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