Subversion Repositories HelenOS

Rev

Rev 4267 | Rev 4312 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
4011 svoboda 1
/*
2
 * Copyright (c) 2001-2004 Jakub Jermar
3
 * All rights reserved.
4
 *
5
 * Redistribution and use in source and binary forms, with or without
6
 * modification, are permitted provided that the following conditions
7
 * are met:
8
 *
9
 * - Redistributions of source code must retain the above copyright
10
 *   notice, this list of conditions and the following disclaimer.
11
 * - Redistributions in binary form must reproduce the above copyright
12
 *   notice, this list of conditions and the following disclaimer in the
13
 *   documentation and/or other materials provided with the distribution.
14
 * - The name of the author may not be used to endorse or promote products
15
 *   derived from this software without specific prior written permission.
16
 *
17
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
18
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
19
 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
20
 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
21
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
22
 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
23
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
24
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
25
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
26
 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
27
 */
28
 
4014 decky 29
/** @addtogroup generic
4011 svoboda 30
 * @{
31
 */
32
 
33
/**
34
 * @file
4209 svoboda 35
 * @brief String functions.
36
 *
37
 * Strings and characters use the Universal Character Set (UCS). The standard
38
 * strings, called just strings are encoded in UTF-8. Wide strings (encoded
39
 * in UTF-32) are supported to a limited degree. A single character is
4212 decky 40
 * represented as wchar_t.@n
4209 svoboda 41
 *
4212 decky 42
 * Overview of the terminology:@n
4209 svoboda 43
 *
4212 decky 44
 *  Term                  Meaning
45
 *  --------------------  ----------------------------------------------------
46
 *  byte                  8 bits stored in uint8_t (unsigned 8 bit integer)
4209 svoboda 47
 *
4212 decky 48
 *  character             UTF-32 encoded Unicode character, stored in wchar_t
49
 *                        (signed 32 bit integer), code points 0 .. 1114111
50
 *                        are valid
4209 svoboda 51
 *
4212 decky 52
 *  ASCII character       7 bit encoded ASCII character, stored in char
53
 *                        (usually signed 8 bit integer), code points 0 .. 127
54
 *                        are valid
4209 svoboda 55
 *
4212 decky 56
 *  string                UTF-8 encoded NULL-terminated Unicode string, char *
57
 *
58
 *  wide string           UTF-32 encoded NULL-terminated Unicode string,
59
 *                        wchar_t *
60
 *
61
 *  [wide] string size    number of BYTES in a [wide] string (excluding
62
 *                        the NULL-terminator), size_t
63
 *
64
 *  [wide] string length  number of CHARACTERS in a [wide] string (excluding
65
 *                        the NULL-terminator), count_t
66
 *
67
 *  [wide] string width   number of display cells on a monospace display taken
68
 *                        by a [wide] string, count_t
69
 *
70
 *
71
 * Overview of string metrics:@n
72
 *
73
 *  Metric  Abbrev.  Type     Meaning
74
 *  ------  ------   ------   -------------------------------------------------
75
 *  size    n        size_t   number of BYTES in a string (excluding the
76
 *                            NULL-terminator)
77
 *
78
 *  length  l        count_t  number of CHARACTERS in a string (excluding the
79
 *                            null terminator)
80
 *
81
 *  width  w         count_t  number of display cells on a monospace display
82
 *                            taken by a string
83
 *
84
 *
85
 * Function naming prefixes:@n
86
 *
87
 *  chr_    operate on characters
88
 *  ascii_  operate on ASCII characters
89
 *  str_    operate on strings
90
 *  wstr_   operate on wide strings
91
 *
92
 *  [w]str_[n|l|w]  operate on a prefix limited by size, length
93
 *                  or width
94
 *
95
 *
96
 * A specific character inside a [wide] string can be referred to by:@n
97
 *
98
 *  pointer (char *, wchar_t *)
99
 *  byte offset (size_t)
100
 *  character index (count_t)
101
 *
4011 svoboda 102
 */
103
 
104
#include <string.h>
105
#include <print.h>
106
#include <cpu.h>
107
#include <arch/asm.h>
108
#include <arch.h>
4208 svoboda 109
#include <errno.h>
4212 decky 110
#include <align.h>
4281 svoboda 111
#include <debug.h>
4011 svoboda 112
 
4212 decky 113
/** Byte mask consisting of lowest @n bits (out of 8) */
114
#define LO_MASK_8(n)  ((uint8_t) ((1 << (n)) - 1))
4196 svoboda 115
 
4212 decky 116
/** Byte mask consisting of lowest @n bits (out of 32) */
117
#define LO_MASK_32(n)  ((uint32_t) ((1 << (n)) - 1))
4198 svoboda 118
 
4212 decky 119
/** Byte mask consisting of highest @n bits (out of 8) */
120
#define HI_MASK_8(n)  (~LO_MASK_8(8 - (n)))
4198 svoboda 121
 
4212 decky 122
/** Number of data bits in a UTF-8 continuation byte */
123
#define CONT_BITS  6
4196 svoboda 124
 
4212 decky 125
/** Decode a single character from a string.
4175 decky 126
 *
4212 decky 127
 * Decode a single character from a string of size @a size. Decoding starts
4200 svoboda 128
 * at @a offset and this offset is moved to the beginning of the next
129
 * character. In case of decoding error, offset generally advances at least
4212 decky 130
 * by one. However, offset is never moved beyond size.
4175 decky 131
 *
4212 decky 132
 * @param str    String (not necessarily NULL-terminated).
133
 * @param offset Byte offset in string where to start decoding.
134
 * @param size   Size of the string (in bytes).
4175 decky 135
 *
4223 decky 136
 * @return Value of decoded character, U_SPECIAL on decoding error or
4212 decky 137
 *         NULL if attempt to decode beyond @a size.
138
 *
4175 decky 139
 */
4212 decky 140
wchar_t str_decode(const char *str, size_t *offset, size_t size)
4175 decky 141
{
4212 decky 142
    if (*offset + 1 > size)
143
        return 0;
144
 
145
    /* First byte read from string */
146
    uint8_t b0 = (uint8_t) str[(*offset)++];
147
 
148
    /* Determine code length */
149
 
150
    unsigned int b0_bits;  /* Data bits in first byte */
151
    unsigned int cbytes;   /* Number of continuation bytes */
152
 
4196 svoboda 153
    if ((b0 & 0x80) == 0) {
154
        /* 0xxxxxxx (Plain ASCII) */
155
        b0_bits = 7;
156
        cbytes = 0;
157
    } else if ((b0 & 0xe0) == 0xc0) {
158
        /* 110xxxxx 10xxxxxx */
159
        b0_bits = 5;
160
        cbytes = 1;
161
    } else if ((b0 & 0xf0) == 0xe0) {
162
        /* 1110xxxx 10xxxxxx 10xxxxxx */
163
        b0_bits = 4;
164
        cbytes = 2;
165
    } else if ((b0 & 0xf8) == 0xf0) {
166
        /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
167
        b0_bits = 3;
168
        cbytes = 3;
169
    } else {
4212 decky 170
        /* 10xxxxxx -- unexpected continuation byte */
4223 decky 171
        return U_SPECIAL;
4179 decky 172
    }
4212 decky 173
 
174
    if (*offset + cbytes > size)
4223 decky 175
        return U_SPECIAL;
4212 decky 176
 
177
    wchar_t ch = b0 & LO_MASK_8(b0_bits);
178
 
179
    /* Decode continuation bytes */
4196 svoboda 180
    while (cbytes > 0) {
4212 decky 181
        uint8_t b = (uint8_t) str[(*offset)++];
182
 
183
        /* Must be 10xxxxxx */
184
        if ((b & 0xc0) != 0x80)
4223 decky 185
            return U_SPECIAL;
4212 decky 186
 
187
        /* Shift data bits to ch */
4196 svoboda 188
        ch = (ch << CONT_BITS) | (wchar_t) (b & LO_MASK_8(CONT_BITS));
4212 decky 189
        cbytes--;
4179 decky 190
    }
4212 decky 191
 
4196 svoboda 192
    return ch;
4175 decky 193
}
194
 
4200 svoboda 195
/** Encode a single character to string representation.
4011 svoboda 196
 *
4200 svoboda 197
 * Encode a single character to string representation (i.e. UTF-8) and store
198
 * it into a buffer at @a offset. Encoding starts at @a offset and this offset
199
 * is moved to the position where the next character can be written to.
4011 svoboda 200
 *
4212 decky 201
 * @param ch     Input character.
202
 * @param str    Output buffer.
203
 * @param offset Byte offset where to start writing.
204
 * @param size   Size of the output buffer (in bytes).
4179 decky 205
 *
4208 svoboda 206
 * @return EOK if the character was encoded successfully, EOVERFLOW if there
207
 *     was not enough space in the output buffer or EINVAL if the character
208
 *     code was invalid.
4179 decky 209
 */
4234 svoboda 210
int chr_encode(wchar_t ch, char *str, size_t *offset, size_t size)
4179 decky 211
{
4212 decky 212
    if (*offset >= size)
4208 svoboda 213
        return EOVERFLOW;
4212 decky 214
 
215
    if (!chr_check(ch))
4208 svoboda 216
        return EINVAL;
4212 decky 217
 
218
    /* Unsigned version of ch (bit operations should only be done
219
       on unsigned types). */
220
    uint32_t cc = (uint32_t) ch;
221
 
222
    /* Determine how many continuation bytes are needed */
223
 
224
    unsigned int b0_bits;  /* Data bits in first byte */
225
    unsigned int cbytes;   /* Number of continuation bytes */
226
 
4198 svoboda 227
    if ((cc & ~LO_MASK_32(7)) == 0) {
228
        b0_bits = 7;
229
        cbytes = 0;
230
    } else if ((cc & ~LO_MASK_32(11)) == 0) {
231
        b0_bits = 5;
232
        cbytes = 1;
233
    } else if ((cc & ~LO_MASK_32(16)) == 0) {
234
        b0_bits = 4;
235
        cbytes = 2;
236
    } else if ((cc & ~LO_MASK_32(21)) == 0) {
237
        b0_bits = 3;
238
        cbytes = 3;
239
    } else {
4212 decky 240
        /* Codes longer than 21 bits are not supported */
4208 svoboda 241
        return EINVAL;
4179 decky 242
    }
4212 decky 243
 
244
    /* Check for available space in buffer */
245
    if (*offset + cbytes >= size)
4208 svoboda 246
        return EOVERFLOW;
4212 decky 247
 
248
    /* Encode continuation bytes */
249
    unsigned int i;
250
    for (i = cbytes; i > 0; i--) {
4200 svoboda 251
        str[*offset + i] = 0x80 | (cc & LO_MASK_32(CONT_BITS));
4198 svoboda 252
        cc = cc >> CONT_BITS;
4179 decky 253
    }
4212 decky 254
 
255
    /* Encode first byte */
4200 svoboda 256
    str[*offset] = (cc & LO_MASK_32(b0_bits)) | HI_MASK_8(8 - b0_bits - 1);
4179 decky 257
 
4212 decky 258
    /* Advance offset */
259
    *offset += cbytes + 1;
260
 
4208 svoboda 261
    return EOK;
4179 decky 262
}
263
 
4212 decky 264
/** Get size of string.
4209 svoboda 265
 *
4212 decky 266
 * Get the number of bytes which are used by the string @a str (excluding the
267
 * NULL-terminator).
268
 *
269
 * @param str String to consider.
270
 *
271
 * @return Number of bytes used by the string
272
 *
4209 svoboda 273
 */
4212 decky 274
size_t str_size(const char *str)
4209 svoboda 275
{
4212 decky 276
    size_t size = 0;
277
 
278
    while (*str++ != 0)
279
        size++;
280
 
281
    return size;
4209 svoboda 282
}
283
 
4212 decky 284
/** Get size of wide string.
4179 decky 285
 *
4212 decky 286
 * Get the number of bytes which are used by the wide string @a str (excluding the
287
 * NULL-terminator).
288
 *
289
 * @param str Wide string to consider.
290
 *
291
 * @return Number of bytes used by the wide string
292
 *
293
 */
294
size_t wstr_size(const wchar_t *str)
295
{
296
    return (wstr_length(str) * sizeof(wchar_t));
297
}
298
 
299
/** Get size of string with length limit.
300
 *
4205 svoboda 301
 * Get the number of bytes which are used by up to @a max_len first
302
 * characters in the string @a str. If @a max_len is greater than
4212 decky 303
 * the length of @a str, the entire string is measured (excluding the
304
 * NULL-terminator).
4179 decky 305
 *
4212 decky 306
 * @param str     String to consider.
307
 * @param max_len Maximum number of characters to measure.
4179 decky 308
 *
4212 decky 309
 * @return Number of bytes used by the characters.
310
 *
4179 decky 311
 */
4205 svoboda 312
size_t str_lsize(const char *str, count_t max_len)
4179 decky 313
{
4205 svoboda 314
    count_t len = 0;
4212 decky 315
    size_t offset = 0;
316
 
317
    while (len < max_len) {
318
        if (str_decode(str, &offset, STR_NO_LIMIT) == 0)
4199 svoboda 319
            break;
4212 decky 320
 
4205 svoboda 321
        len++;
4179 decky 322
    }
4212 decky 323
 
324
    return offset;
4179 decky 325
}
326
 
4212 decky 327
/** Get size of wide string with length limit.
4209 svoboda 328
 *
4212 decky 329
 * Get the number of bytes which are used by up to @a max_len first
330
 * wide characters in the wide string @a str. If @a max_len is greater than
331
 * the length of @a str, the entire wide string is measured (excluding the
332
 * NULL-terminator).
4209 svoboda 333
 *
4212 decky 334
 * @param str     Wide string to consider.
335
 * @param max_len Maximum number of wide characters to measure.
4209 svoboda 336
 *
4212 decky 337
 * @return Number of bytes used by the wide characters.
338
 *
4209 svoboda 339
 */
4212 decky 340
size_t wstr_lsize(const wchar_t *str, count_t max_len)
4209 svoboda 341
{
4212 decky 342
    return (wstr_nlength(str, max_len * sizeof(wchar_t)) * sizeof(wchar_t));
4209 svoboda 343
}
344
 
4212 decky 345
/** Get number of characters in a string.
4209 svoboda 346
 *
4212 decky 347
 * @param str NULL-terminated string.
4209 svoboda 348
 *
4212 decky 349
 * @return Number of characters in string.
4209 svoboda 350
 *
351
 */
4212 decky 352
count_t str_length(const char *str)
4209 svoboda 353
{
4212 decky 354
    count_t len = 0;
355
    size_t offset = 0;
356
 
357
    while (str_decode(str, &offset, STR_NO_LIMIT) != 0)
358
        len++;
359
 
360
    return len;
4209 svoboda 361
}
362
 
4212 decky 363
/** Get number of characters in a wide string.
4179 decky 364
 *
4212 decky 365
 * @param str NULL-terminated wide string.
4179 decky 366
 *
4212 decky 367
 * @return Number of characters in @a str.
368
 *
4179 decky 369
 */
4212 decky 370
count_t wstr_length(const wchar_t *wstr)
4179 decky 371
{
4212 decky 372
    count_t len = 0;
4179 decky 373
 
4212 decky 374
    while (*wstr++ != 0)
375
        len++;
376
 
377
    return len;
4179 decky 378
}
379
 
4212 decky 380
/** Get number of characters in a string with size limit.
4179 decky 381
 *
4212 decky 382
 * @param str  NULL-terminated string.
383
 * @param size Maximum number of bytes to consider.
384
 *
385
 * @return Number of characters in string.
386
 *
4179 decky 387
 */
4212 decky 388
count_t str_nlength(const char *str, size_t size)
4179 decky 389
{
4212 decky 390
    count_t len = 0;
391
    size_t offset = 0;
4179 decky 392
 
4212 decky 393
    while (str_decode(str, &offset, size) != 0)
394
        len++;
395
 
396
    return len;
4179 decky 397
}
398
 
4212 decky 399
/** Get number of characters in a string with size limit.
4179 decky 400
 *
4212 decky 401
 * @param str  NULL-terminated string.
402
 * @param size Maximum number of bytes to consider.
4179 decky 403
 *
4205 svoboda 404
 * @return Number of characters in string.
4212 decky 405
 *
4179 decky 406
 */
4212 decky 407
count_t wstr_nlength(const wchar_t *str, size_t size)
4179 decky 408
{
4205 svoboda 409
    count_t len = 0;
4212 decky 410
    count_t limit = ALIGN_DOWN(size, sizeof(wchar_t));
411
    count_t offset = 0;
412
 
413
    while ((offset < limit) && (*str++ != 0)) {
4205 svoboda 414
        len++;
4212 decky 415
        offset += sizeof(wchar_t);
4179 decky 416
    }
4212 decky 417
 
4205 svoboda 418
    return len;
4011 svoboda 419
}
420
 
4212 decky 421
/** Check whether character is plain ASCII.
4179 decky 422
 *
4212 decky 423
 * @return True if character is plain ASCII.
424
 *
4179 decky 425
 */
4234 svoboda 426
bool ascii_check(wchar_t ch)
4179 decky 427
{
4212 decky 428
    if ((ch >= 0) && (ch <= 127))
429
        return true;
430
 
431
    return false;
432
}
4205 svoboda 433
 
4212 decky 434
/** Check whether character is valid
435
 *
436
 * @return True if character is a valid Unicode code point.
437
 *
438
 */
4234 svoboda 439
bool chr_check(wchar_t ch)
4212 decky 440
{
441
    if ((ch >= 0) && (ch <= 1114111))
442
        return true;
443
 
444
    return false;
4179 decky 445
}
446
 
4212 decky 447
/** Compare two NULL terminated strings.
4011 svoboda 448
 *
4212 decky 449
 * Do a char-by-char comparison of two NULL-terminated strings.
4011 svoboda 450
 * The strings are considered equal iff they consist of the same
451
 * characters on the minimum of their lengths.
452
 *
4212 decky 453
 * @param s1 First string to compare.
454
 * @param s2 Second string to compare.
4011 svoboda 455
 *
4212 decky 456
 * @return 0 if the strings are equal, -1 if first is smaller,
457
 *         1 if second smaller.
4011 svoboda 458
 *
459
 */
4212 decky 460
int str_cmp(const char *s1, const char *s2)
4011 svoboda 461
{
4224 svoboda 462
    wchar_t c1 = 0;
463
    wchar_t c2 = 0;
4212 decky 464
 
465
    size_t off1 = 0;
466
    size_t off2 = 0;
4224 svoboda 467
 
468
    while (true) {
469
        c1 = str_decode(s1, &off1, STR_NO_LIMIT);
470
        c2 = str_decode(s2, &off2, STR_NO_LIMIT);
471
 
4212 decky 472
        if (c1 < c2)
4011 svoboda 473
            return -1;
4212 decky 474
 
475
        if (c1 > c2)
4011 svoboda 476
            return 1;
4224 svoboda 477
 
478
        if (c1 == 0 || c2 == 0)
479
            break;     
4011 svoboda 480
    }
4224 svoboda 481
 
482
    return 0;
4011 svoboda 483
}
484
 
4212 decky 485
/** Compare two NULL terminated strings with length limit.
4011 svoboda 486
 *
4212 decky 487
 * Do a char-by-char comparison of two NULL-terminated strings.
4011 svoboda 488
 * The strings are considered equal iff they consist of the same
4212 decky 489
 * characters on the minimum of their lengths and the length limit.
4011 svoboda 490
 *
4212 decky 491
 * @param s1      First string to compare.
492
 * @param s2      Second string to compare.
493
 * @param max_len Maximum number of characters to consider.
4011 svoboda 494
 *
4212 decky 495
 * @return 0 if the strings are equal, -1 if first is smaller,
496
 *         1 if second smaller.
497
 *
4011 svoboda 498
 */
4212 decky 499
int str_lcmp(const char *s1, const char *s2, count_t max_len)
4011 svoboda 500
{
4212 decky 501
    wchar_t c1 = 0;
502
    wchar_t c2 = 0;
4011 svoboda 503
 
4212 decky 504
    size_t off1 = 0;
505
    size_t off2 = 0;
506
 
507
    count_t len = 0;
4224 svoboda 508
 
509
    while (true) {
510
        if (len >= max_len)
4212 decky 511
            break;
4224 svoboda 512
 
513
        c1 = str_decode(s1, &off1, STR_NO_LIMIT);
514
        c2 = str_decode(s2, &off2, STR_NO_LIMIT);
515
 
4212 decky 516
        if (c1 < c2)
4011 svoboda 517
            return -1;
4224 svoboda 518
 
4212 decky 519
        if (c1 > c2)
4011 svoboda 520
            return 1;
4224 svoboda 521
 
522
        if (c1 == 0 || c2 == 0)
523
            break;
524
 
525
        ++len; 
4011 svoboda 526
    }
4224 svoboda 527
 
528
    return 0;
529
 
4011 svoboda 530
}
531
 
4267 svoboda 532
/** Copy string.
4212 decky 533
 *
4267 svoboda 534
 * Copy source string @a src to destination buffer @a dest.
535
 * No more than @a size bytes are written. If the size of the output buffer
536
 * is at least one byte, the output string will always be well-formed, i.e.
537
 * null-terminated and containing only complete characters.
4212 decky 538
 *
539
 * @param dst   Destination buffer.
4281 svoboda 540
 * @param count Size of the destination buffer (must be > 0).
4267 svoboda 541
 * @param src   Source string.
4212 decky 542
 */
4267 svoboda 543
void str_cpy(char *dest, size_t size, const char *src)
4212 decky 544
{
4267 svoboda 545
    wchar_t ch;
546
    size_t src_off;
547
    size_t dest_off;
548
 
4281 svoboda 549
    /* There must be space for a null terminator in the buffer. */
550
    ASSERT(size > 0);
4212 decky 551
 
4267 svoboda 552
    src_off = 0;
553
    dest_off = 0;
554
 
555
    while ((ch = str_decode(src, &src_off, STR_NO_LIMIT)) != 0) {
556
        if (chr_encode(ch, dest, &dest_off, size - 1) != EOK)
557
            break;
558
    }
559
 
560
    dest[dest_off] = '\0';
561
}
562
 
563
/** Copy size-limited substring.
564
 *
4281 svoboda 565
 * Copy prefix of string @a src of max. size @a size to destination buffer
566
 * @a dest. No more than @a size bytes are written. The output string will
567
 * always be well-formed, i.e. null-terminated and containing only complete
568
 * characters.
4267 svoboda 569
 *
570
 * No more than @a n bytes are read from the input string, so it does not
571
 * have to be null-terminated.
572
 *
573
 * @param dst   Destination buffer.
4281 svoboda 574
 * @param count Size of the destination buffer (must be > 0).
4267 svoboda 575
 * @param src   Source string.
4281 svoboda 576
 * @param n Maximum number of bytes to read from @a src.
4267 svoboda 577
 */
578
void str_ncpy(char *dest, size_t size, const char *src, size_t n)
579
{
4212 decky 580
    wchar_t ch;
4267 svoboda 581
    size_t src_off;
582
    size_t dest_off;
583
 
4281 svoboda 584
    /* There must be space for a null terminator in the buffer. */
585
    ASSERT(size > 0);
4212 decky 586
 
4267 svoboda 587
    src_off = 0;
588
    dest_off = 0;
589
 
590
    while ((ch = str_decode(src, &src_off, n)) != 0) {
591
        if (chr_encode(ch, dest, &dest_off, size - 1) != EOK)
4212 decky 592
            break;
593
    }
4267 svoboda 594
 
595
    dest[dest_off] = '\0';
4212 decky 596
}
4011 svoboda 597
 
4212 decky 598
/** Copy NULL-terminated wide string to string
4011 svoboda 599
 *
4212 decky 600
 * Copy source wide string @a src to destination buffer @a dst.
601
 * No more than @a size bytes are written. NULL-terminator is always
602
 * written after the last succesfully copied character (i.e. if the
603
 * destination buffer is has at least 1 byte, it will be always
604
 * NULL-terminated).
4011 svoboda 605
 *
4212 decky 606
 * @param src   Source wide string.
607
 * @param dst   Destination buffer.
608
 * @param count Size of the destination buffer.
609
 *
4011 svoboda 610
 */
4212 decky 611
void wstr_nstr(char *dst, const wchar_t *src, size_t size)
4011 svoboda 612
{
4212 decky 613
    /* No space for the NULL-terminator in the buffer */
614
    if (size == 0)
615
        return;
4014 decky 616
 
4212 decky 617
    wchar_t ch;
618
    count_t src_idx = 0;
619
    size_t dst_off = 0;
620
 
621
    while ((ch = src[src_idx++]) != 0) {
622
        if (chr_encode(ch, dst, &dst_off, size) != EOK)
623
            break;
4011 svoboda 624
    }
4014 decky 625
 
4212 decky 626
    if (dst_off >= size)
627
        dst[size - 1] = 0;
628
    else
629
        dst[dst_off] = 0;
4011 svoboda 630
}
631
 
4012 svoboda 632
/** Find first occurence of character in string.
633
 *
4212 decky 634
 * @param str String to search.
635
 * @param ch  Character to look for.
4012 svoboda 636
 *
4212 decky 637
 * @return Pointer to character in @a str or NULL if not found.
638
 *
4012 svoboda 639
 */
4212 decky 640
const char *str_chr(const char *str, wchar_t ch)
4012 svoboda 641
{
4212 decky 642
    wchar_t acc;
643
    size_t off = 0;
644
 
4224 svoboda 645
    while ((acc = str_decode(str, &off, STR_NO_LIMIT)) != 0) {
4212 decky 646
        if (acc == ch)
647
            return (str + off);
4012 svoboda 648
    }
4014 decky 649
 
4012 svoboda 650
    return NULL;
651
}
652
 
4212 decky 653
/** Insert a wide character into a wide string.
654
 *
655
 * Insert a wide character into a wide string at position
656
 * @a pos. The characters after the position are shifted.
657
 *
658
 * @param str     String to insert to.
659
 * @param ch      Character to insert to.
660
 * @param pos     Character index where to insert.
661
 @ @param max_pos Characters in the buffer.
662
 *
663
 * @return True if the insertion was sucessful, false if the position
664
 *         is out of bounds.
665
 *
666
 */
667
bool wstr_linsert(wchar_t *str, wchar_t ch, count_t pos, count_t max_pos)
668
{
669
    count_t len = wstr_length(str);
670
 
671
    if ((pos > len) || (pos + 1 > max_pos))
672
        return false;
673
 
674
    count_t i;
675
    for (i = len; i + 1 > pos; i--)
676
        str[i + 1] = str[i];
677
 
678
    str[pos] = ch;
679
 
680
    return true;
681
}
682
 
683
/** Remove a wide character from a wide string.
684
 *
685
 * Remove a wide character from a wide string at position
686
 * @a pos. The characters after the position are shifted.
687
 *
688
 * @param str String to remove from.
689
 * @param pos Character index to remove.
690
 *
691
 * @return True if the removal was sucessful, false if the position
692
 *         is out of bounds.
693
 *
694
 */
695
bool wstr_remove(wchar_t *str, count_t pos)
696
{
697
    count_t len = wstr_length(str);
698
 
699
    if (pos >= len)
700
        return false;
701
 
702
    count_t i;
703
    for (i = pos + 1; i <= len; i++)
704
        str[i - 1] = str[i];
705
 
706
    return true;
707
}
708
 
4011 svoboda 709
/** @}
710
 */