@memberjunction/record-set-processor-base

Client-safe foundation for the MemberJunction Record Set Processing substrate: the shared types, the three pluggable seams, and the built-in source adapters. The server engine that drives these lives in @memberjunction/record-set-processor.

What's in here

The three seams

A processing job is a composition of three independent choices:

• Source — what records to iterate (a User View, a List, an ad-hoc filter, an in-memory array, or a keyset sweep). Sources paginate themselves and hand back an opaque ProcessCursor the engine round-trips for resume.
• Processor — what to do with each record (an Action, an Agent, an Infer-&-Write-Back step, or a function), returning a RecordResult (Succeeded / Failed / Skipped).
• Tracker — where to persist run + per-record audit. The default writes the generic MJ: Process Runs / MJ: Process Run Details tables; domain consumers can supply their own.

Source adapters

Implementing a source

import { IRecordSetSource, RecordBatch, ProcessCursor, SourceDescriptor } from '@memberjunction/record-set-processor-base';

export class MySource implements IRecordSetSource {
    Describe(): SourceDescriptor { return { SourceType: 'Filter', SourceFilter: '...' }; }
    async NextBatch(cursor: ProcessCursor | undefined, batchSize: number, contextUser, provider): Promise<RecordBatch> {
        // fetch up to `batchSize` records after `cursor`, return them plus the next cursor + exhausted flag
    }
}
Copy

This package carries no server-only dependencies and is safe to import on the client.