src/basic/utf8.c

   1 /* SPDX-License-Identifier: LGPL-2.1-or-later */
   2
   3 /* Parts of this file are based on the GLIB utf8 validation functions. The
   4  * original license text follows. */
   5
   6 /* gutf8.c - Operations on UTF-8 strings.
   7  *
   8  * Copyright (C) 1999 Tom Tromey
   9  * Copyright (C) 2000 Red Hat, Inc.
  10  *
  11  * This library is free software; you can redistribute it and/or
  12  * modify it under the terms of the GNU Library General Public
  13  * License as published by the Free Software Foundation; either
  14  * version 2 of the License, or (at your option) any later version.
  15  *
  16  * This library is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19  * Library General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU Library General Public
  22  * License along with this library; if not, write to the Free Software
  23  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
  24  */
  25
  26 #include <errno.h>
  27 #include <stdbool.h>
  28 #include <stdlib.h>
  29
  30 #include "alloc-util.h"
  31 #include "gunicode.h"
  32 #include "hexdecoct.h"
  33 #include "macro.h"
  34 #include "string-util.h"
  35 #include "utf8.h"
  36
  37 bool unichar_is_valid(char32_t ch) {
  38
  39         if (ch >= 0x110000) /* End of unicode space */
  40                 return false;
  41         if ((ch & 0xFFFFF800) == 0xD800) /* Reserved area for UTF-16 */
  42                 return false;
  43         if ((ch >= 0xFDD0) && (ch <= 0xFDEF)) /* Reserved */
  44                 return false;
  45         if ((ch & 0xFFFE) == 0xFFFE) /* BOM (Byte Order Mark) */
  46                 return false;
  47
  48         return true;
  49 }
  50
  51 static bool unichar_is_control(char32_t ch) {
  52
  53         /*
  54           0 to ' '-1 is the C0 range.
  55           DEL=0x7F, and DEL+1 to 0x9F is C1 range.
  56           '\t' is in C0 range, but more or less harmless and commonly used.
  57         */
  58
  59         return (ch < ' ' && !IN_SET(ch, '\t', '\n')) ||
  60                 (0x7F <= ch && ch <= 0x9F);
  61 }
  62
  63 /* count of characters used to encode one unicode char */
  64 static size_t utf8_encoded_expected_len(uint8_t c) {
  65         if (c < 0x80)
  66                 return 1;
  67         if ((c & 0xe0) == 0xc0)
  68                 return 2;
  69         if ((c & 0xf0) == 0xe0)
  70                 return 3;
  71         if ((c & 0xf8) == 0xf0)
  72                 return 4;
  73         if ((c & 0xfc) == 0xf8)
  74                 return 5;
  75         if ((c & 0xfe) == 0xfc)
  76                 return 6;
  77
  78         return 0;
  79 }
  80
  81 /* decode one unicode char */
  82 int utf8_encoded_to_unichar(const char *str, char32_t *ret_unichar) {
  83         char32_t unichar;
  84         size_t len;
  85
  86         assert(str);
  87
  88         len = utf8_encoded_expected_len(str[0]);
  89
  90         switch (len) {
  91         case 1:
  92                 *ret_unichar = (char32_t)str[0];
  93                 return 1;
  94         case 2:
  95                 unichar = str[0] & 0x1f;
  96                 break;
  97         case 3:
  98                 unichar = (char32_t)str[0] & 0x0f;
  99                 break;
 100         case 4:
 101                 unichar = (char32_t)str[0] & 0x07;
 102                 break;
 103         case 5:
 104                 unichar = (char32_t)str[0] & 0x03;
 105                 break;
 106         case 6:
 107                 unichar = (char32_t)str[0] & 0x01;
 108                 break;
 109         default:
 110                 return -EINVAL;
 111         }
 112
 113         for (size_t i = 1; i < len; i++) {
 114                 if (((char32_t)str[i] & 0xc0) != 0x80)
 115                         return -EINVAL;
 116
 117                 unichar <<= 6;
 118                 unichar |= (char32_t)str[i] & 0x3f;
 119         }
 120
 121         *ret_unichar = unichar;
 122         return len;
 123 }
 124
 125 bool utf8_is_printable_newline(const char* str, size_t length, bool allow_newline) {
 126         assert(str);
 127
 128         for (const char *p = str; length > 0;) {
 129                 int encoded_len;
 130                 char32_t val;
 131
 132                 encoded_len = utf8_encoded_valid_unichar(p, length);
 133                 if (encoded_len < 0)
 134                         return false;
 135                 assert(encoded_len > 0 && (size_t) encoded_len <= length);
 136
 137                 if (utf8_encoded_to_unichar(p, &val) < 0 ||
 138                     unichar_is_control(val) ||
 139                     (!allow_newline && val == '\n'))
 140                         return false;
 141
 142                 length -= encoded_len;
 143                 p += encoded_len;
 144         }
 145
 146         return true;
 147 }
 148
 149 char *utf8_is_valid_n(const char *str, size_t len_bytes) {
 150         /* Check if the string is composed of valid utf8 characters. If length len_bytes is given, stop after
 151          * len_bytes. Otherwise, stop at NUL. */
 152
 153         assert(str);
 154
 155         for (const char *p = str; len_bytes != SIZE_MAX ? (size_t) (p - str) < len_bytes : *p != '\0'; ) {
 156                 int len;
 157
 158                 if (_unlikely_(*p == '\0') && len_bytes != SIZE_MAX)
 159                         return NULL; /* embedded NUL */
 160
 161                 len = utf8_encoded_valid_unichar(p,
 162                                                  len_bytes != SIZE_MAX ? len_bytes - (p - str) : SIZE_MAX);
 163                 if (_unlikely_(len < 0))
 164                         return NULL; /* invalid character */
 165
 166                 p += len;
 167         }
 168
 169         return (char*) str;
 170 }
 171
 172 char *utf8_escape_invalid(const char *str) {
 173         char *p, *s;
 174
 175         assert(str);
 176
 177         p = s = malloc(strlen(str) * 4 + 1);
 178         if (!p)
 179                 return NULL;
 180
 181         while (*str) {
 182                 int len;
 183
 184                 len = utf8_encoded_valid_unichar(str, SIZE_MAX);
 185                 if (len > 0) {
 186                         s = mempcpy(s, str, len);
 187                         str += len;
 188                 } else {
 189                         s = stpcpy(s, UTF8_REPLACEMENT_CHARACTER);
 190                         str += 1;
 191                 }
 192         }
 193
 194         *s = '\0';
 195         return str_realloc(p);
 196 }
 197
 198 static int utf8_char_console_width(const char *str) {
 199         char32_t c;
 200         int r;
 201
 202         r = utf8_encoded_to_unichar(str, &c);
 203         if (r < 0)
 204                 return r;
 205
 206         /* TODO: we should detect combining characters */
 207
 208         return unichar_iswide(c) ? 2 : 1;
 209 }
 210
 211 char *utf8_escape_non_printable_full(const char *str, size_t console_width, bool force_ellipsis) {
 212         char *p, *s, *prev_s;
 213         size_t n = 0; /* estimated print width */
 214
 215         assert(str);
 216
 217         if (console_width == 0)
 218                 return strdup("");
 219
 220         p = s = prev_s = malloc(strlen(str) * 4 + 1);
 221         if (!p)
 222                 return NULL;
 223
 224         for (;;) {
 225                 int len;
 226                 char *saved_s = s;
 227
 228                 if (!*str) { /* done! */
 229                         if (force_ellipsis)
 230                                 goto truncation;
 231                         else
 232                                 goto finish;
 233                 }
 234
 235                 len = utf8_encoded_valid_unichar(str, SIZE_MAX);
 236                 if (len > 0) {
 237                         if (utf8_is_printable(str, len)) {
 238                                 int w;
 239
 240                                 w = utf8_char_console_width(str);
 241                                 assert(w >= 0);
 242                                 if (n + w > console_width)
 243                                         goto truncation;
 244
 245                                 s = mempcpy(s, str, len);
 246                                 str += len;
 247                                 n += w;
 248
 249                         } else {
 250                                 for (; len > 0; len--) {
 251                                         if (n + 4 > console_width)
 252                                                 goto truncation;
 253
 254                                         *(s++) = '\\';
 255                                         *(s++) = 'x';
 256                                         *(s++) = hexchar((int) *str >> 4);
 257                                         *(s++) = hexchar((int) *str);
 258
 259                                         str += 1;
 260                                         n += 4;
 261                                 }
 262                         }
 263                 } else {
 264                         if (n + 1 > console_width)
 265                                 goto truncation;
 266
 267                         s = mempcpy(s, UTF8_REPLACEMENT_CHARACTER, strlen(UTF8_REPLACEMENT_CHARACTER));
 268                         str += 1;
 269                         n += 1;
 270                 }
 271
 272                 prev_s = saved_s;
 273         }
 274
 275  truncation:
 276         /* Try to go back one if we don't have enough space for the ellipsis */
 277         if (n + 1 > console_width)
 278                 s = prev_s;
 279
 280         s = mempcpy(s, "…", strlen("…"));
 281
 282  finish:
 283         *s = '\0';
 284         return str_realloc(p);
 285 }
 286
 287 char *ascii_is_valid(const char *str) {
 288         /* Check whether the string consists of valid ASCII bytes,
 289          * i.e values between 0 and 127, inclusive. */
 290
 291         assert(str);
 292
 293         for (const char *p = str; *p; p++)
 294                 if ((unsigned char) *p >= 128)
 295                         return NULL;
 296
 297         return (char*) str;
 298 }
 299
 300 char *ascii_is_valid_n(const char *str, size_t len) {
 301         /* Very similar to ascii_is_valid(), but checks exactly len
 302          * bytes and rejects any NULs in that range. */
 303
 304         assert(str);
 305
 306         for (size_t i = 0; i < len; i++)
 307                 if ((unsigned char) str[i] >= 128 || str[i] == 0)
 308                         return NULL;
 309
 310         return (char*) str;
 311 }
 312
 313 int utf8_to_ascii(const char *str, char replacement_char, char **ret) {
 314         /* Convert to a string that has only ASCII chars, replacing anything that is not ASCII
 315          * by replacement_char. */
 316
 317         _cleanup_free_ char *ans = new(char, strlen(str) + 1);
 318         if (!ans)
 319                 return -ENOMEM;
 320
 321         char *q = ans;
 322
 323         for (const char *p = str; *p; q++) {
 324                 int l;
 325
 326                 l = utf8_encoded_valid_unichar(p, SIZE_MAX);
 327                 if (l < 0)  /* Non-UTF-8, let's not even try to propagate the garbage */
 328                         return l;
 329
 330                 if (l == 1)
 331                         *q = *p;
 332                 else
 333                         /* non-ASCII, we need to replace it */
 334                         *q = replacement_char;
 335
 336                 p += l;
 337         }
 338         *q = '\0';
 339
 340         *ret = TAKE_PTR(ans);
 341         return 0;
 342 }
 343
 344 /**
 345  * utf8_encode_unichar() - Encode single UCS-4 character as UTF-8
 346  * @out_utf8: output buffer of at least 4 bytes or NULL
 347  * @g: UCS-4 character to encode
 348  *
 349  * This encodes a single UCS-4 character as UTF-8 and writes it into @out_utf8.
 350  * The length of the character is returned. It is not zero-terminated! If the
 351  * output buffer is NULL, only the length is returned.
 352  *
 353  * Returns: The length in bytes that the UTF-8 representation does or would
 354  *          occupy.
 355  */
 356 size_t utf8_encode_unichar(char *out_utf8, char32_t g) {
 357
 358         if (g < (1 << 7)) {
 359                 if (out_utf8)
 360                         out_utf8[0] = g & 0x7f;
 361                 return 1;
 362         } else if (g < (1 << 11)) {
 363                 if (out_utf8) {
 364                         out_utf8[0] = 0xc0 | ((g >> 6) & 0x1f);
 365                         out_utf8[1] = 0x80 | (g & 0x3f);
 366                 }
 367                 return 2;
 368         } else if (g < (1 << 16)) {
 369                 if (out_utf8) {
 370                         out_utf8[0] = 0xe0 | ((g >> 12) & 0x0f);
 371                         out_utf8[1] = 0x80 | ((g >> 6) & 0x3f);
 372                         out_utf8[2] = 0x80 | (g & 0x3f);
 373                 }
 374                 return 3;
 375         } else if (g < (1 << 21)) {
 376                 if (out_utf8) {
 377                         out_utf8[0] = 0xf0 | ((g >> 18) & 0x07);
 378                         out_utf8[1] = 0x80 | ((g >> 12) & 0x3f);
 379                         out_utf8[2] = 0x80 | ((g >> 6) & 0x3f);
 380                         out_utf8[3] = 0x80 | (g & 0x3f);
 381                 }
 382                 return 4;
 383         }
 384
 385         return 0;
 386 }
 387
 388 char *utf16_to_utf8(const char16_t *s, size_t length /* bytes! */) {
 389         const uint8_t *f;
 390         char *r, *t;
 391
 392         assert(s);
 393
 394         /* Input length is in bytes, i.e. the shortest possible character takes 2 bytes. Each unicode character may
 395          * take up to 4 bytes in UTF-8. Let's also account for a trailing NUL byte. */
 396         if (length * 2 < length)
 397                 return NULL; /* overflow */
 398
 399         r = new(char, length * 2 + 1);
 400         if (!r)
 401                 return NULL;
 402
 403         f = (const uint8_t*) s;
 404         t = r;
 405
 406         while (f + 1 < (const uint8_t*) s + length) {
 407                 char16_t w1, w2;
 408
 409                 /* see RFC 2781 section 2.2 */
 410
 411                 w1 = f[1] << 8 | f[0];
 412                 f += 2;
 413
 414                 if (!utf16_is_surrogate(w1)) {
 415                         t += utf8_encode_unichar(t, w1);
 416                         continue;
 417                 }
 418
 419                 if (utf16_is_trailing_surrogate(w1))
 420                         continue; /* spurious trailing surrogate, ignore */
 421
 422                 if (f + 1 >= (const uint8_t*) s + length)
 423                         break;
 424
 425                 w2 = f[1] << 8 | f[0];
 426                 f += 2;
 427
 428                 if (!utf16_is_trailing_surrogate(w2)) {
 429                         f -= 2;
 430                         continue; /* surrogate missing its trailing surrogate, ignore */
 431                 }
 432
 433                 t += utf8_encode_unichar(t, utf16_surrogate_pair_to_unichar(w1, w2));
 434         }
 435
 436         *t = 0;
 437         return r;
 438 }
 439
 440 size_t utf16_encode_unichar(char16_t *out, char32_t c) {
 441
 442         /* Note that this encodes as little-endian. */
 443
 444         switch (c) {
 445
 446         case 0 ... 0xd7ffU:
 447         case 0xe000U ... 0xffffU:
 448                 out[0] = htole16(c);
 449                 return 1;
 450
 451         case 0x10000U ... 0x10ffffU:
 452                 c -= 0x10000U;
 453                 out[0] = htole16((c >> 10) + 0xd800U);
 454                 out[1] = htole16((c & 0x3ffU) + 0xdc00U);
 455                 return 2;
 456
 457         default: /* A surrogate (invalid) */
 458                 return 0;
 459         }
 460 }
 461
 462 char16_t *utf8_to_utf16(const char *s, size_t length) {
 463         char16_t *n, *p;
 464         int r;
 465
 466         assert(s);
 467
 468         n = new(char16_t, length + 1);
 469         if (!n)
 470                 return NULL;
 471
 472         p = n;
 473
 474         for (size_t i = 0; i < length;) {
 475                 char32_t unichar;
 476                 size_t e;
 477
 478                 e = utf8_encoded_expected_len(s[i]);
 479                 if (e <= 1) /* Invalid and single byte characters are copied as they are */
 480                         goto copy;
 481
 482                 if (i + e > length) /* sequence longer than input buffer, then copy as-is */
 483                         goto copy;
 484
 485                 r = utf8_encoded_to_unichar(s + i, &unichar);
 486                 if (r < 0) /* sequence invalid, then copy as-is */
 487                         goto copy;
 488
 489                 p += utf16_encode_unichar(p, unichar);
 490                 i += e;
 491                 continue;
 492
 493         copy:
 494                 *(p++) = htole16(s[i++]);
 495         }
 496
 497         *p = 0;
 498         return n;
 499 }
 500
 501 size_t char16_strlen(const char16_t *s) {
 502         size_t n = 0;
 503
 504         assert(s);
 505
 506         while (*s != 0)
 507                 n++, s++;
 508
 509         return n;
 510 }
 511
 512 /* expected size used to encode one unicode char */
 513 static int utf8_unichar_to_encoded_len(char32_t unichar) {
 514
 515         if (unichar < 0x80)
 516                 return 1;
 517         if (unichar < 0x800)
 518                 return 2;
 519         if (unichar < 0x10000)
 520                 return 3;
 521         if (unichar < 0x200000)
 522                 return 4;
 523         if (unichar < 0x4000000)
 524                 return 5;
 525
 526         return 6;
 527 }
 528
 529 /* validate one encoded unicode char and return its length */
 530 int utf8_encoded_valid_unichar(const char *str, size_t length /* bytes */) {
 531         char32_t unichar;
 532         size_t len;
 533         int r;
 534
 535         assert(str);
 536         assert(length > 0);
 537
 538         /* We read until NUL, at most length bytes. SIZE_MAX may be used to disable the length check. */
 539
 540         len = utf8_encoded_expected_len(str[0]);
 541         if (len == 0)
 542                 return -EINVAL;
 543
 544         /* Do we have a truncated multi-byte character? */
 545         if (len > length)
 546                 return -EINVAL;
 547
 548         /* ascii is valid */
 549         if (len == 1)
 550                 return 1;
 551
 552         /* check if expected encoded chars are available */
 553         for (size_t i = 0; i < len; i++)
 554                 if ((str[i] & 0x80) != 0x80)
 555                         return -EINVAL;
 556
 557         r = utf8_encoded_to_unichar(str, &unichar);
 558         if (r < 0)
 559                 return r;
 560
 561         /* check if encoded length matches encoded value */
 562         if (utf8_unichar_to_encoded_len(unichar) != (int) len)
 563                 return -EINVAL;
 564
 565         /* check if value has valid range */
 566         if (!unichar_is_valid(unichar))
 567                 return -EINVAL;
 568
 569         return (int) len;
 570 }
 571
 572 size_t utf8_n_codepoints(const char *str) {
 573         size_t n = 0;
 574
 575         /* Returns the number of UTF-8 codepoints in this string, or SIZE_MAX if the string is not valid UTF-8. */
 576
 577         while (*str != 0) {
 578                 int k;
 579
 580                 k = utf8_encoded_valid_unichar(str, SIZE_MAX);
 581                 if (k < 0)
 582                         return SIZE_MAX;
 583
 584                 str += k;
 585                 n++;
 586         }
 587
 588         return n;
 589 }
 590
 591 size_t utf8_console_width(const char *str) {
 592         size_t n = 0;
 593
 594         /* Returns the approximate width a string will take on screen when printed on a character cell
 595          * terminal/console. */
 596
 597         while (*str) {
 598                 int w;
 599
 600                 w = utf8_char_console_width(str);
 601                 if (w < 0)
 602                         return SIZE_MAX;
 603
 604                 n += w;
 605                 str = utf8_next_char(str);
 606         }
 607
 608         return n;
 609 }