cmpstr 3.2.2 → 3.3.0

package/dist/esm/utils/OptionsValidator.mjs

@@ -0,0 +1,210 @@
+// CmpStr v3.3.0 build-3699f85-260318 by Paul Köhler @komed3 / MIT License
+import { CmpStrValidationError } from './Errors.mjs';
+import '../metric/Cosine.mjs';
+import '../metric/DamerauLevenshtein.mjs';
+import '../metric/DiceSorensen.mjs';
+import '../metric/Hamming.mjs';
+import '../metric/Jaccard.mjs';
+import '../metric/JaroWinkler.mjs';
+import '../metric/LCS.mjs';
+import '../metric/Levenshtein.mjs';
+import '../metric/NeedlemanWunsch.mjs';
+import '../metric/QGram.mjs';
+import '../metric/SmithWaterman.mjs';
+import { MetricRegistry } from '../metric/Metric.mjs';
+import '../phonetic/Caverphone.mjs';
+import '../phonetic/Cologne.mjs';
+import '../phonetic/Metaphone.mjs';
+import '../phonetic/Soundex.mjs';
+import { PhoneticRegistry } from '../phonetic/Phonetic.mjs';
+
+class OptionsValidator {
+  static ALLOWED_FLAGS = new Set([
+    'd',
+    'u',
+    'x',
+    'w',
+    't',
+    'r',
+    's',
+    'k',
+    'n',
+    'i'
+  ]);
+  static ALLOWED_OUTPUT = new Set(['orig', 'prep']);
+  static ALLOWED_MODES = new Set(['default', 'batch', 'single', 'pairwise']);
+  static ALLOWED_SORT = new Set(['asc', 'desc']);
+  static PROCESSORS = {
+    phonetic: (opt) => {
+      if (!opt) return;
+      OptionsValidator.validatePhoneticName(opt.algo);
+      OptionsValidator.validatePhoneticOptions(opt.opt);
+    }
+  };
+  static METRIC_OPT_MAP = {
+    mode: (v) => OptionsValidator.validateMode(v),
+    delimiter: (v) => OptionsValidator.validateString(v, 'opt.delimiter'),
+    pad: (v) => OptionsValidator.validateString(v, 'opt.pad'),
+    q: (v) => OptionsValidator.validateNumber(v, 'opt.q'),
+    match: (v) => OptionsValidator.validateNumber(v, 'opt.match'),
+    mismatch: (v) => OptionsValidator.validateNumber(v, 'opt.mismatch'),
+    gap: (v) => OptionsValidator.validateNumber(v, 'opt.gap')
+  };
+  static PHONETIC_OPT_MAP = {
+    map: (v) =>
+      OptionsValidator.validateString(v, 'processors.phonetic.opt.map'),
+    delimiter: (v) =>
+      OptionsValidator.validateString(v, 'processors.phonetic.opt.delimiter'),
+    length: (v) =>
+      OptionsValidator.validateNumber(v, 'processors.phonetic.opt.length'),
+    pad: (v) =>
+      OptionsValidator.validateString(v, 'processors.phonetic.opt.pad'),
+    dedupe: (v) =>
+      OptionsValidator.validateBoolean(v, 'processors.phonetic.opt.dedupe'),
+    fallback: (v) =>
+      OptionsValidator.validateString(v, 'processors.phonetic.opt.fallback')
+  };
+  static CMPSTR_OPT_MAP = {
+    raw: (v) => OptionsValidator.validateBoolean(v, 'raw'),
+    removeZero: (v) => OptionsValidator.validateBoolean(v, 'removeZero'),
+    safeEmpty: (v) => OptionsValidator.validateBoolean(v, 'safeEmpty'),
+    flags: (v) => OptionsValidator.validateFlags(v),
+    metric: (v) => OptionsValidator.validateMetricName(v),
+    output: (v) => OptionsValidator.validateOutput(v),
+    opt: (v) => OptionsValidator.validateMetricOptions(v),
+    processors: (v) => OptionsValidator.validateProcessors(v),
+    sort: (v) => OptionsValidator.validateSort(v, 'sort'),
+    objectsOnly: (v) => OptionsValidator.validateBoolean(v, 'objectsOnly')
+  };
+  static set2string(set) {
+    return Array.from(set).join(' | ');
+  }
+  static validateType(value, name, type) {
+    if (value === undefined) return;
+    if (typeof value !== type || (type === 'number' && Number.isNaN(value))) {
+      throw new CmpStrValidationError(
+        `Invalid option <${name}>: expected ${type}`,
+        { name, value }
+      );
+    }
+  }
+  static validateEnum(value, name, set) {
+    if (value === undefined) return;
+    if (typeof value !== 'string' || !set.has(value)) {
+      throw new CmpStrValidationError(
+        `Invalid option <${name}>: expected ${OptionsValidator.set2string(set)}`,
+        { name, value }
+      );
+    }
+  }
+  static validateMap(opt, map) {
+    if (!opt) return;
+    for (const k in opt) {
+      const fn = map[k];
+      if (!fn)
+        throw new CmpStrValidationError(`Invalid option <${k}>`, {
+          option: k,
+          value: map[k]
+        });
+      fn(opt[k]);
+    }
+  }
+  static validateRegistryName(value, name, label, has, list) {
+    if (value === undefined) return;
+    if (typeof value !== 'string' || value.length === 0)
+      throw new CmpStrValidationError(
+        `Invalid option <${name}>: expected non-empty string`,
+        { name, value }
+      );
+    if (!has(value))
+      throw new CmpStrValidationError(`${label} <${value}> is not registered`, {
+        name,
+        value,
+        available: list()
+      });
+  }
+  static validateBoolean(value, name) {
+    OptionsValidator.validateType(value, name, 'boolean');
+  }
+  static validateNumber(value, name) {
+    OptionsValidator.validateType(value, name, 'number');
+  }
+  static validateString(value, name) {
+    OptionsValidator.validateType(value, name, 'string');
+  }
+  static validateFlags(value) {
+    if (value === undefined) return;
+    if (typeof value !== 'string')
+      throw new CmpStrValidationError(
+        `Invalid option <flags>: expected string`,
+        { flags: value }
+      );
+    for (let i = 0; i < value.length; i++) {
+      const ch = value[i];
+      if (!OptionsValidator.ALLOWED_FLAGS.has(ch))
+        throw new CmpStrValidationError(
+          `Invalid normalization flag <${ch}> in <flags>: expected ${OptionsValidator.set2string(OptionsValidator.ALLOWED_FLAGS)}`,
+          { flags: value, invalid: ch }
+        );
+    }
+  }
+  static validateOutput(value) {
+    OptionsValidator.validateEnum(
+      value,
+      'output',
+      OptionsValidator.ALLOWED_OUTPUT
+    );
+  }
+  static validateMode(value) {
+    OptionsValidator.validateEnum(
+      value,
+      'mode',
+      OptionsValidator.ALLOWED_MODES
+    );
+  }
+  static validateSort(value, name) {
+    if (value === undefined || typeof value === 'boolean') return;
+    OptionsValidator.validateEnum(value, name, OptionsValidator.ALLOWED_SORT);
+  }
+  static validateMetricName(value) {
+    OptionsValidator.validateRegistryName(
+      value,
+      'metric',
+      'Comparison metric',
+      MetricRegistry.has,
+      MetricRegistry.list
+    );
+  }
+  static validatePhoneticName(value) {
+    OptionsValidator.validateRegistryName(
+      value,
+      'phonetic',
+      'Phonetic algorithm',
+      PhoneticRegistry.has,
+      PhoneticRegistry.list
+    );
+  }
+  static validateMetricOptions(opt) {
+    OptionsValidator.validateMap(opt, OptionsValidator.METRIC_OPT_MAP);
+  }
+  static validatePhoneticOptions(opt) {
+    OptionsValidator.validateMap(opt, OptionsValidator.PHONETIC_OPT_MAP);
+  }
+  static validateProcessors(opt) {
+    if (!opt) return;
+    for (const key in opt) {
+      const fn = OptionsValidator.PROCESSORS[key];
+      if (!fn)
+        throw new CmpStrValidationError(
+          `Invalid processor type <${key}> in <processors>: expected ${Object.keys(OptionsValidator.PROCESSORS).join(' | ')}`,
+          { processors: opt, invalid: key }
+        );
+      fn(opt[key]);
+    }
+  }
+  static validateOptions(opt) {
+    OptionsValidator.validateMap(opt, OptionsValidator.CMPSTR_OPT_MAP);
+  }
+}
+
+export { OptionsValidator };

package/dist/esm/utils/Pool.mjs

@@ -1,4 +1,4 @@
-// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
+// CmpStr v3.3.0 build-3699f85-260318 by Paul Köhler @komed3 / MIT License
 import { CmpStrUsageError, ErrorUtil } from './Errors.mjs';
 
 class RingPool {
@@ -11,15 +11,14 @@ class RingPool {
   acquire(minSize, allowOversize) {
     return ErrorUtil.wrap(
       () => {
-        const len = this.buffers.length;
+        const buffers = this.buffers;
+        const len = buffers.length;
         for (let i = 0; i < len; i++) {
-          const idx = (this.pointer + i) & (len - 1);
-          const item = this.buffers[idx];
-          if (
-            item.size >= minSize &&
-            (allowOversize || item.size === minSize)
-          ) {
-            this.pointer = (idx + 1) & (len - 1);
+          const idx = (this.pointer + i) % len;
+          const item = buffers[idx];
+          const size = item.size;
+          if (size >= minSize && (allowOversize || size === minSize)) {
+            this.pointer = (idx + 1) % len;
             return item;
           }
         }
@@ -32,9 +31,12 @@ class RingPool {
   release(item) {
     ErrorUtil.wrap(
       () => {
-        if (this.buffers.length < this.maxSize)
-          return void [this.buffers.push(item)];
-        this.buffers[this.pointer] = item;
+        const buffers = this.buffers;
+        if (buffers.length < this.maxSize) {
+          buffers.push(item);
+          return;
+        }
+        buffers[this.pointer] = item;
         this.pointer = (this.pointer + 1) % this.maxSize;
       },
       `Failed to release buffer back to pool`,
@@ -54,6 +56,12 @@ class Pool {
       maxItemSize: 2048,
       allowOversize: true
     },
+    'arr[]': {
+      type: 'arr[]',
+      maxSize: 4,
+      maxItemSize: 1024,
+      allowOversize: false
+    },
     'number[]': {
       type: 'number[]',
       maxSize: 16,
@@ -71,6 +79,7 @@ class Pool {
   };
   static POOLS = {
     int32: new RingPool(64),
+    'arr[]': new RingPool(4),
     'number[]': new RingPool(16),
     'string[]': new RingPool(2),
     set: new RingPool(8),
@@ -80,6 +89,8 @@ class Pool {
     switch (type) {
       case 'int32':
         return new Int32Array(size);
+      case 'arr[]':
+        return new Array(size);
       case 'number[]':
         return new Float64Array(size);
       case 'string[]':
@@ -101,7 +112,10 @@ class Pool {
     return this.allocate(type, size);
   }
   static acquireMany(type, sizes) {
-    return sizes.map((size) => this.acquire(type, size));
+    const out = new Array(sizes.length);
+    for (let i = 0; i < sizes.length; i++)
+      out[i] = this.acquire(type, sizes[i]);
+    return out;
   }
   static release(type, buffer, size) {
     const CONFIG = this.CONFIG[type];

package/dist/esm/utils/Profiler.mjs

@@ -1,15 +1,17 @@
-// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
+// CmpStr v3.3.0 build-3699f85-260318 by Paul Köhler @komed3 / MIT License
 class Profiler {
   active;
   static ENV;
   static instance;
   nowFn;
   memFn;
-  store = new Set();
+  store = [];
+  last;
   totalTime = 0;
   totalMem = 0;
   static detectEnv() {
-    if (typeof process !== 'undefined') Profiler.ENV = 'nodejs';
+    if (typeof process !== 'undefined' && process.versions?.node)
+      Profiler.ENV = 'nodejs';
     else if (typeof performance !== 'undefined') Profiler.ENV = 'browser';
     else Profiler.ENV = 'unknown';
   }
@@ -21,7 +23,7 @@ class Profiler {
     this.active = active;
     switch (Profiler.ENV) {
       case 'nodejs':
-        this.nowFn = () => Number(process.hrtime.bigint()) / 1e6;
+        this.nowFn = () => Number(process.hrtime.bigint()) * 1e-6;
         this.memFn = () => process.memoryUsage().heapUsed;
         break;
       case 'browser':
@@ -34,40 +36,52 @@ class Profiler {
         break;
     }
   }
-  now = () => this.nowFn();
-  mem = () => this.memFn();
-  profile(fn, meta) {
-    const startTime = this.now(),
-      startMem = this.mem();
-    const res = fn();
-    const deltaTime = this.now() - startTime,
-      deltaMem = this.mem() - startMem;
-    this.store.add({ time: deltaTime, mem: deltaMem, res, meta });
-    ((this.totalTime += deltaTime), (this.totalMem += deltaMem));
-    return res;
+  storeRes(entry) {
+    this.store.push((this.last = entry));
+    this.totalTime += entry.time;
+    this.totalMem += entry.mem;
   }
-  enable = () => {
+  enable() {
     this.active = true;
-  };
-  disable = () => {
+  }
+  disable() {
     this.active = false;
-  };
+  }
   clear() {
-    this.store.clear();
+    this.store.length = 0;
+    this.last = undefined;
     this.totalTime = 0;
     this.totalMem = 0;
   }
   run(fn, meta = {}) {
-    return this.active ? this.profile(fn, meta) : fn();
+    if (!this.active) return fn();
+    const startTime = this.nowFn(),
+      startMem = this.memFn();
+    const res = fn();
+    const deltaTime = this.nowFn() - startTime,
+      deltaMem = this.memFn() - startMem;
+    this.storeRes({ time: deltaTime, mem: deltaMem, res, meta });
+    return res;
   }
   async runAsync(fn, meta = {}) {
-    return this.active
-      ? this.profile(async () => await fn(), meta)
-      : await fn();
+    if (!this.active) return fn();
+    const startTime = this.nowFn(),
+      startMem = this.memFn();
+    const res = await fn();
+    const deltaTime = this.nowFn() - startTime,
+      deltaMem = this.memFn() - startMem;
+    this.storeRes({ time: deltaTime, mem: deltaMem, res, meta });
+    return res;
+  }
+  getAll() {
+    return [...this.store];
+  }
+  getLast() {
+    return this.last;
+  }
+  getTotal() {
+    return { time: this.totalTime, mem: this.totalMem };
   }
-  getAll = () => [...this.store];
-  getLast = () => this.getAll().pop();
-  getTotal = () => ({ time: this.totalTime, mem: this.totalMem });
   services = Object.freeze({
     enable: this.enable.bind(this),
     disable: this.disable.bind(this),

package/dist/esm/utils/Registry.mjs

@@ -1,4 +1,4 @@
-// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
+// CmpStr v3.3.0 build-3699f85-260318 by Paul Köhler @komed3 / MIT License
 import { ErrorUtil, CmpStrNotFoundError } from './Errors.mjs';
 
 const registry = Object.create(null);
@@ -66,13 +66,13 @@ function resolveCls(reg, cls) {
     throw new CmpStrNotFoundError(`Registry <${reg}> does not exist`, {
       registry: reg
     });
-  return typeof cls === 'string' ? registry[reg]?.get(cls) : cls;
+  return typeof cls === 'string' ? registry[reg].get(cls) : cls;
 }
 function createFromRegistry(reg, cls, ...args) {
-  cls = resolveCls(reg, cls);
+  const ctor = resolveCls(reg, cls);
   return ErrorUtil.wrap(
-    () => new cls(...args),
-    `Failed to create instance of class <${cls.name ?? cls}> from registry <${reg}>`,
+    () => new ctor(...args),
+    `Failed to create instance of class <${ctor.name ?? cls}> from registry <${reg}>`,
     { registry: reg, class: cls, args }
   );
 }

package/dist/esm/utils/StructuredData.mjs

@@ -1,10 +1,12 @@
-// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
+// CmpStr v3.3.0 build-3699f85-260318 by Paul Köhler @komed3 / MIT License
 import { CmpStrValidationError, ErrorUtil } from './Errors.mjs';
 import { Pool } from './Pool.mjs';
 
 class StructuredData {
   data;
   key;
+  static SORT_ASC = (a, b) => a.res - b.res;
+  static SORT_DESC = (a, b) => b.res - a.res;
   static create(data, key) {
     return new StructuredData(data, key);
   }
@@ -13,14 +15,17 @@ class StructuredData {
     this.key = key;
   }
   extractFrom(arr, key) {
-    const result = Pool.acquire('string[]', arr.length);
-    for (let i = 0; i < arr.length; i++) {
+    const n = arr.length;
+    const result = new Array(n);
+    for (let i = 0; i < n; i++) {
       const val = arr[i][key];
-      result[i] = typeof val === 'string' ? val : String(val ?? '');
+      result[i] = val != null ? String(val) : '';
     }
     return result;
   }
-  extract = () => this.extractFrom(this.data, this.key);
+  extract() {
+    return this.extractFrom(this.data, this.key);
+  }
   isMetricResult(v) {
     return (
       typeof v === 'object' && v !== null && 'a' in v && 'b' in v && 'res' in v
@@ -38,64 +43,89 @@ class StructuredData {
   normalizeResults(results) {
     if (!Array.isArray(results) || results.length === 0) return [];
     const first = results[0];
-    let normalized = [];
-    if (this.isMetricResult(first)) normalized = results;
-    else if (this.isCmpStrResult(first))
-      normalized = results.map((r) => ({
-        metric: 'unknown',
-        a: r.source,
-        b: r.target,
-        res: r.match,
-        raw: r.raw
-      }));
-    else
+    let out = new Array(results.length);
+    if (this.isMetricResult(first)) {
+      const src = results;
+      for (let i = 0; i < src.length; i++) out[i] = { ...src[i], __idx: i };
+    } else if (this.isCmpStrResult(first)) {
+      const src = results;
+      for (let i = 0; i < src.length; i++) {
+        const r = src[i];
+        out[i] = {
+          metric: 'unknown',
+          a: r.source,
+          b: r.target,
+          res: r.match,
+          raw: r.raw,
+          __idx: i
+        };
+      }
+    } else
       throw new CmpStrValidationError(
         'Unsupported result format for StructuredData normalization.'
       );
-    return normalized.map((r, idx) => ({ ...r, __idx: idx }));
+    return out;
   }
   rebuild(results, sourceData, extractedStrings, removeZero, objectsOnly) {
-    const stringToIndices = new Map();
-    for (let i = 0; i < extractedStrings.length; i++) {
-      const str = extractedStrings[i];
-      if (!stringToIndices.has(str)) stringToIndices.set(str, []);
-      stringToIndices.get(str).push(i);
-    }
-    const output = new Array(results.length);
-    const occurrenceCount = new Map();
-    let out = 0;
-    for (let i = 0; i < results.length; i++) {
-      const result = results[i];
-      if (removeZero && result.res === 0) continue;
-      const targetStr = result.b || '';
-      const indices = stringToIndices.get(targetStr);
-      let dataIndex;
-      if (indices && indices.length > 0) {
-        const occurrence = occurrenceCount.get(targetStr) ?? 0;
-        occurrenceCount.set(targetStr, occurrence + 1);
-        dataIndex = indices[occurrence % indices.length];
-      } else {
-        dataIndex = result.__idx ?? i;
+    const m = extractedStrings.length,
+      n = results.length;
+    const stringToIndices = Pool.acquire('map', m);
+    const occurrenceCount = Pool.acquire('map', n);
+    const output = new Array(n);
+    stringToIndices.clear();
+    occurrenceCount.clear();
+    try {
+      for (let i = 0; i < m; i++) {
+        const str = extractedStrings[i];
+        let arr = stringToIndices.get(str);
+        if (!arr) {
+          arr = [];
+          stringToIndices.set(str, arr);
+        }
+        arr.push(i);
       }
-      if (dataIndex < 0 || dataIndex >= sourceData.length) continue;
-      const sourceObj = sourceData[dataIndex];
-      const mappedTarget = extractedStrings[dataIndex] || targetStr;
-      if (objectsOnly) output[out++] = sourceObj;
-      else
-        output[out++] = {
-          obj: sourceObj,
-          key: this.key,
-          result: { source: result.a, target: mappedTarget, match: result.res },
-          ...(result.raw ? { raw: result.raw } : null)
-        };
+      let out = 0;
+      for (let i = 0; i < n; i++) {
+        const result = results[i];
+        if (removeZero && result.res === 0) continue;
+        const targetStr = result.b || '';
+        const indices = stringToIndices.get(targetStr);
+        let dataIndex;
+        if (indices && indices.length > 0) {
+          const occurrence = occurrenceCount.get(targetStr) ?? 0;
+          occurrenceCount.set(targetStr, occurrence + 1);
+          dataIndex = indices[occurrence % indices.length];
+        } else {
+          dataIndex = result.__idx ?? i;
+        }
+        if (dataIndex < 0 || dataIndex >= sourceData.length) continue;
+        const sourceObj = sourceData[dataIndex];
+        const mappedTarget = extractedStrings[dataIndex] || targetStr;
+        if (objectsOnly) output[out++] = sourceObj;
+        else
+          output[out++] = {
+            obj: sourceObj,
+            key: this.key,
+            result: {
+              source: result.a,
+              target: mappedTarget,
+              match: result.res
+            },
+            ...(result.raw ? { raw: result.raw } : null)
+          };
+      }
+      output.length = out;
+      return output;
+    } finally {
+      Pool.release('map', stringToIndices, m);
+      Pool.release('map', occurrenceCount, n);
     }
-    output.length = out;
-    return output;
   }
   sort(results, sort) {
     if (!sort || results.length <= 1) return results;
-    const asc = sort === 'asc';
-    return results.sort((a, b) => (asc ? a.res - b.res : b.res - a.res));
+    return results.sort(
+      sort === 'asc' ? StructuredData.SORT_ASC : StructuredData.SORT_DESC
+    );
   }
   finalizeLookup(results, extractedStrings, opt) {
     return this.rebuild(

package/dist/esm/utils/TextAnalyzer.mjs

@@ -1,4 +1,4 @@
-// CmpStr v3.2.2 build-bb61120-260311 by Paul Köhler @komed3 / MIT License
+// CmpStr v3.3.0 build-3699f85-260318 by Paul Köhler @komed3 / MIT License
 class TextAnalyzer {
   static REGEX = {
     number: /\d/,