{ "module": "dagpipe.engine", "content": { "path": "dagpipe.engine", "docstring": "# Summary\n\nExecution engine responsible for running pipelines and graphs.\n\nThe `Engine` executes `Node` objects and propagates immutable `State` instances\nthrough either a linear sequence or a directed acyclic graph (`Graph`).\nIt orchestrates execution order, branching, and state propagation.\n\n---\n\n# Guarantees\n\n- Deterministic execution and consistent state lineage.\n- Orchestrates execution without modifying `Graph`, `Node`, or `State` objects.", "objects": { "Graph": { "name": "Graph", "kind": "class", "path": "dagpipe.engine.Graph", "signature": "Graph()", "docstring": "Directed Acyclic Graph defining execution topology of `Node` objects.\n\nNotes:\n **Responsibilities:**\n\n - Stores node connectivity and validates that the topology remains acyclic.\n - Structure determines how `State` flows between nodes during execution.\n\n **Guarantees:**\n\n - Topology is acyclic. Node relationships remain consistent.\n - Thread-safe for concurrent reads after construction.", "members": { "add_edge": { "name": "add_edge", "kind": "function", "path": "dagpipe.engine.Graph.add_edge", "signature": "add_edge(src: Node, dst: Node)", "docstring": "Add a directed edge from `src` to `dst`.\n\nArgs:\n src (Node):\n Source node.\n\n dst (Node):\n Destination node.\n\nRaises:\n TypeError:\n If `src` or `dst` is not a `Node`.\n\n ValueError:\n If the edge would create a cycle or if `src` and `dst` are common.\n\nNotes:\n - Validates node types.\n - Prevents cycles.\n - Registers nodes if not present.\n - Updates parent and child mappings." }, "add_root": { "name": "add_root", "kind": "function", "path": "dagpipe.engine.Graph.add_root", "signature": "add_root(node: Node)", "docstring": "Add a root node with no parents.\n\nArgs:\n node (Node):\n Node to add as a root.\n\nRaises:\n TypeError:\n If node is not a Node instance." }, "children": { "name": "children", "kind": "function", "path": "dagpipe.engine.Graph.children", "signature": "children(node: Node)", "docstring": "Return child nodes of a node.\n\nArgs:\n node (Node):\n Node to query.\n\nReturns:\n tuple[Node, ...]:\n Outgoing neighbors." }, "parents": { "name": "parents", "kind": "function", "path": "dagpipe.engine.Graph.parents", "signature": "parents(node: Node)", "docstring": "Return parent nodes of a node.\n\nArgs:\n node (Node):\n Node to query.\n\nReturns:\n tuple[Node, ...]:\n Incoming neighbors." }, "roots": { "name": "roots", "kind": "function", "path": "dagpipe.engine.Graph.roots", "signature": "roots()", "docstring": "Return root nodes (nodes with no incoming edges).\n\nReturns:\n tuple[Node, ...]:\n Entry point nodes." }, "nodes": { "name": "nodes", "kind": "function", "path": "dagpipe.engine.Graph.nodes", "signature": "nodes()", "docstring": "Return all nodes in the graph.\n\nReturns:\n tuple[Node, ...]:\n All registered nodes." } } }, "AsyncNode": { "name": "AsyncNode", "kind": "class", "path": "dagpipe.engine.AsyncNode", "signature": null, "docstring": "Base class for nodes whose execution is asynchronous.\n\nSubclasses implement `resolve_async` (an async generator yielding derived\n`State` objects). The engine dispatches to `resolve_async` when running an\nasync traversal (see `Engine.run_async`).\n\nSync-only engines (and the base `Node.run`) treat an `AsyncNode` as a no-op\nconsumer: calling `run` on an `AsyncNode` returns no states, signalling that\nan async engine is required.", "members": { "resolve": { "name": "resolve", "kind": "function", "path": "dagpipe.engine.AsyncNode.resolve", "signature": "resolve(state: State)", "docstring": "Execute no-op resolution in sync contexts.\n\nSync-only engines (and the base `Node.run`) treat an `AsyncNode` as a\nno-op consumer: this returns no states, signalling that an async engine\nis required.\n\nArgs:\n state (State):\n Input execution state.\n\nReturns:\n Iterable[State]:\n Empty tuple, since async execution is handled by\n `resolve_async`." }, "resolve_async": { "name": "resolve_async", "kind": "function", "path": "dagpipe.engine.AsyncNode.resolve_async", "signature": "resolve_async(state: State)", "docstring": "Execute node logic asynchronously.\n\nArgs:\n state (State):\n Input execution state.\n\nReturns:\n Iterable[State]:\n Derived execution state(s).\n\nNotes:\n **Guarantees:**\n\n - Subclasses implement this.\n - Must not mutate the input state.\n - Should use `fork()` to create child states." }, "run_async": { "name": "run_async", "kind": "function", "path": "dagpipe.engine.AsyncNode.run_async", "signature": "run_async(state: State)", "docstring": "Execute this node asynchronously on a state, validating outputs.\n\nArgs:\n state (State):\n Input execution state.\n\nReturns:\n tuple[State, ...]:\n Derived execution states.\n\nRaises:\n TypeError:\n If `resolve_async()` yields a non-`State` object." } } }, "Node": { "name": "Node", "kind": "class", "path": "dagpipe.engine.Node", "signature": null, "docstring": "Base class for all dagpipe execution nodes.\n\nAttributes:\n id (str):\n Unique identifier of the node (snake_case dotted format).\n\n name (str):\n Human-readable display name.\n\nNotes:\n **Responsibilities:**\n\n - Represents a deterministic unit of execution in the pipeline graph.\n - Consumes one `State` and produces zero, one, or many derived states.\n - Defines execution logic and enables branching, filtering, and transformation.\n\n **Guarantees:**\n\n - Nodes must never mutate the input `State`.\n - Instances are singletons per subclass and reused across executions.", "members": { "id": { "name": "id", "kind": "attribute", "path": "dagpipe.engine.Node.id", "signature": null, "docstring": null }, "name": { "name": "name", "kind": "attribute", "path": "dagpipe.engine.Node.name", "signature": null, "docstring": null }, "node_id_to_name": { "name": "node_id_to_name", "kind": "function", "path": "dagpipe.engine.Node.node_id_to_name", "signature": "node_id_to_name(node_id: str)", "docstring": "Convert a dotted snake_case node ID into a human-readable name.\n\nArgs:\n node_id (str):\n Unique node identifier (e.g., 'entity.resolve.numeric_merchant').\n\nReturns:\n str:\n Human-readable display name (e.g., 'Entity › Resolve › Numeric Merchant')." }, "clean_id_and_name": { "name": "clean_id_and_name", "kind": "function", "path": "dagpipe.engine.Node.clean_id_and_name", "signature": "clean_id_and_name()", "docstring": "Normalize and validate node ID and display name.\n\nRaises:\n TypeError:\n If ID is not a string.\n ValueError:\n If ID format is invalid.\n\nNotes:\n **Guarantees:**\n\n - Generates ID from module and class name if missing.\n - Validates ID format.\n - Generates human-readable name if missing." }, "run": { "name": "run", "kind": "function", "path": "dagpipe.engine.Node.run", "signature": "run(state: State)", "docstring": "Execute this node on a `State`.\n\nArgs:\n state (State):\n Input execution state.\n\nReturns:\n tuple[State, ...]:\n Derived execution states.\n\nRaises:\n TypeError:\n If `resolve()` yields a non-`State` object." }, "fork": { "name": "fork", "kind": "function", "path": "dagpipe.engine.Node.fork", "signature": "fork(state: State, *, payload_update: Any = None, confidence_delta: float = 0.0, metadata_update: Any = None)", "docstring": "Create a child `State` attributed to this node.\n\nArgs:\n state (State):\n Parent execution state.\n\n payload_update (Any, optional):\n Dot-path payload updates.\n\n confidence_delta (float, optional):\n Confidence adjustment.\n\n metadata_update (Any, optional):\n Metadata updates.\n\nReturns:\n State:\n New child execution state.\n\nNotes:\n **Responsibilities:**\n\n - Convenience wrapper around `State.fork()` that automatically\n records this node's ID in state history." }, "resolve": { "name": "resolve", "kind": "function", "path": "dagpipe.engine.Node.resolve", "signature": "resolve(state: State)", "docstring": "Execute node logic.\n\nArgs:\n state (State):\n Input execution state.\n\nReturns:\n Iterable[State]:\n Derived execution state(s).\n\nNotes:\n **Responsibilities:**\n\n - Subclasses implement specific resolution behavior.\n - Must not mutate input state.\n - Should use `fork()` to create child states.\n - May yield zero states to terminate a branch." }, "is_async": { "name": "is_async", "kind": "function", "path": "dagpipe.engine.Node.is_async", "signature": "is_async()", "docstring": "Return whether this node executes asynchronously.\n\nReturns:\n bool:\n True if the node is an `AsyncNode` instance." } } }, "State": { "name": "State", "kind": "class", "path": "dagpipe.engine.State", "signature": "State(payload: Payload, confidence: float = 1.0, parent: State | None = None, depth: int = 0, history: tuple[str, ...] = tuple(), metadata: dict[str, Any] = dict())", "docstring": "Immutable execution state propagated through dagpipe pipeline.\n\nAttributes:\n payload (Payload):\n Execution data container.\n\n schema (ClassVar[Schema]):\n Payload validation schema.\n\n confidence (float):\n Execution confidence score.\n\n parent (Optional[State]):\n Parent state reference.\n\n depth (int):\n Execution depth.\n\n history (Tuple[str, ...]):\n Ordered node execution lineage.\n\n metadata (Dict[str, Any]):\n Execution metadata.\n\nNotes:\n **Responsibilities:**\n\n - Represents a complete execution snapshot at a specific point in\n pipeline traversal.\n - Fundamental unit of execution in `dagpipe`.\n - Fully thread-safe due to immutability.", "members": { "payload": { "name": "payload", "kind": "attribute", "path": "dagpipe.engine.State.payload", "signature": null, "docstring": null }, "schema": { "name": "schema", "kind": "attribute", "path": "dagpipe.engine.State.schema", "signature": null, "docstring": null }, "confidence": { "name": "confidence", "kind": "attribute", "path": "dagpipe.engine.State.confidence", "signature": null, "docstring": null }, "parent": { "name": "parent", "kind": "attribute", "path": "dagpipe.engine.State.parent", "signature": null, "docstring": null }, "depth": { "name": "depth", "kind": "attribute", "path": "dagpipe.engine.State.depth", "signature": null, "docstring": null }, "history": { "name": "history", "kind": "attribute", "path": "dagpipe.engine.State.history", "signature": null, "docstring": null }, "metadata": { "name": "metadata", "kind": "attribute", "path": "dagpipe.engine.State.metadata", "signature": null, "docstring": null }, "fork": { "name": "fork", "kind": "function", "path": "dagpipe.engine.State.fork", "signature": "fork(*, payload_update: Mapping[str, Any] | None = None, confidence_delta: float = 0.0, node_id: str | None = None, metadata_update: Mapping[str, Any] | None = None)", "docstring": "Create a new child `State` derived from this state.\n\nArgs:\n payload_update (Mapping[str, Any] | None, optional):\n Dot-path updates applied to the payload.\n\n confidence_delta (float, optional):\n Adjustment applied to current confidence.\n\n node_id (str | None, optional):\n Identifier of the node creating this state.\n\n metadata_update (Mapping[str, Any] | None, optional):\n Updates merged into state metadata.\n\nReturns:\n State:\n A new immutable `State` instance.\n\nNotes:\n **Guarantees:**\n\n - This is the only supported mechanism for modifying execution data.\n - Validates payload updates, preserves lineage, increments depth,\n and appends to history." }, "lineage": { "name": "lineage", "kind": "function", "path": "dagpipe.engine.State.lineage", "signature": "lineage()", "docstring": "Return lineage from root to this State.\n\nReturns:\n tuple[State, ...]:\n Ordered execution lineage (root first)." }, "get": { "name": "get", "kind": "function", "path": "dagpipe.engine.State.get", "signature": "get(key: str, default: Any = None)", "docstring": "Retrieve payload value.\n\nArgs:\n key (str):\n Dot-path key.\n default (Any, optional):\n Fallback value.\n\nReturns:\n Any:\n Stored value or default." }, "has": { "name": "has", "kind": "function", "path": "dagpipe.engine.State.has", "signature": "has(key: str)", "docstring": "Check whether payload contains key.\n\nArgs:\n key (str):\n Dot-path key.\n\nReturns:\n bool:\n Existence of the key." } } }, "StepHook": { "name": "StepHook", "kind": "attribute", "path": "dagpipe.engine.StepHook", "signature": null, "docstring": null }, "AsyncStepHook": { "name": "AsyncStepHook", "kind": "attribute", "path": "dagpipe.engine.AsyncStepHook", "signature": null, "docstring": null }, "ProgressMessage": { "name": "ProgressMessage", "kind": "class", "path": "dagpipe.engine.ProgressMessage", "signature": "ProgressMessage(*, lines: int | None = None, blocks: int | None = None, count: int | None = None, unit: str | None = None, raw_ocr_line: str | None = None, error: str | None = None, step: str = '', status: str = '')", "docstring": "Lightweight progress payload emitted by engine step hooks.\n\nMirrors the imperative `ProgressMessage` used by the legacy orchestrator so\ncallers can surface counts/lines/errors without coupling the engine to pydantic.\n\nAttributes:\n lines (int | None):\n Optional count of processed lines.\n\n blocks (int | None):\n Optional count of processed blocks.\n\n count (int | None):\n Optional generic item count.\n\n unit (str | None):\n Optional unit for the count (e.g., 'pages').\n\n raw_ocr_line (str | None):\n Optional raw OCR line payload.\n\n error (str | None):\n Optional error description.\n\n step (str):\n Identifier of the step that emitted the message.\n\n status (str):\n Status label associated with the step.\n\nNotes:\n **Guarantees:**\n\n - Immutable after construction (attributes are never reassigned).\n - Independent of pydantic; safe to construct in the engine core.", "members": { "lines": { "name": "lines", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.lines", "signature": null, "docstring": null }, "blocks": { "name": "blocks", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.blocks", "signature": null, "docstring": null }, "count": { "name": "count", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.count", "signature": null, "docstring": null }, "unit": { "name": "unit", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.unit", "signature": null, "docstring": null }, "raw_ocr_line": { "name": "raw_ocr_line", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.raw_ocr_line", "signature": null, "docstring": null }, "error": { "name": "error", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.error", "signature": null, "docstring": null }, "step": { "name": "step", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.step", "signature": null, "docstring": null }, "status": { "name": "status", "kind": "attribute", "path": "dagpipe.engine.ProgressMessage.status", "signature": null, "docstring": null }, "as_dict": { "name": "as_dict", "kind": "function", "path": "dagpipe.engine.ProgressMessage.as_dict", "signature": "as_dict() -> dict[str, Any]", "docstring": "Return the message as a plain dictionary.\n\nReturns:\n dict[str, Any]:\n All attribute values keyed by their attribute name." } } }, "StepResult": { "name": "StepResult", "kind": "class", "path": "dagpipe.engine.StepResult", "signature": "StepResult(index: int, node_id: str, states: tuple[State, ...], completed: bool)", "docstring": "A single checkpointed step within an async/resumable engine run.\n\nAttributes:\n index (int): Ordinal index of the step.\n node_id (str): Identifier of the node associated with this step.\n states (tuple[State, ...]): States produced by running this step.\n completed (bool): Whether this step succeeded (vs. paused/interrupted).", "members": { "index": { "name": "index", "kind": "attribute", "path": "dagpipe.engine.StepResult.index", "signature": null, "docstring": null }, "node_id": { "name": "node_id", "kind": "attribute", "path": "dagpipe.engine.StepResult.node_id", "signature": null, "docstring": null }, "states": { "name": "states", "kind": "attribute", "path": "dagpipe.engine.StepResult.states", "signature": null, "docstring": null }, "completed": { "name": "completed", "kind": "attribute", "path": "dagpipe.engine.StepResult.completed", "signature": null, "docstring": null } } }, "Engine": { "name": "Engine", "kind": "class", "path": "dagpipe.engine.Engine", "signature": "Engine(nodes_or_graph: Sequence[Node] | Graph, *, on_step: StepHook | None = None)", "docstring": "Execution engine responsible for running pipeline logic.\n\nNotes:\n **Responsibilities:**\n\n - Accepts either a linear sequence of `Node` objects or a `Graph`\n defining execution topology.\n - Propagates immutable `State` objects through `Node` objects and\n collects terminal states.\n - Supports synchronous (`run`) and asynchronous (`run_async`)\n execution, dispatching per-node.\n - Supports step-wise / resumable execution and progress hooks.\n\n **Guarantees:**\n\n - Never mutates `State`, `Node`, or `Graph` instances.\n - `State` objects are never modified in place; each branch produces\n independent instances.\n - Execution order is deterministic and follows graph or pipeline topology.\n - Thread-safe for concurrent execution.", "members": { "MODE_LINEAR": { "name": "MODE_LINEAR", "kind": "attribute", "path": "dagpipe.engine.Engine.MODE_LINEAR", "signature": null, "docstring": null }, "MODE_GRAPH": { "name": "MODE_GRAPH", "kind": "attribute", "path": "dagpipe.engine.Engine.MODE_GRAPH", "signature": null, "docstring": null }, "run": { "name": "run", "kind": "function", "path": "dagpipe.engine.Engine.run", "signature": "run(root: State) -> list[State]", "docstring": "Execute the pipeline starting from a root `State`.\n\nArgs:\n root (State):\n Initial execution state.\n\nReturns:\n list[State]:\n Terminal execution states produced by the pipeline.\n\nRaises:\n TypeError:\n If `root` is not a `State` instance.\n\n RuntimeError:\n If the engine execution mode is invalid.\n\nNotes:\n **Responsibilities:**\n\n - Selects execution mode, propagates state through nodes, creates\n new instances for branches, and collects terminal states." }, "run_async": { "name": "run_async", "kind": "function", "path": "dagpipe.engine.Engine.run_async", "signature": "run_async(root: State) -> list[State]", "docstring": "Execute the pipeline starting from `root`, dispatching sync vs async nodes.\n\nArgs:\n root (State):\n Initial execution state.\n\nReturns:\n list[State]:\n Terminal execution states produced by the pipeline.\n\nNotes:\n Each node is executed with `Node.run` when synchronous and\n `AsyncNode.run_async` when asynchronous. Linear and graph topologies\n are both supported." }, "run_steps": { "name": "run_steps", "kind": "function", "path": "dagpipe.engine.Engine.run_steps", "signature": "run_steps(root: State, *, resume_from: int | None = None, on_step: StepHook | None = None) -> Iterator[StepResult]", "docstring": "Execute the pipeline step-by-step, yielding one `StepResult` per step.\n\nArgs:\n root (State):\n Initial execution state.\n\n resume_from (int | None, optional):\n Skip steps at index < `resume_from` (for resume-after-partial).\n Steps are 0-indexed.\n\n on_step (StepHook | None, optional):\n Callback `(step, status, message)` invoked per step; falls back\n to the engine-level hook when unset.\n\nYields:\n StepResult:\n One per executed node/step, carrying the produced states.\n\nNotes:\n This is a synchronous, generator-based checkpoint interface compatible\n with the imperative resume-by-step behaviour of the legacy\n orchestrator. Use `run_steps_async` for async nodes." }, "run_steps_async": { "name": "run_steps_async", "kind": "function", "path": "dagpipe.engine.Engine.run_steps_async", "signature": "run_steps_async(root: State, *, resume_from: int | None = None, on_step: AsyncStepHook | None = None) -> AsyncIterator[StepResult]", "docstring": "Async variant of `run_steps` supporting `AsyncNode` execution.\n\nArgs:\n root (State):\n Initial execution state.\n\n resume_from (int | None, optional):\n Skip steps at index < `resume_from`.\n\n on_step (AsyncStepHook | None, optional):\n Async callback `(step, status, message)` invoked per step.\n\nYields:\n StepResult:\n One per executed node/step." }, "nodes": { "name": "nodes", "kind": "attribute", "path": "dagpipe.engine.Engine.nodes", "signature": null, "docstring": "Return nodes managed by this engine.\n\nReturns:\n tuple[Node, ...]:\n Ordered sequence in linear mode or all nodes in graph mode." } } } } } }