cmpstr/dist/esm/utils/HashTable.mjs

CmpStr is a lightweight, fast and well performing package for calculating string similarity

// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
class Hasher {
  static FNV_PRIME = 0x01000193;
  static HASH_OFFSET = 0x811c9dc5;
  static fastFNV1a(str) {
    const len = str.length;
    let hash = this.HASH_OFFSET;
    const chunks = Math.floor(len / 4);
    for (let i = 0; i < chunks; i++) {
      const pos = i * 4;
      const chunk =
        str.charCodeAt(pos) |
        (str.charCodeAt(pos + 1) << 8) |
        (str.charCodeAt(pos + 2) << 16) |
        (str.charCodeAt(pos + 3) << 24);
      hash ^= chunk;
      hash = Math.imul(hash, this.FNV_PRIME);
    }
    const remaining = len % 4;
    if (remaining > 0) {
      const pos = chunks * 4;
      for (let i = 0; i < remaining; i++) {
        hash ^= str.charCodeAt(pos + i);
        hash = Math.imul(hash, this.FNV_PRIME);
      }
    }
    hash ^= hash >>> 16;
    hash *= 0x85ebca6b;
    hash ^= hash >>> 13;
    hash *= 0xc2b2ae35;
    hash ^= hash >>> 16;
    return hash >>> 0;
  }
}
class HashTable {
  LRU;
  static MAX_LEN = 2048;
  static TABLE_SIZE = 10_000;
  table = new Map();
  constructor(LRU = true) {
    this.LRU = LRU;
  }
  key(label, strs, sorted = false) {
    for (const str of strs) if (str.length > HashTable.MAX_LEN) return false;
    const hashes = strs.map((s) => Hasher.fastFNV1a(s));
    return [label, ...(sorted ? hashes.sort() : hashes)].join('-');
  }
  has = (key) => this.table.has(key);
  get = (key) => this.table.get(key);
  set(key, entry, update = true) {
    if (!update && this.table.has(key)) return false;
    while (!this.table.has(key) && this.table.size >= HashTable.TABLE_SIZE) {
      if (!this.LRU) return false;
      this.table.delete(this.table.keys().next().value);
    }
    this.table.set(key, entry);
    return true;
  }
  delete = (key) => this.table.delete(key);
  clear = () => this.table.clear();
  size = () => this.table.size;
}

export { HashTable, Hasher };