += hexlen;
381 }
382 }
383 p += l;
384 } else {
385 const char *const tb = (const char *)transchar_byte((uint8_t)(*p++));
386 const size_t tb_len = strlen(tb);
387 if (buf_p + tb_len > buf_e) {
388 break; // Exceeded `buf` size.
389 }
390 memmove(buf_p, tb, tb_len);
391 buf_p += tb_len;
392 }
393 }
394 *buf_p = NUL;
395 assert(buf_p <= buf_e);
396 return (size_t)(buf_p - buf);
397}
398
399/// Copy string and replace special characters with printable characters
400///
401/// Works like `strtrans()` does, used for that and in some other places.
402///
403/// @param[in] s String to replace characters from.
404///
405/// @return [allocated] translated string
406char *transstr(const char *const s)
407 FUNC_ATTR_NONNULL_RET
408{
409 // Compute the length of the result, taking account of unprintable
410 // multi-byte characters.
411 const size_t len = transstr_len((const char *)s) + 1;
412 char *const buf = xmalloc(len);
413 transstr_buf(s, buf, len);
414 return buf;
415}
416
417/// Convert the string "str[orglen]" to do ignore-case comparing.
418/// Use the current locale.
419///
420/// When "buf" is NULL, return an allocated string.
421/// Otherwise, put the result in buf, limited by buflen, and return buf.
422char_u* str_foldcase(char_u *str, int orglen, char_u *buf, int buflen)
423 FUNC_ATTR_NONNULL_RET
424{
425 garray_T ga;
426 int i;
427 int len = orglen;
428
429#define GA_CHAR(i) ((char_u *)ga.ga_data)[i]
430#define GA_PTR(i) ((char_u *)ga.ga_data + i)
431#define STR_CHAR(i) (buf == NULL ? GA_CHAR(i) : buf[i])
432#define STR_PTR(i) (buf == NULL ? GA_PTR(i) : buf + i)
433
434 // Copy "str" into "buf" or allocated memory, unmodified.
435 if (buf == NULL) {
436 ga_init(&ga, 1, 10);
437
438 ga_grow(&ga, len + 1);
439 memmove(ga.ga_data, str, (size_t)len);
440 ga.ga_len = len;
441 } else {
442 if (len >= buflen) {
443 // Ugly!
444 len = buflen - 1;
445 }
446 memmove(buf, str, (size_t)len);
447 }
448
449 if (buf == NULL) {
450 GA_CHAR(len) = NUL;
451 } else {
452 buf[len] = NUL;
453 }
454
455 // Make each character lower case.
456 i = 0;
457 while (STR_CHAR(i) != NUL) {
458 int c = utf_ptr2char(STR_PTR(i));
459 int olen = utf_ptr2len(STR_PTR(i));
460 int lc = mb_tolower(c);
461
462 // Only replace the character when it is not an invalid
463 // sequence (ASCII character or more than one byte) and
464 // mb_tolower() doesn't return the original character.
465 if (((c < 0x80) || (olen > 1)) && (c != lc)) {
466 int nlen = utf_char2len(lc);
467
468 // If the byte length changes need to shift the following
469 // characters forward or backward.
470 if (olen != nlen) {
471 if (nlen > olen) {
472 if (buf == NULL) {
473 ga_grow(&ga, nlen - olen + 1);
474 } else {
475 if (len + nlen - olen >= buflen) {
476 // out of memory, keep old char
477 lc = c;
478 nlen = olen;
479 }
480 }
481 }
482
483 if (olen != nlen) {
484 if (buf == NULL) {
485 STRMOVE(GA_PTR(i) + nlen, GA_PTR(i) + olen);
486 ga.ga_len += nlen - olen;
487 } else {
488 STRMOVE(buf + i + nlen, buf + i + olen);
489 len += nlen - olen;
490 }
491 }
492 }
493 (void)utf_char2bytes(lc, STR_PTR(i));
494 }
495
496 // skip to next multi-byte char
497 i += (*mb_ptr2len)(STR_PTR(i));
498 }
499
500
501 if (buf == NULL) {
502 return (char_u *)ga.ga_data;
503 }
504 return buf;
505}
506
507// Catch 22: g_chartab[] can't be initialized before the options are
508// initialized, and initializing options may cause transchar() to be called!
509// When chartab_initialized == false don't use g_chartab[].
510// Does NOT work for multi-byte characters, c must be <= 255.
511// Also doesn't work for the first byte of a multi-byte, "c" must be a
512// character!
513static char_u transchar_buf[11];
514
515/// Translate a character into a printable one, leaving printable ASCII intact
516///
517/// All unicode characters are considered non-printable in this function.
518///
519/// @param[in] c Character to translate.
520///
521/// @return translated character into a static buffer.
522char_u *transchar(int c)
523{
524 int i = 0;
525 if (IS_SPECIAL(c)) {
526 // special key code, display as ~@ char
527 transchar_buf[0] = '~';
528 transchar_buf[1] = '@';
529 i = 2;
530 c = K_SECOND(c);
531 }
532
533 if ((!chartab_initialized && (((c >= ' ') && (c <= '~'))))
534 || ((c <= 0xFF) && vim_isprintc_strict(c))) {
535 // printable character
536 transchar_buf[i] = (char_u)c;
537 transchar_buf[i + 1] = NUL;
538 } else if (c <= 0xFF) {
539 transchar_nonprint(transchar_buf + i, c);
540 } else {
541 transchar_hex((char *)transchar_buf + i, c);
542 }
543 return transchar_buf;
544}
545
546/// Like transchar(), but called with a byte instead of a character
547///
548/// Checks for an illegal UTF-8 byte.
549///
550/// @param[in] c Byte to translate.
551///
552/// @return pointer to translated character in transchar_buf.
553char_u *transchar_byte(const int c)
554 FUNC_ATTR_WARN_UNUSED_RESULT
555{
556 if (c >= 0x80) {
557 transchar_nonprint(transchar_buf, c);
558 return transchar_buf;
559 }
560 return transchar(c);
561}
562
563/// Convert non-printable characters to 2..4 printable ones
564///
565/// @warning Does not work for multi-byte characters, c must be <= 255.
566///
567/// @param[out] buf Buffer to store result in, must be able to hold at least
568/// 5 bytes (conversion result + NUL).
569/// @param[in] c Character to convert. NUL is assumed to be NL according to
570/// `:h NL-used-for-NUL`.
571void transchar_nonprint(char_u *buf, int c)
572{
573 if (c == NL) {
574 // we use newline in place of a NUL
575 c = NUL;
576 } else if ((c == CAR) && (get_fileformat(curbuf) == EOL_MAC)) {
577 // we use CR in place of NL in this case
578 c = NL;
579 }
580 assert(c <= 0xff);
581
582 if (dy_flags & DY_UHEX || c > 0x7f) {
583 // 'display' has "uhex"
584 transchar_hex((char *)buf, c);
585 } else {
586 // 0x00 - 0x1f and 0x7f
587 buf[0] = '^';
588 // DEL displayed as ^?
589 buf[1] = (char_u)(c ^ 0x40);
590
591 buf[2] = NUL;
592 }
593}
594
595/// Convert a non-printable character to hex C string like "<FFFF>"
596///
597/// @param[out] buf Buffer to store result in.
598/// @param[in] c Character to convert.
599///
600/// @return Number of bytes stored in buffer, excluding trailing NUL byte.
601size_t transchar_hex(char *const buf, const int c)
602 FUNC_ATTR_NONNULL_ALL
603{
604 size_t i = 0;
605
606 buf[i++] = '<';
607 if (c > 255) {
608 if (c > 255 * 256) {
609 buf[i++] = (char)nr2hex((unsigned)c >> 20);
610 buf[i++] = (char)nr2hex((unsigned)c >> 16);
611 }
612 buf[i++] = (char)nr2hex((unsigned)c >> 12);
613 buf[i++] = (char)nr2hex((unsigned)c >> 8);
614 }
615 buf[i++] = (char)(nr2hex((unsigned)c >> 4));
616 buf[i++] = (char)(nr2hex((unsigned)c));
617 buf[i++] = '>';
618 buf[i] = NUL;
619 return i;
620}
621
622/// Convert the lower 4 bits of byte "c" to its hex character
623///
624/// Lower case letters are used to avoid the confusion of <F1> being 0xf1 or
625/// function key 1.
626///
627/// @param[in] n Number to convert.
628///
629/// @return the hex character.
630static inline unsigned nr2hex(unsigned n)
631 FUNC_ATTR_CONST FUNC_ATTR_WARN_UNUSED_RESULT
632{
633 if ((n & 0xf) <= 9) {
634 return (n & 0xf) + '0';
635 }
636 return (n & 0xf) - 10 + 'a';
637}
638
639/// Return number of display cells occupied by byte "b".
640///
641/// Caller must make sure 0 <= b <= 255.
642/// For multi-byte mode "b" must be the first byte of a character.
643/// A TAB is counted as two cells: "^I".
644/// This will return 0 for bytes >= 0x80, because the number of
645/// cells depends on further bytes in UTF-8.
646///
647/// @param b
648///
649/// @reeturn Number of display cells.
650int byte2cells(int b)
651{
652 if (b >= 0x80) {
653 return 0;
654 }
655 return g_chartab[b] & CT_CELL_MASK;
656}
657
658/// Return number of display cells occupied by character "c".
659///
660/// "c" can be a special key (negative number) in which case 3 or 4 is returned.
661/// A TAB is counted as two cells: "^I" or four: "<09>".
662///
663/// @param c
664///
665/// @return Number of display cells.
666int char2cells(int c)
667{
668 if (IS_SPECIAL(c)) {
669 return char2cells(K_SECOND(c)) + 2;
670 }
671
672 if (c >= 0x80) {
673 // UTF-8: above 0x80 need to check the value
674 return utf_char2cells(c);
675 }
676 return g_chartab[c & 0xff] & CT_CELL_MASK;
677}
678
679/// Return number of display cells occupied by character at "*p".
680/// A TAB is counted as two cells: "^I" or four: "<09>".
681///
682/// @param p
683///
684/// @return number of display cells.
685int ptr2cells(const char_u *p)
686{
687 // For UTF-8 we need to look at more bytes if the first byte is >= 0x80.
688 if (*p >= 0x80) {
689 return utf_ptr2cells(p);
690 }
691
692 // For DBCS we can tell the cell count from the first byte.
693 return g_chartab[*p] & CT_CELL_MASK;
694}
695
696/// Return the number of character cells string "s" will take on the screen,
697/// counting TABs as two characters: "^I".
698///
699/// 's' must be non-null.
700///
701/// @param s
702///
703/// @return number of character cells.
704int vim_strsize(char_u *s)
705{
706 return vim_strnsize(s, (int)MAXCOL);
707}
708
709/// Return the number of character cells string "s[len]" will take on the
710/// screen, counting TABs as two characters: "^I".
711///
712/// 's' must be non-null.
713///
714/// @param s
715/// @param len
716///
717/// @return Number of character cells.
718int vim_strnsize(char_u *s, int len)
719{
720 assert(s != NULL);
721 int size = 0;
722 while (*s != NUL && --len >= 0) {
723 int l = (*mb_ptr2len)(s);
724 size += ptr2cells(s);
725 s += l;
726 len -= l - 1;
727 }
728 return size;
729}
730
731/// Return the number of characters 'c' will take on the screen, taking
732/// into account the size of a tab.
733/// Use a define to make it fast, this is used very often!!!
734/// Also see getvcol() below.
735///
736/// @param p
737/// @param col
738///
739/// @return Number of characters.
740#define RET_WIN_BUF_CHARTABSIZE(wp, buf, p, col) \
741 if (*(p) == TAB && (!(wp)->w_p_list || wp->w_p_lcs_chars.tab1)) { \
742 const int ts = (int)(buf)->b_p_ts; \
743 return (ts - (int)(col % ts)); \
744 } else { \
745 return ptr2cells(p); \
746 }
747
748int chartabsize(char_u *p, colnr_T col)
749{
750 RET_WIN_BUF_CHARTABSIZE(curwin, curbuf, p, col)
751}
752
753static int win_chartabsize(win_T *wp, char_u *p, colnr_T col)
754{
755 RET_WIN_BUF_CHARTABSIZE(wp, wp->w_buffer, p, col)
756}
757
758/// Return the number of characters the string 's' will take on the screen,
759/// taking into account the size of a tab.
760///
761/// @param s
762///
763/// @return Number of characters the string will take on the screen.
764int linetabsize(char_u *s)
765{
766 return linetabsize_col(0, s);
767}
768
769/// Like linetabsize(), but starting at column "startcol".
770///
771/// @param startcol
772/// @param s
773///
774/// @return Number of characters the string will take on the screen.
775int linetabsize_col(int startcol, char_u *s)
776{
777 colnr_T col = startcol;
778 char_u *line = s; /* pointer to start of line, for breakindent */
779
780 while (*s != NUL) {
781 col += lbr_chartabsize_adv(line, &s, col);
782 }
783 return (int)col;
784}
785
786/// Like linetabsize(), but for a given window instead of the current one.
787///
788/// @param wp
789/// @param line
790/// @param len
791///
792/// @return Number of characters the string will take on the screen.
793unsigned int win_linetabsize(win_T *wp, char_u *line, colnr_T len)
794{
795 colnr_T col = 0;
796
797 for (char_u *s = line;
798 *s != NUL && (len == MAXCOL || s < line + len);
799 MB_PTR_ADV(s)) {
800 col += win_lbr_chartabsize(wp, line, s, col, NULL);
801 }
802
803 return (unsigned int)col;
804}
805
806/// Check that "c" is a normal identifier character:
807/// Letters and characters from the 'isident' option.
808///
809/// @param c character to check
810bool vim_isIDc(int c)
811 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT
812{
813 return c > 0 && c < 0x100 && (g_chartab[c] & CT_ID_CHAR);
814}
815
816/// Check that "c" is a keyword character:
817/// Letters and characters from 'iskeyword' option for the current buffer.
818/// For multi-byte characters mb_get_class() is used (builtin rules).
819///
820/// @param c character to check
821bool vim_iswordc(const int c)
822 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT
823{
824 return vim_iswordc_buf(c, curbuf);
825}
826
827/// Check that "c" is a keyword character
828/// Letters and characters from 'iskeyword' option for given buffer.
829/// For multi-byte characters mb_get_class() is used (builtin rules).
830///
831/// @param[in] c Character to check.
832/// @param[in] chartab Buffer chartab.
833bool vim_iswordc_tab(const int c, const uint64_t *const chartab)
834 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ALL
835{
836 return (c >= 0x100
837 ? (utf_class_tab(c, chartab) >= 2)
838 : (c > 0 && GET_CHARTAB_TAB(chartab, c) != 0));
839}
840
841/// Check that "c" is a keyword character:
842/// Letters and characters from 'iskeyword' option for given buffer.
843/// For multi-byte characters mb_get_class() is used (builtin rules).
844///
845/// @param c character to check
846/// @param buf buffer whose keywords to use
847bool vim_iswordc_buf(const int c, buf_T *const buf)
848 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ARG(2)
849{
850 return vim_iswordc_tab(c, buf->b_chartab);
851}
852
853/// Just like vim_iswordc() but uses a pointer to the (multi-byte) character.
854///
855/// @param p pointer to the multi-byte character
856///
857/// @return true if "p" points to a keyword character.
858bool vim_iswordp(const char_u *const p)
859 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ALL
860{
861 return vim_iswordp_buf(p, curbuf);
862}
863
864/// Just like vim_iswordc_buf() but uses a pointer to the (multi-byte)
865/// character.
866///
867/// @param p pointer to the multi-byte character
868/// @param buf buffer whose keywords to use
869///
870/// @return true if "p" points to a keyword character.
871bool vim_iswordp_buf(const char_u *const p, buf_T *const buf)
872 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ALL
873{
874 int c = *p;
875
876 if (MB_BYTE2LEN(c) > 1) {
877 c = utf_ptr2char(p);
878 }
879 return vim_iswordc_buf(c, buf);
880}
881
882/// Check that "c" is a valid file-name character.
883/// Assume characters above 0x100 are valid (multi-byte).
884///
885/// @param c character to check
886bool vim_isfilec(int c)
887 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT
888{
889 return c >= 0x100 || (c > 0 && (g_chartab[c] & CT_FNAME_CHAR));
890}
891
892/// Check that "c" is a valid file-name character or a wildcard character
893/// Assume characters above 0x100 are valid (multi-byte).
894/// Explicitly interpret ']' as a wildcard character as path_has_wildcard("]")
895/// returns false.
896///
897/// @param c character to check
898bool vim_isfilec_or_wc(int c)
899 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT
900{
901 char_u buf[2];
902 buf[0] = (char_u)c;
903 buf[1] = NUL;
904 return vim_isfilec(c) || c == ']' || path_has_wildcard(buf);
905}
906
907/// Check that "c" is a printable character.
908/// Assume characters above 0x100 are printable for double-byte encodings.
909///
910/// @param c character to check
911bool vim_isprintc(int c)
912 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT
913{
914 if (c >= 0x100) {
915 return utf_printable(c);
916 }
917 return c > 0 && (g_chartab[c] & CT_PRINT_CHAR);
918}
919
920/// Strict version of vim_isprintc(c), don't return true if "c" is the head
921/// byte of a double-byte character.
922///
923/// @param c character to check
924///
925/// @return true if "c" is a printable character.
926bool vim_isprintc_strict(int c)
927 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT
928{
929 if (c >= 0x100) {
930 return utf_printable(c);
931 }
932 return c > 0 && (g_chartab[c] & CT_PRINT_CHAR);
933}
934
935/// like chartabsize(), but also check for line breaks on the screen
936///
937/// @param line
938/// @param s
939/// @param col
940///
941/// @return The number of characters taken up on the screen.
942int lbr_chartabsize(char_u *line, unsigned char *s, colnr_T col)
943{
944 if (!curwin->w_p_lbr && (*p_sbr == NUL) && !curwin->w_p_bri) {
945 if (curwin->w_p_wrap) {
946 return win_nolbr_chartabsize(curwin, s, col, NULL);
947 }
948 RET_WIN_BUF_CHARTABSIZE(curwin, curbuf, s, col)
949 }
950 return win_lbr_chartabsize(curwin, line == NULL ? s: line, s, col, NULL);
951}
952
953/// Call lbr_chartabsize() and advance the pointer.
954///
955/// @param line
956/// @param s
957/// @param col
958///
959/// @return The number of characters take up on the screen.
960int lbr_chartabsize_adv(char_u *line, char_u **s, colnr_T col)
961{
962 int retval;
963
964 retval = lbr_chartabsize(line, *s, col);
965 MB_PTR_ADV(*s);
966 return retval;
967}
968
969/// This function is used very often, keep it fast!!!!
970///
971/// If "headp" not NULL, set *headp to the size of what we for 'showbreak'
972/// string at start of line. Warning: *headp is only set if it's a non-zero
973/// value, init to 0 before calling.
974///
975/// @param wp
976/// @param line
977/// @param s
978/// @param col
979/// @param headp
980///
981/// @return The number of characters taken up on the screen.
982int win_lbr_chartabsize(win_T *wp, char_u *line, char_u *s, colnr_T col, int *headp)
983{
984 colnr_T col2;
985 colnr_T col_adj = 0; /* col + screen size of tab */
986 colnr_T colmax;
987 int added;
988 int mb_added = 0;
989 int numberextra;
990 char_u *ps;
991 int n;
992
993 // No 'linebreak', 'showbreak' and 'breakindent': return quickly.
994 if (!wp->w_p_lbr && !wp->w_p_bri && (*p_sbr == NUL)) {
995 if (wp->w_p_wrap) {
996 return win_nolbr_chartabsize(wp, s, col, headp);
997 }
998 RET_WIN_BUF_CHARTABSIZE(wp, wp->w_buffer, s, col)
999 }
1000
1001 // First get normal size, without 'linebreak'
1002 int size = win_chartabsize(wp, s, col);
1003 int c = *s;
1004 if (*s == TAB) {
1005 col_adj = size - 1;
1006 }
1007
1008 // If 'linebreak' set check at a blank before a non-blank if the line
1009 // needs a break here
1010 if (wp->w_p_lbr
1011 && vim_isbreak(c)
1012 && !vim_isbreak((int)s[1])
1013 && wp->w_p_wrap
1014 && (wp->w_width_inner != 0)) {
1015 // Count all characters from first non-blank after a blank up to next
1016 // non-blank after a blank.
1017 numberextra = win_col_off(wp);
1018 col2 = col;
1019 colmax = (colnr_T)(wp->w_width_inner - numberextra - col_adj);
1020
1021 if (col >= colmax) {
1022 colmax += col_adj;
1023 n = colmax + win_col_off2(wp);
1024
1025 if (n > 0) {
1026 colmax += (((col - colmax) / n) + 1) * n - col_adj;
1027 }
1028 }
1029
1030 for (;;) {
1031 ps = s;
1032 MB_PTR_ADV(s);
1033 c = *s;
1034
1035 if (!(c != NUL
1036 && (vim_isbreak(c) || col2 == col || !vim_isbreak((int)(*ps))))) {
1037 break;
1038 }
1039
1040 col2 += win_chartabsize(wp, s, col2);
1041
1042 if (col2 >= colmax) { /* doesn't fit */
1043 size = colmax - col + col_adj;
1044 break;
1045 }
1046 }
1047 } else if ((size == 2)
1048 && (MB_BYTE2LEN(*s) > 1)
1049 && wp->w_p_wrap
1050 && in_win_border(wp, col)) {
1051 // Count the ">" in the last column.
1052 ++size;
1053 mb_added = 1;
1054 }
1055
1056 // May have to add something for 'breakindent' and/or 'showbreak'
1057 // string at start of line.
1058 // Set *headp to the size of what we add.
1059 added = 0;
1060
1061 if ((*p_sbr != NUL || wp->w_p_bri) && wp->w_p_wrap && (col != 0)) {
1062 colnr_T sbrlen = 0;
1063 int numberwidth = win_col_off(wp);
1064
1065 numberextra = numberwidth;
1066 col += numberextra + mb_added;
1067
1068 if (col >= (colnr_T)wp->w_width_inner) {
1069 col -= wp->w_width_inner;
1070 numberextra = wp->w_width_inner - (numberextra - win_col_off2(wp));
1071 if (col >= numberextra && numberextra > 0) {
1072 col %= numberextra;
1073 }
1074 if (*p_sbr != NUL) {
1075 sbrlen = (colnr_T)MB_CHARLEN(p_sbr);
1076 if (col >= sbrlen) {
1077 col -= sbrlen;
1078 }
1079 }
1080 if (col >= numberextra && numberextra > 0) {
1081 col %= numberextra;
1082 } else if (col > 0 && numberextra > 0) {
1083 col += numberwidth - win_col_off2(wp);
1084 }
1085
1086 numberwidth -= win_col_off2(wp);
1087 }
1088
1089 if (col == 0 || (col + size + sbrlen > (colnr_T)wp->w_width_inner)) {
1090 added = 0;
1091
1092 if (*p_sbr != NUL) {
1093 if (size + sbrlen + numberwidth > (colnr_T)wp->w_width_inner) {
1094 // Calculate effective window width.
1095 int width = (colnr_T)wp->w_width_inner - sbrlen - numberwidth;
1096 int prev_width = col ? ((colnr_T)wp->w_width_inner - (sbrlen + col))
1097 : 0;
1098
1099 if (width <= 0) {
1100 width = 1;
1101 }
1102 added += ((size - prev_width) / width) * vim_strsize(p_sbr);
1103 if ((size - prev_width) % width) {
1104 // Wrapped, add another length of 'sbr'.
1105 added += vim_strsize(p_sbr);
1106 }
1107 } else {
1108 added += vim_strsize(p_sbr);
1109 }
1110 }
1111
1112 if (wp->w_p_bri)
1113 added += get_breakindent_win(wp, line);
1114
1115 size += added;
1116 if (col != 0) {
1117 added = 0;
1118 }
1119 }
1120 }
1121
1122 if (headp != NULL) {
1123 *headp = added + mb_added;
1124 }
1125 return size;
1126}
1127
1128/// Like win_lbr_chartabsize(), except that we know 'linebreak' is off and
1129/// 'wrap' is on. This means we need to check for a double-byte character that
1130/// doesn't fit at the end of the screen line.
1131///
1132/// @param wp
1133/// @param s
1134/// @param col
1135/// @param headp
1136///
1137/// @return The number of characters take up on the screen.
1138static int win_nolbr_chartabsize(win_T *wp, char_u *s, colnr_T col, int *headp)
1139{
1140 int n;
1141
1142 if ((*s == TAB) && (!wp->w_p_list || wp->w_p_lcs_chars.tab1)) {
1143 n = (int)wp->w_buffer->b_p_ts;
1144 return n - (col % n);
1145 }
1146 n = ptr2cells(s);
1147
1148 // Add one cell for a double-width character in the last column of the
1149 // window, displayed with a ">".
1150 if ((n == 2) && (MB_BYTE2LEN(*s) > 1) && in_win_border(wp, col)) {
1151 if (headp != NULL) {
1152 *headp = 1;
1153 }
1154 return 3;
1155 }
1156 return n;
1157}
1158
1159/// Check that virtual column "vcol" is in the rightmost column of window "wp".
1160///
1161/// @param wp window
1162/// @param vcol column number
1163bool in_win_border(win_T *wp, colnr_T vcol)
1164 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ARG(1)
1165{
1166 int width1; // width of first line (after line number)
1167 int width2; // width of further lines
1168
1169 if (wp->w_width_inner == 0) {
1170 // there is no border
1171 return false;
1172 }
1173 width1 = wp->w_width_inner - win_col_off(wp);
1174
1175 if ((int)vcol < width1 - 1) {
1176 return false;
1177 }
1178
1179 if ((int)vcol == width1 - 1) {
1180 return true;
1181 }
1182 width2 = width1 + win_col_off2(wp);
1183
1184 if (width2 <= 0) {
1185 return false;
1186 }
1187 return (vcol - width1) % width2 == width2 - 1;
1188}
1189
1190/// Get virtual column number of pos.
1191/// start: on the first position of this character (TAB, ctrl)
1192/// cursor: where the cursor is on this character (first char, except for TAB)
1193/// end: on the last position of this character (TAB, ctrl)
1194///
1195/// This is used very often, keep it fast!
1196///
1197/// @param wp
1198/// @param pos
1199/// @param start
1200/// @param cursor
1201/// @param end
1202void getvcol(win_T *wp, pos_T *pos, colnr_T *start, colnr_T *cursor,
1203 colnr_T *end)
1204{
1205 colnr_T vcol;
1206 char_u *ptr; // points to current char
1207 char_u *posptr; // points to char at pos->col
1208 char_u *line; // start of the line
1209 int incr;
1210 int head;
1211 int ts = (int)wp->w_buffer->b_p_ts;
1212 int c;
1213
1214 vcol = 0;
1215 line = ptr = ml_get_buf(wp->w_buffer, pos->lnum, false);
1216
1217 if (pos->col == MAXCOL) {
1218 // continue until the NUL
1219 posptr = NULL;
1220 } else {
1221 // Special check for an empty line, which can happen on exit, when
1222 // ml_get_buf() always returns an empty string.
1223 if (*ptr == NUL) {
1224 pos->col = 0;
1225 }
1226 posptr = ptr + pos->col;
1227 posptr -= utf_head_off(line, posptr);
1228 }
1229
1230 // This function is used very often, do some speed optimizations.
1231 // When 'list', 'linebreak', 'showbreak' and 'breakindent' are not set
1232 // use a simple loop.
1233 // Also use this when 'list' is set but tabs take their normal size.
1234 if ((!wp->w_p_list || (wp->w_p_lcs_chars.tab1 != NUL))
1235 && !wp->w_p_lbr
1236 && (*p_sbr == NUL)
1237 && !wp->w_p_bri ) {
1238 for (;;) {
1239 head = 0;
1240 c = *ptr;
1241
1242 // make sure we don't go past the end of the line
1243 if (c == NUL) {
1244 // NUL at end of line only takes one column
1245 incr = 1;
1246 break;
1247 }
1248
1249 // A tab gets expanded, depending on the current column
1250 if (c == TAB) {
1251 incr = ts - (vcol % ts);
1252 } else {
1253 // For utf-8, if the byte is >= 0x80, need to look at
1254 // further bytes to find the cell width.
1255 if (c >= 0x80) {
1256 incr = utf_ptr2cells(ptr);
1257 } else {
1258 incr = g_chartab[c] & CT_CELL_MASK;
1259 }
1260
1261 // If a double-cell char doesn't fit at the end of a line
1262 // it wraps to the next line, it's like this char is three
1263 // cells wide.
1264 if ((incr == 2)
1265 && wp->w_p_wrap
1266 && (MB_BYTE2LEN(*ptr) > 1)
1267 && in_win_border(wp, vcol)) {
1268 incr++;
1269 head = 1;
1270 }
1271 }
1272
1273 if ((posptr != NULL) && (ptr >= posptr)) {
1274 // character at pos->col
1275 break;
1276 }
1277
1278 vcol += incr;
1279 MB_PTR_ADV(ptr);
1280 }
1281 } else {
1282 for (;;) {
1283 // A tab gets expanded, depending on the current column
1284 head = 0;
1285 incr = win_lbr_chartabsize(wp, line, ptr, vcol, &head);
1286
1287 // make sure we don't go past the end of the line
1288 if (*ptr == NUL) {
1289 // NUL at end of line only takes one column
1290 incr = 1;
1291 break;
1292 }
1293
1294 if ((posptr != NULL) && (ptr >= posptr)) {
1295 // character at pos->col
1296 break;
1297 }
1298
1299 vcol += incr;
1300 MB_PTR_ADV(ptr);
1301 }
1302 }
1303
1304 if (start != NULL) {
1305 *start = vcol + head;
1306 }
1307
1308 if (end != NULL) {
1309 *end = vcol + incr - 1;
1310 }
1311
1312 if (cursor != NULL) {
1313 if ((*ptr == TAB)
1314 && (State & NORMAL)
1315 && !wp->w_p_list
1316 && !virtual_active()
1317 && !(VIsual_active && ((*p_sel == 'e') || ltoreq(*pos, VIsual)))) {
1318 // cursor at end
1319 *cursor = vcol + incr - 1;
1320 } else {
1321 // cursor at start
1322 *cursor = vcol + head;
1323 }
1324 }
1325}
1326
1327/// Get virtual cursor column in the current window, pretending 'list' is off.
1328///
1329/// @param posp
1330///
1331/// @retujrn The virtual cursor column.
1332colnr_T getvcol_nolist(pos_T *posp)
1333{
1334 int list_save = curwin->w_p_list;
1335 colnr_T vcol;
1336
1337 curwin->w_p_list = false;
1338 if (posp->coladd) {
1339 getvvcol(curwin, posp, NULL, &vcol, NULL);
1340 } else {
1341 getvcol(curwin, posp, NULL, &vcol, NULL);
1342 }
1343 curwin->w_p_list = list_save;
1344 return vcol;
1345}
1346
1347/// Get virtual column in virtual mode.
1348///
1349/// @param wp
1350/// @param pos
1351/// @param start
1352/// @param cursor
1353/// @param end
1354void getvvcol(win_T *wp, pos_T *pos, colnr_T *start, colnr_T *cursor,
1355 colnr_T *end)
1356{
1357 colnr_T col;
1358 colnr_T coladd;
1359 colnr_T endadd;
1360 char_u *ptr;
1361
1362 if (virtual_active()) {
1363 // For virtual mode, only want one value
1364 getvcol(wp, pos, &col, NULL, NULL);
1365
1366 coladd = pos->coladd;
1367 endadd = 0;
1368
1369 // Cannot put the cursor on part of a wide character.
1370 ptr = ml_get_buf(wp->w_buffer, pos->lnum, false);
1371
1372 if (pos->col < (colnr_T)STRLEN(ptr)) {
1373 int c = utf_ptr2char(ptr + pos->col);
1374 if ((c != TAB) && vim_isprintc(c)) {
1375 endadd = (colnr_T)(char2cells(c) - 1);
1376 if (coladd > endadd) {
1377 // past end of line
1378 endadd = 0;
1379 } else {
1380 coladd = 0;
1381 }
1382 }
1383 }
1384 col += coladd;
1385
1386 if (start != NULL) {
1387 *start = col;
1388 }
1389
1390 if (cursor != NULL) {
1391 *cursor = col;
1392 }
1393
1394 if (end != NULL) {
1395 *end = col + endadd;
1396 }
1397 } else {
1398 getvcol(wp, pos, start, cursor, end);
1399 }
1400}
1401
1402/// Get the leftmost and rightmost virtual column of pos1 and pos2.
1403/// Used for Visual block mode.
1404///
1405/// @param wp
1406/// @param pos1
1407/// @param pos2
1408/// @param left
1409/// @param right
1410void getvcols(win_T *wp, pos_T *pos1, pos_T *pos2, colnr_T *left,
1411 colnr_T *right)
1412{
1413 colnr_T from1;
1414 colnr_T from2;
1415 colnr_T to1;
1416 colnr_T to2;
1417
1418 if (lt(*pos1, *pos2)) {
1419 getvvcol(wp, pos1, &from1, NULL, &to1);
1420 getvvcol(wp, pos2, &from2, NULL, &to2);
1421 } else {
1422 getvvcol(wp, pos2, &from1, NULL, &to1);
1423 getvvcol(wp, pos1, &from2, NULL, &to2);
1424 }
1425
1426 if (from2 < from1) {
1427 *left = from2;
1428 } else {
1429 *left = from1;
1430 }
1431
1432 if (to2 > to1) {
1433 if ((*p_sel == 'e') && (from2 - 1 >= to1)) {
1434 *right = from2 - 1;
1435 } else {
1436 *right = to2;
1437 }
1438 } else {
1439 *right = to1;
1440 }
1441}
1442
1443/// skipwhite: skip over ' ' and '\t'.
1444///
1445/// @param[in] q String to skip in.
1446///
1447/// @return Pointer to character after the skipped whitespace.
1448char_u *skipwhite(const char_u *q)
1449 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ALL
1450 FUNC_ATTR_NONNULL_RET
1451{
1452 const char_u *p = q;
1453 while (ascii_iswhite(*p)) {
1454 p++;
1455 }
1456 return (char_u *)p;
1457}
1458
1459// getwhitecols: return the number of whitespace
1460// columns (bytes) at the start of a given line
1461intptr_t getwhitecols_curline(void)
1462{
1463 return getwhitecols(get_cursor_line_ptr());
1464}
1465
1466intptr_t getwhitecols(const char_u *p)
1467{
1468 return skipwhite(p) - p;
1469}
1470
1471/// Skip over digits
1472///
1473/// @param[in] q String to skip digits in.
1474///
1475/// @return Pointer to the character after the skipped digits.
1476char_u *skipdigits(const char_u *q)
1477 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ALL
1478 FUNC_ATTR_NONNULL_RET
1479{
1480 const char_u *p = q;
1481 while (ascii_isdigit(*p)) {
1482 // skip to next non-digit
1483 p++;
1484 }
1485 return (char_u *)p;
1486}
1487
1488/// skip over binary digits
1489///
1490/// @param q pointer to string
1491///
1492/// @return Pointer to the character after the skipped digits.
1493const char* skipbin(const char *q)
1494 FUNC_ATTR_PURE
1495 FUNC_ATTR_NONNULL_ALL
1496 FUNC_ATTR_NONNULL_RET
1497{
1498 const char *p = q;
1499 while (ascii_isbdigit(*p)) {
1500 // skip to next non-digit
1501 p++;
1502 }
1503 return p;
1504}
1505
1506/// skip over digits and hex characters
1507///
1508/// @param q
1509///
1510/// @return Pointer to the character after the skipped digits and hex
1511/// characters.
1512char_u* skiphex(char_u *q)
1513{
1514 char_u *p = q;
1515 while (ascii_isxdigit(*p)) {
1516 // skip to next non-digit
1517 p++;
1518 }
1519 return p;
1520}
1521
1522/// skip to digit (or NUL after the string)
1523///
1524/// @param q
1525///
1526/// @return Pointer to the digit or (NUL after the string).
1527char_u* skiptodigit(char_u *q)
1528{
1529 char_u *p = q;
1530 while (*p != NUL && !ascii_isdigit(*p)) {
1531 // skip to next digit
1532 p++;
1533 }
1534 return p;
1535}
1536
1537/// skip to binary character (or NUL after the string)
1538///
1539/// @param q pointer to string
1540///
1541/// @return Pointer to the binary character or (NUL after the string).
1542const char* skiptobin(const char *q)
1543 FUNC_ATTR_PURE
1544 FUNC_ATTR_NONNULL_ALL
1545 FUNC_ATTR_NONNULL_RET
1546{
1547 const char *p = q;
1548 while (*p != NUL && !ascii_isbdigit(*p)) {
1549 // skip to next digit
1550 p++;
1551 }
1552 return p;
1553}
1554
1555/// skip to hex character (or NUL after the string)
1556///
1557/// @param q
1558///
1559/// @return Pointer to the hex character or (NUL after the string).
1560char_u* skiptohex(char_u *q)
1561{
1562 char_u *p = q;
1563 while (*p != NUL && !ascii_isxdigit(*p)) {
1564 // skip to next digit
1565 p++;
1566 }
1567 return p;
1568}
1569
1570/// Skip over text until ' ' or '\t' or NUL
1571///
1572/// @param[in] p Text to skip over.
1573///
1574/// @return Pointer to the next whitespace or NUL character.
1575char_u *skiptowhite(const char_u *p)
1576{
1577 while (*p != ' ' && *p != '\t' && *p != NUL) {
1578 p++;
1579 }
1580 return (char_u *)p;
1581}
1582
1583/// skiptowhite_esc: Like skiptowhite(), but also skip escaped chars
1584///
1585/// @param p
1586///
1587/// @return Pointer to the next whitespace character.
1588char_u* skiptowhite_esc(char_u *p) {
1589 while (*p != ' ' && *p != '\t' && *p != NUL) {
1590 if (((*p == '\\') || (*p == Ctrl_V)) && (*(p + 1) != NUL)) {
1591 ++p;
1592 }
1593 ++p;
1594 }
1595 return p;
1596}
1597
1598/// Gets a number from a string and skips over it, signalling overflow.
1599///
1600/// @param[out] pp A pointer to a pointer to char_u.
1601/// It will be advanced past the read number.
1602/// @param[out] nr Number read from the string.
1603///
1604/// @return true on success, false on error/overflow
1605bool try_getdigits(char_u **pp, intmax_t *nr)
1606{
1607 errno = 0;
1608 *nr = strtoimax((char *)(*pp), (char **)pp, 10);
1609 if (errno == ERANGE && (*nr == INTMAX_MIN || *nr == INTMAX_MAX)) {
1610 return false;
1611 }
1612 return true;
1613}
1614
1615/// Gets a number from a string and skips over it.
1616///
1617/// @param[out] pp Pointer to a pointer to char_u.
1618/// It will be advanced past the read number.
1619/// @param strict Abort on overflow.
1620/// @param def Default value, if parsing fails or overflow occurs.
1621///
1622/// @return Number read from the string, or `def` on parse failure or overflow.
1623intmax_t getdigits(char_u **pp, bool strict, intmax_t def)
1624{
1625 intmax_t number;
1626 int ok = try_getdigits(pp, &number);
1627 if (strict && !ok) {
1628 abort();
1629 }
1630 return ok ? number : def;
1631}
1632
1633/// Gets an int number from a string.
1634///
1635/// @see getdigits
1636int getdigits_int(char_u **pp, bool strict, int def)
1637{
1638 intmax_t number = getdigits(pp, strict, def);
1639#if SIZEOF_INTMAX_T > SIZEOF_INT
1640 if (strict) {
1641 assert(number >= INT_MIN && number <= INT_MAX);
1642 } else if (!(number >= INT_MIN && number <= INT_MAX)) {
1643 return def;
1644 }
1645#endif
1646 return (int)number;
1647}
1648
1649/// Gets a long number from a string.
1650///
1651/// @see getdigits
1652long getdigits_long(char_u **pp, bool strict, long def)
1653{
1654 intmax_t number = getdigits(pp, strict, def);
1655#if SIZEOF_INTMAX_T > SIZEOF_LONG
1656 if (strict) {
1657 assert(number >= LONG_MIN && number <= LONG_MAX);
1658 } else if (!(number >= LONG_MIN && number <= LONG_MAX)) {
1659 return def;
1660 }
1661#endif
1662 return (long)number;
1663}
1664
1665/// Check that "lbuf" is empty or only contains blanks.
1666///
1667/// @param lbuf line buffer to check
1668bool vim_isblankline(char_u *lbuf)
1669{
1670 char_u *p = skipwhite(lbuf);
1671 return *p == NUL || *p == '\r' || *p == '\n';
1672}
1673
1674/// Convert a string into a long and/or unsigned long, taking care of
1675/// hexadecimal, octal and binary numbers. Accepts a '-' sign.
1676/// If "prep" is not NULL, returns a flag to indicate the type of the number:
1677/// 0 decimal
1678/// '0' octal
1679/// 'B' bin
1680/// 'b' bin
1681/// 'X' hex
1682/// 'x' hex
1683/// If "len" is not NULL, the length of the number in characters is returned.
1684/// If "nptr" is not NULL, the signed result is returned in it.
1685/// If "unptr" is not NULL, the unsigned result is returned in it.
1686/// If "what" contains STR2NR_BIN recognize binary numbers.
1687/// If "what" contains STR2NR_OCT recognize octal numbers.
1688/// If "what" contains STR2NR_HEX recognize hex numbers.
1689/// If "what" contains STR2NR_FORCE always assume bin/oct/hex.
1690/// If maxlen > 0, check at a maximum maxlen chars.
1691///
1692/// @param start
1693/// @param prep Returns guessed type of number 0 = decimal, 'x' or 'X' is
1694/// hexadecimal, '0' = octal, 'b' or 'B' is binary. When using
1695/// STR2NR_FORCE is always zero.
1696/// @param len Returns the detected length of number.
1697/// @param what Recognizes what number passed, @see ChStr2NrFlags.
1698/// @param nptr Returns the signed result.
1699/// @param unptr Returns the unsigned result.
1700/// @param maxlen Max length of string to check.
1701void vim_str2nr(const char_u *const start, int *const prep, int *const len,
1702 const int what, varnumber_T *const nptr,
1703 uvarnumber_T *const unptr, const int maxlen)
1704 FUNC_ATTR_NONNULL_ARG(1)
1705{
1706 const char *ptr = (const char *)start;
1707#define STRING_ENDED(ptr) \
1708 (!(maxlen == 0 || (int)((ptr) - (const char *)start) < maxlen))
1709 int pre = 0; // default is decimal
1710 const bool negative = (ptr[0] == '-');
1711 uvarnumber_T un = 0;
1712
1713 if (negative) {
1714 ptr++;
1715 }
1716
1717 if (what & STR2NR_FORCE) {
1718 // When forcing main consideration is skipping the prefix. Octal and decimal
1719 // numbers have no prefixes to skip. pre is not set.
1720 switch ((unsigned)what & (~(unsigned)STR2NR_FORCE)) {
1721 case STR2NR_HEX: {
1722 if (!STRING_ENDED(ptr + 2)
1723 && ptr[0] == '0'
1724 && (ptr[1] == 'x' || ptr[1] == 'X')
1725 && ascii_isxdigit(ptr[2])) {
1726 ptr += 2;
1727 }
1728 goto vim_str2nr_hex;
1729 }
1730 case STR2NR_BIN: {
1731 if (!STRING_ENDED(ptr + 2)
1732 && ptr[0] == '0'
1733 && (ptr[1] == 'b' || ptr[1] == 'B')
1734 && ascii_isbdigit(ptr[2])) {
1735 ptr += 2;
1736 }
1737 goto vim_str2nr_bin;
1738 }
1739 case STR2NR_OCT: {
1740 goto vim_str2nr_oct;
1741 }
1742 case 0: {
1743 goto vim_str2nr_dec;
1744 }
1745 default: {
1746 assert(false);
1747 }
1748 }
1749 } else if ((what & (STR2NR_HEX|STR2NR_OCT|STR2NR_BIN))
1750 && !STRING_ENDED(ptr + 1)
1751 && ptr[0] == '0' && ptr[1] != '8' && ptr[1] != '9') {
1752 pre = ptr[1];
1753 // Detect hexadecimal: 0x or 0X followed by hex digit.
1754 if ((what & STR2NR_HEX)
1755 && !STRING_ENDED(ptr + 2)
1756 && (pre == 'X' || pre == 'x')
1757 && ascii_isxdigit(ptr[2])) {
1758 ptr += 2;
1759 goto vim_str2nr_hex;
1760 }
1761 // Detect binary: 0b or 0B followed by 0 or 1.
1762 if ((what & STR2NR_BIN)
1763 && !STRING_ENDED(ptr + 2)
1764 && (pre == 'B' || pre == 'b')
1765 && ascii_isbdigit(ptr[2])) {
1766 ptr += 2;
1767 goto vim_str2nr_bin;
1768 }
1769 // Detect octal number: zero followed by octal digits without '8' or '9'.
1770 pre = 0;
1771 if (!(what & STR2NR_OCT)
1772 || !('0' <= ptr[1] && ptr[1] <= '7')) {
1773 goto vim_str2nr_dec;
1774 }
1775 for (int i = 2; !STRING_ENDED(ptr + i) && ascii_isdigit(ptr[i]); i++) {
1776 if (ptr[i] > '7') {
1777 goto vim_str2nr_dec;
1778 }
1779 }
1780 pre = '0';
1781 goto vim_str2nr_oct;
1782 } else {
1783 goto vim_str2nr_dec;
1784 }
1785
1786 // Do the string-to-numeric conversion "manually" to avoid sscanf quirks.
1787 assert(false); // Should’ve used goto earlier.
1788#define PARSE_NUMBER(base, cond, conv) \
1789 do { \
1790 while (!STRING_ENDED(ptr) && (cond)) { \
1791 const uvarnumber_T digit = (uvarnumber_T)(conv); \
1792 /* avoid ubsan error for overflow */ \
1793 if (un < UVARNUMBER_MAX / base \
1794 || (un == UVARNUMBER_MAX / base \
1795 && (base != 10 || digit <= UVARNUMBER_MAX % 10))) { \
1796 un = base * un + digit; \
1797 } else { \
1798 un = UVARNUMBER_MAX; \
1799 } \
1800 ptr++; \
1801 } \
1802 } while (0)
1803vim_str2nr_bin:
1804 PARSE_NUMBER(2, (*ptr == '0' || *ptr == '1'), (*ptr - '0'));
1805 goto vim_str2nr_proceed;
1806vim_str2nr_oct:
1807 PARSE_NUMBER(8, ('0' <= *ptr && *ptr <= '7'), (*ptr - '0'));
1808 goto vim_str2nr_proceed;
1809vim_str2nr_dec:
1810 PARSE_NUMBER(10, (ascii_isdigit(*ptr)), (*ptr - '0'));
1811 goto vim_str2nr_proceed;
1812vim_str2nr_hex:
1813 PARSE_NUMBER(16, (ascii_isxdigit(*ptr)), (hex2nr(*ptr)));
1814 goto vim_str2nr_proceed;
1815#undef PARSE_NUMBER
1816
1817vim_str2nr_proceed:
1818 if (prep != NULL) {
1819 *prep = pre;
1820 }
1821
1822 if (len != NULL) {
1823 *len = (int)(ptr - (const char *)start);
1824 }
1825
1826 if (nptr != NULL) {
1827 if (negative) { // account for leading '-' for decimal numbers
1828 // avoid ubsan error for overflow
1829 if (un > VARNUMBER_MAX) {
1830 *nptr = VARNUMBER_MIN;
1831 } else {
1832 *nptr = -(varnumber_T)un;
1833 }
1834 } else {
1835 if (un > VARNUMBER_MAX) {
1836 un = VARNUMBER_MAX;
1837 }
1838 *nptr = (varnumber_T)un;
1839 }
1840 }
1841
1842 if (unptr != NULL) {
1843 *unptr = un;
1844 }
1845#undef STRING_ENDED
1846}
1847
1848/// Return the value of a single hex character.
1849/// Only valid when the argument is '0' - '9', 'A' - 'F' or 'a' - 'f'.
1850///
1851/// @param c
1852///
1853/// @return The value of the hex character.
1854int hex2nr(int c)
1855{
1856 if ((c >= 'a') && (c <= 'f')) {
1857 return c - 'a' + 10;
1858 }
1859
1860 if ((c >= 'A') && (c <= 'F')) {
1861 return c - 'A' + 10;
1862 }
1863 return c - '0';
1864}
1865
1866/// Check that "str" starts with a backslash that should be removed.
1867/// For Windows this is only done when the character after the
1868/// backslash is not a normal file name character.
1869/// '$' is a valid file name character, we don't remove the backslash before
1870/// it. This means it is not possible to use an environment variable after a
1871/// backslash. "C:\$VIM\doc" is taken literally, only "$VIM\doc" works.
1872/// Although "\ name" is valid, the backslash in "Program\ files" must be
1873/// removed. Assume a file name doesn't start with a space.
1874/// For multi-byte names, never remove a backslash before a non-ascii
1875/// character, assume that all multi-byte characters are valid file name
1876/// characters.
1877///
1878/// @param str file path string to check
1879bool rem_backslash(const char_u *str)
1880 FUNC_ATTR_PURE FUNC_ATTR_WARN_UNUSED_RESULT FUNC_ATTR_NONNULL_ALL
1881{
1882#ifdef BACKSLASH_IN_FILENAME
1883 return str[0] == '\\'
1884 && str[1] < 0x80
1885 && (str[1] == ' '
1886 || (str[1] != NUL
1887 && str[1] != '*'
1888 && str[1] != '?'
1889 && !vim_isfilec(str[1])));
1890
1891#else // ifdef BACKSLASH_IN_FILENAME
1892 return str[0] == '\\' && str[1] != NUL;
1893#endif // ifdef BACKSLASH_IN_FILENAME
1894}
1895
1896/// Halve the number of backslashes in a file name argument.
1897///
1898/// @param p
1899void backslash_halve(char_u *p)
1900{
1901 for (; *p; ++p) {
1902 if (rem_backslash(p)) {
1903 STRMOVE(p, p + 1);
1904 }
1905 }
1906}
1907
1908/// backslash_halve() plus save the result in allocated memory.
1909///
1910/// @param p
1911///
1912/// @return String with the number of backslashes halved.
1913char_u *backslash_halve_save(const char_u *p)
1914 FUNC_ATTR_NONNULL_ALL FUNC_ATTR_NONNULL_RET
1915{
1916 // TODO(philix): simplify and improve backslash_halve_save algorithm
1917 char_u *res = vim_strsave(p);
1918 backslash_halve(res);
1919 return res;
1920}
1921