@goshenkata/dryscan-core 1.2.5 → 1.2.7

package/dist/chunk-ZZC4V5LV.js

@@ -0,0 +1,52 @@
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __decorateClass = (decorators, target, key, kind) => {
+  var result = kind > 1 ? void 0 : kind ? __getOwnPropDesc(target, key) : target;
+  for (var i = decorators.length - 1, decorator; i >= 0; i--)
+    if (decorator = decorators[i])
+      result = (kind ? decorator(target, key, result) : decorator(result)) || result;
+  if (kind && result) __defProp(target, key, result);
+  return result;
+};
+
+// src/services/ParallelSimilarity.ts
+import os from "os";
+import { Worker } from "worker_threads";
+import { cosineSimilarity } from "@langchain/core/utils/math";
+var MIN_PARALLEL_ROWS = 50;
+async function parallelCosineSimilarity(A, B) {
+  if (A.length === 0 || B.length === 0) return [];
+  if (A.length < MIN_PARALLEL_ROWS) return cosineSimilarity(A, B);
+  const dims = A[0].length;
+  const chunkSize = Math.ceil(A.length / os.cpus().length);
+  const sharedB = new SharedArrayBuffer(B.length * dims * 8);
+  const bView = new Float64Array(sharedB);
+  B.forEach((row, i) => bView.set(row, i * dims));
+  const workerUrl = new URL("./services/cosineSimilarityWorker.js", import.meta.url);
+  const execArgv = workerUrl.pathname.endsWith(".ts") ? ["--import", "tsx/esm"] : [];
+  const chunks = Array.from(
+    { length: Math.ceil(A.length / chunkSize) },
+    (_, i) => A.slice(i * chunkSize, (i + 1) * chunkSize)
+  );
+  const results = await Promise.all(chunks.map((chunk) => runWorker(chunk, sharedB, B.length, dims, workerUrl, execArgv)));
+  return results.flat();
+}
+function runWorker(chunk, sharedB, bCount, dims, workerUrl, execArgv) {
+  return new Promise((resolve, reject) => {
+    const rowsFlat = new Float64Array(chunk.length * dims);
+    chunk.forEach((row, i) => rowsFlat.set(row, i * dims));
+    const worker = new Worker(workerUrl, {
+      workerData: { rowsBuffer: rowsFlat.buffer, rowCount: chunk.length, allBuffer: sharedB, allCount: bCount, dims },
+      transferList: [rowsFlat.buffer],
+      execArgv
+    });
+    worker.once("message", ({ result }) => resolve(result));
+    worker.once("error", reject);
+  });
+}
+
+export {
+  __decorateClass,
+  parallelCosineSimilarity
+};
+//# sourceMappingURL=chunk-ZZC4V5LV.js.map

package/dist/chunk-ZZC4V5LV.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/services/ParallelSimilarity.ts"],"sourcesContent":["import os from \"node:os\";\nimport { Worker } from \"node:worker_threads\";\nimport { cosineSimilarity } from \"@langchain/core/utils/math\";\n\n/** Minimum row count below which synchronous is faster than worker overhead. */\nconst MIN_PARALLEL_ROWS = 50;\n\n/**\n * Computes cosineSimilarity(A, B) using worker threads for large inputs,\n * falling back to the synchronous library call for small ones.\n * B is packed into a SharedArrayBuffer shared across all workers — no copies.\n */\nexport async function parallelCosineSimilarity(A: number[][], B: number[][]): Promise<number[][]> {\n  if (A.length === 0 || B.length === 0) return [];\n  if (A.length < MIN_PARALLEL_ROWS) return cosineSimilarity(A, B);\n\n  const dims = A[0].length;\n  const chunkSize = Math.ceil(A.length / os.cpus().length);\n\n  const sharedB = new SharedArrayBuffer(B.length * dims * 8);\n  const bView = new Float64Array(sharedB);\n  B.forEach((row, i) => bView.set(row, i * dims));\n\n  // import.meta.resolve respects the active module loader:\n  // under tsx it remaps .js → .ts; in compiled output it stays .js.\n  const workerUrl = new URL(\"./services/cosineSimilarityWorker.js\", import.meta.url);\n  const execArgv = workerUrl.pathname.endsWith(\".ts\") ? [\"--import\", \"tsx/esm\"] : [];\n\n  const chunks = Array.from(\n    { length: Math.ceil(A.length / chunkSize) },\n    (_, i) => A.slice(i * chunkSize, (i + 1) * chunkSize),\n  );\n\n  const results = await Promise.all(chunks.map(chunk => runWorker(chunk, sharedB, B.length, dims, workerUrl, execArgv)));\n  return results.flat();\n}\n\nfunction runWorker(\n  chunk: number[][],\n  sharedB: SharedArrayBuffer,\n  bCount: number,\n  dims: number,\n  workerUrl: URL,\n  execArgv: string[],\n): Promise<number[][]> {\n  return new Promise((resolve, reject) => {\n    const rowsFlat = new Float64Array(chunk.length * dims);\n    chunk.forEach((row, i) => rowsFlat.set(row, i * dims));\n\n    const worker = new Worker(workerUrl, {\n      workerData: { rowsBuffer: rowsFlat.buffer, rowCount: chunk.length, allBuffer: sharedB, allCount: bCount, dims },\n      transferList: [rowsFlat.buffer],\n      execArgv,\n    });\n\n    worker.once(\"message\", ({ result }) => resolve(result));\n    worker.once(\"error\", reject);\n  });\n}\n"],"mappings":";;;;;;;;;;;;AAAA,OAAO,QAAQ;AACf,SAAS,cAAc;AACvB,SAAS,wBAAwB;AAGjC,IAAM,oBAAoB;AAO1B,eAAsB,yBAAyB,GAAe,GAAoC;AAChG,MAAI,EAAE,WAAW,KAAK,EAAE,WAAW,EAAG,QAAO,CAAC;AAC9C,MAAI,EAAE,SAAS,kBAAmB,QAAO,iBAAiB,GAAG,CAAC;AAE9D,QAAM,OAAO,EAAE,CAAC,EAAE;AAClB,QAAM,YAAY,KAAK,KAAK,EAAE,SAAS,GAAG,KAAK,EAAE,MAAM;AAEvD,QAAM,UAAU,IAAI,kBAAkB,EAAE,SAAS,OAAO,CAAC;AACzD,QAAM,QAAQ,IAAI,aAAa,OAAO;AACtC,IAAE,QAAQ,CAAC,KAAK,MAAM,MAAM,IAAI,KAAK,IAAI,IAAI,CAAC;AAI9C,QAAM,YAAY,IAAI,IAAI,wCAAwC,YAAY,GAAG;AACjF,QAAM,WAAW,UAAU,SAAS,SAAS,KAAK,IAAI,CAAC,YAAY,SAAS,IAAI,CAAC;AAEjF,QAAM,SAAS,MAAM;AAAA,IACnB,EAAE,QAAQ,KAAK,KAAK,EAAE,SAAS,SAAS,EAAE;AAAA,IAC1C,CAAC,GAAG,MAAM,EAAE,MAAM,IAAI,YAAY,IAAI,KAAK,SAAS;AAAA,EACtD;AAEA,QAAM,UAAU,MAAM,QAAQ,IAAI,OAAO,IAAI,WAAS,UAAU,OAAO,SAAS,EAAE,QAAQ,MAAM,WAAW,QAAQ,CAAC,CAAC;AACrH,SAAO,QAAQ,KAAK;AACtB;AAEA,SAAS,UACP,OACA,SACA,QACA,MACA,WACA,UACqB;AACrB,SAAO,IAAI,QAAQ,CAAC,SAAS,WAAW;AACtC,UAAM,WAAW,IAAI,aAAa,MAAM,SAAS,IAAI;AACrD,UAAM,QAAQ,CAAC,KAAK,MAAM,SAAS,IAAI,KAAK,IAAI,IAAI,CAAC;AAErD,UAAM,SAAS,IAAI,OAAO,WAAW;AAAA,MACnC,YAAY,EAAE,YAAY,SAAS,QAAQ,UAAU,MAAM,QAAQ,WAAW,SAAS,UAAU,QAAQ,KAAK;AAAA,MAC9G,cAAc,CAAC,SAAS,MAAM;AAAA,MAC9B;AAAA,IACF,CAAC;AAED,WAAO,KAAK,WAAW,CAAC,EAAE,OAAO,MAAM,QAAQ,MAAM,CAAC;AACtD,WAAO,KAAK,SAAS,MAAM;AAAA,EAC7B,CAAC;AACH;","names":[]}

package/dist/index.d.ts

@@ -221,7 +221,7 @@ declare class DryScan {
      * 6. Recompute embeddings for affected units
      * 7. Update file tracking metadata
      */
-    updateIndex(): Promise<void>;
+    updateIndex(): Promise<string[]>;
     /**
      * Runs duplicate detection and returns a normalized report payload ready for persistence or display.
      */