@autobe/agent/lib/utils/divideArray.js

AI backend server code generator

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.divideArray = divideArray;
/**
 * Divides an array into chunks of at most `capacity` elements each for parallel
 * batch processing.
 *
 * Used with `executeCachedBatch` to split large task lists into smaller batches
 * that can be processed concurrently. The `capacity` parameter controls the
 * maximum number of elements per chunk, not the number of chunks. This enables
 * balancing parallelism against prompt cache efficiency.
 *
 * For example, dividing 100 operations with capacity=3 creates 34 chunks of 3
 * operations each (the last chunk may have fewer elements if the array length
 * is not a multiple of capacity). This allows processing 100 operations in 34
 * parallel batches of up to 3 operations each.
 *
 * @param props Object containing the array to divide and the maximum chunk size
 *   (`capacity`)
 * @returns Array of chunks with at most `capacity` elements each
 * @throws Error if capacity is non-positive, NaN, or Infinity
 */
function divideArray(props) {
    if (props.capacity <= 0) {
        throw new Error("Capacity must be a positive integer");
    }
    if (Number.isNaN(props.capacity)) {
        throw new TypeError("Capacity must be a positive integer");
    }
    if (props.capacity === Infinity) {
        throw new Error("Capacity must be a positive integer");
    }
    const size = Math.ceil(props.array.length / props.capacity);
    const capacity = Math.ceil(props.array.length / size);
    const replica = props.array.slice();
    return Array.from({ length: size }, () => replica.splice(0, capacity));
}
//# sourceMappingURL=divideArray.js.map