src/basic/utf8.c

   1 /* SPDX-License-Identifier: LGPL-2.1+ */
   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 #include <string.h>
  30
  31 #include "alloc-util.h"
  32 #include "gunicode.h"
  33 #include "hexdecoct.h"
  34 #include "macro.h"
  35 #include "string-util.h"
  36 #include "utf8.h"
  37
  38 bool unichar_is_valid(char32_t ch) {
  39
  40         if (ch >= 0x110000) /* End of unicode space */
  41                 return false;
  42         if ((ch & 0xFFFFF800) == 0xD800) /* Reserved area for UTF-16 */
  43                 return false;
  44         if ((ch >= 0xFDD0) && (ch <= 0xFDEF)) /* Reserved */
  45                 return false;
  46         if ((ch & 0xFFFE) == 0xFFFE) /* BOM (Byte Order Mark) */
  47                 return false;
  48
  49         return true;
  50 }
  51
  52 static bool unichar_is_control(char32_t ch) {
  53
  54         /*
  55           0 to ' '-1 is the C0 range.
  56           DEL=0x7F, and DEL+1 to 0x9F is C1 range.
  57           '\t' is in C0 range, but more or less harmless and commonly used.
  58         */
  59
  60         return (ch < ' ' && !IN_SET(ch, '\t', '\n')) ||
  61                 (0x7F <= ch && ch <= 0x9F);
  62 }
  63
  64 /* count of characters used to encode one unicode char */
  65 static size_t utf8_encoded_expected_len(uint8_t c) {
  66         if (c < 0x80)
  67                 return 1;
  68         if ((c & 0xe0) == 0xc0)
  69                 return 2;
  70         if ((c & 0xf0) == 0xe0)
  71                 return 3;
  72         if ((c & 0xf8) == 0xf0)
  73                 return 4;
  74         if ((c & 0xfc) == 0xf8)
  75                 return 5;
  76         if ((c & 0xfe) == 0xfc)
  77                 return 6;
  78
  79         return 0;
  80 }
  81
  82 /* decode one unicode char */
  83 int utf8_encoded_to_unichar(const char *str, char32_t *ret_unichar) {
  84         char32_t unichar;
  85         size_t len, i;
  86
  87         assert(str);
  88
  89         len = utf8_encoded_expected_len(str[0]);
  90
  91         switch (len) {
  92         case 1:
  93                 *ret_unichar = (char32_t)str[0];
  94                 return 0;
  95         case 2:
  96                 unichar = str[0] & 0x1f;
  97                 break;
  98         case 3:
  99                 unichar = (char32_t)str[0] & 0x0f;
 100                 break;
 101         case 4:
 102                 unichar = (char32_t)str[0] & 0x07;
 103                 break;
 104         case 5:
 105                 unichar = (char32_t)str[0] & 0x03;
 106                 break;
 107         case 6:
 108                 unichar = (char32_t)str[0] & 0x01;
 109                 break;
 110         default:
 111                 return -EINVAL;
 112         }
 113
 114         for (i = 1; i < len; i++) {
 115                 if (((char32_t)str[i] & 0xc0) != 0x80)
 116                         return -EINVAL;
 117
 118                 unichar <<= 6;
 119                 unichar |= (char32_t)str[i] & 0x3f;
 120         }
 121
 122         *ret_unichar = unichar;
 123
 124         return 0;
 125 }
 126
 127 bool utf8_is_printable_newline(const char* str, size_t length, bool newline) {
 128         const char *p;
 129
 130         assert(str);
 131
 132         for (p = str; length > 0;) {
 133                 int encoded_len, r;
 134                 char32_t val;
 135
 136                 encoded_len = utf8_encoded_valid_unichar(p, length);
 137                 if (encoded_len < 0)
 138                         return false;
 139                 assert(encoded_len > 0 && (size_t) encoded_len <= length);
 140
 141                 r = utf8_encoded_to_unichar(p, &val);
 142                 if (r < 0 ||
 143                     unichar_is_control(val) ||
 144                     (!newline && val == '\n'))
 145                         return false;
 146
 147                 length -= encoded_len;
 148                 p += encoded_len;
 149         }
 150
 151         return true;
 152 }
 153
 154 char *utf8_is_valid(const char *str) {
 155         const char *p;
 156
 157         assert(str);
 158
 159         p = str;
 160         while (*p) {
 161                 int len;
 162
 163                 len = utf8_encoded_valid_unichar(p, (size_t) -1);
 164                 if (len < 0)
 165                         return NULL;
 166
 167                 p += len;
 168         }
 169
 170         return (char*) str;
 171 }
 172
 173 char *utf8_escape_invalid(const char *str) {
 174         char *p, *s;
 175
 176         assert(str);
 177
 178         p = s = malloc(strlen(str) * 4 + 1);
 179         if (!p)
 180                 return NULL;
 181
 182         while (*str) {
 183                 int len;
 184
 185                 len = utf8_encoded_valid_unichar(str, (size_t) -1);
 186                 if (len > 0) {
 187                         s = mempcpy(s, str, len);
 188                         str += len;
 189                 } else {
 190                         s = stpcpy(s, UTF8_REPLACEMENT_CHARACTER);
 191                         str += 1;
 192                 }
 193         }
 194
 195         *s = '\0';
 196         (void) str_realloc(&p);
 197         return p;
 198 }
 199
 200 static int utf8_char_console_width(const char *str) {
 201         char32_t c;
 202         int r;
 203
 204         r = utf8_encoded_to_unichar(str, &c);
 205         if (r < 0)
 206                 return r;
 207
 208         /* TODO: we should detect combining characters */
 209
 210         return unichar_iswide(c) ? 2 : 1;
 211 }
 212
 213 char *utf8_escape_non_printable_full(const char *str, size_t console_width) {
 214         char *p, *s, *prev_s;
 215         size_t n = 0; /* estimated print width */
 216
 217         assert(str);
 218
 219         if (console_width == 0)
 220                 return strdup("");
 221
 222         p = s = prev_s = malloc(strlen(str) * 4 + 1);
 223         if (!p)
 224                 return NULL;
 225
 226         for (;;) {
 227                 int len;
 228                 char *saved_s = s;
 229
 230                 if (!*str) /* done! */
 231                         goto finish;
 232
 233                 len = utf8_encoded_valid_unichar(str, (size_t) -1);
 234                 if (len > 0) {
 235                         if (utf8_is_printable(str, len)) {
 236                                 int w;
 237
 238                                 w = utf8_char_console_width(str);
 239                                 assert(w >= 0);
 240                                 if (n + w > console_width)
 241                                         goto truncation;
 242
 243                                 s = mempcpy(s, str, len);
 244                                 str += len;
 245                                 n += w;
 246
 247                         } else {
 248                                 for (; len > 0; len--) {
 249                                         if (n + 4 > console_width)
 250                                                 goto truncation;
 251
 252                                         *(s++) = '\\';
 253                                         *(s++) = 'x';
 254                                         *(s++) = hexchar((int) *str >> 4);
 255                                         *(s++) = hexchar((int) *str);
 256
 257                                         str += 1;
 258                                         n += 4;
 259                                 }
 260                         }
 261                 } else {
 262                         if (n + 1 > console_width)
 263                                 goto truncation;
 264
 265                         s = mempcpy(s, UTF8_REPLACEMENT_CHARACTER, strlen(UTF8_REPLACEMENT_CHARACTER));
 266                         str += 1;
 267                         n += 1;
 268                 }
 269
 270                 prev_s = saved_s;
 271         }
 272
 273  truncation:
 274         /* Try to go back one if we don't have enough space for the ellipsis */
 275         if (n + 1 >= console_width)
 276                 s = prev_s;
 277
 278         s = mempcpy(s, "…", strlen("…"));
 279
 280  finish:
 281         *s = '\0';
 282         (void) str_realloc(&p);
 283         return p;
 284 }
 285
 286 char *ascii_is_valid(const char *str) {
 287         const char *p;
 288
 289         /* Check whether the string consists of valid ASCII bytes,
 290          * i.e values between 0 and 127, inclusive. */
 291
 292         assert(str);
 293
 294         for (p = str; *p; p++)
 295                 if ((unsigned char) *p >= 128)
 296                         return NULL;
 297
 298         return (char*) str;
 299 }
 300
 301 char *ascii_is_valid_n(const char *str, size_t len) {
 302         size_t i;
 303
 304         /* Very similar to ascii_is_valid(), but checks exactly len
 305          * bytes and rejects any NULs in that range. */
 306
 307         assert(str);
 308
 309         for (i = 0; i < len; i++)
 310                 if ((unsigned char) str[i] >= 128 || str[i] == 0)
 311                         return NULL;
 312
 313         return (char*) str;
 314 }
 315
 316 /**
 317  * utf8_encode_unichar() - Encode single UCS-4 character as UTF-8
 318  * @out_utf8: output buffer of at least 4 bytes or NULL
 319  * @g: UCS-4 character to encode
 320  *
 321  * This encodes a single UCS-4 character as UTF-8 and writes it into @out_utf8.
 322  * The length of the character is returned. It is not zero-terminated! If the
 323  * output buffer is NULL, only the length is returned.
 324  *
 325  * Returns: The length in bytes that the UTF-8 representation does or would
 326  *          occupy.
 327  */
 328 size_t utf8_encode_unichar(char *out_utf8, char32_t g) {
 329
 330         if (g < (1 << 7)) {
 331                 if (out_utf8)
 332                         out_utf8[0] = g & 0x7f;
 333                 return 1;
 334         } else if (g < (1 << 11)) {
 335                 if (out_utf8) {
 336                         out_utf8[0] = 0xc0 | ((g >> 6) & 0x1f);
 337                         out_utf8[1] = 0x80 | (g & 0x3f);
 338                 }
 339                 return 2;
 340         } else if (g < (1 << 16)) {
 341                 if (out_utf8) {
 342                         out_utf8[0] = 0xe0 | ((g >> 12) & 0x0f);
 343                         out_utf8[1] = 0x80 | ((g >> 6) & 0x3f);
 344                         out_utf8[2] = 0x80 | (g & 0x3f);
 345                 }
 346                 return 3;
 347         } else if (g < (1 << 21)) {
 348                 if (out_utf8) {
 349                         out_utf8[0] = 0xf0 | ((g >> 18) & 0x07);
 350                         out_utf8[1] = 0x80 | ((g >> 12) & 0x3f);
 351                         out_utf8[2] = 0x80 | ((g >> 6) & 0x3f);
 352                         out_utf8[3] = 0x80 | (g & 0x3f);
 353                 }
 354                 return 4;
 355         }
 356
 357         return 0;
 358 }
 359
 360 char *utf16_to_utf8(const char16_t *s, size_t length /* bytes! */) {
 361         const uint8_t *f;
 362         char *r, *t;
 363
 364         assert(s);
 365
 366         /* Input length is in bytes, i.e. the shortest possible character takes 2 bytes. Each unicode character may
 367          * take up to 4 bytes in UTF-8. Let's also account for a trailing NUL byte. */
 368         if (length * 2 < length)
 369                 return NULL; /* overflow */
 370
 371         r = new(char, length * 2 + 1);
 372         if (!r)
 373                 return NULL;
 374
 375         f = (const uint8_t*) s;
 376         t = r;
 377
 378         while (f + 1 < (const uint8_t*) s + length) {
 379                 char16_t w1, w2;
 380
 381                 /* see RFC 2781 section 2.2 */
 382
 383                 w1 = f[1] << 8 | f[0];
 384                 f += 2;
 385
 386                 if (!utf16_is_surrogate(w1)) {
 387                         t += utf8_encode_unichar(t, w1);
 388                         continue;
 389                 }
 390
 391                 if (utf16_is_trailing_surrogate(w1))
 392                         continue; /* spurious trailing surrogate, ignore */
 393
 394                 if (f + 1 >= (const uint8_t*) s + length)
 395                         break;
 396
 397                 w2 = f[1] << 8 | f[0];
 398                 f += 2;
 399
 400                 if (!utf16_is_trailing_surrogate(w2)) {
 401                         f -= 2;
 402                         continue; /* surrogate missing its trailing surrogate, ignore */
 403                 }
 404
 405                 t += utf8_encode_unichar(t, utf16_surrogate_pair_to_unichar(w1, w2));
 406         }
 407
 408         *t = 0;
 409         return r;
 410 }
 411
 412 size_t utf16_encode_unichar(char16_t *out, char32_t c) {
 413
 414         /* Note that this encodes as little-endian. */
 415
 416         switch (c) {
 417
 418         case 0 ... 0xd7ffU:
 419         case 0xe000U ... 0xffffU:
 420                 out[0] = htole16(c);
 421                 return 1;
 422
 423         case 0x10000U ... 0x10ffffU:
 424                 c -= 0x10000U;
 425                 out[0] = htole16((c >> 10) + 0xd800U);
 426                 out[1] = htole16((c & 0x3ffU) + 0xdc00U);
 427                 return 2;
 428
 429         default: /* A surrogate (invalid) */
 430                 return 0;
 431         }
 432 }
 433
 434 char16_t *utf8_to_utf16(const char *s, size_t length) {
 435         char16_t *n, *p;
 436         size_t i;
 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 (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, i;
 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 (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 }