databonk 0.0.1 → 0.0.2

package/dist/core/column.d.ts

@@ -1,4 +1,5 @@
-import { DataType } from '../utils/types.js';
+import { DataType, TypedArrayInstance } from '../utils/types.js';
+import { BitSet } from '../utils/bitset.js';
 export declare class Column<T = any> {
     readonly name: string;
     readonly dataType: DataType;
@@ -8,6 +9,24 @@ export declare class Column<T = any> {
     constructor(name: string, values: T[], dataType?: DataType);
     private createDataArray;
     get(index: number): T | null;
+    /**
+     * Get a value without bounds checking or null handling.
+     * Use only when caller ensures valid index and handles nulls separately.
+     * @internal
+     */
+    getRaw(index: number): T;
+    /**
+     * Get direct reference to the underlying data array.
+     * Use for batch operations that need raw access.
+     * @internal
+     */
+    getDataRef(): TypedArrayInstance | any[];
+    /**
+     * Get direct reference to the null bitmap.
+     * Use for batch null checking.
+     * @internal
+     */
+    getNullBitmapRef(): BitSet;
     isNull(index: number): boolean;
     slice(start?: number, end?: number): Column<T>;
     filter(predicate: (value: T | null, index: number) => boolean): Column<T>;
@@ -21,5 +40,16 @@ export declare class Column<T = any> {
     values(): Iterator<T | null>;
     toArray(): (T | null)[];
     static from<T>(name: string, values: T[], dataType?: DataType): Column<T>;
+    /**
+     * Create a Column directly from raw data without copying.
+     * Use for optimized construction when data is already in the correct format.
+     * @internal
+     */
+    static fromRaw<T>(name: string, data: TypedArrayInstance | any[], nullBitmap: BitSet, dataType: DataType): Column<T>;
+    /**
+     * Select rows by indices with optimized batch copying.
+     * Much faster than calling get() for each index.
+     */
+    selectIndices(indices: number[]): Column<T>;
 }
 //# sourceMappingURL=column.d.ts.map

package/dist/core/column.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"column.d.ts","sourceRoot":"","sources":["../../src/core/column.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAA6D,MAAM,mBAAmB,CAAC;AAGxG,qBAAa,MAAM,CAAC,CAAC,GAAG,GAAG;IACzB,SAAgB,IAAI,EAAE,MAAM,CAAC;IAC7B,SAAgB,QAAQ,EAAE,QAAQ,CAAC;IACnC,OAAO,CAAC,IAAI,CAA6B;IACzC,OAAO,CAAC,UAAU,CAAS;IAC3B,SAAgB,MAAM,EAAE,MAAM,CAAC;gBAEnB,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,QAAQ,CAAC,EAAE,QAAQ;IAS1D,OAAO,CAAC,eAAe;IAsBvB,GAAG,CAAC,KAAK,EAAE,MAAM,GAAG,CAAC,GAAG,IAAI;IAY5B,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO;IAI9B,KAAK,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,GAAG,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC;IAY9C,MAAM,CAAC,SAAS,EAAE,CAAC,KAAK,EAAE,CAAC,GAAG,IAAI,EAAE,KAAK,EAAE,MAAM,KAAK,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC;IAazE,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,KAAK,EAAE,CAAC,GAAG,IAAI,EAAE,KAAK,EAAE,MAAM,KAAK,CAAC,EAAE,WAAW,CAAC,EAAE,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC;IAWpF,GAAG,IAAI,MAAM;IAeb,IAAI,IAAI,MAAM;IASd,GAAG,IAAI,MAAM;IAqBb,GAAG,IAAI,MAAM;IAqBb,KAAK,IAAI,MAAM;IAIf,MAAM,IAAI,CAAC,EAAE;IAYZ,MAAM,IAAI,QAAQ,CAAC,CAAC,GAAG,IAAI,CAAC;IAM7B,OAAO,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC,EAAE;IAQvB,MAAM,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,QAAQ,CAAC,EAAE,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC;CAG1E"}
+{"version":3,"file":"column.d.ts","sourceRoot":"","sources":["../../src/core/column.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,kBAAkB,EAAyC,MAAM,mBAAmB,CAAC;AACxG,OAAO,EAAE,MAAM,EAAE,MAAM,oBAAoB,CAAC;AAE5C,qBAAa,MAAM,CAAC,CAAC,GAAG,GAAG;IACzB,SAAgB,IAAI,EAAE,MAAM,CAAC;IAC7B,SAAgB,QAAQ,EAAE,QAAQ,CAAC;IACnC,OAAO,CAAC,IAAI,CAA6B;IACzC,OAAO,CAAC,UAAU,CAAS;IAC3B,SAAgB,MAAM,EAAE,MAAM,CAAC;gBAEnB,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,QAAQ,CAAC,EAAE,QAAQ;IAS1D,OAAO,CAAC,eAAe;IAsBvB,GAAG,CAAC,KAAK,EAAE,MAAM,GAAG,CAAC,GAAG,IAAI;IAY5B;;;;OAIG;IACH,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,CAAC;IAIxB;;;;OAIG;IACH,UAAU,IAAI,kBAAkB,GAAG,GAAG,EAAE;IAIxC;;;;OAIG;IACH,gBAAgB,IAAI,MAAM;IAI1B,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO;IAI9B,KAAK,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,GAAG,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC;IAY9C,MAAM,CAAC,SAAS,EAAE,CAAC,KAAK,EAAE,CAAC,GAAG,IAAI,EAAE,KAAK,EAAE,MAAM,KAAK,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC;IAazE,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,KAAK,EAAE,CAAC,GAAG,IAAI,EAAE,KAAK,EAAE,MAAM,KAAK,CAAC,EAAE,WAAW,CAAC,EAAE,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC;IAWpF,GAAG,IAAI,MAAM;IAeb,IAAI,IAAI,MAAM;IASd,GAAG,IAAI,MAAM;IAqBb,GAAG,IAAI,MAAM;IAqBb,KAAK,IAAI,MAAM;IAIf,MAAM,IAAI,CAAC,EAAE;IAYZ,MAAM,IAAI,QAAQ,CAAC,CAAC,GAAG,IAAI,CAAC;IAM7B,OAAO,IAAI,CAAC,CAAC,GAAG,IAAI,CAAC,EAAE;IAQvB,MAAM,CAAC,IAAI,CAAC,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,CAAC,EAAE,EAAE,QAAQ,CAAC,EAAE,QAAQ,GAAG,MAAM,CAAC,CAAC,CAAC;IAIzE;;;;OAIG;IACH,MAAM,CAAC,OAAO,CAAC,CAAC,EACd,IAAI,EAAE,MAAM,EACZ,IAAI,EAAE,kBAAkB,GAAG,GAAG,EAAE,EAChC,UAAU,EAAE,MAAM,EAClB,QAAQ,EAAE,QAAQ,GACjB,MAAM,CAAC,CAAC,CAAC;IAWZ;;;OAGG;IACH,aAAa,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC;CAmC5C"}

package/dist/core/dataframe.d.ts

@@ -1,5 +1,35 @@
 import { Column } from './column';
 export type RowObject = Record<string, any>;
+/**
+ * RowProxy provides zero-allocation row access for iteration.
+ * Reuses a single object while iterating, avoiding object creation per row.
+ */
+export declare class RowProxy {
+    private columnCache;
+    private index;
+    constructor(df: DataFrame);
+    /**
+     * Set the current row index.
+     * @returns this for chaining
+     */
+    setIndex(i: number): this;
+    /**
+     * Get a value from the current row.
+     */
+    get(col: string): any;
+    /**
+     * Get a value without null checking (faster for non-null columns).
+     */
+    getRaw(col: string): any;
+    /**
+     * Check if a column value is null at the current row.
+     */
+    isNull(col: string): boolean;
+    /**
+     * Get the current row index.
+     */
+    getIndex(): number;
+}
 export declare class DataFrame {
     private columns;
     readonly length: number;
@@ -14,6 +44,16 @@ export declare class DataFrame {
     filter(predicate: (row: RowObject, index: number) => boolean): DataFrame;
     slice(start?: number, end?: number): DataFrame;
     selectRows(indices: number[]): DataFrame;
+    /**
+     * Filter rows using a predicate function that receives a RowProxy.
+     * More efficient than filter() as it avoids creating a new object per row.
+     */
+    filterByIndex(predicate: (index: number, proxy: RowProxy) => boolean): DataFrame;
+    /**
+     * Create a RowProxy for efficient iteration.
+     * Use this when you need to access multiple columns per row without allocation.
+     */
+    createRowProxy(): RowProxy;
     getRow(index: number): RowObject;
     rows(): IterableIterator<RowObject>;
     head(n?: number): DataFrame;

package/dist/core/dataframe.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"dataframe.d.ts","sourceRoot":"","sources":["../../src/core/dataframe.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,UAAU,CAAC;AAGlC,MAAM,MAAM,SAAS,GAAG,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;AAE5C,qBAAa,SAAS;IACpB,OAAO,CAAC,OAAO,CAAkC;IACjD,SAAgB,MAAM,EAAE,MAAM,CAAC;gBAEnB,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,MAAM,EAAE;IAiBnD,IAAI,WAAW,IAAI,MAAM,EAAE,CAE1B;IAED,IAAI,WAAW,IAAI,MAAM,CAExB;IAED,MAAM,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM;IAQ5B,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO;IAIhC,SAAS,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS;IAUpC,YAAY,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS;IAUrC,MAAM,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,SAAS;IAapC,MAAM,CAAC,SAAS,EAAE,CAAC,GAAG,EAAE,SAAS,EAAE,KAAK,EAAE,MAAM,KAAK,OAAO,GAAG,SAAS;IAaxE,KAAK,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS;IAa9C,UAAU,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,SAAS;IAWxC,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS;IAa/B,IAAI,IAAI,gBAAgB,CAAC,SAAS,CAAC;IAMpC,IAAI,CAAC,CAAC,GAAE,MAAU,GAAG,SAAS;IAI9B,IAAI,CAAC,CAAC,GAAE,MAAU,GAAG,SAAS;IAI9B,IAAI,CAAC,UAAU,EAAE,MAAM,EAAE,SAAS,GAAE,OAAc,GAAG,SAAS;IA6B9D,IAAI,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,SAAS;IAYlC,MAAM,CAAC,aAAa,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,SAAS;IAWxD,OAAO,IAAI,SAAS,EAAE;IAQtB,SAAS,IAAI,MAAM,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC;IAQlC,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,MAAM,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,GAAG,SAAS;IAQjE,MAAM,CAAC,QAAQ,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,SAAS;IAgB7C,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,GAAG,SAAS;CAS3D"}
+{"version":3,"file":"dataframe.d.ts","sourceRoot":"","sources":["../../src/core/dataframe.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,UAAU,CAAC;AAGlC,MAAM,MAAM,SAAS,GAAG,MAAM,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC;AAE5C;;;GAGG;AACH,qBAAa,QAAQ;IACnB,OAAO,CAAC,WAAW,CAAkC;IACrD,OAAO,CAAC,KAAK,CAAa;gBAEd,EAAE,EAAE,SAAS;IAMzB;;;OAGG;IACH,QAAQ,CAAC,CAAC,EAAE,MAAM,GAAG,IAAI;IAKzB;;OAEG;IACH,GAAG,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG;IAQrB;;OAEG;IACH,MAAM,CAAC,GAAG,EAAE,MAAM,GAAG,GAAG;IAIxB;;OAEG;IACH,MAAM,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO;IAI5B;;OAEG;IACH,QAAQ,IAAI,MAAM;CAGnB;AAED,qBAAa,SAAS;IACpB,OAAO,CAAC,OAAO,CAAkC;IACjD,SAAgB,MAAM,EAAE,MAAM,CAAC;gBAEnB,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,MAAM,EAAE;IAiBnD,IAAI,WAAW,IAAI,MAAM,EAAE,CAE1B;IAED,IAAI,WAAW,IAAI,MAAM,CAExB;IAED,MAAM,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM;IAQ5B,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO;IAIhC,SAAS,CAAC,MAAM,EAAE,MAAM,GAAG,SAAS;IAUpC,YAAY,CAAC,IAAI,EAAE,MAAM,GAAG,SAAS;IAUrC,MAAM,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,SAAS;IAapC,MAAM,CAAC,SAAS,EAAE,CAAC,GAAG,EAAE,SAAS,EAAE,KAAK,EAAE,MAAM,KAAK,OAAO,GAAG,SAAS;IAuBxE,KAAK,CAAC,KAAK,CAAC,EAAE,MAAM,EAAE,GAAG,CAAC,EAAE,MAAM,GAAG,SAAS;IAa9C,UAAU,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,SAAS;IAWxC;;;OAGG;IACH,aAAa,CAAC,SAAS,EAAE,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,QAAQ,KAAK,OAAO,GAAG,SAAS;IAahF;;;OAGG;IACH,cAAc,IAAI,QAAQ;IAI1B,MAAM,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS;IAa/B,IAAI,IAAI,gBAAgB,CAAC,SAAS,CAAC;IAMpC,IAAI,CAAC,CAAC,GAAE,MAAU,GAAG,SAAS;IAI9B,IAAI,CAAC,CAAC,GAAE,MAAU,GAAG,SAAS;IAI9B,IAAI,CAAC,UAAU,EAAE,MAAM,EAAE,SAAS,GAAE,OAAc,GAAG,SAAS;IA6B9D,IAAI,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,SAAS;IAYlC,MAAM,CAAC,aAAa,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,GAAG,SAAS;IAWxD,OAAO,IAAI,SAAS,EAAE;IAQtB,SAAS,IAAI,MAAM,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC;IAQlC,MAAM,CAAC,IAAI,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,MAAM,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,GAAG,SAAS;IAQjE,MAAM,CAAC,QAAQ,CAAC,IAAI,EAAE,SAAS,EAAE,GAAG,SAAS;IAgB7C,MAAM,CAAC,WAAW,CAAC,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,GAAG,EAAE,CAAC,GAAG,SAAS;CAS3D"}

package/dist/core/index-cache.d.ts

@@ -0,0 +1,44 @@
+import { DataFrame } from './dataframe.js';
+/**
+ * CachedIndex stores a hash index along with metadata about when it was created.
+ */
+export interface CachedIndex {
+    columns: string[];
+    index: Map<string, number[]>;
+    createdAt: number;
+}
+/**
+ * IndexCache provides caching for hash indices used in join and groupBy operations.
+ * Uses WeakMap to allow garbage collection of DataFrames.
+ */
+export declare class IndexCache {
+    private cache;
+    private maxAge;
+    constructor(maxAge?: number);
+    /**
+     * Generate a cache key from column names.
+     */
+    private getCacheKey;
+    /**
+     * Get a cached index for the given DataFrame and columns.
+     * Returns null if not cached or expired.
+     */
+    getIndex(df: DataFrame, columns: string[]): Map<string, number[]> | null;
+    /**
+     * Store an index in the cache.
+     */
+    setIndex(df: DataFrame, columns: string[], index: Map<string, number[]>): void;
+    /**
+     * Invalidate all cached indices for a DataFrame.
+     */
+    invalidate(df: DataFrame): void;
+    /**
+     * Clear all cached indices.
+     */
+    clear(): void;
+}
+/**
+ * Global index cache instance for shared use across operations.
+ */
+export declare const globalIndexCache: IndexCache;
+//# sourceMappingURL=index-cache.d.ts.map

package/dist/core/index-cache.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index-cache.d.ts","sourceRoot":"","sources":["../../src/core/index-cache.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAE3C;;GAEG;AACH,MAAM,WAAW,WAAW;IAC1B,OAAO,EAAE,MAAM,EAAE,CAAC;IAClB,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,CAAC,CAAC;IAC7B,SAAS,EAAE,MAAM,CAAC;CACnB;AAED;;;GAGG;AACH,qBAAa,UAAU;IACrB,OAAO,CAAC,KAAK,CAA+D;IAC5E,OAAO,CAAC,MAAM,CAAS;gBAEX,MAAM,GAAE,MAAc;IAIlC;;OAEG;IACH,OAAO,CAAC,WAAW;IAInB;;;OAGG;IACH,QAAQ,CAAC,EAAE,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,EAAE,GAAG,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,CAAC,GAAG,IAAI;IAkBxE;;OAEG;IACH,QAAQ,CAAC,EAAE,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,EAAE,EAAE,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,MAAM,EAAE,CAAC,GAAG,IAAI;IAe9E;;OAEG;IACH,UAAU,CAAC,EAAE,EAAE,SAAS,GAAG,IAAI;IAI/B;;OAEG;IACH,KAAK,IAAI,IAAI;CAId;AAED;;GAEG;AACH,eAAO,MAAM,gBAAgB,YAAmB,CAAC"}