document-dataply 0.0.10-alpha.6 → 0.0.10-alpha.7

package/dist/cjs/index.js

@@ -10482,7 +10482,7 @@ var SELECTIVITY = {
   /** FTS 통계 없을 때 보수적 추정 */
   FTS_DEFAULT: 0.5,
   /** 정렬 비용 가중치 (orderBy 미지원 시) */
-  SORT_PENALTY: 0.3,
+  SORT_PENALTY: 0.5,
   /** 인메모리 정렬이 유의미해지는 임계 문서 수 */
   SORT_THRESHOLD: 1e4
 };
@@ -10668,9 +10668,12 @@ var Optimizer = class {
    * 비용 계산: effectiveScanCost + sortPenalty
    * - effectiveScanCost: 인덱스 순서 지원 + limit 존재 시 조기 종료 이점 반영
    * - sortPenalty: 인메모리 정렬의 절대 문서 수 기반 비용
+   * - hasUncoveredFilters: 드라이버가 커버하지 못하는 비-FTS 필터 존재 여부
+   *   true일 경우 topK/N 조기 종료를 적용하지 않음
+   *   (uncovered 필터가 행을 탈락시킬 수 있어 topK개 이상 스캔 필요)
    */
-  calculateCost(selectivity, isIndexOrderSupported, orderByField, N, topK) {
-    const effectiveScanCost = isIndexOrderSupported && isFinite(topK) && N > 0 ? Math.min(topK / N, selectivity) : selectivity;
+  calculateCost(selectivity, isIndexOrderSupported, orderByField, N, topK, hasUncoveredFilters = false) {
+    const effectiveScanCost = isIndexOrderSupported && isFinite(topK) && N > 0 && !hasUncoveredFilters ? Math.min(topK / N, selectivity) : selectivity;
     const estimatedSortDocs = selectivity * N;
     const sortPenalty = orderByField && !isIndexOrderSupported ? Math.min(estimatedSortDocs / SELECTIVITY.SORT_THRESHOLD, 1) * SELECTIVITY.SORT_PENALTY : 0;
     return effectiveScanCost + sortPenalty;
@@ -10699,9 +10702,19 @@ var Optimizer = class {
           orderByField
         );
         if (candidate) {
+          const hasUncoveredFilters = ![...queryFields].every(
+            (f) => candidate.coveredFields.includes(f)
+          );
           candidates.push({
             ...candidate,
-            cost: this.calculateCost(candidate.selectivity, candidate.isIndexOrderSupported, orderByField, N, topK)
+            cost: this.calculateCost(
+              candidate.selectivity,
+              candidate.isIndexOrderSupported,
+              orderByField,
+              N,
+              topK,
+              hasUncoveredFilters
+            )
           });
         }
       } else if (config.type === "fts") {
@@ -10716,7 +10729,14 @@ var Optimizer = class {
         if (candidate) {
           candidates.push({
             ...candidate,
-            cost: this.calculateCost(candidate.selectivity, candidate.isIndexOrderSupported, orderByField, N, topK)
+            cost: this.calculateCost(
+              candidate.selectivity,
+              candidate.isIndexOrderSupported,
+              orderByField,
+              N,
+              topK,
+              true
+            )
           });
         }
       }
@@ -11035,12 +11055,24 @@ var QueryManager = class {
     }
     return true;
   }
-  adjustChunkSize(currentChunkSize, chunkTotalSize) {
-    if (chunkTotalSize <= 0) return currentChunkSize;
-    const { verySmallChunkSize, smallChunkSize } = this.getFreeMemoryChunkSize();
-    if (chunkTotalSize < verySmallChunkSize) return currentChunkSize * 2;
-    if (chunkTotalSize > smallChunkSize) return Math.max(Math.floor(currentChunkSize / 2), 20);
-    return currentChunkSize;
+  /**
+   * 최적화 공식: x = x * Math.sqrt(z / n)
+   * @param currentChunkSize 현재 청크 크기
+   * @param matchedCount 매칭된 문서 개수
+   * @param limit 최대 문서 개수
+   * @param chunkTotalSize 청크 내 문서 총 크기
+   * @returns 
+   */
+  adjustChunkSize(currentChunkSize, matchedCount, limit, chunkTotalSize) {
+    if (matchedCount <= 0 || chunkTotalSize <= 0) return currentChunkSize;
+    const n = Math.max(matchedCount, 1);
+    const z = isFinite(limit) ? limit : currentChunkSize * 10;
+    const nextChunkSize = Math.ceil(currentChunkSize * Math.sqrt(z / n));
+    const { smallChunkSize } = this.getFreeMemoryChunkSize();
+    const avgDocSize = chunkTotalSize / currentChunkSize;
+    const maxSafeChunkSize = Math.max(Math.floor(smallChunkSize / avgDocSize), 20);
+    const finalChunkSize = Math.max(Math.min(nextChunkSize, maxSafeChunkSize), 20);
+    return finalChunkSize;
   }
   async *processChunkedKeysWithVerify(keysStream, startIdx, initialChunkSize, limit, ftsConditions, compositeVerifyConditions, others, tx) {
     const verifyOthers = others.filter((o) => !o.isFtsMatch);
@@ -11056,6 +11088,7 @@ var QueryManager = class {
     let chunk = [];
     let chunkSize = 0;
     let dropped = 0;
+    let nAccumulated = 0;
     const processChunk = async (pks) => {
       const docs = [];
       const rawResults = await this.api.selectMany(new Float64Array(pks), false, tx);
@@ -11107,8 +11140,9 @@ var QueryManager = class {
         }
         docs.push(doc);
       }
+      nAccumulated += docs.length;
       if (!isReadQuotaLimited) {
-        currentChunkSize = this.adjustChunkSize(currentChunkSize, chunkTotalSize);
+        currentChunkSize = this.adjustChunkSize(currentChunkSize, nAccumulated, limit, chunkTotalSize);
       }
       return docs;
     };

package/dist/types/core/Optimizer.d.ts

@@ -38,6 +38,9 @@ export declare class Optimizer<T extends Record<string, any>> {
      * 비용 계산: effectiveScanCost + sortPenalty
      * - effectiveScanCost: 인덱스 순서 지원 + limit 존재 시 조기 종료 이점 반영
      * - sortPenalty: 인메모리 정렬의 절대 문서 수 기반 비용
+     * - hasUncoveredFilters: 드라이버가 커버하지 못하는 비-FTS 필터 존재 여부
+     *   true일 경우 topK/N 조기 종료를 적용하지 않음
+     *   (uncovered 필터가 행을 탈락시킬 수 있어 topK개 이상 스캔 필요)
      */
     private calculateCost;
     /**

package/dist/types/core/QueryManager.d.ts

@@ -45,7 +45,15 @@ export declare class QueryManager<T extends DocumentJSON> {
         condition: any;
     }[]): boolean;
     verifyValue(value: Primitive, condition: any): boolean;
-    adjustChunkSize(currentChunkSize: number, chunkTotalSize: number): number;
+    /**
+     * 최적화 공식: x = x * Math.sqrt(z / n)
+     * @param currentChunkSize 현재 청크 크기
+     * @param matchedCount 매칭된 문서 개수
+     * @param limit 최대 문서 개수
+     * @param chunkTotalSize 청크 내 문서 총 크기
+     * @returns
+     */
+    adjustChunkSize(currentChunkSize: number, matchedCount: number, limit: number, chunkTotalSize: number): number;
     processChunkedKeysWithVerify(keysStream: AsyncIterableIterator<number>, startIdx: number, initialChunkSize: number, limit: number, ftsConditions: {
         field: string;
         matchTokens: string[];

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "document-dataply",
-  "version": "0.0.10-alpha.6",
+  "version": "0.0.10-alpha.7",
   "description": "Simple and powerful JSON document database supporting complex queries and flexible indexing policies.",
   "license": "MIT",
   "author": "izure <admin@izure.org>",