| 1 | "use strict";
|
|---|
| 2 | Object.defineProperty(exports, "__esModule", { value: true });
|
|---|
| 3 | exports.Trie = exports.createTrieFromBase64 = exports.UTRIE2_INDEX_2_MASK = exports.UTRIE2_INDEX_2_BLOCK_LENGTH = exports.UTRIE2_OMITTED_BMP_INDEX_1_LENGTH = exports.UTRIE2_INDEX_1_OFFSET = exports.UTRIE2_UTF8_2B_INDEX_2_LENGTH = exports.UTRIE2_UTF8_2B_INDEX_2_OFFSET = exports.UTRIE2_INDEX_2_BMP_LENGTH = exports.UTRIE2_LSCP_INDEX_2_LENGTH = exports.UTRIE2_DATA_MASK = exports.UTRIE2_DATA_BLOCK_LENGTH = exports.UTRIE2_LSCP_INDEX_2_OFFSET = exports.UTRIE2_SHIFT_1_2 = exports.UTRIE2_INDEX_SHIFT = exports.UTRIE2_SHIFT_1 = exports.UTRIE2_SHIFT_2 = void 0;
|
|---|
| 4 | var Util_1 = require("./Util");
|
|---|
| 5 | /** Shift size for getting the index-2 table offset. */
|
|---|
| 6 | exports.UTRIE2_SHIFT_2 = 5;
|
|---|
| 7 | /** Shift size for getting the index-1 table offset. */
|
|---|
| 8 | exports.UTRIE2_SHIFT_1 = 6 + 5;
|
|---|
| 9 | /**
|
|---|
| 10 | * Shift size for shifting left the index array values.
|
|---|
| 11 | * Increases possible data size with 16-bit index values at the cost
|
|---|
| 12 | * of compactability.
|
|---|
| 13 | * This requires data blocks to be aligned by UTRIE2_DATA_GRANULARITY.
|
|---|
| 14 | */
|
|---|
| 15 | exports.UTRIE2_INDEX_SHIFT = 2;
|
|---|
| 16 | /**
|
|---|
| 17 | * Difference between the two shift sizes,
|
|---|
| 18 | * for getting an index-1 offset from an index-2 offset. 6=11-5
|
|---|
| 19 | */
|
|---|
| 20 | exports.UTRIE2_SHIFT_1_2 = exports.UTRIE2_SHIFT_1 - exports.UTRIE2_SHIFT_2;
|
|---|
| 21 | /**
|
|---|
| 22 | * The part of the index-2 table for U+D800..U+DBFF stores values for
|
|---|
| 23 | * lead surrogate code _units_ not code _points_.
|
|---|
| 24 | * Values for lead surrogate code _points_ are indexed with this portion of the table.
|
|---|
| 25 | * Length=32=0x20=0x400>>UTRIE2_SHIFT_2. (There are 1024=0x400 lead surrogates.)
|
|---|
| 26 | */
|
|---|
| 27 | exports.UTRIE2_LSCP_INDEX_2_OFFSET = 0x10000 >> exports.UTRIE2_SHIFT_2;
|
|---|
| 28 | /** Number of entries in a data block. 32=0x20 */
|
|---|
| 29 | exports.UTRIE2_DATA_BLOCK_LENGTH = 1 << exports.UTRIE2_SHIFT_2;
|
|---|
| 30 | /** Mask for getting the lower bits for the in-data-block offset. */
|
|---|
| 31 | exports.UTRIE2_DATA_MASK = exports.UTRIE2_DATA_BLOCK_LENGTH - 1;
|
|---|
| 32 | exports.UTRIE2_LSCP_INDEX_2_LENGTH = 0x400 >> exports.UTRIE2_SHIFT_2;
|
|---|
| 33 | /** Count the lengths of both BMP pieces. 2080=0x820 */
|
|---|
| 34 | exports.UTRIE2_INDEX_2_BMP_LENGTH = exports.UTRIE2_LSCP_INDEX_2_OFFSET + exports.UTRIE2_LSCP_INDEX_2_LENGTH;
|
|---|
| 35 | /**
|
|---|
| 36 | * The 2-byte UTF-8 version of the index-2 table follows at offset 2080=0x820.
|
|---|
| 37 | * Length 32=0x20 for lead bytes C0..DF, regardless of UTRIE2_SHIFT_2.
|
|---|
| 38 | */
|
|---|
| 39 | exports.UTRIE2_UTF8_2B_INDEX_2_OFFSET = exports.UTRIE2_INDEX_2_BMP_LENGTH;
|
|---|
| 40 | exports.UTRIE2_UTF8_2B_INDEX_2_LENGTH = 0x800 >> 6; /* U+0800 is the first code point after 2-byte UTF-8 */
|
|---|
| 41 | /**
|
|---|
| 42 | * The index-1 table, only used for supplementary code points, at offset 2112=0x840.
|
|---|
| 43 | * Variable length, for code points up to highStart, where the last single-value range starts.
|
|---|
| 44 | * Maximum length 512=0x200=0x100000>>UTRIE2_SHIFT_1.
|
|---|
| 45 | * (For 0x100000 supplementary code points U+10000..U+10ffff.)
|
|---|
| 46 | *
|
|---|
| 47 | * The part of the index-2 table for supplementary code points starts
|
|---|
| 48 | * after this index-1 table.
|
|---|
| 49 | *
|
|---|
| 50 | * Both the index-1 table and the following part of the index-2 table
|
|---|
| 51 | * are omitted completely if there is only BMP data.
|
|---|
| 52 | */
|
|---|
| 53 | exports.UTRIE2_INDEX_1_OFFSET = exports.UTRIE2_UTF8_2B_INDEX_2_OFFSET + exports.UTRIE2_UTF8_2B_INDEX_2_LENGTH;
|
|---|
| 54 | /**
|
|---|
| 55 | * Number of index-1 entries for the BMP. 32=0x20
|
|---|
| 56 | * This part of the index-1 table is omitted from the serialized form.
|
|---|
| 57 | */
|
|---|
| 58 | exports.UTRIE2_OMITTED_BMP_INDEX_1_LENGTH = 0x10000 >> exports.UTRIE2_SHIFT_1;
|
|---|
| 59 | /** Number of entries in an index-2 block. 64=0x40 */
|
|---|
| 60 | exports.UTRIE2_INDEX_2_BLOCK_LENGTH = 1 << exports.UTRIE2_SHIFT_1_2;
|
|---|
| 61 | /** Mask for getting the lower bits for the in-index-2-block offset. */
|
|---|
| 62 | exports.UTRIE2_INDEX_2_MASK = exports.UTRIE2_INDEX_2_BLOCK_LENGTH - 1;
|
|---|
| 63 | var slice16 = function (view, start, end) {
|
|---|
| 64 | if (view.slice) {
|
|---|
| 65 | return view.slice(start, end);
|
|---|
| 66 | }
|
|---|
| 67 | return new Uint16Array(Array.prototype.slice.call(view, start, end));
|
|---|
| 68 | };
|
|---|
| 69 | var slice32 = function (view, start, end) {
|
|---|
| 70 | if (view.slice) {
|
|---|
| 71 | return view.slice(start, end);
|
|---|
| 72 | }
|
|---|
| 73 | return new Uint32Array(Array.prototype.slice.call(view, start, end));
|
|---|
| 74 | };
|
|---|
| 75 | var createTrieFromBase64 = function (base64, _byteLength) {
|
|---|
| 76 | var buffer = Util_1.decode(base64);
|
|---|
| 77 | var view32 = Array.isArray(buffer) ? Util_1.polyUint32Array(buffer) : new Uint32Array(buffer);
|
|---|
| 78 | var view16 = Array.isArray(buffer) ? Util_1.polyUint16Array(buffer) : new Uint16Array(buffer);
|
|---|
| 79 | var headerLength = 24;
|
|---|
| 80 | var index = slice16(view16, headerLength / 2, view32[4] / 2);
|
|---|
| 81 | var data = view32[5] === 2
|
|---|
| 82 | ? slice16(view16, (headerLength + view32[4]) / 2)
|
|---|
| 83 | : slice32(view32, Math.ceil((headerLength + view32[4]) / 4));
|
|---|
| 84 | return new Trie(view32[0], view32[1], view32[2], view32[3], index, data);
|
|---|
| 85 | };
|
|---|
| 86 | exports.createTrieFromBase64 = createTrieFromBase64;
|
|---|
| 87 | var Trie = /** @class */ (function () {
|
|---|
| 88 | function Trie(initialValue, errorValue, highStart, highValueIndex, index, data) {
|
|---|
| 89 | this.initialValue = initialValue;
|
|---|
| 90 | this.errorValue = errorValue;
|
|---|
| 91 | this.highStart = highStart;
|
|---|
| 92 | this.highValueIndex = highValueIndex;
|
|---|
| 93 | this.index = index;
|
|---|
| 94 | this.data = data;
|
|---|
| 95 | }
|
|---|
| 96 | /**
|
|---|
| 97 | * Get the value for a code point as stored in the Trie.
|
|---|
| 98 | *
|
|---|
| 99 | * @param codePoint the code point
|
|---|
| 100 | * @return the value
|
|---|
| 101 | */
|
|---|
| 102 | Trie.prototype.get = function (codePoint) {
|
|---|
| 103 | var ix;
|
|---|
| 104 | if (codePoint >= 0) {
|
|---|
| 105 | if (codePoint < 0x0d800 || (codePoint > 0x0dbff && codePoint <= 0x0ffff)) {
|
|---|
| 106 | // Ordinary BMP code point, excluding leading surrogates.
|
|---|
| 107 | // BMP uses a single level lookup. BMP index starts at offset 0 in the Trie2 index.
|
|---|
| 108 | // 16 bit data is stored in the index array itself.
|
|---|
| 109 | ix = this.index[codePoint >> exports.UTRIE2_SHIFT_2];
|
|---|
| 110 | ix = (ix << exports.UTRIE2_INDEX_SHIFT) + (codePoint & exports.UTRIE2_DATA_MASK);
|
|---|
| 111 | return this.data[ix];
|
|---|
| 112 | }
|
|---|
| 113 | if (codePoint <= 0xffff) {
|
|---|
| 114 | // Lead Surrogate Code Point. A Separate index section is stored for
|
|---|
| 115 | // lead surrogate code units and code points.
|
|---|
| 116 | // The main index has the code unit data.
|
|---|
| 117 | // For this function, we need the code point data.
|
|---|
| 118 | // Note: this expression could be refactored for slightly improved efficiency, but
|
|---|
| 119 | // surrogate code points will be so rare in practice that it's not worth it.
|
|---|
| 120 | ix = this.index[exports.UTRIE2_LSCP_INDEX_2_OFFSET + ((codePoint - 0xd800) >> exports.UTRIE2_SHIFT_2)];
|
|---|
| 121 | ix = (ix << exports.UTRIE2_INDEX_SHIFT) + (codePoint & exports.UTRIE2_DATA_MASK);
|
|---|
| 122 | return this.data[ix];
|
|---|
| 123 | }
|
|---|
| 124 | if (codePoint < this.highStart) {
|
|---|
| 125 | // Supplemental code point, use two-level lookup.
|
|---|
| 126 | ix = exports.UTRIE2_INDEX_1_OFFSET - exports.UTRIE2_OMITTED_BMP_INDEX_1_LENGTH + (codePoint >> exports.UTRIE2_SHIFT_1);
|
|---|
| 127 | ix = this.index[ix];
|
|---|
| 128 | ix += (codePoint >> exports.UTRIE2_SHIFT_2) & exports.UTRIE2_INDEX_2_MASK;
|
|---|
| 129 | ix = this.index[ix];
|
|---|
| 130 | ix = (ix << exports.UTRIE2_INDEX_SHIFT) + (codePoint & exports.UTRIE2_DATA_MASK);
|
|---|
| 131 | return this.data[ix];
|
|---|
| 132 | }
|
|---|
| 133 | if (codePoint <= 0x10ffff) {
|
|---|
| 134 | return this.data[this.highValueIndex];
|
|---|
| 135 | }
|
|---|
| 136 | }
|
|---|
| 137 | // Fall through. The code point is outside of the legal range of 0..0x10ffff.
|
|---|
| 138 | return this.errorValue;
|
|---|
| 139 | };
|
|---|
| 140 | return Trie;
|
|---|
| 141 | }());
|
|---|
| 142 | exports.Trie = Trie;
|
|---|
| 143 | //# sourceMappingURL=Trie.js.map |
|---|
