cspell-trie-lib 8.3.2 → 8.4.0

package/dist/lib/TrieBlob/TrieBlob.js

@@ -1,6 +1,7 @@
 import { defaultTrieInfo } from '../constants.js';
 import { findNode } from '../ITrieNode/trie-util.js';
 import { mergeOptionalWithDefaults } from '../utils/mergeOptionalWithDefaults.js';
+import { NumberSequenceByteDecoderAccumulator, NumberSequenceByteEncoderDecoder, } from './NumberSequenceByteDecoderAccumulator.js';
 import { TrieBlobInternals, TrieBlobIRoot } from './TrieBlobIRoot.js';
 const NodeHeaderNumChildrenBits = 8;
 const NodeHeaderNumChildrenShift = 0;
@@ -35,11 +36,14 @@ export class TrieBlob {
     _forbidIdx;
     _size;
     _iTrieRoot;
+    wordToCharacters;
     constructor(nodes, charIndex, info) {
         this.nodes = nodes;
         this.charIndex = charIndex;
         this.info = mergeOptionalWithDefaults(info);
+        this.wordToCharacters = (word) => [...word];
         this.charToIndexMap = Object.create(null);
+        Object.freeze(this.charIndex);
         for (let i = 0; i < charIndex.length; ++i) {
             const char = charIndex[i];
             this.charToIndexMap[char.normalize('NFC')] = i;
@@ -47,6 +51,15 @@ export class TrieBlob {
         }
         this._forbidIdx = this._lookupNode(0, this.info.forbiddenWordPrefix);
     }
+    _lookUpCharIndex = (char) => {
+        return this.charToIndexMap[char] || 0;
+    };
+    wordToNodeCharIndexSequence(word) {
+        return TrieBlob.charactersToCharIndexSequence(this.wordToCharacters(word), this._lookUpCharIndex);
+    }
+    letterToNodeCharIndexSequence(letter) {
+        return TrieBlob.toCharIndexSequence(this._lookUpCharIndex(letter));
+    }
     has(word) {
         return this._has(0, word);
     }
@@ -76,11 +89,11 @@ export class TrieBlob {
         const NodeMaskChildCharIndex = TrieBlob.NodeMaskChildCharIndex;
         const NodeChildRefShift = TrieBlob.NodeChildRefShift;
         const nodes = this.nodes;
-        const len = word.length;
-        const charToIndexMap = this.charToIndexMap;
+        const wordIndexes = this.wordToNodeCharIndexSequence(word);
+        const len = wordIndexes.length;
         let node = nodes[nodeIdx];
         for (let p = 0; p < len; ++p, node = nodes[nodeIdx]) {
-            const letterIdx = charToIndexMap[word[p]];
+            const letterIdx = wordIndexes[p];
             const count = node & NodeMaskNumChildren;
             let i = count;
             for (; i > 0; --i) {
@@ -94,14 +107,39 @@ export class TrieBlob {
         }
         return (node & TrieBlob.NodeMaskEOW) === TrieBlob.NodeMaskEOW;
     }
+    /**
+     * Find the node index for the given character.
+     * @param nodeIdx - node index to start the search
+     * @param char - character to look for
+     * @returns
+     */
     _lookupNode(nodeIdx, char) {
+        const indexSeq = this.letterToNodeCharIndexSequence(char);
+        const len = indexSeq.length;
+        if (!len)
+            return undefined;
+        let currNodeIdx = nodeIdx;
+        for (let i = 0; i < len; ++i) {
+            currNodeIdx = this._lookupNodeByCharIndexSeq(currNodeIdx, indexSeq[i]);
+            if (currNodeIdx === undefined) {
+                return undefined;
+            }
+        }
+        return currNodeIdx;
+    }
+    /**
+     * Find the node index for the given character.
+     * @param nodeIdx - node index to start the search
+     * @param char - character to look for
+     * @returns
+     */
+    _lookupNodeByCharIndexSeq(nodeIdx, index) {
         const NodeMaskNumChildren = TrieBlob.NodeMaskNumChildren;
         const NodeMaskChildCharIndex = TrieBlob.NodeMaskChildCharIndex;
         const NodeChildRefShift = TrieBlob.NodeChildRefShift;
         const nodes = this.nodes;
-        const charToIndexMap = this.charToIndexMap;
         const node = nodes[nodeIdx];
-        const letterIdx = charToIndexMap[char];
+        const letterIdx = index;
         const count = node & NodeMaskNumChildren;
         let i = count;
         for (; i > 0; --i) {
@@ -117,10 +155,12 @@ export class TrieBlob {
         const NodeMaskChildCharIndex = TrieBlob.NodeMaskChildCharIndex;
         const NodeChildRefShift = TrieBlob.NodeChildRefShift;
         const nodes = this.nodes;
-        const stack = [{ nodeIdx: 0, pos: 0, word: '' }];
+        const stack = [
+            { nodeIdx: 0, pos: 0, word: '', acc: NumberSequenceByteDecoderAccumulator.create() },
+        ];
         let depth = 0;
         while (depth >= 0) {
-            const { nodeIdx, pos, word } = stack[depth];
+            const { nodeIdx, pos, word, acc } = stack[depth];
             const node = nodes[nodeIdx];
             // pos is 0 when first entering a node
             if (!pos && node & NodeMaskEOW) {
@@ -133,13 +173,15 @@ export class TrieBlob {
             }
             const nextPos = ++stack[depth].pos;
             const entry = nodes[nodeIdx + nextPos];
-            const charIdx = entry & NodeMaskChildCharIndex;
-            const letter = this.charIndex[charIdx];
+            const nAcc = acc.clone();
+            const charIdx = nAcc.decode(entry & NodeMaskChildCharIndex);
+            const letter = (charIdx && this.charIndex[charIdx]) || '';
             ++depth;
             stack[depth] = {
                 nodeIdx: entry >>> NodeChildRefShift,
                 pos: 0,
                 word: word + letter,
+                acc: nAcc,
             };
         }
     }
@@ -159,14 +201,14 @@ export class TrieBlob {
     }
     toJSON() {
         return {
-            charIndex: this.charIndex,
             options: this.info,
-            nodes: splitString(Buffer.from(this.nodes.buffer, 128).toString('base64')),
+            nodes: nodesToJson(this.nodes),
+            charIndex: this.charIndex,
         };
     }
     encodeBin() {
         const charIndex = Buffer.from(this.charIndex.join('\n'));
-        const charIndexLen = (charIndex.byteLength + 3) & ~3;
+        const charIndexLen = (charIndex.byteLength + 3) & ~3; // round up to the nearest 4 byte boundary.
         const nodeOffset = HEADER_SIZE + charIndexLen;
         const size = nodeOffset + this.nodes.length * 4;
         const useLittle = isLittleEndian();
@@ -182,7 +224,8 @@ export class TrieBlob {
         header.setUint32(HEADER.charIndexLen, charIndex.length, useLittle);
         buffer.set(charIndex, HEADER_SIZE);
         buffer.set(nodeData, nodeOffset);
-        // dumpBin(nodeData);
+        // console.log('encodeBin: %o', this.toJSON());
+        // console.log('encodeBin: buf %o nodes %o', buffer, this.nodes);
         return buffer;
     }
     static decodeBin(blob) {
@@ -205,14 +248,54 @@ export class TrieBlob {
         const charIndex = Buffer.from(blob.subarray(offsetCharIndex, offsetCharIndex + lenCharIndex))
             .toString('utf8')
             .split('\n');
-        const nodes = new Uint32Array(blob.buffer).subarray(offsetNodes / 4, offsetNodes / 4 + lenNodes);
-        return new TrieBlob(nodes, charIndex, defaultTrieInfo);
+        const nodes = new Uint32Array(blob.buffer, offsetNodes, lenNodes);
+        const trieBlob = new TrieBlob(nodes, charIndex, defaultTrieInfo);
+        // console.log('decodeBin: %o', trieBlob.toJSON());
+        return trieBlob;
     }
     static NodeMaskEOW = 0x00000100;
     static NodeMaskNumChildren = (1 << NodeHeaderNumChildrenBits) - 1;
     static NodeMaskNumChildrenShift = NodeHeaderNumChildrenShift;
     static NodeChildRefShift = 8;
+    /**
+     * Only 8 bits are reserved for the character index.
+     * The max index is {@link TrieBlob.SpecialCharIndexMask} - 1.
+     * Node chaining is used to reference higher character indexes.
+     * - @see {@link TrieBlob.SpecialCharIndexMask}
+     * - @see {@link TrieBlob.MaxCharIndex}
+     */
     static NodeMaskChildCharIndex = 0x000000ff;
+    /** SpecialCharIndexMask is used to indicate a node chain */
+    static SpecialCharIndexMask = 0xf8;
+    static MaxCharIndex = this.SpecialCharIndexMask - 1;
+    /**
+     * SpecialCharIndex8bit is used to indicate a node chain. Where the final character index is 248 + the index found in the next node.
+     */
+    static SpecialCharIndex8bit = this.SpecialCharIndexMask | 0x01;
+    static SpecialCharIndex16bit = this.SpecialCharIndexMask | 0x02;
+    static SpecialCharIndex24bit = this.SpecialCharIndexMask | 0x03;
+    /**
+     * Since it is only possible to store single byte indexes, a multi-byte index is stored as a sequence of indexes chained between nodes.
+     * @param charIndex - character index to convert to a sequence of indexes
+     * @returns encoded index values.
+     */
+    static toCharIndexSequence(charIndex) {
+        return NumberSequenceByteEncoderDecoder.encode(charIndex);
+    }
+    static fromCharIndexSequence(charIndexes) {
+        return NumberSequenceByteEncoderDecoder.decodeSequence(charIndexes);
+    }
+    static charactersToCharIndexSequence(chars, charToIndexMap) {
+        const fn = typeof charToIndexMap === 'function' ? charToIndexMap : (c) => charToIndexMap[c];
+        return chars.map(fn).flatMap((c) => this.toCharIndexSequence(c));
+    }
+    static charIndexSequenceToCharacters(charIndexSequence, charIndex) {
+        const chars = [...this.fromCharIndexSequence(charIndexSequence)].map((c) => charIndex[c]);
+        return chars;
+    }
+    static nodesView(trie) {
+        return new Uint32Array(trie.nodes);
+    }
 }
 function isLittleEndian() {
     const buf = new Uint8Array([1, 2, 3, 4]);
@@ -234,11 +317,27 @@ class ErrorDecodeTrieBlob extends Error {
         super(message);
     }
 }
-function splitString(s, len = 64) {
-    const splits = [];
-    for (let i = 0; i < s.length; i += len) {
-        splits.push(s.slice(i, i + len));
+function nodesToJson(nodes) {
+    function nodeElement(offset) {
+        const node = nodes[offset];
+        const numChildren = node & TrieBlob.NodeMaskNumChildren;
+        const eow = !!(node & TrieBlob.NodeMaskEOW);
+        const children = [];
+        for (let i = 1; i <= numChildren; ++i) {
+            children.push({
+                c: ('00' + (nodes[offset + i] & TrieBlob.NodeMaskChildCharIndex).toString(16)).slice(-2),
+                o: nodes[offset + i] >>> TrieBlob.NodeChildRefShift,
+            });
+        }
+        return { id: offset, eow, n: offset + numChildren + 1, c: children };
+    }
+    const elements = [];
+    let offset = 0;
+    while (offset < nodes.length) {
+        const e = nodeElement(offset);
+        elements.push(e);
+        offset = e.n;
     }
-    return splits;
+    return elements;
 }
 //# sourceMappingURL=TrieBlob.js.map

package/dist/lib/TrieBlob/TrieBlobIRoot.d.ts

@@ -6,26 +6,34 @@ interface BitMaskInfo {
     readonly NodeMaskChildCharIndex: number;
     readonly NodeChildRefShift: number;
 }
+type Node = number;
+type NodeIndex = number;
 export declare class TrieBlobInternals implements BitMaskInfo {
     readonly nodes: Uint32Array;
-    readonly charIndex: string[];
+    readonly charIndex: readonly string[];
     readonly charToIndexMap: Readonly<Record<string, number>>;
     readonly NodeMaskEOW: number;
     readonly NodeMaskNumChildren: number;
     readonly NodeMaskChildCharIndex: number;
     readonly NodeChildRefShift: number;
-    constructor(nodes: Uint32Array, charIndex: string[], charToIndexMap: Readonly<Record<string, number>>, maskInfo: BitMaskInfo);
+    readonly isIndexDecoderNeeded: boolean;
+    constructor(nodes: Uint32Array, charIndex: readonly string[], charToIndexMap: Readonly<Record<string, number>>, maskInfo: BitMaskInfo);
 }
 declare class TrieBlobINode implements ITrieNode {
     readonly trie: TrieBlobInternals;
-    readonly nodeIdx: number;
+    readonly nodeIdx: NodeIndex;
     readonly id: number;
-    readonly size: number;
-    readonly node: number;
+    readonly node: Node;
     readonly eow: boolean;
     private _keys;
-    charToIdx: Record<string, number> | undefined;
-    constructor(trie: TrieBlobInternals, nodeIdx: number);
+    private _count;
+    private _size;
+    private _chained;
+    private _nodesEntries;
+    private _entries;
+    private _values;
+    protected charToIdx: Readonly<Record<string, number>> | undefined;
+    constructor(trie: TrieBlobInternals, nodeIdx: NodeIndex);
     /** get keys to children */
     keys(): readonly string[];
     values(): readonly ITrieNode[];
@@ -36,6 +44,10 @@ declare class TrieBlobINode implements ITrieNode {
     hasChildren(): boolean;
     child(keyIdx: number): ITrieNode;
     getCharToIdxMap(): Record<string, number>;
+    private containsChainedIndexes;
+    private getNodesEntries;
+    private walkChainedIndexes;
+    get size(): number;
 }
 export declare class TrieBlobIRoot extends TrieBlobINode implements ITrieNodeRoot {
     readonly info: Readonly<TrieInfo>;

package/dist/lib/TrieBlob/TrieBlobIRoot.js

@@ -1,3 +1,5 @@
+import { NumberSequenceByteDecoderAccumulator, NumberSequenceByteEncoderDecoder, } from './NumberSequenceByteDecoderAccumulator.js';
+const SpecialCharIndexMask = NumberSequenceByteEncoderDecoder.SpecialCharIndexMask;
 export class TrieBlobInternals {
     nodes;
     charIndex;
@@ -6,6 +8,7 @@ export class TrieBlobInternals {
     NodeMaskNumChildren;
     NodeMaskChildCharIndex;
     NodeChildRefShift;
+    isIndexDecoderNeeded;
     constructor(nodes, charIndex, charToIndexMap, maskInfo) {
         this.nodes = nodes;
         this.charIndex = charIndex;
@@ -15,18 +18,25 @@ export class TrieBlobInternals {
         this.NodeMaskNumChildren = NodeMaskNumChildren;
         this.NodeMaskChildCharIndex = NodeMaskChildCharIndex;
         this.NodeChildRefShift = NodeChildRefShift;
+        this.isIndexDecoderNeeded = charIndex.length > NumberSequenceByteEncoderDecoder.MaxCharIndex;
     }
 }
 const EmptyKeys = Object.freeze([]);
 const EmptyNodes = Object.freeze([]);
+const EmptyEntries = Object.freeze([]);
 class TrieBlobINode {
     trie;
     nodeIdx;
     id;
-    size;
     node;
     eow;
     _keys;
+    _count;
+    _size;
+    _chained;
+    _nodesEntries;
+    _entries;
+    _values;
     charToIdx;
     constructor(trie, nodeIdx) {
         this.trie = trie;
@@ -34,46 +44,34 @@ class TrieBlobINode {
         const node = trie.nodes[nodeIdx];
         this.node = node;
         this.eow = !!(node & trie.NodeMaskEOW);
-        this.size = node & trie.NodeMaskNumChildren;
+        this._count = node & trie.NodeMaskNumChildren;
         this.id = nodeIdx;
     }
     /** get keys to children */
     keys() {
         if (this._keys)
             return this._keys;
-        if (!this.size)
+        if (!this._count)
             return EmptyKeys;
-        const NodeMaskChildCharIndex = this.trie.NodeMaskChildCharIndex;
-        const charIndex = this.trie.charIndex;
-        const keys = Array(this.size);
-        const offset = this.nodeIdx + 1;
-        const len = this.size;
-        for (let i = 0; i < len; ++i) {
-            const entry = this.trie.nodes[i + offset];
-            const charIdx = entry & NodeMaskChildCharIndex;
-            keys[i] = charIndex[charIdx];
-        }
-        this._keys = keys;
-        return keys;
+        this._keys = this.getNodesEntries().map(([key]) => key);
+        return this._keys;
     }
     values() {
-        if (!this.size)
+        if (!this._count)
             return EmptyNodes;
-        const nodes = Array(this.size);
-        for (let i = 0; i < this.size; ++i) {
-            nodes[i] = this.child(i);
-        }
-        return nodes;
+        if (this._values)
+            return this._values;
+        this._values = this.entries().map(([, value]) => value);
+        return this._values;
     }
     entries() {
-        const keys = this.keys();
-        const values = this.values();
-        const len = keys.length;
-        const entries = Array(len);
-        for (let i = 0; i < len; ++i) {
-            entries[i] = [keys[i], values[i]];
-        }
-        return entries;
+        if (this._entries)
+            return this._entries;
+        if (!this._count)
+            return EmptyEntries;
+        const entries = this.getNodesEntries();
+        this._entries = entries.map(([key, value]) => [key, new TrieBlobINode(this.trie, value)]);
+        return this._entries;
     }
     /** get child ITrieNode */
     get(char) {
@@ -87,12 +85,15 @@ class TrieBlobINode {
         return idx !== undefined;
     }
     hasChildren() {
-        return this.size > 0;
+        return this._count > 0;
     }
     child(keyIdx) {
-        const n = this.trie.nodes[this.nodeIdx + keyIdx + 1];
-        const nodeIdx = n >>> this.trie.NodeChildRefShift;
-        return new TrieBlobINode(this.trie, nodeIdx);
+        if (!this._values && !this.containsChainedIndexes()) {
+            const n = this.trie.nodes[this.nodeIdx + keyIdx + 1];
+            const nodeIdx = n >>> this.trie.NodeChildRefShift;
+            return new TrieBlobINode(this.trie, nodeIdx);
+        }
+        return this.values()[keyIdx];
     }
     getCharToIdxMap() {
         const m = this.charToIdx;
@@ -106,6 +107,101 @@ class TrieBlobINode {
         this.charToIdx = map;
         return map;
     }
+    containsChainedIndexes() {
+        if (this._chained !== undefined)
+            return this._chained;
+        if (!this._count || !this.trie.isIndexDecoderNeeded) {
+            this._chained = false;
+            return false;
+        }
+        // scan the node to see if there are encoded entries.
+        let found = false;
+        const NodeMaskChildCharIndex = this.trie.NodeMaskChildCharIndex;
+        const offset = this.nodeIdx + 1;
+        const nodes = this.trie.nodes;
+        const len = this._count;
+        for (let i = 0; i < len && !found; ++i) {
+            const entry = nodes[i + offset];
+            const charIdx = entry & NodeMaskChildCharIndex;
+            found = (charIdx & SpecialCharIndexMask) === SpecialCharIndexMask;
+        }
+        this._chained = !!found;
+        return this._chained;
+    }
+    getNodesEntries() {
+        if (this._nodesEntries)
+            return this._nodesEntries;
+        if (!this.containsChainedIndexes()) {
+            const entries = Array(this._count);
+            const nodes = this.trie.nodes;
+            const offset = this.nodeIdx + 1;
+            const charIndex = this.trie.charIndex;
+            const NodeMaskChildCharIndex = this.trie.NodeMaskChildCharIndex;
+            const RefShift = this.trie.NodeChildRefShift;
+            for (let i = 0; i < this._count; ++i) {
+                const entry = nodes[offset + i];
+                const charIdx = entry & NodeMaskChildCharIndex;
+                entries[i] = [charIndex[charIdx], entry >>> RefShift];
+            }
+            this._nodesEntries = entries;
+            return entries;
+        }
+        this._nodesEntries = this.walkChainedIndexes();
+        return this._nodesEntries;
+    }
+    walkChainedIndexes() {
+        const NodeMaskChildCharIndex = this.trie.NodeMaskChildCharIndex;
+        const NodeChildRefShift = this.trie.NodeChildRefShift;
+        const NodeMaskNumChildren = this.trie.NodeMaskNumChildren;
+        const nodes = this.trie.nodes;
+        const acc = NumberSequenceByteDecoderAccumulator.create();
+        const stack = [{ nodeIdx: this.nodeIdx + 1, lastIdx: this.nodeIdx + this._count, acc }];
+        let depth = 0;
+        const entries = Array(this._count);
+        let eIdx = 0;
+        const charIndex = this.trie.charIndex;
+        while (depth >= 0) {
+            const s = stack[depth];
+            const { nodeIdx, lastIdx } = s;
+            if (nodeIdx > lastIdx) {
+                --depth;
+                continue;
+            }
+            ++s.nodeIdx;
+            const entry = nodes[nodeIdx];
+            const charIdx = entry & NodeMaskChildCharIndex;
+            const acc = s.acc.clone();
+            const letterIdx = acc.decode(charIdx);
+            if (letterIdx !== undefined) {
+                const char = charIndex[letterIdx];
+                const nodeIdx = entry >>> NodeChildRefShift;
+                entries[eIdx++] = [char, nodeIdx];
+                continue;
+            }
+            const idx = entry >>> NodeChildRefShift;
+            const lIdx = idx + (nodes[idx] & NodeMaskNumChildren);
+            const ss = stack[++depth];
+            if (ss) {
+                ss.nodeIdx = idx + 1;
+                ss.lastIdx = lIdx;
+                ss.acc = acc;
+            }
+            else {
+                stack[depth] = { nodeIdx: idx + 1, lastIdx: lIdx, acc };
+            }
+        }
+        return entries;
+    }
+    get size() {
+        if (this._size === undefined) {
+            if (!this.containsChainedIndexes()) {
+                this._size = this._count;
+                return this._size;
+            }
+            this._size = this.getNodesEntries().length;
+        }
+        return this._size;
+    }
 }
 export class TrieBlobIRoot extends TrieBlobINode {
     info;

package/dist/lib/TrieBlob/createTrieBlob.d.ts

@@ -2,7 +2,7 @@ import type { ITrieNodeRoot } from '../ITrieNode/ITrieNode.js';
 import type { PartialTrieInfo } from '../ITrieNode/TrieInfo.js';
 import type { TrieData } from '../TrieData.js';
 import { TrieBlob } from './TrieBlob.js';
-export declare function createTrieBlob(words: string[], options?: PartialTrieInfo): TrieBlob;
+export declare function createTrieBlob(words: readonly string[], options?: PartialTrieInfo): TrieBlob;
 export declare function createTrieBlobFromITrieNodeRoot(root: ITrieNodeRoot): TrieBlob;
 export declare function createTrieBlobFromTrieData(trie: TrieData): TrieBlob;
 //# sourceMappingURL=createTrieBlob.d.ts.map

package/dist/lib/TrieData.d.ts

@@ -2,6 +2,8 @@ import type { ITrieNode, ITrieNodeRoot } from './ITrieNode/ITrieNode.js';
 import type { TrieInfo } from './ITrieNode/TrieInfo.js';
 export interface TrieData {
     info: Readonly<TrieInfo>;
+    /** Method used to split words into individual characters. */
+    wordToCharacters(word: string): readonly string[];
     words(): Iterable<string>;
     getRoot(): ITrieNodeRoot;
     getNode(prefix: string): ITrieNode | undefined;

package/dist/lib/TrieNode/TrieNodeBuilder.d.ts

@@ -5,6 +5,7 @@ import { TrieNodeTrie } from './TrieNodeTrie.js';
 export declare class TrieNodeBuilder implements TrieBuilder<TrieNodeTrie> {
     private _cursor;
     root: TrieRoot;
+    wordToCharacters: (word: string) => string[];
     setOptions(options: Readonly<PartialTrieOptions>): Readonly<TrieOptions>;
     build(): TrieNodeTrie;
     getCursor(): BuilderCursor;

package/dist/lib/TrieNode/TrieNodeBuilder.js

@@ -7,6 +7,7 @@ const EOW = Object.freeze({ f: 1, k: true });
 export class TrieNodeBuilder {
     _cursor;
     root = { ...defaultTrieInfo, c: Object.create(null) };
+    wordToCharacters = (word) => word.split('');
     setOptions(options) {
         const opts = mergeOptionalWithDefaults(options, this.root);
         Object.assign(this.root, opts);

package/dist/lib/TrieNode/TrieNodeTrie.d.ts

@@ -8,6 +8,7 @@ export declare class TrieNodeTrie implements TrieData {
     readonly info: TrieOptions;
     private _size;
     constructor(root: TrieRoot);
+    wordToCharacters: (word: string) => string[];
     get iTrieRoot(): ITrieNodeRoot;
     getRoot(): ITrieNodeRoot;
     getNode(prefix: string): ITrieNode | undefined;

package/dist/lib/TrieNode/TrieNodeTrie.js

@@ -13,6 +13,7 @@ export class TrieNodeTrie {
         this.root = root;
         this.info = mergeOptionalWithDefaults(root);
     }
+    wordToCharacters = (word) => word.split('');
     get iTrieRoot() {
         return this._iTrieRoot || (this._iTrieRoot = trieRootToITrieRoot(this.root));
     }

package/dist/lib/utils/text.d.ts

@@ -50,4 +50,6 @@ export declare function stripAccents(characters: string): string;
  * @returns - only the accents.
  */
 export declare function stripNonAccents(characters: string): string;
+export declare function isValidUtf16Character(char: string): boolean;
+export declare function assertValidUtf16Character(char: string): void;
 //# sourceMappingURL=text.d.ts.map

package/dist/lib/utils/text.js

@@ -99,4 +99,31 @@ export function stripAccents(characters) {
 export function stripNonAccents(characters) {
     return characters.normalize('NFD').replace(/[^\p{M}]/gu, '');
 }
+export function isValidUtf16Character(char) {
+    const len = char.length;
+    const code = char.charCodeAt(0) & 0xfc00;
+    const valid = (len === 1 && (code & 0xf800) !== 0xd800) ||
+        (len === 2 && (code & 0xfc00) === 0xd800 && (char.charCodeAt(1) & 0xfc00) === 0xdc00);
+    return valid;
+}
+export function assertValidUtf16Character(char) {
+    if (!isValidUtf16Character(char)) {
+        const len = char.length;
+        const codes = char
+            .slice(0, 2)
+            .split('')
+            .map((c) => '0x' + ('0000' + c.charCodeAt(0).toString(16)).slice(-4));
+        let message;
+        if (len == 1) {
+            message = `Invalid utf16 character, lone surrogate: ${codes[0]}`;
+        }
+        else if (len == 2) {
+            message = `Invalid utf16 character, not a valid surrogate pair: [${codes.join(', ')}]`;
+        }
+        else {
+            message = `Invalid utf16 character, must be a single character, found: ${len}`;
+        }
+        throw new Error(message);
+    }
+}
 //# sourceMappingURL=text.js.map

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "cspell-trie-lib",
-  "version": "8.3.2",
+  "version": "8.4.0",
   "description": "Trie Data Structure to support cspell.",
   "type": "module",
   "sideEffects": false,
@@ -47,18 +47,18 @@
   },
   "homepage": "https://github.com/streetsidesoftware/cspell#readme",
   "dependencies": {
-    "@cspell/cspell-pipe": "8.3.2",
-    "@cspell/cspell-types": "8.3.2",
+    "@cspell/cspell-pipe": "8.4.0",
+    "@cspell/cspell-types": "8.4.0",
     "gensequence": "^6.0.0"
   },
   "engines": {
     "node": ">=18"
   },
   "devDependencies": {
-    "@cspell/dict-en_us": "^4.3.13",
+    "@cspell/dict-en_us": "^4.3.16",
     "@cspell/dict-es-es": "^2.3.1",
     "@cspell/dict-nl-nl": "^2.3.0",
     "import-meta-resolve": "^4.0.0"
   },
-  "gitHead": "98f622b2b12529f2d1ccf0f3a57991e4c08b3e3a"
+  "gitHead": "f9ad457ca2102c6642c377417a95a4415f5ec3d8"
 }