1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779
// This file is part of ICU4X. For terms of use, please see the file
// called LICENSE at the top level of the ICU4X source tree
// (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).
// https://github.com/unicode-org/icu4x/blob/main/documents/process/boilerplate.md#library-annotations
#![cfg_attr(not(any(test, feature = "std")), no_std)]
#![cfg_attr(
not(test),
deny(
clippy::indexing_slicing,
clippy::unwrap_used,
clippy::expect_used,
clippy::panic,
clippy::exhaustive_structs,
clippy::exhaustive_enums,
missing_debug_implementations,
)
)]
#![warn(missing_docs)]
//! Normalizing text into Unicode Normalization Forms.
//!
//! This module is published as its own crate ([`icu_normalizer`](https://docs.rs/icu_normalizer/latest/icu_normalizer/))
//! and as part of the [`icu`](https://docs.rs/icu/latest/icu/) crate. See the latter for more details on the ICU4X project.
//!
//! # Implementation notes
//!
//! The normalizer operates on a lazy iterator over Unicode scalar values (Rust `char`) internally
//! and iterating over guaranteed-valid UTF-8, potentially-invalid UTF-8, and potentially-invalid
//! UTF-16 is a step that doesn’t leak into the normalizer internals. Ill-formed byte sequences are
//! treated as U+FFFD.
//!
//! The normalizer data layout is not based on the ICU4C design at all. Instead, the normalization
//! data layout is a clean-slate design optimized for the concept of fusing the NFD decomposition
//! into the collator. That is, the decomposing normalizer is a by-product of the collator-motivated
//! data layout.
//!
//! Notably, the decomposition data structure is optimized for a starter decomposing to itself,
//! which is the most common case, and for a starter decomposing to a starter and a non-starter
//! on the Basic Multilingual Plane. Notably, in this case, the collator makes use of the
//! knowledge that the second character of such a decomposition is a non-starter. Therefore,
//! decomposition into two starters is handled by generic fallback path that looks the
//! decomposition from an array by offset and length instead of baking a BMP starter pair directly
//! into a trie value.
//!
//! The decompositions into non-starters are hard-coded. At present in Unicode, these appear
//! to be special cases falling into three categories:
//!
//! 1. Deprecated combining marks.
//! 2. Particular Tibetan vowel sings.
//! 3. NFKD only: half-width kana voicing marks.
//!
//! Hopefully Unicode never adds more decompositions into non-starters (other than a character
//! decomposing to itself), but if it does, a code update is needed instead of a mere data update.
//!
//! The composing normalizer builds on the decomposing normalizer by performing the canonical
//! composition post-processing per spec. As an optimization, though, the composing normalizer
//! attempts to pass through already-normalized text consisting of starters that never combine
//! backwards and that map to themselves if followed by a character whose decomposition starts
//! with a starter that never combines backwards.
//!
//! As a difference with ICU4C, the composing normalizer has only the simplest possible
//! passthrough (only one inversion list lookup per character in the best case) and the full
//! decompose-then-canonically-compose behavior, whereas ICU4C has other paths between these
//! extremes. The ICU4X collator doesn't make use of the FCD concept at all in order to avoid
//! doing the work of checking whether the FCD condition holds.
extern crate alloc;
mod error;
pub mod properties;
pub mod provider;
pub mod uts46;
pub use crate::error::NormalizerError;
#[doc(no_inline)]
pub use NormalizerError as Error;
use crate::provider::CanonicalDecompositionDataV1Marker;
use crate::provider::CompatibilityDecompositionSupplementV1Marker;
use crate::provider::DecompositionDataV1;
use crate::provider::Uts46DecompositionSupplementV1Marker;
use alloc::string::String;
use alloc::vec::Vec;
use core::char::REPLACEMENT_CHARACTER;
use core::str::from_utf8_unchecked;
use icu_collections::char16trie::Char16Trie;
use icu_collections::char16trie::Char16TrieIterator;
use icu_collections::char16trie::TrieResult;
use icu_collections::codepointtrie::CodePointTrie;
use icu_properties::CanonicalCombiningClass;
use icu_provider::prelude::*;
use provider::CanonicalCompositionsV1Marker;
use provider::CanonicalDecompositionTablesV1Marker;
use provider::CompatibilityDecompositionTablesV1Marker;
use provider::DecompositionSupplementV1;
use provider::DecompositionTablesV1;
use smallvec::SmallVec;
use utf16_iter::Utf16CharsEx;
use utf8_iter::Utf8CharsEx;
use write16::Write16;
use zerofrom::ZeroFrom;
use zerovec::{zeroslice, ZeroSlice};
#[derive(Debug)]
enum SupplementPayloadHolder {
Compatibility(DataPayload<CompatibilityDecompositionSupplementV1Marker>),
Uts46(DataPayload<Uts46DecompositionSupplementV1Marker>),
}
impl SupplementPayloadHolder {
fn get(&self) -> &DecompositionSupplementV1 {
match self {
SupplementPayloadHolder::Compatibility(d) => d.get(),
SupplementPayloadHolder::Uts46(d) => d.get(),
}
}
}
/// Treatment of the ignorable marker (0xFFFFFFFF) in data.
#[derive(Debug, PartialEq, Eq)]
enum IgnorableBehavior {
/// 0xFFFFFFFF in data is not supported.
Unsupported,
/// Ignorables are ignored.
Ignored,
/// Ignorables are treated as singleton decompositions
/// to the REPLACEMENT CHARACTER.
ReplacementCharacter,
}
/// Number of iterations allowed on the fast path before flushing.
/// Since a typical UTF-16 iteration advances over a 2-byte BMP
/// character, this means two memory pages.
/// Intel Core i7-4770 had the best results between 2 and 4 pages
/// when testing powers of two. Apple M1 didn't seem to care
/// about 1, 2, 4, or 8 pages.
///
/// Curiously, the `str` case does not appear to benefit from
/// similar flushing, though the tested monomorphization never
/// passes an error through from `Write`.
const UTF16_FAST_PATH_FLUSH_THRESHOLD: usize = 4096;
/// Marker for UTS 46 ignorables.
const IGNORABLE_MARKER: u32 = 0xFFFFFFFF;
/// Marker for starters that decompose to themselves but may
/// combine backwards under canonical composition.
/// (Main trie only; not used in the supplementary trie.)
const BACKWARD_COMBINING_STARTER_MARKER: u32 = 1;
/// Magic marker trie value for characters whose decomposition
/// starts with a non-starter. The actual decomposition is
/// hard-coded.
const SPECIAL_NON_STARTER_DECOMPOSITION_MARKER: u32 = 2;
/// `u16` version of the previous marker value.
const SPECIAL_NON_STARTER_DECOMPOSITION_MARKER_U16: u16 = 2;
/// Marker that a complex decomposition isn't round-trippable
/// under re-composition.
const NON_ROUND_TRIP_MARKER: u16 = 1;
/// Checks if a trie value carries a (non-zero) canonical
/// combining class.
fn trie_value_has_ccc(trie_value: u32) -> bool {
(trie_value & 0xFFFFFF00) == 0xD800
}
/// Checks if the trie signifies a special non-starter decomposition.
fn trie_value_indicates_special_non_starter_decomposition(trie_value: u32) -> bool {
trie_value == SPECIAL_NON_STARTER_DECOMPOSITION_MARKER
}
/// Checks if a trie value signifies a character whose decomposition
/// starts with a non-starter.
fn decomposition_starts_with_non_starter(trie_value: u32) -> bool {
trie_value_has_ccc(trie_value)
|| trie_value_indicates_special_non_starter_decomposition(trie_value)
}
/// Extracts a canonical combining class (possibly zero) from a trie value.
///
/// # Panics
///
/// The trie value must not be one that signifies a special non-starter
/// decomposition. (Debug-only)
fn ccc_from_trie_value(trie_value: u32) -> CanonicalCombiningClass {
if trie_value_has_ccc(trie_value) {
CanonicalCombiningClass(trie_value as u8)
} else {
debug_assert_ne!(trie_value, SPECIAL_NON_STARTER_DECOMPOSITION_MARKER);
CanonicalCombiningClass::NotReordered
}
}
/// The tail (everything after the first character) of the NFKD form U+FDFA
/// as 16-bit units.
static FDFA_NFKD: [u16; 17] = [
0x644, 0x649, 0x20, 0x627, 0x644, 0x644, 0x647, 0x20, 0x639, 0x644, 0x64A, 0x647, 0x20, 0x648,
0x633, 0x644, 0x645,
];
/// Marker value for U+FDFA in NFKD
const FDFA_MARKER: u16 = 3;
// These constants originate from page 143 of Unicode 14.0
/// Syllable base
const HANGUL_S_BASE: u32 = 0xAC00;
/// Lead jamo base
const HANGUL_L_BASE: u32 = 0x1100;
/// Vowel jamo base
const HANGUL_V_BASE: u32 = 0x1161;
/// Trail jamo base (deliberately off by one to account for the absence of a trail)
const HANGUL_T_BASE: u32 = 0x11A7;
/// Lead jamo count
const HANGUL_L_COUNT: u32 = 19;
/// Vowel jamo count
const HANGUL_V_COUNT: u32 = 21;
/// Trail jamo count (deliberately off by one to account for the absence of a trail)
const HANGUL_T_COUNT: u32 = 28;
/// Vowel jamo count times trail jamo count
const HANGUL_N_COUNT: u32 = 588;
/// Syllable count
const HANGUL_S_COUNT: u32 = 11172;
/// One past the conjoining jamo block
const HANGUL_JAMO_LIMIT: u32 = 0x1200;
/// If `opt` is `Some`, unwrap it. If `None`, panic if debug assertions
/// are enabled and return `default` if debug assertions are not enabled.
///
/// Use this only if the only reason why `opt` could be `None` is bogus
/// data from the provider.
#[inline(always)]
fn unwrap_or_gigo<T>(opt: Option<T>, default: T) -> T {
if let Some(val) = opt {
val
} else {
// GIGO case
debug_assert!(false);
default
}
}
/// Convert a `u32` _obtained from data provider data_ to `char`.
#[inline(always)]
fn char_from_u32(u: u32) -> char {
unwrap_or_gigo(core::char::from_u32(u), REPLACEMENT_CHARACTER)
}
/// Convert a `u16` _obtained from data provider data_ to `char`.
#[inline(always)]
fn char_from_u16(u: u16) -> char {
char_from_u32(u32::from(u))
}
const EMPTY_U16: &ZeroSlice<u16> = zeroslice![];
const EMPTY_CHAR: &ZeroSlice<char> = zeroslice![];
#[inline(always)]
fn in_inclusive_range(c: char, start: char, end: char) -> bool {
u32::from(c).wrapping_sub(u32::from(start)) <= (u32::from(end) - u32::from(start))
}
#[inline(always)]
fn in_inclusive_range32(u: u32, start: u32, end: u32) -> bool {
u.wrapping_sub(start) <= (end - start)
}
#[inline(always)]
fn in_inclusive_range16(u: u16, start: u16, end: u16) -> bool {
u.wrapping_sub(start) <= (end - start)
}
/// Performs canonical composition (including Hangul) on a pair of
/// characters or returns `None` if these characters don't compose.
/// Composition exclusions are taken into account.
#[inline]
fn compose(iter: Char16TrieIterator, starter: char, second: char) -> Option<char> {
let v = u32::from(second).wrapping_sub(HANGUL_V_BASE);
if v >= HANGUL_JAMO_LIMIT - HANGUL_V_BASE {
return compose_non_hangul(iter, starter, second);
}
if v < HANGUL_V_COUNT {
let l = u32::from(starter).wrapping_sub(HANGUL_L_BASE);
if l < HANGUL_L_COUNT {
let lv = l * HANGUL_N_COUNT + v * HANGUL_T_COUNT;
// Safe, because the inputs are known to be in range.
return Some(unsafe { char::from_u32_unchecked(HANGUL_S_BASE + lv) });
}
return None;
}
if in_inclusive_range(second, '\u{11A8}', '\u{11C2}') {
let lv = u32::from(starter).wrapping_sub(HANGUL_S_BASE);
if lv < HANGUL_S_COUNT && lv % HANGUL_T_COUNT == 0 {
let lvt = lv + (u32::from(second) - HANGUL_T_BASE);
// Safe, because the inputs are known to be in range.
return Some(unsafe { char::from_u32_unchecked(HANGUL_S_BASE + lvt) });
}
}
None
}
/// Performs (non-Hangul) canonical composition on a pair of characters
/// or returns `None` if these characters don't compose. Composition
/// exclusions are taken into account.
fn compose_non_hangul(mut iter: Char16TrieIterator, starter: char, second: char) -> Option<char> {
// To make the trie smaller, the pairs are stored second character first.
// Given how this method is used in ways where it's known that `second`
// is or isn't a starter. We could potentially split the trie into two
// tries depending on whether `second` is a starter.
match iter.next(second) {
TrieResult::NoMatch => None,
TrieResult::NoValue => match iter.next(starter) {
TrieResult::NoMatch => None,
TrieResult::FinalValue(i) => {
if let Some(c) = char::from_u32(i as u32) {
Some(c)
} else {
// GIGO case
debug_assert!(false);
None
}
}
TrieResult::NoValue | TrieResult::Intermediate(_) => {
// GIGO case
debug_assert!(false);
None
}
},
TrieResult::FinalValue(_) | TrieResult::Intermediate(_) => {
// GIGO case
debug_assert!(false);
None
}
}
}
/// Struct for holding together a character and the value
/// looked up for it from the NFD trie in a more explicit
/// way than an anonymous pair.
/// Also holds a flag about the supplementary-trie provenance.
#[derive(Debug, PartialEq, Eq)]
struct CharacterAndTrieValue {
character: char,
trie_val: u32,
from_supplement: bool,
}
impl CharacterAndTrieValue {
#[inline(always)]
pub fn new(c: char, trie_value: u32) -> Self {
CharacterAndTrieValue {
character: c,
trie_val: trie_value,
from_supplement: false,
}
}
#[inline(always)]
pub fn new_from_supplement(c: char, trie_value: u32) -> Self {
CharacterAndTrieValue {
character: c,
trie_val: trie_value,
from_supplement: true,
}
}
#[inline(always)]
pub fn starter_and_decomposes_to_self(&self) -> bool {
if self.trie_val > BACKWARD_COMBINING_STARTER_MARKER {
return false;
}
// Hangul syllables get 0 as their trie value
u32::from(self.character).wrapping_sub(HANGUL_S_BASE) >= HANGUL_S_COUNT
}
#[inline(always)]
pub fn can_combine_backwards(&self) -> bool {
decomposition_starts_with_non_starter(self.trie_val)
|| self.trie_val == BACKWARD_COMBINING_STARTER_MARKER
|| in_inclusive_range32(self.trie_val, 0x1161, 0x11C2)
}
#[inline(always)]
pub fn potential_passthrough(&self) -> bool {
self.potential_passthrough_impl(BACKWARD_COMBINING_STARTER_MARKER)
}
#[inline(always)]
pub fn potential_passthrough_and_cannot_combine_backwards(&self) -> bool {
self.potential_passthrough_impl(0)
}
#[inline(always)]
fn potential_passthrough_impl(&self, bound: u32) -> bool {
// This methods looks badly branchy, but most characters
// take the first return.
if self.trie_val <= bound {
return true;
}
if self.from_supplement {
return false;
}
let trail_or_complex = (self.trie_val >> 16) as u16;
if trail_or_complex == 0 {
return false;
}
let lead = self.trie_val as u16;
if lead == 0 {
return true;
}
if lead == NON_ROUND_TRIP_MARKER {
return false;
}
if (trail_or_complex & 0x7F) == 0x3C
&& in_inclusive_range16(trail_or_complex, 0x0900, 0x0BFF)
{
// Nukta
return false;
}
if in_inclusive_range(self.character, '\u{FB1D}', '\u{FB4E}') {
// Hebrew presentation forms
return false;
}
if in_inclusive_range(self.character, '\u{1F71}', '\u{1FFB}') {
// Polytonic Greek with oxia
return false;
}
// To avoid more branchiness, 4 characters that decompose to
// a BMP starter followed by a BMP non-starter are excluded
// from being encoded directly into the trie value and are
// handled as complex decompositions instead. These are:
// U+0F76 TIBETAN VOWEL SIGN VOCALIC R
// U+0F78 TIBETAN VOWEL SIGN VOCALIC L
// U+212B ANGSTROM SIGN
// U+2ADC FORKING
true
}
}
/// Pack a `char` and a `CanonicalCombiningClass` in
/// 32 bits (the former in the lower 24 bits and the
/// latter in the high 8 bits). The latter can be
/// initialized to 0xFF upon creation, in which case
/// it can be actually set later by calling
/// `set_ccc_from_trie_if_not_already_set`. This is
/// a micro optimization to avoid the Canonical
/// Combining Class trie lookup when there is only
/// one combining character in a sequence. This type
/// is intentionally non-`Copy` to get compiler help
/// in making sure that the class is set on the
/// instance on which it is intended to be set
/// and not on a temporary copy.
///
/// Note that 0xFF is won't be assigned to an actual
/// canonical combining class per definition D104
/// in The Unicode Standard.
//
// NOTE: The Pernosco debugger has special knowledge
// of this struct. Please do not change the bit layout
// or the crate-module-qualified name of this struct
// without coordination.
#[derive(Debug)]
struct CharacterAndClass(u32);
impl CharacterAndClass {
pub fn new(c: char, ccc: CanonicalCombiningClass) -> Self {
CharacterAndClass(u32::from(c) | (u32::from(ccc.0) << 24))
}
pub fn new_with_placeholder(c: char) -> Self {
CharacterAndClass(u32::from(c) | ((0xFF) << 24))
}
pub fn new_with_trie_value(c_tv: CharacterAndTrieValue) -> Self {
Self::new(c_tv.character, ccc_from_trie_value(c_tv.trie_val))
}
pub fn new_starter(c: char) -> Self {
CharacterAndClass(u32::from(c))
}
pub fn character(&self) -> char {
// Safe, because the low 24 bits came from a `char`
// originally.
unsafe { char::from_u32_unchecked(self.0 & 0xFFFFFF) }
}
pub fn ccc(&self) -> CanonicalCombiningClass {
CanonicalCombiningClass((self.0 >> 24) as u8)
}
pub fn character_and_ccc(&self) -> (char, CanonicalCombiningClass) {
(self.character(), self.ccc())
}
pub fn set_ccc_from_trie_if_not_already_set(&mut self, trie: &CodePointTrie<u32>) {
if self.0 >> 24 != 0xFF {
return;
}
let scalar = self.0 & 0xFFFFFF;
self.0 = ((ccc_from_trie_value(trie.get32_u32(scalar)).0 as u32) << 24) | scalar;
}
}
// This function exists as a borrow check helper.
#[inline(always)]
fn sort_slice_by_ccc(slice: &mut [CharacterAndClass], trie: &CodePointTrie<u32>) {
// We don't look up the canonical combining class for starters
// of for single combining characters between starters. When
// there's more than one combining character between starters,
// we look up the canonical combining class for each character
// exactly once.
if slice.len() < 2 {
return;
}
slice
.iter_mut()
.for_each(|cc| cc.set_ccc_from_trie_if_not_already_set(trie));
slice.sort_by_key(|cc| cc.ccc());
}
/// An iterator adaptor that turns an `Iterator` over `char` into
/// a lazily-decomposed `char` sequence.
#[derive(Debug)]
pub struct Decomposition<'data, I>
where
I: Iterator<Item = char>,
{
delegate: I,
buffer: SmallVec<[CharacterAndClass; 17]>, // Enough to hold NFKD for U+FDFA
/// The index of the next item to be read from `buffer`.
/// The purpose if this index is to avoid having to move
/// the rest upon every read.
buffer_pos: usize,
// At the start of `next()` if not `None`, this is a pending unnormalized
// starter. When `Decomposition` appears alone, this is never a non-starter.
// However, when `Decomposition` appears inside a `Composition`, this
// may become a non-starter before `decomposing_next()` is called.
pending: Option<CharacterAndTrieValue>, // None at end of stream
trie: &'data CodePointTrie<'data, u32>,
supplementary_trie: Option<&'data CodePointTrie<'data, u32>>,
scalars16: &'data ZeroSlice<u16>,
scalars24: &'data ZeroSlice<char>,
supplementary_scalars16: &'data ZeroSlice<u16>,
supplementary_scalars24: &'data ZeroSlice<char>,
half_width_voicing_marks_become_non_starters: bool,
/// The lowest character for which either of the following does
/// not hold:
/// 1. Decomposes to self.
/// 2. Decomposition starts with a non-starter
decomposition_passthrough_bound: u32, // never above 0xC0
ignorable_behavior: IgnorableBehavior, // Arguably should be a type parameter
}
impl<'data, I> Decomposition<'data, I>
where
I: Iterator<Item = char>,
{
/// Constructs a decomposing iterator adapter from a delegate
/// iterator and references to the necessary data, without
/// supplementary data.
///
/// Use `DecomposingNormalizer::normalize_iter()` instead unless
/// there's a good reason to use this constructor directly.
///
/// Public but hidden in order to be able to use this from the
/// collator.
#[doc(hidden)]
pub fn new(
delegate: I,
decompositions: &'data DecompositionDataV1,
tables: &'data DecompositionTablesV1,
) -> Self {
Self::new_with_supplements(
delegate,
decompositions,
None,
tables,
None,
0xC0,
IgnorableBehavior::Unsupported,
)
}
/// Constructs a decomposing iterator adapter from a delegate
/// iterator and references to the necessary data, including
/// supplementary data.
///
/// Use `DecomposingNormalizer::normalize_iter()` instead unless
/// there's a good reason to use this constructor directly.
fn new_with_supplements(
delegate: I,
decompositions: &'data DecompositionDataV1,
supplementary_decompositions: Option<&'data DecompositionSupplementV1>,
tables: &'data DecompositionTablesV1,
supplementary_tables: Option<&'data DecompositionTablesV1>,
decomposition_passthrough_bound: u8,
ignorable_behavior: IgnorableBehavior,
) -> Self {
let half_width_voicing_marks_become_non_starters =
if let Some(supplementary) = supplementary_decompositions {
supplementary.half_width_voicing_marks_become_non_starters()
} else {
false
};
let mut ret = Decomposition::<I> {
delegate,
buffer: SmallVec::new(), // Normalized
buffer_pos: 0,
// Initialize with a placeholder starter in case
// the real stream starts with a non-starter.
pending: Some(CharacterAndTrieValue::new('\u{FFFF}', 0)),
trie: &decompositions.trie,
supplementary_trie: supplementary_decompositions.map(|s| &s.trie),
scalars16: &tables.scalars16,
scalars24: &tables.scalars24,
supplementary_scalars16: if let Some(supplementary) = supplementary_tables {
&supplementary.scalars16
} else {
EMPTY_U16
},
supplementary_scalars24: if let Some(supplementary) = supplementary_tables {
&supplementary.scalars24
} else {
EMPTY_CHAR
},
half_width_voicing_marks_become_non_starters,
decomposition_passthrough_bound: u32::from(decomposition_passthrough_bound),
ignorable_behavior,
};
let _ = ret.next(); // Remove the U+FFFF placeholder
ret
}
fn push_decomposition16(
&mut self,
low: u16,
offset: usize,
slice16: &ZeroSlice<u16>,
) -> (char, usize) {
let len = usize::from(low >> 13) + 2;
let (starter, tail) = slice16
.get_subslice(offset..offset + len)
.and_then(|slice| slice.split_first())
.map_or_else(
|| {
// GIGO case
debug_assert!(false);
(REPLACEMENT_CHARACTER, EMPTY_U16)
},
|(first, trail)| (char_from_u16(first), trail),
);
if low & 0x1000 != 0 {
// All the rest are combining
self.buffer.extend(
tail.iter()
.map(|u| CharacterAndClass::new_with_placeholder(char_from_u16(u))),
);
(starter, 0)
} else {
let mut i = 0;
let mut combining_start = 0;
for u in tail.iter() {
let ch = char_from_u16(u);
let trie_value = self.trie.get(ch);
self.buffer.push(CharacterAndClass::new_with_trie_value(
CharacterAndTrieValue::new(ch, trie_value),
));
i += 1;
// Half-width kana and iota subscript don't occur in the tails
// of these multicharacter decompositions.
if !decomposition_starts_with_non_starter(trie_value) {
combining_start = i;
}
}
(starter, combining_start)
}
}
fn push_decomposition32(
&mut self,
low: u16,
offset: usize,
slice32: &ZeroSlice<char>,
) -> (char, usize) {
let len = usize::from(low >> 13) + 1;
let (starter, tail) = slice32
.get_subslice(offset..offset + len)
.and_then(|slice| slice.split_first())
.unwrap_or_else(|| {
// GIGO case
debug_assert!(false);
(REPLACEMENT_CHARACTER, EMPTY_CHAR)
});
if low & 0x1000 != 0 {
// All the rest are combining
self.buffer
.extend(tail.iter().map(CharacterAndClass::new_with_placeholder));
(starter, 0)
} else {
let mut i = 0;
let mut combining_start = 0;
for ch in tail.iter() {
let trie_value = self.trie.get(ch);
self.buffer.push(CharacterAndClass::new_with_trie_value(
CharacterAndTrieValue::new(ch, trie_value),
));
i += 1;
// Half-width kana and iota subscript don't occur in the tails
// of these multicharacter decompositions.
if !decomposition_starts_with_non_starter(trie_value) {
combining_start = i;
}
}
(starter, combining_start)
}
}
#[inline(always)]
fn attach_trie_value(&self, c: char) -> CharacterAndTrieValue {
if let Some(supplementary) = self.supplementary_trie {
if let Some(value) = self.attach_supplementary_trie_value(c, supplementary) {
return value;
}
}
CharacterAndTrieValue::new(c, self.trie.get(c))
}
#[inline(never)]
fn attach_supplementary_trie_value(
&self,
c: char,
supplementary: &CodePointTrie<u32>,
) -> Option<CharacterAndTrieValue> {
let voicing_mark = u32::from(c).wrapping_sub(0xFF9E);
if voicing_mark <= 1 && self.half_width_voicing_marks_become_non_starters {
return Some(CharacterAndTrieValue::new(
if voicing_mark == 0 {
'\u{3099}'
} else {
'\u{309A}'
},
0xD800 | u32::from(CanonicalCombiningClass::KanaVoicing.0),
));
}
let trie_value = supplementary.get32(u32::from(c));
if trie_value != 0 {
return Some(CharacterAndTrieValue::new_from_supplement(c, trie_value));
}
None
}
fn delegate_next_no_pending(&mut self) -> Option<CharacterAndTrieValue> {
debug_assert!(self.pending.is_none());
loop {
let c = self.delegate.next()?;
// TODO(#2384): Measure if this check is actually an optimization even in the
// non-supplementary case of if this should go inside the supplementary
// `if` below.
if u32::from(c) < self.decomposition_passthrough_bound {
return Some(CharacterAndTrieValue::new(c, 0));
}
if let Some(supplementary) = self.supplementary_trie {
if let Some(value) = self.attach_supplementary_trie_value(c, supplementary) {
if value.trie_val == IGNORABLE_MARKER {
match self.ignorable_behavior {
IgnorableBehavior::Unsupported => {
debug_assert!(false);
}
IgnorableBehavior::ReplacementCharacter => {
return Some(CharacterAndTrieValue::new(
c,
u32::from(REPLACEMENT_CHARACTER),
));
}
IgnorableBehavior::Ignored => {
// Else ignore this character by reading the next one from the delegate.
continue;
}
}
}
return Some(value);
}
}
let trie_val = self.trie.get(c);
debug_assert_ne!(trie_val, IGNORABLE_MARKER);
return Some(CharacterAndTrieValue::new(c, trie_val));
}
}
fn delegate_next(&mut self) -> Option<CharacterAndTrieValue> {
if let Some(pending) = self.pending.take() {
// Only happens as part of `Composition` and as part of
// the contiguous-buffer methods of `DecomposingNormalizer`.
// I.e. does not happen as part of standalone iterator
// usage of `Decomposition`.
Some(pending)
} else {
self.delegate_next_no_pending()
}
}
fn decomposing_next(&mut self, c_and_trie_val: CharacterAndTrieValue) -> char {
let (starter, combining_start) = {
let c = c_and_trie_val.character;
let hangul_offset = u32::from(c).wrapping_sub(HANGUL_S_BASE); // SIndex in the spec
if hangul_offset >= HANGUL_S_COUNT {
let decomposition = c_and_trie_val.trie_val;
if decomposition <= BACKWARD_COMBINING_STARTER_MARKER {
// The character is its own decomposition
(c, 0)
} else {
let trail_or_complex = (decomposition >> 16) as u16;
let lead = decomposition as u16;
if lead > NON_ROUND_TRIP_MARKER && trail_or_complex != 0 {
// Decomposition into two BMP characters: starter and non-starter
let starter = char_from_u16(lead);
let combining = char_from_u16(trail_or_complex);
self.buffer
.push(CharacterAndClass::new_with_placeholder(combining));
(starter, 0)
} else if lead > NON_ROUND_TRIP_MARKER {
if lead != FDFA_MARKER {
debug_assert_ne!(
lead, SPECIAL_NON_STARTER_DECOMPOSITION_MARKER_U16,
"Should not reach this point with non-starter marker"
);
// Decomposition into one BMP character
let starter = char_from_u16(lead);
(starter, 0)
} else {
// Special case for the NFKD form of U+FDFA.
self.buffer.extend(FDFA_NFKD.map(|u| {
// Safe, because `FDFA_NFKD` is known not to contain
// surrogates.
CharacterAndClass::new_starter(unsafe {
core::char::from_u32_unchecked(u32::from(u))
})
}));
('\u{0635}', 17)
}
} else {
// Complex decomposition
// Format for 16-bit value:
// 15..13: length minus two for 16-bit case and length minus one for
// the 32-bit case. Length 8 needs to fit in three bits in
// the 16-bit case, and this way the value is future-proofed
// up to 9 in the 16-bit case. Zero is unused and length one
// in the 16-bit case goes directly into the trie.
// 12: 1 if all trailing characters are guaranteed non-starters,
// 0 if no guarantees about non-starterness.
// Note: The bit choice is this way around to allow for
// dynamically falling back to not having this but instead
// having one more bit for length by merely choosing
// different masks.
// 11..0: Start offset in storage. The offset is to the logical
// sequence of scalars16, scalars32, supplementary_scalars16,
// supplementary_scalars32.
let offset = usize::from(trail_or_complex & 0xFFF);
if offset < self.scalars16.len() {
self.push_decomposition16(trail_or_complex, offset, self.scalars16)
} else if offset < self.scalars16.len() + self.scalars24.len() {
self.push_decomposition32(
trail_or_complex,
offset - self.scalars16.len(),
self.scalars24,
)
} else if offset
< self.scalars16.len()
+ self.scalars24.len()
+ self.supplementary_scalars16.len()
{
self.push_decomposition16(
trail_or_complex,
offset - (self.scalars16.len() + self.scalars24.len()),
self.supplementary_scalars16,
)
} else {
self.push_decomposition32(
trail_or_complex,
offset
- (self.scalars16.len()
+ self.scalars24.len()
+ self.supplementary_scalars16.len()),
self.supplementary_scalars24,
)
}
}
}
} else {
// Hangul syllable
// The math here comes from page 144 of Unicode 14.0
let l = hangul_offset / HANGUL_N_COUNT;
let v = (hangul_offset % HANGUL_N_COUNT) / HANGUL_T_COUNT;
let t = hangul_offset % HANGUL_T_COUNT;
// The unsafe blocks here are OK, because the values stay
// within the Hangul jamo block and, therefore, the scalar
// value range by construction.
self.buffer.push(CharacterAndClass::new_starter(unsafe {
core::char::from_u32_unchecked(HANGUL_V_BASE + v)
}));
let first = unsafe { core::char::from_u32_unchecked(HANGUL_L_BASE + l) };
if t != 0 {
self.buffer.push(CharacterAndClass::new_starter(unsafe {
core::char::from_u32_unchecked(HANGUL_T_BASE + t)
}));
(first, 2)
} else {
(first, 1)
}
}
};
// Either we're inside `Composition` or `self.pending.is_none()`.
self.gather_and_sort_combining(combining_start);
starter
}
fn gather_and_sort_combining(&mut self, combining_start: usize) {
// Not a `for` loop to avoid holding a mutable reference to `self` across
// the loop body.
while let Some(ch_and_trie_val) = self.delegate_next() {
if trie_value_has_ccc(ch_and_trie_val.trie_val) {
self.buffer
.push(CharacterAndClass::new_with_trie_value(ch_and_trie_val));
} else if trie_value_indicates_special_non_starter_decomposition(
ch_and_trie_val.trie_val,
) {
// The Tibetan special cases are starters that decompose into non-starters.
let mapped = match ch_and_trie_val.character {
'\u{0340}' => {
// COMBINING GRAVE TONE MARK
CharacterAndClass::new('\u{0300}', CanonicalCombiningClass::Above)
}
'\u{0341}' => {
// COMBINING ACUTE TONE MARK
CharacterAndClass::new('\u{0301}', CanonicalCombiningClass::Above)
}
'\u{0343}' => {
// COMBINING GREEK KORONIS
CharacterAndClass::new('\u{0313}', CanonicalCombiningClass::Above)
}
'\u{0344}' => {
// COMBINING GREEK DIALYTIKA TONOS
self.buffer.push(CharacterAndClass::new(
'\u{0308}',
CanonicalCombiningClass::Above,
));
CharacterAndClass::new('\u{0301}', CanonicalCombiningClass::Above)
}
'\u{0F73}' => {
// TIBETAN VOWEL SIGN II
self.buffer.push(CharacterAndClass::new(
'\u{0F71}',
CanonicalCombiningClass::CCC129,
));
CharacterAndClass::new('\u{0F72}', CanonicalCombiningClass::CCC130)
}
'\u{0F75}' => {
// TIBETAN VOWEL SIGN UU
self.buffer.push(CharacterAndClass::new(
'\u{0F71}',
CanonicalCombiningClass::CCC129,
));
CharacterAndClass::new('\u{0F74}', CanonicalCombiningClass::CCC132)
}
'\u{0F81}' => {
// TIBETAN VOWEL SIGN REVERSED II
self.buffer.push(CharacterAndClass::new(
'\u{0F71}',
CanonicalCombiningClass::CCC129,
));
CharacterAndClass::new('\u{0F80}', CanonicalCombiningClass::CCC130)
}
_ => {
// GIGO case
debug_assert!(false);
CharacterAndClass::new_with_placeholder(REPLACEMENT_CHARACTER)
}
};
self.buffer.push(mapped);
} else {
self.pending = Some(ch_and_trie_val);
break;
}
}
// Slicing succeeds by construction; we've always ensured that `combining_start`
// is in permissible range.
#[allow(clippy::indexing_slicing)]
sort_slice_by_ccc(&mut self.buffer[combining_start..], self.trie);
}
}
impl<'data, I> Iterator for Decomposition<'data, I>
where
I: Iterator<Item = char>,
{
type Item = char;
fn next(&mut self) -> Option<char> {
if let Some(ret) = self.buffer.get(self.buffer_pos).map(|c| c.character()) {
self.buffer_pos += 1;
if self.buffer_pos == self.buffer.len() {
self.buffer.clear();
self.buffer_pos = 0;
}
return Some(ret);
}
debug_assert_eq!(self.buffer_pos, 0);
let c_and_trie_val = self.pending.take()?;
Some(self.decomposing_next(c_and_trie_val))
}
}
/// An iterator adaptor that turns an `Iterator` over `char` into
/// a lazily-decomposed and then canonically composed `char` sequence.
#[derive(Debug)]
pub struct Composition<'data, I>
where
I: Iterator<Item = char>,
{
/// The decomposing part of the normalizer than operates before
/// the canonical composition is performed on its output.
decomposition: Decomposition<'data, I>,
/// Non-Hangul canonical composition data.
canonical_compositions: Char16Trie<'data>,
/// To make `next()` yield in cases where there's a non-composing
/// starter in the decomposition buffer, we put it here to let it
/// wait for the next `next()` call (or a jump forward within the
/// `next()` call).
unprocessed_starter: Option<char>,
/// The lowest character for which any one of the following does
/// not hold:
/// 1. Roundtrips via decomposition and recomposition.
/// 2. Decomposition starts with a non-starter
/// 3. Is not a backward-combining starter
composition_passthrough_bound: u32,
}
impl<'data, I> Composition<'data, I>
where
I: Iterator<Item = char>,
{
fn new(
decomposition: Decomposition<'data, I>,
canonical_compositions: Char16Trie<'data>,
composition_passthrough_bound: u16,
) -> Self {
Self {
decomposition,
canonical_compositions,
unprocessed_starter: None,
composition_passthrough_bound: u32::from(composition_passthrough_bound),
}
}
/// Performs canonical composition (including Hangul) on a pair of
/// characters or returns `None` if these characters don't compose.
/// Composition exclusions are taken into account.
#[inline(always)]
pub fn compose(&self, starter: char, second: char) -> Option<char> {
compose(self.canonical_compositions.iter(), starter, second)
}
/// Performs (non-Hangul) canonical composition on a pair of characters
/// or returns `None` if these characters don't compose. Composition
/// exclusions are taken into account.
#[inline(always)]
fn compose_non_hangul(&self, starter: char, second: char) -> Option<char> {
compose_non_hangul(self.canonical_compositions.iter(), starter, second)
}
}
impl<'data, I> Iterator for Composition<'data, I>
where
I: Iterator<Item = char>,
{
type Item = char;
#[inline]
fn next(&mut self) -> Option<char> {
let mut undecomposed_starter = CharacterAndTrieValue::new('\u{0}', 0); // The compiler can't figure out that this gets overwritten before use.
if self.unprocessed_starter.is_none() {
// The loop is only broken out of as goto forward
#[allow(clippy::never_loop)]
loop {
if let Some((character, ccc)) = self
.decomposition
.buffer
.get(self.decomposition.buffer_pos)
.map(|c| c.character_and_ccc())
{
self.decomposition.buffer_pos += 1;
if self.decomposition.buffer_pos == self.decomposition.buffer.len() {
self.decomposition.buffer.clear();
self.decomposition.buffer_pos = 0;
}
if ccc == CanonicalCombiningClass::NotReordered {
// Previous decomposition contains a starter. This must
// now become the `unprocessed_starter` for it to have
// a chance to compose with the upcoming characters.
//
// E.g. parenthesized Hangul in NFKC comes through here,
// but suitable composition exclusion could exercise this
// in NFC.
self.unprocessed_starter = Some(character);
break; // We already have a starter, so skip taking one from `pending`.
}
return Some(character);
}
debug_assert_eq!(self.decomposition.buffer_pos, 0);
undecomposed_starter = self.decomposition.pending.take()?;
if u32::from(undecomposed_starter.character) < self.composition_passthrough_bound
|| undecomposed_starter.potential_passthrough()
{
// TODO(#2385): In the NFC case (moot for NFKC and UTS46), if the upcoming
// character is not below `decomposition_passthrough_bound` but is
// below `composition_passthrough_bound`, we read from the trie
// unnecessarily.
if let Some(upcoming) = self.decomposition.delegate_next_no_pending() {
let cannot_combine_backwards = u32::from(upcoming.character)
< self.composition_passthrough_bound
|| !upcoming.can_combine_backwards();
self.decomposition.pending = Some(upcoming);
if cannot_combine_backwards {
// Fast-track succeeded!
return Some(undecomposed_starter.character);
}
} else {
// End of stream
return Some(undecomposed_starter.character);
}
}
break; // Not actually looping
}
}
let mut starter = '\u{0}'; // The compiler can't figure out this gets overwritten before use.
// The point of having this boolean is to have only one call site to
// `self.decomposition.decomposing_next`, which is hopefully beneficial for
// code size under inlining.
let mut attempt_composition = false;
loop {
if let Some(unprocessed) = self.unprocessed_starter.take() {
debug_assert_eq!(undecomposed_starter, CharacterAndTrieValue::new('\u{0}', 0));
debug_assert_eq!(starter, '\u{0}');
starter = unprocessed;
} else {
debug_assert_eq!(self.decomposition.buffer_pos, 0);
let next_starter = self.decomposition.decomposing_next(undecomposed_starter);
if !attempt_composition {
starter = next_starter;
} else if let Some(composed) = self.compose(starter, next_starter) {
starter = composed;
} else {
// This is our yield point. We'll pick this up above in the
// next call to `next()`.
self.unprocessed_starter = Some(next_starter);
return Some(starter);
}
}
// We first loop by index to avoid moving the contents of `buffer`, but
// if there's a discontiguous match, we'll start modifying `buffer` instead.
loop {
let (character, ccc) = if let Some((character, ccc)) = self
.decomposition
.buffer
.get(self.decomposition.buffer_pos)
.map(|c| c.character_and_ccc())
{
(character, ccc)
} else {
self.decomposition.buffer.clear();
self.decomposition.buffer_pos = 0;
break;
};
if let Some(composed) = self.compose(starter, character) {
starter = composed;
self.decomposition.buffer_pos += 1;
continue;
}
let mut most_recent_skipped_ccc = ccc;
{
let _ = self
.decomposition
.buffer
.drain(0..self.decomposition.buffer_pos);
}
self.decomposition.buffer_pos = 0;
if most_recent_skipped_ccc == CanonicalCombiningClass::NotReordered {
// We failed to compose a starter. Discontiguous match not allowed.
// We leave the starter in `buffer` for `next()` to find.
return Some(starter);
}
let mut i = 1; // We have skipped one non-starter.
while let Some((character, ccc)) = self
.decomposition
.buffer
.get(i)
.map(|c| c.character_and_ccc())
{
if ccc == CanonicalCombiningClass::NotReordered {
// Discontiguous match not allowed.
return Some(starter);
}
debug_assert!(ccc >= most_recent_skipped_ccc);
if ccc != most_recent_skipped_ccc {
// Using the non-Hangul version as a micro-optimization, since
// we already rejected the case where `second` is a starter
// above, and conjoining jamo are starters.
if let Some(composed) = self.compose_non_hangul(starter, character) {
self.decomposition.buffer.remove(i);
starter = composed;
continue;
}
}
most_recent_skipped_ccc = ccc;
i += 1;
}
break;
}
debug_assert_eq!(self.decomposition.buffer_pos, 0);
if !self.decomposition.buffer.is_empty() {
return Some(starter);
}
// Now we need to check if composition with an upcoming starter is possible.
#[allow(clippy::unwrap_used)]
if self.decomposition.pending.is_some() {
// We know that `pending_starter` decomposes to start with a starter.
// Otherwise, it would have been moved to `self.decomposition.buffer`
// by `self.decomposing_next()`. We do this set lookup here in order
// to get an opportunity to go back to the fast track.
// Note that this check has to happen _after_ checking that `pending`
// holds a character, because this flag isn't defined to be meaningful
// when `pending` isn't holding a character.
let pending = self.decomposition.pending.as_ref().unwrap();
if u32::from(pending.character) < self.composition_passthrough_bound
|| !pending.can_combine_backwards()
{
// Won't combine backwards anyway.
return Some(starter);
}
// Consume what we peeked. `unwrap` OK, because we checked `is_some()`
// above.
undecomposed_starter = self.decomposition.pending.take().unwrap();
// The following line is OK, because we're about to loop back
// to `self.decomposition.decomposing_next(c);`, which will
// restore the between-`next()`-calls invariant of `pending`
// before this function returns.
attempt_composition = true;
continue;
}
// End of input
return Some(starter);
}
}
}
macro_rules! composing_normalize_to {
($(#[$meta:meta])*,
$normalize_to:ident,
$write:path,
$slice:ty,
$prolog:block,
$always_valid_utf:literal,
$as_slice:ident,
$fast:block,
$text:ident,
$sink:ident,
$composition:ident,
$composition_passthrough_bound:ident,
$undecomposed_starter:ident,
$pending_slice:ident,
$len_utf:ident,
) => {
$(#[$meta])*
pub fn $normalize_to<W: $write + ?Sized>(
&self,
$text: $slice,
$sink: &mut W,
) -> core::fmt::Result {
$prolog
let mut $composition = self.normalize_iter($text.chars());
debug_assert_eq!($composition.decomposition.ignorable_behavior, IgnorableBehavior::Unsupported);
for cc in $composition.decomposition.buffer.drain(..) {
$sink.write_char(cc.character())?;
}
// Try to get the compiler to hoist the bound to a register.
let $composition_passthrough_bound = $composition.composition_passthrough_bound;
'outer: loop {
debug_assert_eq!($composition.decomposition.buffer_pos, 0);
let mut $undecomposed_starter =
if let Some(pending) = $composition.decomposition.pending.take() {
pending
} else {
return Ok(());
};
// Allowing indexed slicing, because a failure would be a code bug and
// not a data issue.
#[allow(clippy::indexing_slicing)]
if u32::from($undecomposed_starter.character) < $composition_passthrough_bound ||
$undecomposed_starter.potential_passthrough()
{
// We don't know if a `REPLACEMENT_CHARACTER` occurred in the slice or
// was returned in response to an error by the iterator. Assume the
// latter for correctness even though it pessimizes the former.
if $always_valid_utf || $undecomposed_starter.character != REPLACEMENT_CHARACTER {
let $pending_slice = &$text[$text.len() - $composition.decomposition.delegate.$as_slice().len() - $undecomposed_starter.character.$len_utf()..];
// The `$fast` block must either:
// 1. Return due to reaching EOF
// 2. Leave a starter with its trie value in `$undecomposed_starter`
// and, if there is still more input, leave the next character
// and its trie value in `$composition.decomposition.pending`.
$fast
}
}
// Fast track above, full algorithm below
let mut starter = $composition
.decomposition
.decomposing_next($undecomposed_starter);
'bufferloop: loop {
// We first loop by index to avoid moving the contents of `buffer`, but
// if there's a discontiguous match, we'll start modifying `buffer` instead.
loop {
let (character, ccc) = if let Some((character, ccc)) = $composition
.decomposition
.buffer
.get($composition.decomposition.buffer_pos)
.map(|c| c.character_and_ccc())
{
(character, ccc)
} else {
$composition.decomposition.buffer.clear();
$composition.decomposition.buffer_pos = 0;
break;
};
if let Some(composed) = $composition.compose(starter, character) {
starter = composed;
$composition.decomposition.buffer_pos += 1;
continue;
}
let mut most_recent_skipped_ccc = ccc;
if most_recent_skipped_ccc == CanonicalCombiningClass::NotReordered {
// We failed to compose a starter. Discontiguous match not allowed.
// Write the current `starter` we've been composing, make the unmatched
// starter in the buffer the new `starter` (we know it's been decomposed)
// and process the rest of the buffer with that as the starter.
$sink.write_char(starter)?;
starter = character;
$composition.decomposition.buffer_pos += 1;
continue 'bufferloop;
} else {
{
let _ = $composition
.decomposition
.buffer
.drain(0..$composition.decomposition.buffer_pos);
}
$composition.decomposition.buffer_pos = 0;
}
let mut i = 1; // We have skipped one non-starter.
while let Some((character, ccc)) = $composition
.decomposition
.buffer
.get(i)
.map(|c| c.character_and_ccc())
{
if ccc == CanonicalCombiningClass::NotReordered {
// Discontiguous match not allowed.
$sink.write_char(starter)?;
for cc in $composition.decomposition.buffer.drain(..i) {
$sink.write_char(cc.character())?;
}
starter = character;
{
let removed = $composition.decomposition.buffer.remove(0);
debug_assert_eq!(starter, removed.character());
}
debug_assert_eq!($composition.decomposition.buffer_pos, 0);
continue 'bufferloop;
}
debug_assert!(ccc >= most_recent_skipped_ccc);
if ccc != most_recent_skipped_ccc {
// Using the non-Hangul version as a micro-optimization, since
// we already rejected the case where `second` is a starter
// above, and conjoining jamo are starters.
if let Some(composed) =
$composition.compose_non_hangul(starter, character)
{
$composition.decomposition.buffer.remove(i);
starter = composed;
continue;
}
}
most_recent_skipped_ccc = ccc;
i += 1;
}
break;
}
debug_assert_eq!($composition.decomposition.buffer_pos, 0);
if !$composition.decomposition.buffer.is_empty() {
$sink.write_char(starter)?;
for cc in $composition.decomposition.buffer.drain(..) {
$sink.write_char(cc.character())?;
}
// We had non-empty buffer, so can't compose with upcoming.
continue 'outer;
}
// Now we need to check if composition with an upcoming starter is possible.
if $composition.decomposition.pending.is_some() {
// We know that `pending_starter` decomposes to start with a starter.
// Otherwise, it would have been moved to `composition.decomposition.buffer`
// by `composition.decomposing_next()`. We do this set lookup here in order
// to get an opportunity to go back to the fast track.
// Note that this check has to happen _after_ checking that `pending`
// holds a character, because this flag isn't defined to be meaningful
// when `pending` isn't holding a character.
let pending = $composition.decomposition.pending.as_ref().unwrap();
if u32::from(pending.character) < $composition.composition_passthrough_bound
|| !pending.can_combine_backwards()
{
// Won't combine backwards anyway.
$sink.write_char(starter)?;
continue 'outer;
}
let pending_starter = $composition.decomposition.pending.take().unwrap();
let decomposed = $composition.decomposition.decomposing_next(pending_starter);
if let Some(composed) = $composition.compose(starter, decomposed) {
starter = composed;
} else {
$sink.write_char(starter)?;
starter = decomposed;
}
continue 'bufferloop;
}
// End of input
$sink.write_char(starter)?;
return Ok(());
} // 'bufferloop
}
}
};
}
macro_rules! decomposing_normalize_to {
($(#[$meta:meta])*,
$normalize_to:ident,
$write:path,
$slice:ty,
$prolog:block,
$as_slice:ident,
$fast:block,
$text:ident,
$sink:ident,
$decomposition:ident,
$decomposition_passthrough_bound:ident,
$undecomposed_starter:ident,
$pending_slice:ident,
$outer:lifetime, // loop labels use lifetime tokens
) => {
$(#[$meta])*
pub fn $normalize_to<W: $write + ?Sized>(
&self,
$text: $slice,
$sink: &mut W,
) -> core::fmt::Result {
$prolog
let mut $decomposition = self.normalize_iter($text.chars());
debug_assert_eq!($decomposition.ignorable_behavior, IgnorableBehavior::Unsupported);
// Try to get the compiler to hoist the bound to a register.
let $decomposition_passthrough_bound = $decomposition.decomposition_passthrough_bound;
$outer: loop {
for cc in $decomposition.buffer.drain(..) {
$sink.write_char(cc.character())?;
}
debug_assert_eq!($decomposition.buffer_pos, 0);
let mut $undecomposed_starter = if let Some(pending) = $decomposition.pending.take() {
pending
} else {
return Ok(());
};
// Allowing indexed slicing, because a failure would be a code bug and
// not a data issue.
#[allow(clippy::indexing_slicing)]
if $undecomposed_starter.starter_and_decomposes_to_self() {
// Don't bother including `undecomposed_starter` in a contiguous buffer
// write: Just write it right away:
$sink.write_char($undecomposed_starter.character)?;
let $pending_slice = $decomposition.delegate.$as_slice();
$fast
}
let starter = $decomposition.decomposing_next($undecomposed_starter);
$sink.write_char(starter)?;
}
}
};
}
macro_rules! normalizer_methods {
() => {
/// Normalize a string slice into a `String`.
pub fn normalize(&self, text: &str) -> String {
let mut ret = String::new();
ret.reserve(text.len());
let _ = self.normalize_to(text, &mut ret);
ret
}
/// Check whether a string slice is normalized.
pub fn is_normalized(&self, text: &str) -> bool {
let mut sink = IsNormalizedSinkStr::new(text);
if self.normalize_to(text, &mut sink).is_err() {
return false;
}
sink.finished()
}
/// Normalize a slice of potentially-invalid UTF-16 into a `Vec`.
///
/// Unpaired surrogates are mapped to the REPLACEMENT CHARACTER
/// before normalizing.
pub fn normalize_utf16(&self, text: &[u16]) -> Vec<u16> {
let mut ret = Vec::new();
let _ = self.normalize_utf16_to(text, &mut ret);
ret
}
/// Checks whether a slice of potentially-invalid UTF-16 is normalized.
///
/// Unpaired surrogates are treated as the REPLACEMENT CHARACTER.
pub fn is_normalized_utf16(&self, text: &[u16]) -> bool {
let mut sink = IsNormalizedSinkUtf16::new(text);
if self.normalize_utf16_to(text, &mut sink).is_err() {
return false;
}
sink.finished()
}
/// Normalize a slice of potentially-invalid UTF-8 into a `String`.
///
/// Ill-formed byte sequences are mapped to the REPLACEMENT CHARACTER
/// according to the WHATWG Encoding Standard.
pub fn normalize_utf8(&self, text: &[u8]) -> String {
let mut ret = String::new();
ret.reserve(text.len());
let _ = self.normalize_utf8_to(text, &mut ret);
ret
}
/// Check if a slice of potentially-invalid UTF-8 is normalized.
///
/// Ill-formed byte sequences are mapped to the REPLACEMENT CHARACTER
/// according to the WHATWG Encoding Standard before checking.
pub fn is_normalized_utf8(&self, text: &[u8]) -> bool {
let mut sink = IsNormalizedSinkUtf8::new(text);
if self.normalize_utf8_to(text, &mut sink).is_err() {
return false;
}
sink.finished()
}
};
}
/// A normalizer for performing decomposing normalization.
#[derive(Debug)]
pub struct DecomposingNormalizer {
decompositions: DataPayload<CanonicalDecompositionDataV1Marker>,
supplementary_decompositions: Option<SupplementPayloadHolder>,
tables: DataPayload<CanonicalDecompositionTablesV1Marker>,
supplementary_tables: Option<DataPayload<CompatibilityDecompositionTablesV1Marker>>,
decomposition_passthrough_bound: u8, // never above 0xC0
composition_passthrough_bound: u16, // never above 0x0300
}
impl DecomposingNormalizer {
/// NFD constructor using compiled data.
///
/// ✨ *Enabled with the `compiled_data` Cargo feature.*
///
/// [📚 Help choosing a constructor](icu_provider::constructors)
#[cfg(feature = "compiled_data")]
pub const fn new_nfd() -> Self {
const _: () = assert!(
crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1
.scalars16
.const_len()
+ crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1
.scalars24
.const_len()
<= 0xFFF,
"NormalizerError::FutureExtension"
);
DecomposingNormalizer {
decompositions: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFD_V1,
),
supplementary_decompositions: None,
tables: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1,
),
supplementary_tables: None,
decomposition_passthrough_bound: 0xC0,
composition_passthrough_bound: 0x0300,
}
}
icu_provider::gen_any_buffer_data_constructors!(
locale: skip,
options: skip,
error: NormalizerError,
#[cfg(skip)]
functions: [
new_nfd,
try_new_nfd_with_any_provider,
try_new_nfd_with_buffer_provider,
try_new_nfd_unstable,
Self,
]
);
#[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_nfd)]
pub fn try_new_nfd_unstable<D>(provider: &D) -> Result<Self, NormalizerError>
where
D: DataProvider<CanonicalDecompositionDataV1Marker>
+ DataProvider<CanonicalDecompositionTablesV1Marker>
+ ?Sized,
{
let decompositions: DataPayload<CanonicalDecompositionDataV1Marker> =
provider.load(Default::default())?.take_payload()?;
let tables: DataPayload<CanonicalDecompositionTablesV1Marker> =
provider.load(Default::default())?.take_payload()?;
if tables.get().scalars16.len() + tables.get().scalars24.len() > 0xFFF {
// The data is from a future where there exists a normalization flavor whose
// complex decompositions take more than 0xFFF but fewer than 0x1FFF code points
// of space. If a good use case from such a decomposition flavor arises, we can
// dynamically change the bit masks so that the length mask becomes 0x1FFF instead
// of 0xFFF and the all-non-starters mask becomes 0 instead of 0x1000. However,
// since for now the masks are hard-coded, error out.
return Err(NormalizerError::FutureExtension);
}
Ok(DecomposingNormalizer {
decompositions,
supplementary_decompositions: None,
tables,
supplementary_tables: None,
decomposition_passthrough_bound: 0xC0,
composition_passthrough_bound: 0x0300,
})
}
/// NFKD constructor using compiled data.
///
/// ✨ *Enabled with the `compiled_data` Cargo feature.*
///
/// [📚 Help choosing a constructor](icu_provider::constructors)
#[cfg(feature = "compiled_data")]
pub const fn new_nfkd() -> Self {
const _: () = assert!(
crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1
.scalars16
.const_len()
+ crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1
.scalars24
.const_len()
+ crate::provider::Baked::SINGLETON_NORMALIZER_NFKDEX_V1
.scalars16
.const_len()
+ crate::provider::Baked::SINGLETON_NORMALIZER_NFKDEX_V1
.scalars24
.const_len()
<= 0xFFF,
"NormalizerError::FutureExtension"
);
const _: () = assert!(
crate::provider::Baked::SINGLETON_NORMALIZER_NFKD_V1.passthrough_cap <= 0x0300,
"NormalizerError::ValidationError"
);
let decomposition_capped =
if crate::provider::Baked::SINGLETON_NORMALIZER_NFKD_V1.passthrough_cap < 0xC0 {
crate::provider::Baked::SINGLETON_NORMALIZER_NFKD_V1.passthrough_cap
} else {
0xC0
};
let composition_capped =
if crate::provider::Baked::SINGLETON_NORMALIZER_NFKD_V1.passthrough_cap < 0x0300 {
crate::provider::Baked::SINGLETON_NORMALIZER_NFKD_V1.passthrough_cap
} else {
0x0300
};
DecomposingNormalizer {
decompositions: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFD_V1,
),
supplementary_decompositions: Some(SupplementPayloadHolder::Compatibility(
DataPayload::from_static_ref(crate::provider::Baked::SINGLETON_NORMALIZER_NFKD_V1),
)),
tables: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1,
),
supplementary_tables: Some(DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFKDEX_V1,
)),
decomposition_passthrough_bound: decomposition_capped as u8,
composition_passthrough_bound: composition_capped,
}
}
icu_provider::gen_any_buffer_data_constructors!(
locale: skip,
options: skip,
error: NormalizerError,
#[cfg(skip)]
functions: [
new_nfkd,
try_new_nfkd_with_any_provider,
try_new_nfkd_with_buffer_provider,
try_new_nfkd_unstable,
Self,
]
);
#[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_nfkd)]
pub fn try_new_nfkd_unstable<D>(provider: &D) -> Result<Self, NormalizerError>
where
D: DataProvider<CanonicalDecompositionDataV1Marker>
+ DataProvider<CompatibilityDecompositionSupplementV1Marker>
+ DataProvider<CanonicalDecompositionTablesV1Marker>
+ DataProvider<CompatibilityDecompositionTablesV1Marker>
+ ?Sized,
{
let decompositions: DataPayload<CanonicalDecompositionDataV1Marker> =
provider.load(Default::default())?.take_payload()?;
let supplementary_decompositions: DataPayload<
CompatibilityDecompositionSupplementV1Marker,
> = provider.load(Default::default())?.take_payload()?;
let tables: DataPayload<CanonicalDecompositionTablesV1Marker> =
provider.load(Default::default())?.take_payload()?;
let supplementary_tables: DataPayload<CompatibilityDecompositionTablesV1Marker> =
provider.load(Default::default())?.take_payload()?;
if tables.get().scalars16.len()
+ tables.get().scalars24.len()
+ supplementary_tables.get().scalars16.len()
+ supplementary_tables.get().scalars24.len()
> 0xFFF
{
// The data is from a future where there exists a normalization flavor whose
// complex decompositions take more than 0xFFF but fewer than 0x1FFF code points
// of space. If a good use case from such a decomposition flavor arises, we can
// dynamically change the bit masks so that the length mask becomes 0x1FFF instead
// of 0xFFF and the all-non-starters mask becomes 0 instead of 0x1000. However,
// since for now the masks are hard-coded, error out.
return Err(NormalizerError::FutureExtension);
}
let cap = supplementary_decompositions.get().passthrough_cap;
if cap > 0x0300 {
return Err(NormalizerError::ValidationError);
}
let decomposition_capped = cap.min(0xC0);
let composition_capped = cap.min(0x0300);
Ok(DecomposingNormalizer {
decompositions,
supplementary_decompositions: Some(SupplementPayloadHolder::Compatibility(
supplementary_decompositions,
)),
tables,
supplementary_tables: Some(supplementary_tables),
decomposition_passthrough_bound: decomposition_capped as u8,
composition_passthrough_bound: composition_capped,
})
}
#[doc(hidden)]
#[cfg(feature = "compiled_data")]
pub(crate) const fn new_uts46_decomposed() -> Self {
const _: () = assert!(
crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1
.scalars16
.const_len()
+ crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1
.scalars24
.const_len()
+ crate::provider::Baked::SINGLETON_NORMALIZER_NFKDEX_V1
.scalars16
.const_len()
+ crate::provider::Baked::SINGLETON_NORMALIZER_NFKDEX_V1
.scalars24
.const_len()
<= 0xFFF,
"NormalizerError::FutureExtension"
);
const _: () = assert!(
crate::provider::Baked::SINGLETON_NORMALIZER_UTS46D_V1.passthrough_cap <= 0x0300,
"NormalizerError::ValidationError"
);
let decomposition_capped =
if crate::provider::Baked::SINGLETON_NORMALIZER_UTS46D_V1.passthrough_cap < 0xC0 {
crate::provider::Baked::SINGLETON_NORMALIZER_UTS46D_V1.passthrough_cap
} else {
0xC0
};
let composition_capped =
if crate::provider::Baked::SINGLETON_NORMALIZER_UTS46D_V1.passthrough_cap < 0x0300 {
crate::provider::Baked::SINGLETON_NORMALIZER_UTS46D_V1.passthrough_cap
} else {
0x0300
};
DecomposingNormalizer {
decompositions: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFD_V1,
),
supplementary_decompositions: Some(SupplementPayloadHolder::Uts46(
DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_UTS46D_V1,
),
)),
tables: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFDEX_V1,
),
supplementary_tables: Some(DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_NFKDEX_V1,
)),
decomposition_passthrough_bound: decomposition_capped as u8,
composition_passthrough_bound: composition_capped,
}
}
/// UTS 46 decomposed constructor (testing only)
///
/// This is a special building block normalization for IDNA. It is the decomposed counterpart of
/// ICU4C's UTS 46 normalization with two exceptions: characters that UTS 46 disallows and
/// ICU4C maps to U+FFFD and characters that UTS 46 maps to the empty string normalize as in
/// NFD in this normalization. In both cases, the previous UTS 46 processing before using
/// normalization is expected to deal with these characters. Making the disallowed characters
/// behave like this is beneficial to data size, and this normalizer implementation cannot
/// deal with a character normalizing to the empty string, which doesn't happen in NFD or
/// NFKD as of Unicode 14.
///
/// Warning: In this normalization, U+0345 COMBINING GREEK YPOGEGRAMMENI exhibits a behavior
/// that no character in Unicode exhibits in NFD, NFKD, NFC, or NFKC: Case folding turns
/// U+0345 from a reordered character into a non-reordered character before reordering happens.
/// Therefore, the output of this normalization may differ for different inputs that are
/// canonically equivalent with each other if they differ by how U+0345 is ordered relative
/// to other reorderable characters.
///
/// Public for testing only.
#[doc(hidden)]
pub(crate) fn try_new_uts46_decomposed_unstable<D>(
provider: &D,
) -> Result<Self, NormalizerError>
where
D: DataProvider<CanonicalDecompositionDataV1Marker>
+ DataProvider<Uts46DecompositionSupplementV1Marker>
+ DataProvider<CanonicalDecompositionTablesV1Marker>
+ DataProvider<CompatibilityDecompositionTablesV1Marker>
// UTS 46 tables merged into CompatibilityDecompositionTablesV1Marker
+ ?Sized,
{
let decompositions: DataPayload<CanonicalDecompositionDataV1Marker> =
provider.load(Default::default())?.take_payload()?;
let supplementary_decompositions: DataPayload<Uts46DecompositionSupplementV1Marker> =
provider.load(Default::default())?.take_payload()?;
let tables: DataPayload<CanonicalDecompositionTablesV1Marker> =
provider.load(Default::default())?.take_payload()?;
let supplementary_tables: DataPayload<CompatibilityDecompositionTablesV1Marker> =
provider.load(Default::default())?.take_payload()?;
if tables.get().scalars16.len()
+ tables.get().scalars24.len()
+ supplementary_tables.get().scalars16.len()
+ supplementary_tables.get().scalars24.len()
> 0xFFF
{
// The data is from a future where there exists a normalization flavor whose
// complex decompositions take more than 0xFFF but fewer than 0x1FFF code points
// of space. If a good use case from such a decomposition flavor arises, we can
// dynamically change the bit masks so that the length mask becomes 0x1FFF instead
// of 0xFFF and the all-non-starters mask becomes 0 instead of 0x1000. However,
// since for now the masks are hard-coded, error out.
return Err(NormalizerError::FutureExtension);
}
let cap = supplementary_decompositions.get().passthrough_cap;
if cap > 0x0300 {
return Err(NormalizerError::ValidationError);
}
let decomposition_capped = cap.min(0xC0);
let composition_capped = cap.min(0x0300);
Ok(DecomposingNormalizer {
decompositions,
supplementary_decompositions: Some(SupplementPayloadHolder::Uts46(
supplementary_decompositions,
)),
tables,
supplementary_tables: Some(supplementary_tables),
decomposition_passthrough_bound: decomposition_capped as u8,
composition_passthrough_bound: composition_capped,
})
}
/// Wraps a delegate iterator into a decomposing iterator
/// adapter by using the data already held by this normalizer.
pub fn normalize_iter<I: Iterator<Item = char>>(&self, iter: I) -> Decomposition<I> {
Decomposition::new_with_supplements(
iter,
self.decompositions.get(),
self.supplementary_decompositions.as_ref().map(|s| s.get()),
self.tables.get(),
self.supplementary_tables.as_ref().map(|s| s.get()),
self.decomposition_passthrough_bound,
IgnorableBehavior::Unsupported,
)
}
normalizer_methods!();
decomposing_normalize_to!(
/// Normalize a string slice into a `Write` sink.
,
normalize_to,
core::fmt::Write,
&str,
{
},
as_str,
{
let decomposition_passthrough_byte_bound = if decomposition_passthrough_bound == 0xC0 {
0xC3u8
} else {
decomposition_passthrough_bound.min(0x80) as u8
};
// The attribute belongs on an inner statement, but Rust doesn't allow it there.
#[allow(clippy::unwrap_used)]
'fast: loop {
let mut code_unit_iter = decomposition.delegate.as_str().as_bytes().iter();
'fastest: loop {
if let Some(&upcoming_byte) = code_unit_iter.next() {
if upcoming_byte < decomposition_passthrough_byte_bound {
// Fast-track succeeded!
continue 'fastest;
}
decomposition.delegate = pending_slice[pending_slice.len() - code_unit_iter.as_slice().len() - 1..].chars();
break 'fastest;
}
// End of stream
sink.write_str(pending_slice)?;
return Ok(());
}
// `unwrap()` OK, because the slice is valid UTF-8 and we know there
// is an upcoming byte.
let upcoming = decomposition.delegate.next().unwrap();
let upcoming_with_trie_value = decomposition.attach_trie_value(upcoming);
if upcoming_with_trie_value.starter_and_decomposes_to_self() {
continue 'fast;
}
let consumed_so_far_slice = &pending_slice[..pending_slice.len()
- decomposition.delegate.as_str().len()
- upcoming.len_utf8()];
sink.write_str(consumed_so_far_slice)?;
// Now let's figure out if we got a starter or a non-starter.
if decomposition_starts_with_non_starter(
upcoming_with_trie_value.trie_val,
) {
// Let this trie value to be reprocessed in case it is
// one of the rare decomposing ones.
decomposition.pending = Some(upcoming_with_trie_value);
decomposition.gather_and_sort_combining(0);
continue 'outer;
}
undecomposed_starter = upcoming_with_trie_value;
debug_assert!(decomposition.pending.is_none());
break 'fast;
}
},
text,
sink,
decomposition,
decomposition_passthrough_bound,
undecomposed_starter,
pending_slice,
'outer,
);
decomposing_normalize_to!(
/// Normalize a slice of potentially-invalid UTF-8 into a `Write` sink.
///
/// Ill-formed byte sequences are mapped to the REPLACEMENT CHARACTER
/// according to the WHATWG Encoding Standard.
,
normalize_utf8_to,
core::fmt::Write,
&[u8],
{
},
as_slice,
{
let decomposition_passthrough_byte_bound = decomposition_passthrough_bound.min(0x80) as u8;
// The attribute belongs on an inner statement, but Rust doesn't allow it there.
#[allow(clippy::unwrap_used)]
'fast: loop {
let mut code_unit_iter = decomposition.delegate.as_slice().iter();
'fastest: loop {
if let Some(&upcoming_byte) = code_unit_iter.next() {
if upcoming_byte < decomposition_passthrough_byte_bound {
// Fast-track succeeded!
continue 'fastest;
}
break 'fastest;
}
// End of stream
sink.write_str(unsafe { from_utf8_unchecked(pending_slice) })?;
return Ok(());
}
decomposition.delegate = pending_slice[pending_slice.len() - code_unit_iter.as_slice().len() - 1..].chars();
// `unwrap()` OK, because the slice is valid UTF-8 and we know there
// is an upcoming byte.
let upcoming = decomposition.delegate.next().unwrap();
let upcoming_with_trie_value = decomposition.attach_trie_value(upcoming);
if upcoming_with_trie_value.starter_and_decomposes_to_self() {
if upcoming != REPLACEMENT_CHARACTER {
continue 'fast;
}
// We might have an error, so fall out of the fast path.
// Since the U+FFFD might signify an error, we can't
// assume `upcoming.len_utf8()` for the backoff length.
let mut consumed_so_far = pending_slice[..pending_slice.len() - decomposition.delegate.as_slice().len()].chars();
let back = consumed_so_far.next_back();
debug_assert_eq!(back, Some(REPLACEMENT_CHARACTER));
let consumed_so_far_slice = consumed_so_far.as_slice();
sink.write_str(unsafe{from_utf8_unchecked(consumed_so_far_slice)})?;
// We could call `gather_and_sort_combining` here and
// `continue 'outer`, but this should be better for code
// size.
undecomposed_starter = upcoming_with_trie_value;
debug_assert!(decomposition.pending.is_none());
break 'fast;
}
let consumed_so_far_slice = &pending_slice[..pending_slice.len()
- decomposition.delegate.as_slice().len()
- upcoming.len_utf8()];
sink.write_str(unsafe{from_utf8_unchecked(consumed_so_far_slice)})?;
// Now let's figure out if we got a starter or a non-starter.
if decomposition_starts_with_non_starter(
upcoming_with_trie_value.trie_val,
) {
// Let this trie value to be reprocessed in case it is
// one of the rare decomposing ones.
decomposition.pending = Some(upcoming_with_trie_value);
decomposition.gather_and_sort_combining(0);
continue 'outer;
}
undecomposed_starter = upcoming_with_trie_value;
debug_assert!(decomposition.pending.is_none());
break 'fast;
}
},
text,
sink,
decomposition,
decomposition_passthrough_bound,
undecomposed_starter,
pending_slice,
'outer,
);
decomposing_normalize_to!(
/// Normalize a slice of potentially-invalid UTF-16 into a `Write16` sink.
///
/// Unpaired surrogates are mapped to the REPLACEMENT CHARACTER
/// before normalizing.
,
normalize_utf16_to,
write16::Write16,
&[u16],
{
sink.size_hint(text.len())?;
},
as_slice,
{
let mut code_unit_iter = decomposition.delegate.as_slice().iter();
// The purpose of the counter is to flush once in a while. If we flush
// too much, there is too much flushing overhead. If we flush too rarely,
// the flush starts reading from too far behind compared to the hot
// recently-read memory.
let mut counter = UTF16_FAST_PATH_FLUSH_THRESHOLD;
'fast: loop {
counter -= 1;
if let Some(&upcoming_code_unit) = code_unit_iter.next() {
let mut upcoming32 = u32::from(upcoming_code_unit);
if upcoming32 < decomposition_passthrough_bound && counter != 0 {
continue 'fast;
}
// The loop is only broken out of as goto forward
#[allow(clippy::never_loop)]
'surrogateloop: loop {
let surrogate_base = upcoming32.wrapping_sub(0xD800);
if surrogate_base > (0xDFFF - 0xD800) {
// Not surrogate
break 'surrogateloop;
}
if surrogate_base <= (0xDBFF - 0xD800) {
let iter_backup = code_unit_iter.clone();
if let Some(&low) = code_unit_iter.next() {
if in_inclusive_range16(low, 0xDC00, 0xDFFF) {
upcoming32 = (upcoming32 << 10) + u32::from(low)
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32);
break 'surrogateloop;
} else {
code_unit_iter = iter_backup;
}
}
}
// unpaired surrogate
let slice_to_write = &pending_slice
[..pending_slice.len() - code_unit_iter.as_slice().len() - 1];
sink.write_slice(slice_to_write)?;
undecomposed_starter =
CharacterAndTrieValue::new(REPLACEMENT_CHARACTER, 0);
debug_assert!(decomposition.pending.is_none());
// We could instead call `gather_and_sort_combining` and `continue 'outer`, but
// assuming this is better for code size.
break 'fast;
}
// Not unpaired surrogate
let upcoming = unsafe { char::from_u32_unchecked(upcoming32) };
let upcoming_with_trie_value =
decomposition.attach_trie_value(upcoming);
if upcoming_with_trie_value.starter_and_decomposes_to_self() && counter != 0 {
continue 'fast;
}
let consumed_so_far_slice = &pending_slice[..pending_slice.len()
- code_unit_iter.as_slice().len()
- upcoming.len_utf16()];
sink.write_slice(consumed_so_far_slice)?;
// Now let's figure out if we got a starter or a non-starter.
if decomposition_starts_with_non_starter(
upcoming_with_trie_value.trie_val,
) {
// Sync with main iterator
decomposition.delegate = code_unit_iter.as_slice().chars();
// Let this trie value to be reprocessed in case it is
// one of the rare decomposing ones.
decomposition.pending = Some(upcoming_with_trie_value);
decomposition.gather_and_sort_combining(0);
continue 'outer;
}
undecomposed_starter = upcoming_with_trie_value;
debug_assert!(decomposition.pending.is_none());
break 'fast;
}
// End of stream
sink.write_slice(pending_slice)?;
return Ok(());
}
// Sync the main iterator
decomposition.delegate = code_unit_iter.as_slice().chars();
},
text,
sink,
decomposition,
decomposition_passthrough_bound,
undecomposed_starter,
pending_slice,
'outer,
);
}
/// A normalizer for performing composing normalization.
#[derive(Debug)]
pub struct ComposingNormalizer {
decomposing_normalizer: DecomposingNormalizer,
canonical_compositions: DataPayload<CanonicalCompositionsV1Marker>,
}
impl ComposingNormalizer {
/// NFC constructor using compiled data.
///
/// ✨ *Enabled with the `compiled_data` Cargo feature.*
///
/// [📚 Help choosing a constructor](icu_provider::constructors)
#[cfg(feature = "compiled_data")]
pub const fn new_nfc() -> Self {
ComposingNormalizer {
decomposing_normalizer: DecomposingNormalizer::new_nfd(),
canonical_compositions: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_COMP_V1,
),
}
}
icu_provider::gen_any_buffer_data_constructors!(
locale: skip,
options: skip,
error: NormalizerError,
#[cfg(skip)]
functions: [
new_nfc,
try_new_nfc_with_any_provider,
try_new_nfc_with_buffer_provider,
try_new_nfc_unstable,
Self,
]
);
#[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_nfc)]
pub fn try_new_nfc_unstable<D>(provider: &D) -> Result<Self, NormalizerError>
where
D: DataProvider<CanonicalDecompositionDataV1Marker>
+ DataProvider<CanonicalDecompositionTablesV1Marker>
+ DataProvider<CanonicalCompositionsV1Marker>
+ ?Sized,
{
let decomposing_normalizer = DecomposingNormalizer::try_new_nfd_unstable(provider)?;
let canonical_compositions: DataPayload<CanonicalCompositionsV1Marker> =
provider.load(Default::default())?.take_payload()?;
Ok(ComposingNormalizer {
decomposing_normalizer,
canonical_compositions,
})
}
/// NFKC constructor using compiled data.
///
/// ✨ *Enabled with the `compiled_data` Cargo feature.*
///
/// [📚 Help choosing a constructor](icu_provider::constructors)
#[cfg(feature = "compiled_data")]
pub const fn new_nfkc() -> Self {
ComposingNormalizer {
decomposing_normalizer: DecomposingNormalizer::new_nfkd(),
canonical_compositions: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_COMP_V1,
),
}
}
icu_provider::gen_any_buffer_data_constructors!(
locale: skip,
options: skip,
error: NormalizerError,
#[cfg(skip)]
functions: [
new_nfkc,
try_new_nfkc_with_any_provider,
try_new_nfkc_with_buffer_provider,
try_new_nfkc_unstable,
Self,
]
);
#[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_nfkc)]
pub fn try_new_nfkc_unstable<D>(provider: &D) -> Result<Self, NormalizerError>
where
D: DataProvider<CanonicalDecompositionDataV1Marker>
+ DataProvider<CompatibilityDecompositionSupplementV1Marker>
+ DataProvider<CanonicalDecompositionTablesV1Marker>
+ DataProvider<CompatibilityDecompositionTablesV1Marker>
+ DataProvider<CanonicalCompositionsV1Marker>
+ ?Sized,
{
let decomposing_normalizer = DecomposingNormalizer::try_new_nfkd_unstable(provider)?;
let canonical_compositions: DataPayload<CanonicalCompositionsV1Marker> =
provider.load(Default::default())?.take_payload()?;
Ok(ComposingNormalizer {
decomposing_normalizer,
canonical_compositions,
})
}
/// This is a special building block normalization for IDNA that implements parts of the Map
/// step and the following Normalize step.
///
/// Warning: In this normalization, U+0345 COMBINING GREEK YPOGEGRAMMENI exhibits a behavior
/// that no character in Unicode exhibits in NFD, NFKD, NFC, or NFKC: Case folding turns
/// U+0345 from a reordered character into a non-reordered character before reordering happens.
/// Therefore, the output of this normalization may differ for different inputs that are
/// canonically equivalents with each other if they differ by how U+0345 is ordered relative
/// to other reorderable characters.
#[cfg(feature = "compiled_data")]
pub(crate) const fn new_uts46() -> Self {
ComposingNormalizer {
decomposing_normalizer: DecomposingNormalizer::new_uts46_decomposed(),
canonical_compositions: DataPayload::from_static_ref(
crate::provider::Baked::SINGLETON_NORMALIZER_COMP_V1,
),
}
}
#[doc = icu_provider::gen_any_buffer_unstable_docs!(UNSTABLE, Self::new_uts46)]
pub(crate) fn try_new_uts46_unstable<D>(provider: &D) -> Result<Self, NormalizerError>
where
D: DataProvider<CanonicalDecompositionDataV1Marker>
+ DataProvider<Uts46DecompositionSupplementV1Marker>
+ DataProvider<CanonicalDecompositionTablesV1Marker>
+ DataProvider<CompatibilityDecompositionTablesV1Marker>
// UTS 46 tables merged into CompatibilityDecompositionTablesV1Marker
+ DataProvider<CanonicalCompositionsV1Marker>
+ ?Sized,
{
let decomposing_normalizer =
DecomposingNormalizer::try_new_uts46_decomposed_unstable(provider)?;
let canonical_compositions: DataPayload<CanonicalCompositionsV1Marker> =
provider.load(Default::default())?.take_payload()?;
Ok(ComposingNormalizer {
decomposing_normalizer,
canonical_compositions,
})
}
/// Wraps a delegate iterator into a composing iterator
/// adapter by using the data already held by this normalizer.
pub fn normalize_iter<I: Iterator<Item = char>>(&self, iter: I) -> Composition<I> {
self.normalize_iter_private(iter, IgnorableBehavior::Unsupported)
}
fn normalize_iter_private<I: Iterator<Item = char>>(
&self,
iter: I,
ignorable_behavior: IgnorableBehavior,
) -> Composition<I> {
Composition::new(
Decomposition::new_with_supplements(
iter,
self.decomposing_normalizer.decompositions.get(),
self.decomposing_normalizer
.supplementary_decompositions
.as_ref()
.map(|s| s.get()),
self.decomposing_normalizer.tables.get(),
self.decomposing_normalizer
.supplementary_tables
.as_ref()
.map(|s| s.get()),
self.decomposing_normalizer.decomposition_passthrough_bound,
ignorable_behavior,
),
ZeroFrom::zero_from(&self.canonical_compositions.get().canonical_compositions),
self.decomposing_normalizer.composition_passthrough_bound,
)
}
normalizer_methods!();
composing_normalize_to!(
/// Normalize a string slice into a `Write` sink.
,
normalize_to,
core::fmt::Write,
&str,
{},
true,
as_str,
{
// Let's hope LICM hoists this outside `'outer`.
let composition_passthrough_byte_bound = if composition_passthrough_bound == 0x300 {
0xCCu8
} else {
// We can make this fancy if a normalization other than NFC where looking at
// non-ASCII lead bytes is worthwhile is ever introduced.
composition_passthrough_bound.min(0x80) as u8
};
// This is basically an `Option` discriminant for `undecomposed_starter`,
// but making it a boolean so that writes in the tightest loop are as
// simple as possible (and potentially as peel-hoistable as possible).
// Furthermore, this reduces `unwrap()` later.
let mut undecomposed_starter_valid = true;
// Annotation belongs really on inner statements, but Rust doesn't
// allow it there.
#[allow(clippy::unwrap_used)]
'fast: loop {
let mut code_unit_iter = composition.decomposition.delegate.as_str().as_bytes().iter();
'fastest: loop {
if let Some(&upcoming_byte) = code_unit_iter.next() {
if upcoming_byte < composition_passthrough_byte_bound {
// Fast-track succeeded!
undecomposed_starter_valid = false;
continue 'fastest;
}
composition.decomposition.delegate = pending_slice[pending_slice.len() - code_unit_iter.as_slice().len() - 1..].chars();
break 'fastest;
}
// End of stream
sink.write_str(pending_slice)?;
return Ok(());
}
// `unwrap()` OK, because the slice is valid UTF-8 and we know there
// is an upcoming byte.
let upcoming = composition.decomposition.delegate.next().unwrap();
let upcoming_with_trie_value = composition.decomposition.attach_trie_value(upcoming);
if upcoming_with_trie_value.potential_passthrough_and_cannot_combine_backwards() {
// Can't combine backwards, hence a plain (non-backwards-combining)
// starter albeit past `composition_passthrough_bound`
// Fast-track succeeded!
undecomposed_starter = upcoming_with_trie_value;
undecomposed_starter_valid = true;
continue 'fast;
}
// We need to fall off the fast path.
composition.decomposition.pending = Some(upcoming_with_trie_value);
let consumed_so_far_slice = if undecomposed_starter_valid {
&pending_slice[..pending_slice.len() - composition.decomposition.delegate.as_str().len() - upcoming.len_utf8() - undecomposed_starter.character.len_utf8()]
} else {
// slicing and unwrap OK, because we've just evidently read enough previously.
let mut consumed_so_far = pending_slice[..pending_slice.len() - composition.decomposition.delegate.as_str().len() - upcoming.len_utf8()].chars();
// `unwrap` OK, because we've previously manage to read the previous character
undecomposed_starter = composition.decomposition.attach_trie_value(consumed_so_far.next_back().unwrap());
undecomposed_starter_valid = true;
consumed_so_far.as_str()
};
sink.write_str(consumed_so_far_slice)?;
break 'fast;
}
debug_assert!(undecomposed_starter_valid);
},
text,
sink,
composition,
composition_passthrough_bound,
undecomposed_starter,
pending_slice,
len_utf8,
);
composing_normalize_to!(
/// Normalize a slice of potentially-invalid UTF-8 into a `Write` sink.
///
/// Ill-formed byte sequences are mapped to the REPLACEMENT CHARACTER
/// according to the WHATWG Encoding Standard.
,
normalize_utf8_to,
core::fmt::Write,
&[u8],
{},
false,
as_slice,
{
// This is basically an `Option` discriminant for `undecomposed_starter`,
// but making it a boolean so that writes in the tightest loop are as
// simple as possible (and potentially as peel-hoistable as possible).
// Furthermore, this reduces `unwrap()` later.
let mut undecomposed_starter_valid = true;
'fast: loop {
if let Some(upcoming) = composition.decomposition.delegate.next() {
if u32::from(upcoming) < composition_passthrough_bound {
// Fast-track succeeded!
undecomposed_starter_valid = false;
continue 'fast;
}
// TODO(#2006): Annotate as unlikely
if upcoming == REPLACEMENT_CHARACTER {
// Can't tell if this is an error or a literal U+FFFD in
// the input. Assuming the former to be sure.
// Since the U+FFFD might signify an error, we can't
// assume `upcoming.len_utf8()` for the backoff length.
let mut consumed_so_far = pending_slice[..pending_slice.len() - composition.decomposition.delegate.as_slice().len()].chars();
let back = consumed_so_far.next_back();
debug_assert_eq!(back, Some(REPLACEMENT_CHARACTER));
let consumed_so_far_slice = consumed_so_far.as_slice();
sink.write_str(unsafe{ from_utf8_unchecked(consumed_so_far_slice)})?;
undecomposed_starter = CharacterAndTrieValue::new(REPLACEMENT_CHARACTER, 0);
undecomposed_starter_valid = true;
composition.decomposition.pending = None;
break 'fast;
}
let upcoming_with_trie_value = composition.decomposition.attach_trie_value(upcoming);
if upcoming_with_trie_value.potential_passthrough_and_cannot_combine_backwards() {
// Can't combine backwards, hence a plain (non-backwards-combining)
// starter albeit past `composition_passthrough_bound`
// Fast-track succeeded!
undecomposed_starter = upcoming_with_trie_value;
undecomposed_starter_valid = true;
continue 'fast;
}
// We need to fall off the fast path.
composition.decomposition.pending = Some(upcoming_with_trie_value);
// Annotation belongs really on inner statement, but Rust doesn't
// allow it there.
#[allow(clippy::unwrap_used)]
let consumed_so_far_slice = if undecomposed_starter_valid {
&pending_slice[..pending_slice.len() - composition.decomposition.delegate.as_slice().len() - upcoming.len_utf8() - undecomposed_starter.character.len_utf8()]
} else {
// slicing and unwrap OK, because we've just evidently read enough previously.
let mut consumed_so_far = pending_slice[..pending_slice.len() - composition.decomposition.delegate.as_slice().len() - upcoming.len_utf8()].chars();
// `unwrap` OK, because we've previously manage to read the previous character
undecomposed_starter = composition.decomposition.attach_trie_value(consumed_so_far.next_back().unwrap());
undecomposed_starter_valid = true;
consumed_so_far.as_slice()
};
sink.write_str(unsafe { from_utf8_unchecked(consumed_so_far_slice)})?;
break 'fast;
}
// End of stream
sink.write_str(unsafe {from_utf8_unchecked(pending_slice) })?;
return Ok(());
}
debug_assert!(undecomposed_starter_valid);
},
text,
sink,
composition,
composition_passthrough_bound,
undecomposed_starter,
pending_slice,
len_utf8,
);
composing_normalize_to!(
/// Normalize a slice of potentially-invalid UTF-16 into a `Write16` sink.
///
/// Unpaired surrogates are mapped to the REPLACEMENT CHARACTER
/// before normalizing.
,
normalize_utf16_to,
write16::Write16,
&[u16],
{
sink.size_hint(text.len())?;
},
false,
as_slice,
{
let mut code_unit_iter = composition.decomposition.delegate.as_slice().iter();
let mut upcoming32;
// This is basically an `Option` discriminant for `undecomposed_starter`,
// but making it a boolean so that writes to it are are as
// simple as possible.
// Furthermore, this removes the need for `unwrap()` later.
let mut undecomposed_starter_valid;
// The purpose of the counter is to flush once in a while. If we flush
// too much, there is too much flushing overhead. If we flush too rarely,
// the flush starts reading from too far behind compared to the hot
// recently-read memory.
let mut counter = UTF16_FAST_PATH_FLUSH_THRESHOLD;
// The purpose of this trickiness is to avoid writing to
// `undecomposed_starter_valid` from the tightest loop. Writing to it
// from there destroys performance.
let mut counter_reference = counter - 1;
'fast: loop {
counter -= 1;
if let Some(&upcoming_code_unit) = code_unit_iter.next() {
upcoming32 = u32::from(upcoming_code_unit); // may be surrogate
if upcoming32 < composition_passthrough_bound && counter != 0 {
// No need for surrogate or U+FFFD check, because
// `composition_passthrough_bound` cannot be higher than
// U+0300.
// Fast-track succeeded!
continue 'fast;
}
// if `counter` equals `counter_reference`, the `continue 'fast`
// line above has not executed and `undecomposed_starter` is still
// valid.
undecomposed_starter_valid = counter == counter_reference;
// The loop is only broken out of as goto forward
#[allow(clippy::never_loop)]
'surrogateloop: loop {
let surrogate_base = upcoming32.wrapping_sub(0xD800);
if surrogate_base > (0xDFFF - 0xD800) {
// Not surrogate
break 'surrogateloop;
}
if surrogate_base <= (0xDBFF - 0xD800) {
let iter_backup = code_unit_iter.clone();
if let Some(&low) = code_unit_iter.next() {
if in_inclusive_range16(low, 0xDC00, 0xDFFF) {
upcoming32 = (upcoming32 << 10) + u32::from(low)
- (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32);
break 'surrogateloop;
} else {
code_unit_iter = iter_backup;
}
}
}
// unpaired surrogate
let slice_to_write = &pending_slice[..pending_slice.len() - code_unit_iter.as_slice().len() - 1];
sink.write_slice(slice_to_write)?;
undecomposed_starter = CharacterAndTrieValue::new(REPLACEMENT_CHARACTER, 0);
undecomposed_starter_valid = true;
composition.decomposition.pending = None;
break 'fast;
}
// Not unpaired surrogate
let upcoming = unsafe { char::from_u32_unchecked(upcoming32) };
let upcoming_with_trie_value = composition.decomposition.attach_trie_value(upcoming);
if upcoming_with_trie_value.potential_passthrough_and_cannot_combine_backwards() && counter != 0 {
// Can't combine backwards, hence a plain (non-backwards-combining)
// starter albeit past `composition_passthrough_bound`
// Fast-track succeeded!
undecomposed_starter = upcoming_with_trie_value;
// Cause `undecomposed_starter_valid` to be set to true.
// This regresses English performance on Haswell by 11%
// compared to commenting out this assignment to
// `counter_reference`.
counter_reference = counter - 1;
continue 'fast;
}
// We need to fall off the fast path.
composition.decomposition.pending = Some(upcoming_with_trie_value);
// Annotation belongs really on inner statement, but Rust doesn't
// allow it there.
#[allow(clippy::unwrap_used)]
let consumed_so_far_slice = if undecomposed_starter_valid {
&pending_slice[..pending_slice.len() - code_unit_iter.as_slice().len() - upcoming.len_utf16() - undecomposed_starter.character.len_utf16()]
} else {
// slicing and unwrap OK, because we've just evidently read enough previously.
let mut consumed_so_far = pending_slice[..pending_slice.len() - code_unit_iter.as_slice().len() - upcoming.len_utf16()].chars();
// `unwrap` OK, because we've previously manage to read the previous character
undecomposed_starter = composition.decomposition.attach_trie_value(consumed_so_far.next_back().unwrap());
undecomposed_starter_valid = true;
consumed_so_far.as_slice()
};
sink.write_slice(consumed_so_far_slice)?;
break 'fast;
}
// End of stream
sink.write_slice(pending_slice)?;
return Ok(());
}
debug_assert!(undecomposed_starter_valid);
// Sync the main iterator
composition.decomposition.delegate = code_unit_iter.as_slice().chars();
},
text,
sink,
composition,
composition_passthrough_bound,
undecomposed_starter,
pending_slice,
len_utf16,
);
}
struct IsNormalizedSinkUtf16<'a> {
expect: &'a [u16],
}
impl<'a> IsNormalizedSinkUtf16<'a> {
pub fn new(slice: &'a [u16]) -> Self {
IsNormalizedSinkUtf16 { expect: slice }
}
pub fn finished(&self) -> bool {
self.expect.is_empty()
}
}
impl<'a> Write16 for IsNormalizedSinkUtf16<'a> {
fn write_slice(&mut self, s: &[u16]) -> core::fmt::Result {
// We know that if we get a slice, it's a pass-through,
// so we can compare addresses. Indexing is OK, because
// an indexing failure would be a code bug rather than
// an input or data issue.
#[allow(clippy::indexing_slicing)]
if s.as_ptr() == self.expect.as_ptr() {
self.expect = &self.expect[s.len()..];
Ok(())
} else {
Err(core::fmt::Error {})
}
}
fn write_char(&mut self, c: char) -> core::fmt::Result {
let mut iter = self.expect.chars();
if iter.next() == Some(c) {
self.expect = iter.as_slice();
Ok(())
} else {
Err(core::fmt::Error {})
}
}
}
struct IsNormalizedSinkUtf8<'a> {
expect: &'a [u8],
}
impl<'a> IsNormalizedSinkUtf8<'a> {
pub fn new(slice: &'a [u8]) -> Self {
IsNormalizedSinkUtf8 { expect: slice }
}
pub fn finished(&self) -> bool {
self.expect.is_empty()
}
}
impl<'a> core::fmt::Write for IsNormalizedSinkUtf8<'a> {
fn write_str(&mut self, s: &str) -> core::fmt::Result {
// We know that if we get a slice, it's a pass-through,
// so we can compare addresses. Indexing is OK, because
// an indexing failure would be a code bug rather than
// an input or data issue.
#[allow(clippy::indexing_slicing)]
if s.as_ptr() == self.expect.as_ptr() {
self.expect = &self.expect[s.len()..];
Ok(())
} else {
Err(core::fmt::Error {})
}
}
fn write_char(&mut self, c: char) -> core::fmt::Result {
let mut iter = self.expect.chars();
if iter.next() == Some(c) {
self.expect = iter.as_slice();
Ok(())
} else {
Err(core::fmt::Error {})
}
}
}
struct IsNormalizedSinkStr<'a> {
expect: &'a str,
}
impl<'a> IsNormalizedSinkStr<'a> {
pub fn new(slice: &'a str) -> Self {
IsNormalizedSinkStr { expect: slice }
}
pub fn finished(&self) -> bool {
self.expect.is_empty()
}
}
impl<'a> core::fmt::Write for IsNormalizedSinkStr<'a> {
fn write_str(&mut self, s: &str) -> core::fmt::Result {
// We know that if we get a slice, it's a pass-through,
// so we can compare addresses. Indexing is OK, because
// an indexing failure would be a code bug rather than
// an input or data issue.
#[allow(clippy::indexing_slicing)]
if s.as_ptr() == self.expect.as_ptr() {
self.expect = &self.expect[s.len()..];
Ok(())
} else {
Err(core::fmt::Error {})
}
}
fn write_char(&mut self, c: char) -> core::fmt::Result {
let mut iter = self.expect.chars();
if iter.next() == Some(c) {
self.expect = iter.as_str();
Ok(())
} else {
Err(core::fmt::Error {})
}
}
}