`ocpm_engine.standalone`¶

Source-neutral Python facade over the standalone Rust engine. Public request and result values are ordinary mappings following the versioned 1.0 JSON contracts; this module only handles Python-to-JSON conversion, and algorithms execute in Rust with the GIL released.

`StandaloneEngine`¶

Run OCPM operations from files or canonical in-memory data.

Constructors¶

StandaloneEngine(canonical_log: Mapping[str, Any])
StandaloneEngine.from_ocel2_json(value: str | bytes | Path)
StandaloneEngine.from_xes(value: str | bytes | Path)
StandaloneEngine.from_sqlite(path: str | Path)
StandaloneEngine.from_duckdb_parquet(source: Mapping[str, Any])

canonical_log — a canonical log as a JSON object.
from_ocel2_json / from_xes — accept a Path to read, or the document itself as str/bytes.
from_sqlite — path to an OCEL 2.0 SQLite file.
from_duckdb_parquet — opens local or S3 Parquet through a deployment-supplied DuckDB installation; the source object selects the existing catalog, location, snapshot, layout, cache mode, and options. The engine never creates the catalog implicitly.

from ocpm_engine import StandaloneEngine

engine = StandaloneEngine.from_ocel2_json("events.json")

engine = StandaloneEngine.from_duckdb_parquet(
    {
        "database": {
            "kind": "existing",
            "path": "/catalog/analytics.duckdb",
            "read_only": True,
        },
        "location": {"kind": "local", "root": "/data/ocel-parquet"},
        "snapshot": {"kind": "current", "pointer": "CURRENT"},
        "layout": {"kind": "canonical_v1"},
        "cache": {"kind": "direct"},
        "options": {
            "memory_budget_bytes": 536_870_912,
            "result_cache_bytes": 67_108_864,
            "materialize_execution_relation": True,
        },
    }
)

Properties¶

Property	Type	Purpose
`provider_name`	`str`	Name of the active provider
`capabilities`	`list[str]`	Capability surface of the active provider

Methods¶

Every request-taking method accepts a versioned JSON object (a plain mapping) and returns a plain dictionary.

append(batch: Mapping) -> None

Atomically validate and append a canonical columnar batch; the source watermark advances only after the complete batch succeeds.

profile(view: Mapping | None = None) -> dict

Activity profile for a view (object types, time bounds).

query(request: Mapping) -> dict

Typed filtering and binding queries.

discover(request: Mapping) -> dict

DFG, OC-DFG, Alpha, process-tree, Petri-net, OCPN, and declarative discovery.

conformance(request: Mapping) -> dict
enhance(request: Mapping) -> dict

Conformance checking and model enhancement (frequencies, durations, bottlenecks).

fit_prediction(request: Mapping) -> dict
predict(request: Mapping) -> dict
evaluate_prediction(view: Mapping, target: str, *,
                    holdout_fraction: float = 0.2,
                    parameters: Mapping | None = None) -> dict

Prediction: fit a model, predict, or run a temporal holdout evaluation for target on the selected view.

execution_summary(request: Mapping) -> dict

Exact compact lifecycle, variant, DFG, and activity statistics without materializing results.

canonical_json(view: Mapping | None = None) -> dict
ocel2_json(view: Mapping | None = None) -> dict
xes(object_type: str, view: Mapping | None = None) -> str
write_sqlite(path: str | Path, view: Mapping | None = None) -> None
write_parquet_snapshot(root: str | Path, version: str,
                       view: Mapping | None = None) -> dict

Serialization: export a view as canonical JSON, OCEL 2.0 JSON, or XES (one object type per XES document); persist to OCEL SQLite; or write a new immutable canonical Parquet snapshot and CURRENT pointer.

explain(view: Mapping, capability: str) -> dict

Report the selected provider boundary for a capability without exposing provider-specific expressions.

Example¶

from ocpm_engine import StandaloneEngine

engine = StandaloneEngine.from_ocel2_json("events.json")

profile = engine.profile({"object_types": ["Order"]})
model = engine.discover(
    {
        "view": {"object_types": ["Order"]},
        "algorithm": "object_centric_dfg",
    }
)

`serialize_model`¶

serialize_model(artifact: Mapping, format: str = "json") -> str

Serialize a 1.0 model artifact as JSON, DOT, PNML, or SVG.

ocpm_engine.standalone¶

StandaloneEngine¶