databonk 0.0.2 → 0.0.3

package/dist/operations/join.d.ts

@@ -1,7 +1,33 @@
 import { DataFrame } from '../core/dataframe.js';
 export type JoinType = 'inner' | 'left' | 'right' | 'outer';
+/**
+ * Join algorithm selection:
+ * - 'hash': Hash join - good for most cases, O(n+m) with good hash distribution
+ * - 'sort-merge': Sort-merge join - better cache locality for large sorted datasets
+ * - 'auto': Automatically select based on available indices
+ */
+export type JoinAlgorithm = 'hash' | 'sort-merge' | 'auto';
+/**
+ * Options for join operations.
+ */
+export interface JoinOptions {
+    /** Column name suffixes for overlapping non-key columns. Default: ['_x', '_y'] */
+    suffixes?: [string, string];
+    /** Join algorithm to use. Default: 'auto' */
+    algorithm?: JoinAlgorithm;
+    /** Whether to use existing indices. Default: true */
+    useIndices?: boolean;
+}
 export declare class Joiner {
-    static join(left: DataFrame, right: DataFrame, on: string | string[], how?: JoinType, suffixes?: [string, string]): DataFrame;
+    static join(left: DataFrame, right: DataFrame, on: string | string[], how?: JoinType, options?: JoinOptions | [string, string]): DataFrame;
+    /**
+     * Select the best join algorithm based on available indices and data characteristics.
+     */
+    private static selectAlgorithm;
+    /**
+     * Get an existing hash index or build a new one.
+     */
+    private static getOrBuildHashIndex;
     private static validateJoinKeys;
     private static buildHashIndex;
     /**
@@ -12,11 +38,21 @@ export declare class Joiner {
     private static leftJoin;
     private static rightJoin;
     private static outerJoin;
+    /**
+     * Sort-merge join algorithm.
+     * Efficient when both sides have sorted indices - uses sequential access pattern
+     * which is more cache-friendly for large datasets.
+     */
+    private static sortMergeJoin;
+    /**
+     * Get sorted entries from an existing sorted index or build them.
+     */
+    private static getOrBuildSortedEntries;
     private static buildJoinedDataFrame;
 }
 declare module '../core/dataframe.js' {
     interface DataFrame {
-        join(other: DataFrame, on: string | string[], how?: JoinType, suffixes?: [string, string]): DataFrame;
+        join(other: DataFrame, on: string | string[], how?: JoinType, options?: JoinOptions | [string, string]): DataFrame;
     }
 }
 //# sourceMappingURL=join.d.ts.map

package/dist/operations/join.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"join.d.ts","sourceRoot":"","sources":["../../src/operations/join.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAa,MAAM,sBAAsB,CAAC;AAI5D,MAAM,MAAM,QAAQ,GAAG,OAAO,GAAG,MAAM,GAAG,OAAO,GAAG,OAAO,CAAC;AAE5D,qBAAa,MAAM;IACjB,MAAM,CAAC,IAAI,CACT,IAAI,EAAE,SAAS,EACf,KAAK,EAAE,SAAS,EAChB,EAAE,EAAE,MAAM,GAAG,MAAM,EAAE,EACrB,GAAG,GAAE,QAAkB,EACvB,QAAQ,GAAE,CAAC,MAAM,EAAE,MAAM,CAAgB,GACxC,SAAS;IAsBZ,OAAO,CAAC,MAAM,CAAC,gBAAgB;IAW/B,OAAO,CAAC,MAAM,CAAC,cAAc;IAqC7B;;OAEG;IACH,OAAO,CAAC,MAAM,CAAC,aAAa;IAa5B,OAAO,CAAC,MAAM,CAAC,SAAS;IAwCxB,OAAO,CAAC,MAAM,CAAC,QAAQ;IAuCvB,OAAO,CAAC,MAAM,CAAC,SAAS;IAuCxB,OAAO,CAAC,MAAM,CAAC,SAAS;IAoDxB,OAAO,CAAC,MAAM,CAAC,oBAAoB;CA+CpC;AAED,OAAO,QAAQ,sBAAsB,CAAC;IACpC,UAAU,SAAS;QACjB,IAAI,CACF,KAAK,EAAE,SAAS,EAChB,EAAE,EAAE,MAAM,GAAG,MAAM,EAAE,EACrB,GAAG,CAAC,EAAE,QAAQ,EACd,QAAQ,CAAC,EAAE,CAAC,MAAM,EAAE,MAAM,CAAC,GAC1B,SAAS,CAAC;KACd;CACF"}
+{"version":3,"file":"join.d.ts","sourceRoot":"","sources":["../../src/operations/join.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAa,MAAM,sBAAsB,CAAC;AAK5D,MAAM,MAAM,QAAQ,GAAG,OAAO,GAAG,MAAM,GAAG,OAAO,GAAG,OAAO,CAAC;AAE5D;;;;;GAKG;AACH,MAAM,MAAM,aAAa,GAAG,MAAM,GAAG,YAAY,GAAG,MAAM,CAAC;AAE3D;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,kFAAkF;IAClF,QAAQ,CAAC,EAAE,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAC5B,6CAA6C;IAC7C,SAAS,CAAC,EAAE,aAAa,CAAC;IAC1B,qDAAqD;IACrD,UAAU,CAAC,EAAE,OAAO,CAAC;CACtB;AAED,qBAAa,MAAM;IACjB,MAAM,CAAC,IAAI,CACT,IAAI,EAAE,SAAS,EACf,KAAK,EAAE,SAAS,EAChB,EAAE,EAAE,MAAM,GAAG,MAAM,EAAE,EACrB,GAAG,GAAE,QAAkB,EACvB,OAAO,GAAE,WAAW,GAAG,CAAC,MAAM,EAAE,MAAM,CAAM,GAC3C,SAAS;IA4CZ;;OAEG;IACH,OAAO,CAAC,MAAM,CAAC,eAAe;IAkC9B;;OAEG;IACH,OAAO,CAAC,MAAM,CAAC,mBAAmB;IAkBlC,OAAO,CAAC,MAAM,CAAC,gBAAgB;IAW/B,OAAO,CAAC,MAAM,CAAC,cAAc;IAqC7B;;OAEG;IACH,OAAO,CAAC,MAAM,CAAC,aAAa;IAa5B,OAAO,CAAC,MAAM,CAAC,SAAS;IAwCxB,OAAO,CAAC,MAAM,CAAC,QAAQ;IAuCvB,OAAO,CAAC,MAAM,CAAC,SAAS;IAuCxB,OAAO,CAAC,MAAM,CAAC,SAAS;IAoDxB;;;;OAIG;IACH,OAAO,CAAC,MAAM,CAAC,aAAa;IAoF5B;;OAEG;IACH,OAAO,CAAC,MAAM,CAAC,uBAAuB;IA+BtC,OAAO,CAAC,MAAM,CAAC,oBAAoB;CAsDpC;AAED,OAAO,QAAQ,sBAAsB,CAAC;IACpC,UAAU,SAAS;QACjB,IAAI,CACF,KAAK,EAAE,SAAS,EAChB,EAAE,EAAE,MAAM,GAAG,MAAM,EAAE,EACrB,GAAG,CAAC,EAAE,QAAQ,EACd,OAAO,CAAC,EAAE,WAAW,GAAG,CAAC,MAAM,EAAE,MAAM,CAAC,GACvC,SAAS,CAAC;KACd;CACF"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "databonk",
-  "version": "0.0.2",
+  "version": "0.0.3",
   "type": "module",
   "description": "A lightweight, fast data frame library for JavaScript and TypeScript",
   "main": "dist/index.js",