@x12i/memorix-resolvers

Pluggable resolver APIs for Memorix. Resolvers normalize join keys, bridge lookups, and domain-specific relationship signals without baking domain knowledge into the pipeline engine.

Used in:

• Pipeline resolution / enrichment stages — via descriptor processing.resolvers (see memorix-pipeline)
• Associator requests — optional resolvers / normalization blocks before cross-record join rules

Design

• Resolver registry — register built-in and custom resolver types.
• Resolver steps — run one or more resolvers against a source record.
• Domain plug-ins — networking, pharma, marketing, identity, or any future domain implements the same ResolverDefinition API.
• Remote resolvers — HTTP providers registered through Explorer /pipeline/registry/parts and merged at pipeline run time.

The engine does not branch on resolver type. A CIDR containment resolver, pharma SKU resolver, and marketing campaign resolver are all ordinary plug-ins.

Usage

import {
  createResolverRegistry,
  builtinResolvers,
  runResolverSteps,
  type ResolverDefinition,
} from "@x12i/memorix-resolvers";

const registry = createResolverRegistry(builtinResolvers);

const myResolver: ResolverDefinition<{ outputPath: string }> = {
  type: "my-domain",
  create(config) {
    return {
      async resolve(input) {
        return {
          outputs: [{ path: config.outputPath, values: ["resolved"] }],
          stats: { inputsScanned: 1, valuesParsed: 1, valuesRejected: 0, matchesFound: 1, ambiguousMatches: 0 },
          diagnostics: [],
        };
      },
    };
  },
};

registry.register(myResolver);

const result = await runResolverSteps(registry, {
  steps: [{ type: "my-domain", resolverKey: "step-1", config: { outputPath: "data.key" } }],
  record: { data: {} },
});

Built-in resolvers

| Type | Purpose | Typical pipeline stage | |------|---------|------------------------| | exact-match | Copy normalized values from source paths to an output path | resolution | | multi-value | Split comma-separated scalars into array join fields | resolution | | bridge-index | Map readable keys through a bridge collection to target join keys | resolution | | markdown-key-value | Parse **Label:** value markdown strings into plain objects | resolution | | markdown-pipe-records | Parse pipe-delimited **ID:** … \| **Name:** … lines into object arrays | resolution | | path-existence-classifier | Classify path objects into attackPathExistence enum (supported|possible|limited|none|unknown) from pathContext / attackPathMeaning | resolution | | identifier-key | Ensure concept.identifier.identifierKey (kind + values, :-joined) | resolution | | vulnerability-concept | Fix vulnerability snapshot title + hostVulnerability identifier + vulnerabilityGroupIdentifierKey | resolution | | subnet-cidr | Normalize CIDR or resolve asset IPs to canonical subnet CIDRs | resolution | | subnet-key | Derive legacy subnetIp / subnetId from IP or CIDR network portion | resolution | | topology-extractor | Build associatedTopology from associatedData or subnet IP lookup | resolution | | zone-from-topology | Remediate UNKNOWN subnet data.zone from topology-raw graph | enrichment |

Assignments are declared per object type in entity descriptor processing.resolvers. The pipeline parts catalog (loadPipelinePartsCatalog) indexes which types use each resolver.

Related packages

• Running resolvers — operational guide
• @x12i/memorix-pipeline — runs resolver steps during resolution/enrichment stages
• @x12i/memorix-associator — may run resolvers in association request rules before joins
• Operational design note