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 0;
  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
 123         return 0;
 124 }
 125
 126 bool utf8_is_printable_newline(const char* str, size_t length, bool allow_newline) {
 127         const char *p;
 128
 129         assert(str);
 130
 131         for (p = str; length > 0;) {
 132                 int encoded_len, r;
 133                 char32_t val;
 134
 135                 encoded_len = utf8_encoded_valid_unichar(p, length);
 136                 if (encoded_len < 0)
 137                         return false;
 138                 assert(encoded_len > 0 && (size_t) encoded_len <= length);
 139
 140                 r = utf8_encoded_to_unichar(p, &val);
 141                 if (r < 0 ||
 142                     unichar_is_control(val) ||
 143                     (!allow_newline && val == '\n'))
 144                         return false;
 145
 146                 length -= encoded_len;
 147                 p += encoded_len;
 148         }
 149
 150         return true;
 151 }
 152
 153 char *utf8_is_valid_n(const char *str, size_t len_bytes) {
 154         /* Check if the string is composed of valid utf8 characters. If length len_bytes is given, stop after
 155          * len_bytes. Otherwise, stop at NUL. */
 156
 157         assert(str);
 158
 159         for (const char *p = str; len_bytes != (size_t) -1 ? (size_t) (p - str) < len_bytes : *p != '\0'; ) {
 160                 int len;
 161
 162                 if (_unlikely_(*p == '\0') && len_bytes != (size_t) -1)
 163                         return NULL; /* embedded NUL */
 164
 165                 len = utf8_encoded_valid_unichar(p,
 166                                                  len_bytes != (size_t) -1 ? len_bytes - (p - str) : (size_t) -1);
 167                 if (_unlikely_(len < 0))
 168                         return NULL; /* invalid character */
 169
 170                 p += len;
 171         }
 172
 173         return (char*) str;
 174 }
 175
 176 char *utf8_escape_invalid(const char *str) {
 177         char *p, *s;
 178
 179         assert(str);
 180
 181         p = s = malloc(strlen(str) * 4 + 1);
 182         if (!p)
 183                 return NULL;
 184
 185         while (*str) {
 186                 int len;
 187
 188                 len = utf8_encoded_valid_unichar(str, (size_t) -1);
 189                 if (len > 0) {
 190                         s = mempcpy(s, str, len);
 191                         str += len;
 192                 } else {
 193                         s = stpcpy(s, UTF8_REPLACEMENT_CHARACTER);
 194                         str += 1;
 195                 }
 196         }
 197
 198         *s = '\0';
 199         (void) str_realloc(&p);
 200         return p;
 201 }
 202
 203 static int utf8_char_console_width(const char *str) {
 204         char32_t c;
 205         int r;
 206
 207         r = utf8_encoded_to_unichar(str, &c);
 208         if (r < 0)
 209                 return r;
 210
 211         /* TODO: we should detect combining characters */
 212
 213         return unichar_iswide(c) ? 2 : 1;
 214 }
 215
 216 char *utf8_escape_non_printable_full(const char *str, size_t console_width) {
 217         char *p, *s, *prev_s;
 218         size_t n = 0; /* estimated print width */
 219
 220         assert(str);
 221
 222         if (console_width == 0)
 223                 return strdup("");
 224
 225         p = s = prev_s = malloc(strlen(str) * 4 + 1);
 226         if (!p)
 227                 return NULL;
 228
 229         for (;;) {
 230                 int len;
 231                 char *saved_s = s;
 232
 233                 if (!*str) /* done! */
 234                         goto finish;
 235
 236                 len = utf8_encoded_valid_unichar(str, (size_t) -1);
 237                 if (len > 0) {
 238                         if (utf8_is_printable(str, len)) {
 239                                 int w;
 240
 241                                 w = utf8_char_console_width(str);
 242                                 assert(w >= 0);
 243                                 if (n + w > console_width)
 244                                         goto truncation;
 245
 246                                 s = mempcpy(s, str, len);
 247                                 str += len;
 248                                 n += w;
 249
 250                         } else {
 251                                 for (; len > 0; len--) {
 252                                         if (n + 4 > console_width)
 253                                                 goto truncation;
 254
 255                                         *(s++) = '\\';
 256                                         *(s++) = 'x';
 257                                         *(s++) = hexchar((int) *str >> 4);
 258                                         *(s++) = hexchar((int) *str);
 259
 260                                         str += 1;
 261                                         n += 4;
 262                                 }
 263                         }
 264                 } else {
 265                         if (n + 1 > console_width)
 266                                 goto truncation;
 267
 268                         s = mempcpy(s, UTF8_REPLACEMENT_CHARACTER, strlen(UTF8_REPLACEMENT_CHARACTER));
 269                         str += 1;
 270                         n += 1;
 271                 }
 272
 273                 prev_s = saved_s;
 274         }
 275
 276  truncation:
 277         /* Try to go back one if we don't have enough space for the ellipsis */
 278         if (n + 1 >= console_width)
 279                 s = prev_s;
 280
 281         s = mempcpy(s, "…", strlen("…"));
 282
 283  finish:
 284         *s = '\0';
 285         (void) str_realloc(&p);
 286         return p;
 287 }
 288
 289 char *ascii_is_valid(const char *str) {
 290         const char *p;
 291
 292         /* Check whether the string consists of valid ASCII bytes,
 293          * i.e values between 0 and 127, inclusive. */
 294
 295         assert(str);
 296
 297         for (p = str; *p; p++)
 298                 if ((unsigned char) *p >= 128)
 299                         return NULL;
 300
 301         return (char*) str;
 302 }
 303
 304 char *ascii_is_valid_n(const char *str, size_t len) {
 305         /* Very similar to ascii_is_valid(), but checks exactly len
 306          * bytes and rejects any NULs in that range. */
 307
 308         assert(str);
 309
 310         for (size_t i = 0; i < len; i++)
 311                 if ((unsigned char) str[i] >= 128 || str[i] == 0)
 312                         return NULL;
 313
 314         return (char*) str;
 315 }
 316
 317 /**
 318  * utf8_encode_unichar() - Encode single UCS-4 character as UTF-8
 319  * @out_utf8: output buffer of at least 4 bytes or NULL
 320  * @g: UCS-4 character to encode
 321  *
 322  * This encodes a single UCS-4 character as UTF-8 and writes it into @out_utf8.
 323  * The length of the character is returned. It is not zero-terminated! If the
 324  * output buffer is NULL, only the length is returned.
 325  *
 326  * Returns: The length in bytes that the UTF-8 representation does or would
 327  *          occupy.
 328  */
 329 size_t utf8_encode_unichar(char *out_utf8, char32_t g) {
 330
 331         if (g < (1 << 7)) {
 332                 if (out_utf8)
 333                         out_utf8[0] = g & 0x7f;
 334                 return 1;
 335         } else if (g < (1 << 11)) {
 336                 if (out_utf8) {
 337                         out_utf8[0] = 0xc0 | ((g >> 6) & 0x1f);
 338                         out_utf8[1] = 0x80 | (g & 0x3f);
 339                 }
 340                 return 2;
 341         } else if (g < (1 << 16)) {
 342                 if (out_utf8) {
 343                         out_utf8[0] = 0xe0 | ((g >> 12) & 0x0f);
 344                         out_utf8[1] = 0x80 | ((g >> 6) & 0x3f);
 345                         out_utf8[2] = 0x80 | (g & 0x3f);
 346                 }
 347                 return 3;
 348         } else if (g < (1 << 21)) {
 349                 if (out_utf8) {
 350                         out_utf8[0] = 0xf0 | ((g >> 18) & 0x07);
 351                         out_utf8[1] = 0x80 | ((g >> 12) & 0x3f);
 352                         out_utf8[2] = 0x80 | ((g >> 6) & 0x3f);
 353                         out_utf8[3] = 0x80 | (g & 0x3f);
 354                 }
 355                 return 4;
 356         }
 357
 358         return 0;
 359 }
 360
 361 char *utf16_to_utf8(const char16_t *s, size_t length /* bytes! */) {
 362         const uint8_t *f;
 363         char *r, *t;
 364
 365         assert(s);
 366
 367         /* Input length is in bytes, i.e. the shortest possible character takes 2 bytes. Each unicode character may
 368          * take up to 4 bytes in UTF-8. Let's also account for a trailing NUL byte. */
 369         if (length * 2 < length)
 370                 return NULL; /* overflow */
 371
 372         r = new(char, length * 2 + 1);
 373         if (!r)
 374                 return NULL;
 375
 376         f = (const uint8_t*) s;
 377         t = r;
 378
 379         while (f + 1 < (const uint8_t*) s + length) {
 380                 char16_t w1, w2;
 381
 382                 /* see RFC 2781 section 2.2 */
 383
 384                 w1 = f[1] << 8 | f[0];
 385                 f += 2;
 386
 387                 if (!utf16_is_surrogate(w1)) {
 388                         t += utf8_encode_unichar(t, w1);
 389                         continue;
 390                 }
 391
 392                 if (utf16_is_trailing_surrogate(w1))
 393                         continue; /* spurious trailing surrogate, ignore */
 394
 395                 if (f + 1 >= (const uint8_t*) s + length)
 396                         break;
 397
 398                 w2 = f[1] << 8 | f[0];
 399                 f += 2;
 400
 401                 if (!utf16_is_trailing_surrogate(w2)) {
 402                         f -= 2;
 403                         continue; /* surrogate missing its trailing surrogate, ignore */
 404                 }
 405
 406                 t += utf8_encode_unichar(t, utf16_surrogate_pair_to_unichar(w1, w2));
 407         }
 408
 409         *t = 0;
 410         return r;
 411 }
 412
 413 size_t utf16_encode_unichar(char16_t *out, char32_t c) {
 414
 415         /* Note that this encodes as little-endian. */
 416
 417         switch (c) {
 418
 419         case 0 ... 0xd7ffU:
 420         case 0xe000U ... 0xffffU:
 421                 out[0] = htole16(c);
 422                 return 1;
 423
 424         case 0x10000U ... 0x10ffffU:
 425                 c -= 0x10000U;
 426                 out[0] = htole16((c >> 10) + 0xd800U);
 427                 out[1] = htole16((c & 0x3ffU) + 0xdc00U);
 428                 return 2;
 429
 430         default: /* A surrogate (invalid) */
 431                 return 0;
 432         }
 433 }
 434
 435 char16_t *utf8_to_utf16(const char *s, size_t length) {
 436         char16_t *n, *p;
 437         int r;
 438
 439         assert(s);
 440
 441         n = new(char16_t, length + 1);
 442         if (!n)
 443                 return NULL;
 444
 445         p = n;
 446
 447         for (size_t i = 0; i < length;) {
 448                 char32_t unichar;
 449                 size_t e;
 450
 451                 e = utf8_encoded_expected_len(s[i]);
 452                 if (e <= 1) /* Invalid and single byte characters are copied as they are */
 453                         goto copy;
 454
 455                 if (i + e > length) /* sequence longer than input buffer, then copy as-is */
 456                         goto copy;
 457
 458                 r = utf8_encoded_to_unichar(s + i, &unichar);
 459                 if (r < 0) /* sequence invalid, then copy as-is */
 460                         goto copy;
 461
 462                 p += utf16_encode_unichar(p, unichar);
 463                 i += e;
 464                 continue;
 465
 466         copy:
 467                 *(p++) = htole16(s[i++]);
 468         }
 469
 470         *p = 0;
 471         return n;
 472 }
 473
 474 size_t char16_strlen(const char16_t *s) {
 475         size_t n = 0;
 476
 477         assert(s);
 478
 479         while (*s != 0)
 480                 n++, s++;
 481
 482         return n;
 483 }
 484
 485 /* expected size used to encode one unicode char */
 486 static int utf8_unichar_to_encoded_len(char32_t unichar) {
 487
 488         if (unichar < 0x80)
 489                 return 1;
 490         if (unichar < 0x800)
 491                 return 2;
 492         if (unichar < 0x10000)
 493                 return 3;
 494         if (unichar < 0x200000)
 495                 return 4;
 496         if (unichar < 0x4000000)
 497                 return 5;
 498
 499         return 6;
 500 }
 501
 502 /* validate one encoded unicode char and return its length */
 503 int utf8_encoded_valid_unichar(const char *str, size_t length /* bytes */) {
 504         char32_t unichar;
 505         size_t len;
 506         int r;
 507
 508         assert(str);
 509         assert(length > 0);
 510
 511         /* We read until NUL, at most length bytes. (size_t) -1 may be used to disable the length check. */
 512
 513         len = utf8_encoded_expected_len(str[0]);
 514         if (len == 0)
 515                 return -EINVAL;
 516
 517         /* Do we have a truncated multi-byte character? */
 518         if (len > length)
 519                 return -EINVAL;
 520
 521         /* ascii is valid */
 522         if (len == 1)
 523                 return 1;
 524
 525         /* check if expected encoded chars are available */
 526         for (size_t i = 0; i < len; i++)
 527                 if ((str[i] & 0x80) != 0x80)
 528                         return -EINVAL;
 529
 530         r = utf8_encoded_to_unichar(str, &unichar);
 531         if (r < 0)
 532                 return r;
 533
 534         /* check if encoded length matches encoded value */
 535         if (utf8_unichar_to_encoded_len(unichar) != (int) len)
 536                 return -EINVAL;
 537
 538         /* check if value has valid range */
 539         if (!unichar_is_valid(unichar))
 540                 return -EINVAL;
 541
 542         return (int) len;
 543 }
 544
 545 size_t utf8_n_codepoints(const char *str) {
 546         size_t n = 0;
 547
 548         /* Returns the number of UTF-8 codepoints in this string, or (size_t) -1 if the string is not valid UTF-8. */
 549
 550         while (*str != 0) {
 551                 int k;
 552
 553                 k = utf8_encoded_valid_unichar(str, (size_t) -1);
 554                 if (k < 0)
 555                         return (size_t) -1;
 556
 557                 str += k;
 558                 n++;
 559         }
 560
 561         return n;
 562 }
 563
 564 size_t utf8_console_width(const char *str) {
 565         size_t n = 0;
 566
 567         /* Returns the approximate width a string will take on screen when printed on a character cell
 568          * terminal/console. */
 569
 570         while (*str) {
 571                 int w;
 572
 573                 w = utf8_char_console_width(str);
 574                 if (w < 0)
 575                         return (size_t) -1;
 576
 577                 n += w;
 578                 str = utf8_next_char(str);
 579         }
 580
 581         return n;
 582 }