Skip to main content

Architecture

CIMXML is organized around a streaming parser that turns a CIMXML document into triples, a set of CIM-aware graph structures, and a CimDatasetGraph that exposes the model's logical structure. This page describes the core components, the parse pipeline, and the data model the library produces.

Core components​

1. Parser (de.soptim.opencgmes.cimxml.parser)​

  • CimXmlParser — the main entry point. Its parseCimModel(...) overloads accept a Path, InputStream, or Reader, and parseAndRegisterCimProfile(Path) loads profile ontologies into the parser's registry.
  • ReaderCIMXML_StAX_SR — the StAX-based XML reader implementation.
  • ParserCIMXML_StAX_SR — the core parsing logic with CIMXML-specific handling (CIM version detection, UUID normalization, profile-driven datatype resolution, and the IEC 61970-552 processing instruction).

2. Graph structures (de.soptim.opencgmes.cimxml.graph)​

  • CimProfile — represents a CIM profile ontology. Built-in implementations cover versions 16, 17, and 18; the CimNamespaceFactoryRegistry lets you map any other CIM namespace URI to a custom CimProfile implementation.
  • CimModelHeader — a wrapper exposing model header information (getModel, getProfiles, getSupersedes, getDependentOn).
  • FastDeltaGraph — an efficient delta graph used when applying difference models.
  • DisjointMultiUnion — a union graph without duplicate elimination, used to combine header and body into a single graph.

3. Dataset (de.soptim.opencgmes.cimxml.sparql.core)​

  • CimDatasetGraph — an extended Jena DatasetGraph with CIM-specific operations (isFullModel, getBody, getForwardDifferences, differenceModelToFullModel, and so on).
  • LinkedCimDatasetGraph — the implementation supporting the multiple named graphs that make up a CIM model.

4. Profile registry (de.soptim.opencgmes.cimxml.rdfs)​

  • CimProfileRegistry — the interface for registering and querying CIM profiles and resolving property datatypes.
  • CimProfileRegistryStd — the standard implementation with caching. See Profiles & datatypes.

Parse pipeline​

A CIMXML file is read by the StAX reader, parsed into triples by the CIMXML parser, routed into the appropriate graphs, and assembled into a CimDatasetGraph:

Datatype resolution happens during parsing using the profiles registered on the parser, and CIM UUIDs are normalized to canonical urn:uuid: IRIs as triples are produced.

Data model​

The library organizes a CIMXML document into distinct graphs depending on whether it is a full model or a difference model:

FullModel — the header graph carries metadata (model IRI, profiles, supersedes, dependentOn) and the body is the default graph holding the actual model data. fullModelToSingleGraph() combines the two into a single DisjointMultiUnion graph when you need a flat view.

DifferenceModel — the header graph carries metadata, and the forward, reverse, and (optional) preconditions containers each become their own graph. Applying the model with differenceModelToFullModel(...) yields a body graph (see Difference models).

Reaching the header

The header lives in its own named graph (urn:FullModel / urn:DifferenceModel), but you normally reach it through getModelHeader() rather than by graph name.