Coverage for python/lsst/pipe/base/graph/graph.py: 15%
429 statements
« prev ^ index » next coverage.py v7.2.7, created at 2023-06-02 02:17 -0700
« prev ^ index » next coverage.py v7.2.7, created at 2023-06-02 02:17 -0700
1# This file is part of pipe_base.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (http://www.lsst.org).
6# See the COPYRIGHT file at the top-level directory of this distribution
7# for details of code ownership.
8#
9# This program is free software: you can redistribute it and/or modify
10# it under the terms of the GNU General Public License as published by
11# the Free Software Foundation, either version 3 of the License, or
12# (at your option) any later version.
13#
14# This program is distributed in the hope that it will be useful,
15# but WITHOUT ANY WARRANTY; without even the implied warranty of
16# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17# GNU General Public License for more details.
18#
19# You should have received a copy of the GNU General Public License
20# along with this program. If not, see <http://www.gnu.org/licenses/>.
21from __future__ import annotations
23__all__ = ("QuantumGraph", "IncompatibleGraphError")
25import io
26import json
27import lzma
28import os
29import pickle
30import struct
31import time
32import uuid
33import warnings
34from collections import defaultdict, deque
35from itertools import chain
36from types import MappingProxyType
37from typing import (
38 Any,
39 BinaryIO,
40 DefaultDict,
41 Deque,
42 Dict,
43 FrozenSet,
44 Generator,
45 Iterable,
46 List,
47 Mapping,
48 MutableMapping,
49 Optional,
50 Set,
51 Tuple,
52 TypeVar,
53 Union,
54)
56import networkx as nx
57from lsst.daf.butler import DatasetRef, DatasetType, DimensionRecordsAccumulator, DimensionUniverse, Quantum
58from lsst.resources import ResourcePath, ResourcePathExpression
59from lsst.utils.introspection import get_full_type_name
60from networkx.drawing.nx_agraph import write_dot
62from ..connections import iterConnections
63from ..pipeline import TaskDef
64from ._implDetails import DatasetTypeName, _DatasetTracker, _pruner
65from ._loadHelpers import LoadHelper
66from ._versionDeserializers import DESERIALIZER_MAP
67from .quantumNode import BuildId, QuantumNode
69_T = TypeVar("_T", bound="QuantumGraph")
71# modify this constant any time the on disk representation of the save file
72# changes, and update the load helpers to behave properly for each version.
73SAVE_VERSION = 3
75# Strings used to describe the format for the preamble bytes in a file save
76# The base is a big endian encoded unsigned short that is used to hold the
77# file format version. This allows reading version bytes and determine which
78# loading code should be used for the rest of the file
79STRUCT_FMT_BASE = ">H"
80#
81# Version 1
82# This marks a big endian encoded format with an unsigned short, an unsigned
83# long long, and an unsigned long long in the byte stream
84# Version 2
85# A big endian encoded format with an unsigned long long byte stream used to
86# indicate the total length of the entire header.
87STRUCT_FMT_STRING = {1: ">QQ", 2: ">Q"}
89# magic bytes that help determine this is a graph save
90MAGIC_BYTES = b"qgraph4\xf6\xe8\xa9"
93class IncompatibleGraphError(Exception):
94 """Exception class to indicate that a lookup by NodeId is impossible due
95 to incompatibilities
96 """
98 pass
101class QuantumGraph:
102 """QuantumGraph is a directed acyclic graph of `QuantumNode` objects
104 This data structure represents a concrete workflow generated from a
105 `Pipeline`.
107 Parameters
108 ----------
109 quanta : Mapping of `TaskDef` to sets of `Quantum`
110 This maps tasks (and their configs) to the sets of data they are to
111 process.
112 metadata : Optional Mapping of `str` to primitives
113 This is an optional parameter of extra data to carry with the graph.
114 Entries in this mapping should be able to be serialized in JSON.
115 pruneRefs : iterable [ `DatasetRef` ], optional
116 Set of dataset refs to exclude from a graph.
117 universe : `lsst.daf.butler.DimensionUniverse`, optional
118 The dimensions in which quanta can be defined. Need only be provided if
119 no quanta have data IDs.
120 initInputs : `Mapping`, optional
121 Maps tasks to their InitInput dataset refs. Dataset refs can be either
122 resolved or non-resolved. Presently the same dataset refs are included
123 in each `Quantum` for the same task.
124 initOutputs : `Mapping`, optional
125 Maps tasks to their InitOutput dataset refs. Dataset refs can be either
126 resolved or non-resolved. For intermediate resolved refs their dataset
127 ID must match ``initInputs`` and Quantum ``initInputs``.
128 globalInitOutputs : iterable [ `DatasetRef` ], optional
129 Dataset refs for some global objects produced by pipeline. These
130 objects include task configurations and package versions. Typically
131 they have an empty DataId, but there is no real restriction on what
132 can appear here.
133 registryDatasetTypes : iterable [ `DatasetType` ], optional
134 Dataset types which are used by this graph, their definitions must
135 match registry. If registry does not define dataset type yet, then
136 it should match one that will be created later.
138 Raises
139 ------
140 ValueError
141 Raised if the graph is pruned such that some tasks no longer have nodes
142 associated with them.
143 """
145 def __init__(
146 self,
147 quanta: Mapping[TaskDef, Set[Quantum]],
148 metadata: Optional[Mapping[str, Any]] = None,
149 pruneRefs: Optional[Iterable[DatasetRef]] = None,
150 universe: Optional[DimensionUniverse] = None,
151 initInputs: Optional[Mapping[TaskDef, Iterable[DatasetRef]]] = None,
152 initOutputs: Optional[Mapping[TaskDef, Iterable[DatasetRef]]] = None,
153 globalInitOutputs: Optional[Iterable[DatasetRef]] = None,
154 registryDatasetTypes: Optional[Iterable[DatasetType]] = None,
155 ):
156 self._buildGraphs(
157 quanta,
158 metadata=metadata,
159 pruneRefs=pruneRefs,
160 universe=universe,
161 initInputs=initInputs,
162 initOutputs=initOutputs,
163 globalInitOutputs=globalInitOutputs,
164 registryDatasetTypes=registryDatasetTypes,
165 )
167 def _buildGraphs(
168 self,
169 quanta: Mapping[TaskDef, Set[Quantum]],
170 *,
171 _quantumToNodeId: Optional[Mapping[Quantum, uuid.UUID]] = None,
172 _buildId: Optional[BuildId] = None,
173 metadata: Optional[Mapping[str, Any]] = None,
174 pruneRefs: Optional[Iterable[DatasetRef]] = None,
175 universe: Optional[DimensionUniverse] = None,
176 initInputs: Optional[Mapping[TaskDef, Iterable[DatasetRef]]] = None,
177 initOutputs: Optional[Mapping[TaskDef, Iterable[DatasetRef]]] = None,
178 globalInitOutputs: Optional[Iterable[DatasetRef]] = None,
179 registryDatasetTypes: Optional[Iterable[DatasetType]] = None,
180 ) -> None:
181 """Builds the graph that is used to store the relation between tasks,
182 and the graph that holds the relations between quanta
183 """
184 self._metadata = metadata
185 self._buildId = _buildId if _buildId is not None else BuildId(f"{time.time()}-{os.getpid()}")
186 # Data structures used to identify relations between components;
187 # DatasetTypeName -> TaskDef for task,
188 # and DatasetRef -> QuantumNode for the quanta
189 self._datasetDict = _DatasetTracker[DatasetTypeName, TaskDef](createInverse=True)
190 self._datasetRefDict = _DatasetTracker[DatasetRef, QuantumNode]()
192 self._nodeIdMap: Dict[uuid.UUID, QuantumNode] = {}
193 self._taskToQuantumNode: MutableMapping[TaskDef, Set[QuantumNode]] = defaultdict(set)
194 for taskDef, quantumSet in quanta.items():
195 connections = taskDef.connections
197 # For each type of connection in the task, add a key to the
198 # `_DatasetTracker` for the connections name, with a value of
199 # the TaskDef in the appropriate field
200 for inpt in iterConnections(connections, ("inputs", "prerequisiteInputs", "initInputs")):
201 # Have to handle components in inputs.
202 dataset_name, _, _ = inpt.name.partition(".")
203 self._datasetDict.addConsumer(DatasetTypeName(dataset_name), taskDef)
205 for output in iterConnections(connections, ("outputs",)):
206 # Have to handle possible components in outputs.
207 dataset_name, _, _ = output.name.partition(".")
208 self._datasetDict.addProducer(DatasetTypeName(dataset_name), taskDef)
210 # For each `Quantum` in the set of all `Quantum` for this task,
211 # add a key to the `_DatasetTracker` that is a `DatasetRef` for one
212 # of the individual datasets inside the `Quantum`, with a value of
213 # a newly created QuantumNode to the appropriate input/output
214 # field.
215 for quantum in quantumSet:
216 if quantum.dataId is not None:
217 if universe is None:
218 universe = quantum.dataId.universe
219 elif universe != quantum.dataId.universe:
220 raise RuntimeError(
221 "Mismatched dimension universes in QuantumGraph construction: "
222 f"{universe} != {quantum.dataId.universe}. "
223 )
225 if _quantumToNodeId:
226 if (nodeId := _quantumToNodeId.get(quantum)) is None:
227 raise ValueError(
228 "If _quantuMToNodeNumber is not None, all quanta must have an "
229 "associated value in the mapping"
230 )
231 else:
232 nodeId = uuid.uuid4()
234 inits = quantum.initInputs.values()
235 inputs = quantum.inputs.values()
236 value = QuantumNode(quantum, taskDef, nodeId)
237 self._taskToQuantumNode[taskDef].add(value)
238 self._nodeIdMap[nodeId] = value
240 for dsRef in chain(inits, inputs):
241 # unfortunately, `Quantum` allows inits to be individual
242 # `DatasetRef`s or an Iterable of such, so there must
243 # be an instance check here
244 if isinstance(dsRef, Iterable):
245 for sub in dsRef:
246 if sub.isComponent():
247 sub = sub.makeCompositeRef()
248 self._datasetRefDict.addConsumer(sub, value)
249 else:
250 assert isinstance(dsRef, DatasetRef)
251 if dsRef.isComponent():
252 dsRef = dsRef.makeCompositeRef()
253 self._datasetRefDict.addConsumer(dsRef, value)
254 for dsRef in chain.from_iterable(quantum.outputs.values()):
255 self._datasetRefDict.addProducer(dsRef, value)
257 if pruneRefs is not None:
258 # track what refs were pruned and prune the graph
259 prunes: Set[QuantumNode] = set()
260 _pruner(self._datasetRefDict, pruneRefs, alreadyPruned=prunes)
262 # recreate the taskToQuantumNode dict removing nodes that have been
263 # pruned. Keep track of task defs that now have no QuantumNodes
264 emptyTasks: Set[str] = set()
265 newTaskToQuantumNode: DefaultDict[TaskDef, Set[QuantumNode]] = defaultdict(set)
266 # accumulate all types
267 types_ = set()
268 # tracker for any pruneRefs that have caused tasks to have no nodes
269 # This helps the user find out what caused the issues seen.
270 culprits = set()
271 # Find all the types from the refs to prune
272 for r in pruneRefs:
273 types_.add(r.datasetType)
275 # For each of the tasks, and their associated nodes, remove any
276 # any nodes that were pruned. If there are no nodes associated
277 # with a task, record that task, and find out if that was due to
278 # a type from an input ref to prune.
279 for td, taskNodes in self._taskToQuantumNode.items():
280 diff = taskNodes.difference(prunes)
281 if len(diff) == 0:
282 if len(taskNodes) != 0:
283 tp: DatasetType
284 for tp in types_:
285 if (tmpRefs := next(iter(taskNodes)).quantum.inputs.get(tp)) and not set(
286 tmpRefs
287 ).difference(pruneRefs):
288 culprits.add(tp.name)
289 emptyTasks.add(td.label)
290 newTaskToQuantumNode[td] = diff
292 # update the internal dict
293 self._taskToQuantumNode = newTaskToQuantumNode
295 if emptyTasks:
296 raise ValueError(
297 f"{', '.join(emptyTasks)} task(s) have no nodes associated with them "
298 f"after graph pruning; {', '.join(culprits)} caused over-pruning"
299 )
301 # Dimension universe
302 if universe is None:
303 raise RuntimeError(
304 "Dimension universe or at least one quantum with a data ID "
305 "must be provided when constructing a QuantumGraph."
306 )
307 self._universe = universe
309 # Graph of quanta relations
310 self._connectedQuanta = self._datasetRefDict.makeNetworkXGraph()
311 self._count = len(self._connectedQuanta)
313 # Graph of task relations, used in various methods
314 self._taskGraph = self._datasetDict.makeNetworkXGraph()
316 # convert default dict into a regular to prevent accidental key
317 # insertion
318 self._taskToQuantumNode = dict(self._taskToQuantumNode.items())
320 self._initInputRefs: Dict[TaskDef, List[DatasetRef]] = {}
321 self._initOutputRefs: Dict[TaskDef, List[DatasetRef]] = {}
322 self._globalInitOutputRefs: List[DatasetRef] = []
323 self._registryDatasetTypes: List[DatasetType] = []
324 if initInputs is not None:
325 self._initInputRefs = {taskDef: list(refs) for taskDef, refs in initInputs.items()}
326 if initOutputs is not None:
327 self._initOutputRefs = {taskDef: list(refs) for taskDef, refs in initOutputs.items()}
328 if globalInitOutputs is not None:
329 self._globalInitOutputRefs = list(globalInitOutputs)
330 if registryDatasetTypes is not None:
331 self._registryDatasetTypes = list(registryDatasetTypes)
333 @property
334 def taskGraph(self) -> nx.DiGraph:
335 """A graph representing the relations between the tasks inside
336 the quantum graph (`networkx.DiGraph`).
337 """
338 return self._taskGraph
340 @property
341 def graph(self) -> nx.DiGraph:
342 """A graph representing the relations between all the `QuantumNode`
343 objects (`networkx.DiGraph`).
345 The graph should usually be iterated over, or passed to methods of this
346 class, but sometimes direct access to the ``networkx`` object may be
347 helpful.
348 """
349 return self._connectedQuanta
351 @property
352 def inputQuanta(self) -> Iterable[QuantumNode]:
353 """The nodes that are inputs to the graph (iterable [`QuantumNode`]).
355 These are the nodes that do not depend on any other nodes in the
356 graph.
357 """
358 return (q for q, n in self._connectedQuanta.in_degree if n == 0)
360 @property
361 def outputQuanta(self) -> Iterable[QuantumNode]:
362 """The nodes that are outputs of the graph (iterable [`QuantumNode`]).
364 These are the nodes that have no nodes that depend on them in the
365 graph.
366 """
367 return [q for q, n in self._connectedQuanta.out_degree if n == 0]
369 @property
370 def allDatasetTypes(self) -> Tuple[DatasetTypeName, ...]:
371 """All the data set type names that are present in the graph
372 (`tuple` [`str`]).
374 These types do not include global init-outputs.
375 """
376 return tuple(self._datasetDict.keys())
378 @property
379 def isConnected(self) -> bool:
380 """Whether all of the nodes in the graph are connected, ignoring
381 directionality of connections (`bool`).
382 """
383 return nx.is_weakly_connected(self._connectedQuanta)
385 def pruneGraphFromRefs(self: _T, refs: Iterable[DatasetRef]) -> _T:
386 r"""Return a graph pruned of input `~lsst.daf.butler.DatasetRef`\ s
387 and nodes which depend on them.
389 Parameters
390 ----------
391 refs : `Iterable` of `DatasetRef`
392 Refs which should be removed from resulting graph
394 Returns
395 -------
396 graph : `QuantumGraph`
397 A graph that has been pruned of specified refs and the nodes that
398 depend on them.
399 """
400 newInst = object.__new__(type(self))
401 quantumMap = defaultdict(set)
402 for node in self:
403 quantumMap[node.taskDef].add(node.quantum)
405 # convert to standard dict to prevent accidental key insertion
406 quantumDict: Dict[TaskDef, Set[Quantum]] = dict(quantumMap.items())
408 # This should not change set of tasks in a graph, so we can keep the
409 # same registryDatasetTypes as in the original graph.
410 # TODO: Do we need to copy initInputs/initOutputs?
411 newInst._buildGraphs(
412 quantumDict,
413 _quantumToNodeId={n.quantum: n.nodeId for n in self},
414 metadata=self._metadata,
415 pruneRefs=refs,
416 universe=self._universe,
417 globalInitOutputs=self._globalInitOutputRefs,
418 registryDatasetTypes=self._registryDatasetTypes,
419 )
420 return newInst
422 def getQuantumNodeByNodeId(self, nodeId: uuid.UUID) -> QuantumNode:
423 """Lookup a `QuantumNode` from an id associated with the node.
425 Parameters
426 ----------
427 nodeId : `NodeId`
428 The number associated with a node
430 Returns
431 -------
432 node : `QuantumNode`
433 The node corresponding with input number
435 Raises
436 ------
437 KeyError
438 Raised if the requested nodeId is not in the graph.
439 """
440 return self._nodeIdMap[nodeId]
442 def getQuantaForTask(self, taskDef: TaskDef) -> FrozenSet[Quantum]:
443 """Return all the `Quantum` associated with a `TaskDef`.
445 Parameters
446 ----------
447 taskDef : `TaskDef`
448 The `TaskDef` for which `Quantum` are to be queried
450 Returns
451 -------
452 frozenset of `Quantum`
453 The `set` of `Quantum` that is associated with the specified
454 `TaskDef`.
455 """
456 return frozenset(node.quantum for node in self._taskToQuantumNode.get(taskDef, ()))
458 def getNumberOfQuantaForTask(self, taskDef: TaskDef) -> int:
459 """Return all the number of `Quantum` associated with a `TaskDef`.
461 Parameters
462 ----------
463 taskDef : `TaskDef`
464 The `TaskDef` for which `Quantum` are to be queried
466 Returns
467 -------
468 count : int
469 The number of `Quantum` that are associated with the specified
470 `TaskDef`.
471 """
472 return len(self._taskToQuantumNode.get(taskDef, ()))
474 def getNodesForTask(self, taskDef: TaskDef) -> FrozenSet[QuantumNode]:
475 """Return all the `QuantumNodes` associated with a `TaskDef`.
477 Parameters
478 ----------
479 taskDef : `TaskDef`
480 The `TaskDef` for which `Quantum` are to be queried
482 Returns
483 -------
484 frozenset of `QuantumNodes`
485 The `frozenset` of `QuantumNodes` that is associated with the
486 specified `TaskDef`.
487 """
488 return frozenset(self._taskToQuantumNode[taskDef])
490 def findTasksWithInput(self, datasetTypeName: DatasetTypeName) -> Iterable[TaskDef]:
491 """Find all tasks that have the specified dataset type name as an
492 input.
494 Parameters
495 ----------
496 datasetTypeName : `str`
497 A string representing the name of a dataset type to be queried,
498 can also accept a `DatasetTypeName` which is a `NewType` of str for
499 type safety in static type checking.
501 Returns
502 -------
503 tasks : iterable of `TaskDef`
504 `TaskDef` objects that have the specified `DatasetTypeName` as an
505 input, list will be empty if no tasks use specified
506 `DatasetTypeName` as an input.
508 Raises
509 ------
510 KeyError
511 Raised if the `DatasetTypeName` is not part of the `QuantumGraph`
512 """
513 return (c for c in self._datasetDict.getConsumers(datasetTypeName))
515 def findTaskWithOutput(self, datasetTypeName: DatasetTypeName) -> Optional[TaskDef]:
516 """Find all tasks that have the specified dataset type name as an
517 output.
519 Parameters
520 ----------
521 datasetTypeName : `str`
522 A string representing the name of a dataset type to be queried,
523 can also accept a `DatasetTypeName` which is a `NewType` of str for
524 type safety in static type checking.
526 Returns
527 -------
528 `TaskDef` or `None`
529 `TaskDef` that outputs `DatasetTypeName` as an output or None if
530 none of the tasks produce this `DatasetTypeName`.
532 Raises
533 ------
534 KeyError
535 Raised if the `DatasetTypeName` is not part of the `QuantumGraph`
536 """
537 return self._datasetDict.getProducer(datasetTypeName)
539 def tasksWithDSType(self, datasetTypeName: DatasetTypeName) -> Iterable[TaskDef]:
540 """Find all tasks that are associated with the specified dataset type
541 name.
543 Parameters
544 ----------
545 datasetTypeName : `str`
546 A string representing the name of a dataset type to be queried,
547 can also accept a `DatasetTypeName` which is a `NewType` of str for
548 type safety in static type checking.
550 Returns
551 -------
552 result : iterable of `TaskDef`
553 `TaskDef` objects that are associated with the specified
554 `DatasetTypeName`
556 Raises
557 ------
558 KeyError
559 Raised if the `DatasetTypeName` is not part of the `QuantumGraph`
560 """
561 return self._datasetDict.getAll(datasetTypeName)
563 def findTaskDefByName(self, taskName: str) -> List[TaskDef]:
564 """Determine which `TaskDef` objects in this graph are associated
565 with a `str` representing a task name (looks at the taskName property
566 of `TaskDef` objects).
568 Returns a list of `TaskDef` objects as a `PipelineTask` may appear
569 multiple times in a graph with different labels.
571 Parameters
572 ----------
573 taskName : str
574 Name of a task to search for
576 Returns
577 -------
578 result : list of `TaskDef`
579 List of the `TaskDef` objects that have the name specified.
580 Multiple values are returned in the case that a task is used
581 multiple times with different labels.
582 """
583 results = []
584 for task in self._taskToQuantumNode.keys():
585 split = task.taskName.split(".")
586 if split[-1] == taskName:
587 results.append(task)
588 return results
590 def findTaskDefByLabel(self, label: str) -> Optional[TaskDef]:
591 """Determine which `TaskDef` objects in this graph are associated
592 with a `str` representing a tasks label.
594 Parameters
595 ----------
596 taskName : str
597 Name of a task to search for
599 Returns
600 -------
601 result : `TaskDef`
602 `TaskDef` objects that has the specified label.
603 """
604 for task in self._taskToQuantumNode.keys():
605 if label == task.label:
606 return task
607 return None
609 def findQuantaWithDSType(self, datasetTypeName: DatasetTypeName) -> Set[Quantum]:
610 """Return all the `Quantum` that contain a specified `DatasetTypeName`.
612 Parameters
613 ----------
614 datasetTypeName : `str`
615 The name of the dataset type to search for as a string,
616 can also accept a `DatasetTypeName` which is a `NewType` of str for
617 type safety in static type checking.
619 Returns
620 -------
621 result : `set` of `QuantumNode` objects
622 A `set` of `QuantumNode`s that contain specified `DatasetTypeName`
624 Raises
625 ------
626 KeyError
627 Raised if the `DatasetTypeName` is not part of the `QuantumGraph`
629 """
630 tasks = self._datasetDict.getAll(datasetTypeName)
631 result: Set[Quantum] = set()
632 result = result.union(quantum for task in tasks for quantum in self.getQuantaForTask(task))
633 return result
635 def checkQuantumInGraph(self, quantum: Quantum) -> bool:
636 """Check if specified quantum appears in the graph as part of a node.
638 Parameters
639 ----------
640 quantum : `Quantum`
641 The quantum to search for
643 Returns
644 -------
645 `bool`
646 The result of searching for the quantum
647 """
648 for node in self:
649 if quantum == node.quantum:
650 return True
651 return False
653 def writeDotGraph(self, output: Union[str, io.BufferedIOBase]) -> None:
654 """Write out the graph as a dot graph.
656 Parameters
657 ----------
658 output : str or `io.BufferedIOBase`
659 Either a filesystem path to write to, or a file handle object
660 """
661 write_dot(self._connectedQuanta, output)
663 def subset(self: _T, nodes: Union[QuantumNode, Iterable[QuantumNode]]) -> _T:
664 """Create a new graph object that contains the subset of the nodes
665 specified as input. Node number is preserved.
667 Parameters
668 ----------
669 nodes : `QuantumNode` or iterable of `QuantumNode`
671 Returns
672 -------
673 graph : instance of graph type
674 An instance of the type from which the subset was created
675 """
676 if not isinstance(nodes, Iterable):
677 nodes = (nodes,)
678 quantumSubgraph = self._connectedQuanta.subgraph(nodes).nodes
679 quantumMap = defaultdict(set)
681 dataset_type_names: set[str] = set()
682 node: QuantumNode
683 for node in quantumSubgraph:
684 quantumMap[node.taskDef].add(node.quantum)
685 dataset_type_names.update(
686 dstype.name
687 for dstype in chain(
688 node.quantum.inputs.keys(), node.quantum.outputs.keys(), node.quantum.initInputs.keys()
689 )
690 )
692 # May need to trim dataset types from registryDatasetTypes.
693 for taskDef in quantumMap:
694 if refs := self.initOutputRefs(taskDef):
695 dataset_type_names.update(ref.datasetType.name for ref in refs)
696 dataset_type_names.update(ref.datasetType.name for ref in self._globalInitOutputRefs)
697 registryDatasetTypes = [
698 dstype for dstype in self._registryDatasetTypes if dstype.name in dataset_type_names
699 ]
701 # convert to standard dict to prevent accidental key insertion
702 quantumDict: Dict[TaskDef, Set[Quantum]] = dict(quantumMap.items())
703 # Create an empty graph, and then populate it with custom mapping
704 newInst = type(self)({}, universe=self._universe)
705 # TODO: Do we need to copy initInputs/initOutputs?
706 newInst._buildGraphs(
707 quantumDict,
708 _quantumToNodeId={n.quantum: n.nodeId for n in nodes},
709 _buildId=self._buildId,
710 metadata=self._metadata,
711 universe=self._universe,
712 globalInitOutputs=self._globalInitOutputRefs,
713 registryDatasetTypes=registryDatasetTypes,
714 )
715 return newInst
717 def subsetToConnected(self: _T) -> Tuple[_T, ...]:
718 """Generate a list of subgraphs where each is connected.
720 Returns
721 -------
722 result : list of `QuantumGraph`
723 A list of graphs that are each connected
724 """
725 return tuple(
726 self.subset(connectedSet)
727 for connectedSet in nx.weakly_connected_components(self._connectedQuanta)
728 )
730 def determineInputsToQuantumNode(self, node: QuantumNode) -> Set[QuantumNode]:
731 """Return a set of `QuantumNode` that are direct inputs to a specified
732 node.
734 Parameters
735 ----------
736 node : `QuantumNode`
737 The node of the graph for which inputs are to be determined
739 Returns
740 -------
741 set of `QuantumNode`
742 All the nodes that are direct inputs to specified node
743 """
744 return set(pred for pred in self._connectedQuanta.predecessors(node))
746 def determineOutputsOfQuantumNode(self, node: QuantumNode) -> Set[QuantumNode]:
747 """Return a set of `QuantumNode` that are direct outputs of a specified
748 node.
750 Parameters
751 ----------
752 node : `QuantumNode`
753 The node of the graph for which outputs are to be determined
755 Returns
756 -------
757 set of `QuantumNode`
758 All the nodes that are direct outputs to specified node
759 """
760 return set(succ for succ in self._connectedQuanta.successors(node))
762 def determineConnectionsOfQuantumNode(self: _T, node: QuantumNode) -> _T:
763 """Return a graph of `QuantumNode` that are direct inputs and outputs
764 of a specified node.
766 Parameters
767 ----------
768 node : `QuantumNode`
769 The node of the graph for which connected nodes are to be
770 determined.
772 Returns
773 -------
774 graph : graph of `QuantumNode`
775 All the nodes that are directly connected to specified node
776 """
777 nodes = self.determineInputsToQuantumNode(node).union(self.determineOutputsOfQuantumNode(node))
778 nodes.add(node)
779 return self.subset(nodes)
781 def determineAncestorsOfQuantumNode(self: _T, node: QuantumNode) -> _T:
782 """Return a graph of the specified node and all the ancestor nodes
783 directly reachable by walking edges.
785 Parameters
786 ----------
787 node : `QuantumNode`
788 The node for which all ansestors are to be determined
790 Returns
791 -------
792 graph of `QuantumNode`
793 Graph of node and all of its ansestors
794 """
795 predecessorNodes = nx.ancestors(self._connectedQuanta, node)
796 predecessorNodes.add(node)
797 return self.subset(predecessorNodes)
799 def findCycle(self) -> List[Tuple[QuantumNode, QuantumNode]]:
800 """Check a graph for the presense of cycles and returns the edges of
801 any cycles found, or an empty list if there is no cycle.
803 Returns
804 -------
805 result : list of tuple of `QuantumNode`, `QuantumNode`
806 A list of any graph edges that form a cycle, or an empty list if
807 there is no cycle. Empty list to so support if graph.find_cycle()
808 syntax as an empty list is falsy.
809 """
810 try:
811 return nx.find_cycle(self._connectedQuanta)
812 except nx.NetworkXNoCycle:
813 return []
815 def saveUri(self, uri: ResourcePathExpression) -> None:
816 """Save `QuantumGraph` to the specified URI.
818 Parameters
819 ----------
820 uri : convertible to `ResourcePath`
821 URI to where the graph should be saved.
822 """
823 buffer = self._buildSaveObject()
824 path = ResourcePath(uri)
825 if path.getExtension() not in (".qgraph"):
826 raise TypeError(f"Can currently only save a graph in qgraph format not {uri}")
827 path.write(buffer) # type: ignore # Ignore because bytearray is safe to use in place of bytes
829 @property
830 def metadata(self) -> Optional[MappingProxyType[str, Any]]:
831 """Extra data carried with the graph (mapping [`str`] or `None`).
833 The mapping is a dynamic view of this object's metadata. Values should
834 be able to be serialized in JSON.
835 """
836 if self._metadata is None:
837 return None
838 return MappingProxyType(self._metadata)
840 def initInputRefs(self, taskDef: TaskDef) -> Optional[List[DatasetRef]]:
841 """Return DatasetRefs for a given task InitInputs.
843 Parameters
844 ----------
845 taskDef : `TaskDef`
846 Task definition structure.
848 Returns
849 -------
850 refs : `list` [ `DatasetRef` ] or None
851 DatasetRef for the task InitInput, can be `None`. This can return
852 either resolved or non-resolved reference.
853 """
854 return self._initInputRefs.get(taskDef)
856 def initOutputRefs(self, taskDef: TaskDef) -> Optional[List[DatasetRef]]:
857 """Return DatasetRefs for a given task InitOutputs.
859 Parameters
860 ----------
861 taskDef : `TaskDef`
862 Task definition structure.
864 Returns
865 -------
866 refs : `list` [ `DatasetRef` ] or None
867 DatasetRefs for the task InitOutput, can be `None`. This can return
868 either resolved or non-resolved reference. Resolved reference will
869 match Quantum's initInputs if this is an intermediate dataset type.
870 """
871 return self._initOutputRefs.get(taskDef)
873 def globalInitOutputRefs(self) -> List[DatasetRef]:
874 """Return DatasetRefs for global InitOutputs.
876 Returns
877 -------
878 refs : `list` [ `DatasetRef` ]
879 DatasetRefs for global InitOutputs.
880 """
881 return self._globalInitOutputRefs
883 def registryDatasetTypes(self) -> List[DatasetType]:
884 """Return dataset types used by this graph, their definitions match
885 dataset types from registry.
887 Returns
888 -------
889 refs : `list` [ `DatasetType` ]
890 Dataset types for this graph.
891 """
892 return self._registryDatasetTypes
894 @classmethod
895 def loadUri(
896 cls,
897 uri: ResourcePathExpression,
898 universe: Optional[DimensionUniverse] = None,
899 nodes: Optional[Iterable[uuid.UUID]] = None,
900 graphID: Optional[BuildId] = None,
901 minimumVersion: int = 3,
902 ) -> QuantumGraph:
903 """Read `QuantumGraph` from a URI.
905 Parameters
906 ----------
907 uri : convertible to `ResourcePath`
908 URI from where to load the graph.
909 universe: `~lsst.daf.butler.DimensionUniverse` optional
910 DimensionUniverse instance, not used by the method itself but
911 needed to ensure that registry data structures are initialized.
912 If None it is loaded from the QuantumGraph saved structure. If
913 supplied, the DimensionUniverse from the loaded `QuantumGraph`
914 will be validated against the supplied argument for compatibility.
915 nodes: iterable of `int` or None
916 Numbers that correspond to nodes in the graph. If specified, only
917 these nodes will be loaded. Defaults to None, in which case all
918 nodes will be loaded.
919 graphID : `str` or `None`
920 If specified this ID is verified against the loaded graph prior to
921 loading any Nodes. This defaults to None in which case no
922 validation is done.
923 minimumVersion : int
924 Minimum version of a save file to load. Set to -1 to load all
925 versions. Older versions may need to be loaded, and re-saved
926 to upgrade them to the latest format before they can be used in
927 production.
929 Returns
930 -------
931 graph : `QuantumGraph`
932 Resulting QuantumGraph instance.
934 Raises
935 ------
936 TypeError
937 Raised if pickle contains instance of a type other than
938 QuantumGraph.
939 ValueError
940 Raised if one or more of the nodes requested is not in the
941 `QuantumGraph` or if graphID parameter does not match the graph
942 being loaded or if the supplied uri does not point at a valid
943 `QuantumGraph` save file.
944 RuntimeError
945 Raise if Supplied DimensionUniverse is not compatible with the
946 DimensionUniverse saved in the graph
949 Notes
950 -----
951 Reading Quanta from pickle requires existence of singleton
952 DimensionUniverse which is usually instantiated during Registry
953 initialization. To make sure that DimensionUniverse exists this method
954 accepts dummy DimensionUniverse argument.
955 """
956 uri = ResourcePath(uri)
957 # With ResourcePath we have the choice of always using a local file
958 # or reading in the bytes directly. Reading in bytes can be more
959 # efficient for reasonably-sized pickle files when the resource
960 # is remote. For now use the local file variant. For a local file
961 # as_local() does nothing.
963 if uri.getExtension() in (".pickle", ".pkl"):
964 with uri.as_local() as local, open(local.ospath, "rb") as fd:
965 warnings.warn("Pickle graphs are deprecated, please re-save your graph with the save method")
966 qgraph = pickle.load(fd)
967 elif uri.getExtension() in (".qgraph"):
968 with LoadHelper(uri, minimumVersion) as loader:
969 qgraph = loader.load(universe, nodes, graphID)
970 else:
971 raise ValueError("Only know how to handle files saved as `pickle`, `pkl`, or `qgraph`")
972 if not isinstance(qgraph, QuantumGraph):
973 raise TypeError(f"QuantumGraph save file contains unexpected object type: {type(qgraph)}")
974 return qgraph
976 @classmethod
977 def readHeader(cls, uri: ResourcePathExpression, minimumVersion: int = 3) -> Optional[str]:
978 """Read the header of a `QuantumGraph` pointed to by the uri parameter
979 and return it as a string.
981 Parameters
982 ----------
983 uri : convertible to `ResourcePath`
984 The location of the `QuantumGraph` to load. If the argument is a
985 string, it must correspond to a valid `ResourcePath` path.
986 minimumVersion : int
987 Minimum version of a save file to load. Set to -1 to load all
988 versions. Older versions may need to be loaded, and re-saved
989 to upgrade them to the latest format before they can be used in
990 production.
992 Returns
993 -------
994 header : `str` or `None`
995 The header associated with the specified `QuantumGraph` it there is
996 one, else `None`.
998 Raises
999 ------
1000 ValueError
1001 Raised if `QuantuGraph` was saved as a pickle.
1002 Raised if the extention of the file specified by uri is not a
1003 `QuantumGraph` extention.
1004 """
1005 uri = ResourcePath(uri)
1006 if uri.getExtension() in (".pickle", ".pkl"):
1007 raise ValueError("Reading a header from a pickle save is not supported")
1008 elif uri.getExtension() in (".qgraph"):
1009 return LoadHelper(uri, minimumVersion).readHeader()
1010 else:
1011 raise ValueError("Only know how to handle files saved as `qgraph`")
1013 def buildAndPrintHeader(self) -> None:
1014 """Creates a header that would be used in a save of this object and
1015 prints it out to standard out.
1016 """
1017 _, header = self._buildSaveObject(returnHeader=True)
1018 print(json.dumps(header))
1020 def save(self, file: BinaryIO) -> None:
1021 """Save QuantumGraph to a file.
1023 Parameters
1024 ----------
1025 file : `io.BufferedIOBase`
1026 File to write pickle data open in binary mode.
1027 """
1028 buffer = self._buildSaveObject()
1029 file.write(buffer) # type: ignore # Ignore because bytearray is safe to use in place of bytes
1031 def _buildSaveObject(self, returnHeader: bool = False) -> Union[bytearray, Tuple[bytearray, Dict]]:
1032 # make some containers
1033 jsonData: Deque[bytes] = deque()
1034 # node map is a list because json does not accept mapping keys that
1035 # are not strings, so we store a list of key, value pairs that will
1036 # be converted to a mapping on load
1037 nodeMap = []
1038 taskDefMap = {}
1039 headerData: Dict[str, Any] = {}
1041 # Store the QauntumGraph BuildId, this will allow validating BuildIds
1042 # at load time, prior to loading any QuantumNodes. Name chosen for
1043 # unlikely conflicts.
1044 headerData["GraphBuildID"] = self.graphID
1045 headerData["Metadata"] = self._metadata
1047 # Store the universe this graph was created with
1048 universeConfig = self._universe.dimensionConfig
1049 headerData["universe"] = universeConfig.toDict()
1051 # counter for the number of bytes processed thus far
1052 count = 0
1053 # serialize out the task Defs recording the start and end bytes of each
1054 # taskDef
1055 inverseLookup = self._datasetDict.inverse
1056 taskDef: TaskDef
1057 # sort by task label to ensure serialization happens in the same order
1058 for taskDef in self.taskGraph:
1059 # compressing has very little impact on saving or load time, but
1060 # a large impact on on disk size, so it is worth doing
1061 taskDescription: Dict[str, Any] = {}
1062 # save the fully qualified name.
1063 taskDescription["taskName"] = get_full_type_name(taskDef.taskClass)
1064 # save the config as a text stream that will be un-persisted on the
1065 # other end
1066 stream = io.StringIO()
1067 taskDef.config.saveToStream(stream)
1068 taskDescription["config"] = stream.getvalue()
1069 taskDescription["label"] = taskDef.label
1070 if (refs := self._initInputRefs.get(taskDef)) is not None:
1071 taskDescription["initInputRefs"] = [ref.to_json() for ref in refs]
1072 if (refs := self._initOutputRefs.get(taskDef)) is not None:
1073 taskDescription["initOutputRefs"] = [ref.to_json() for ref in refs]
1075 inputs = []
1076 outputs = []
1078 # Determine the connection between all of tasks and save that in
1079 # the header as a list of connections and edges in each task
1080 # this will help in un-persisting, and possibly in a "quick view"
1081 # method that does not require everything to be un-persisted
1082 #
1083 # Typing returns can't be parameter dependent
1084 for connection in inverseLookup[taskDef]: # type: ignore
1085 consumers = self._datasetDict.getConsumers(connection)
1086 producer = self._datasetDict.getProducer(connection)
1087 if taskDef in consumers:
1088 # This checks if the task consumes the connection directly
1089 # from the datastore or it is produced by another task
1090 producerLabel = producer.label if producer is not None else "datastore"
1091 inputs.append((producerLabel, connection))
1092 elif taskDef not in consumers and producer is taskDef:
1093 # If there are no consumers for this tasks produced
1094 # connection, the output will be said to be the datastore
1095 # in which case the for loop will be a zero length loop
1096 if not consumers:
1097 outputs.append(("datastore", connection))
1098 for td in consumers:
1099 outputs.append((td.label, connection))
1101 # dump to json string, and encode that string to bytes and then
1102 # conpress those bytes
1103 dump = lzma.compress(json.dumps(taskDescription).encode())
1104 # record the sizing and relation information
1105 taskDefMap[taskDef.label] = {
1106 "bytes": (count, count + len(dump)),
1107 "inputs": inputs,
1108 "outputs": outputs,
1109 }
1110 count += len(dump)
1111 jsonData.append(dump)
1113 headerData["TaskDefs"] = taskDefMap
1115 # serialize the nodes, recording the start and end bytes of each node
1116 dimAccumulator = DimensionRecordsAccumulator()
1117 for node in self:
1118 # compressing has very little impact on saving or load time, but
1119 # a large impact on on disk size, so it is worth doing
1120 simpleNode = node.to_simple(accumulator=dimAccumulator)
1122 dump = lzma.compress(simpleNode.json().encode())
1123 jsonData.append(dump)
1124 nodeMap.append(
1125 (
1126 str(node.nodeId),
1127 {
1128 "bytes": (count, count + len(dump)),
1129 "inputs": [str(n.nodeId) for n in self.determineInputsToQuantumNode(node)],
1130 "outputs": [str(n.nodeId) for n in self.determineOutputsOfQuantumNode(node)],
1131 },
1132 )
1133 )
1134 count += len(dump)
1136 headerData["DimensionRecords"] = {
1137 key: value.dict() for key, value in dimAccumulator.makeSerializedDimensionRecordMapping().items()
1138 }
1140 # need to serialize this as a series of key,value tuples because of
1141 # a limitation on how json cant do anything but strings as keys
1142 headerData["Nodes"] = nodeMap
1144 if self._globalInitOutputRefs:
1145 headerData["GlobalInitOutputRefs"] = [ref.to_json() for ref in self._globalInitOutputRefs]
1147 if self._registryDatasetTypes:
1148 headerData["RegistryDatasetTypes"] = [dstype.to_json() for dstype in self._registryDatasetTypes]
1150 # dump the headerData to json
1151 header_encode = lzma.compress(json.dumps(headerData).encode())
1153 # record the sizes as 2 unsigned long long numbers for a total of 16
1154 # bytes
1155 save_bytes = struct.pack(STRUCT_FMT_BASE, SAVE_VERSION)
1157 fmt_string = DESERIALIZER_MAP[SAVE_VERSION].FMT_STRING()
1158 map_lengths = struct.pack(fmt_string, len(header_encode))
1160 # write each component of the save out in a deterministic order
1161 # buffer = io.BytesIO()
1162 # buffer.write(map_lengths)
1163 # buffer.write(taskDef_pickle)
1164 # buffer.write(map_pickle)
1165 buffer = bytearray()
1166 buffer.extend(MAGIC_BYTES)
1167 buffer.extend(save_bytes)
1168 buffer.extend(map_lengths)
1169 buffer.extend(header_encode)
1170 # Iterate over the length of pickleData, and for each element pop the
1171 # leftmost element off the deque and write it out. This is to save
1172 # memory, as the memory is added to the buffer object, it is removed
1173 # from from the container.
1174 #
1175 # Only this section needs to worry about memory pressue because
1176 # everything else written to the buffer prior to this pickle data is
1177 # only on the order of kilobytes to low numbers of megabytes.
1178 while jsonData:
1179 buffer.extend(jsonData.popleft())
1180 if returnHeader:
1181 return buffer, headerData
1182 else:
1183 return buffer
1185 @classmethod
1186 def load(
1187 cls,
1188 file: BinaryIO,
1189 universe: Optional[DimensionUniverse] = None,
1190 nodes: Optional[Iterable[uuid.UUID]] = None,
1191 graphID: Optional[BuildId] = None,
1192 minimumVersion: int = 3,
1193 ) -> QuantumGraph:
1194 """Read QuantumGraph from a file that was made by `save`.
1196 Parameters
1197 ----------
1198 file : `io.IO` of bytes
1199 File with pickle data open in binary mode.
1200 universe: `~lsst.daf.butler.DimensionUniverse`, optional
1201 DimensionUniverse instance, not used by the method itself but
1202 needed to ensure that registry data structures are initialized.
1203 If None it is loaded from the QuantumGraph saved structure. If
1204 supplied, the DimensionUniverse from the loaded `QuantumGraph`
1205 will be validated against the supplied argument for compatibility.
1206 nodes: iterable of `int` or None
1207 Numbers that correspond to nodes in the graph. If specified, only
1208 these nodes will be loaded. Defaults to None, in which case all
1209 nodes will be loaded.
1210 graphID : `str` or `None`
1211 If specified this ID is verified against the loaded graph prior to
1212 loading any Nodes. This defaults to None in which case no
1213 validation is done.
1214 minimumVersion : int
1215 Minimum version of a save file to load. Set to -1 to load all
1216 versions. Older versions may need to be loaded, and re-saved
1217 to upgrade them to the latest format before they can be used in
1218 production.
1220 Returns
1221 -------
1222 graph : `QuantumGraph`
1223 Resulting QuantumGraph instance.
1225 Raises
1226 ------
1227 TypeError
1228 Raised if pickle contains instance of a type other than
1229 QuantumGraph.
1230 ValueError
1231 Raised if one or more of the nodes requested is not in the
1232 `QuantumGraph` or if graphID parameter does not match the graph
1233 being loaded or if the supplied uri does not point at a valid
1234 `QuantumGraph` save file.
1236 Notes
1237 -----
1238 Reading Quanta from pickle requires existence of singleton
1239 DimensionUniverse which is usually instantiated during Registry
1240 initialization. To make sure that DimensionUniverse exists this method
1241 accepts dummy DimensionUniverse argument.
1242 """
1243 # Try to see if the file handle contains pickle data, this will be
1244 # removed in the future
1245 try:
1246 qgraph = pickle.load(file)
1247 warnings.warn("Pickle graphs are deprecated, please re-save your graph with the save method")
1248 except pickle.UnpicklingError:
1249 with LoadHelper(file, minimumVersion) as loader:
1250 qgraph = loader.load(universe, nodes, graphID)
1251 if not isinstance(qgraph, QuantumGraph):
1252 raise TypeError(f"QuantumGraph pickle file has contains unexpected object type: {type(qgraph)}")
1253 return qgraph
1255 def iterTaskGraph(self) -> Generator[TaskDef, None, None]:
1256 """Iterate over the `taskGraph` attribute in topological order
1258 Yields
1259 ------
1260 taskDef : `TaskDef`
1261 `TaskDef` objects in topological order
1262 """
1263 yield from nx.topological_sort(self.taskGraph)
1265 def updateRun(self, run: str, *, metadata_key: str | None = None, update_graph_id: bool = False) -> None:
1266 """Change output run and dataset ID for each output dataset.
1268 Parameters
1269 ----------
1270 run : `str`
1271 New output run name.
1272 metadata_key : `str` or `None`
1273 Specifies matadata key corresponding to output run name to update
1274 with new run name. If `None` or if metadata is missing it is not
1275 updated. If metadata is present but key is missing, it will be
1276 added.
1277 update_graph_id : `bool`, optional
1278 If `True` then also update graph ID with a new unique value.
1279 """
1280 dataset_id_map = {}
1282 def _update_output_refs_in_place(refs: list[DatasetRef], run: str) -> None:
1283 """Updated list of DatasetRef with new run and dataset IDs."""
1284 new_refs = []
1285 for ref in refs:
1286 new_ref = DatasetRef(ref.datasetType, ref.dataId, run=run, conform=False)
1287 dataset_id_map[ref.id] = new_ref.id
1288 new_refs.append(new_ref)
1289 refs[:] = new_refs
1291 def _update_input_refs_in_place(refs: list[DatasetRef], run: str) -> None:
1292 """Updated list of DatasetRef with IDs from dataset_id_map."""
1293 new_refs = []
1294 for ref in refs:
1295 if (new_id := dataset_id_map.get(ref.id)) is not None:
1296 new_ref = DatasetRef(ref.datasetType, ref.dataId, id=new_id, run=run, conform=False)
1297 new_refs.append(new_ref)
1298 else:
1299 new_refs.append(ref)
1300 refs[:] = new_refs
1302 # Loop through all outputs and update their datasets.
1303 for node in self._connectedQuanta:
1304 for refs in node.quantum.outputs.values():
1305 _update_output_refs_in_place(refs, run)
1307 for refs in self._initOutputRefs.values():
1308 _update_output_refs_in_place(refs, run)
1310 _update_output_refs_in_place(self._globalInitOutputRefs, run)
1312 # Update all intermediates from their matching outputs.
1313 for node in self._connectedQuanta:
1314 for refs in node.quantum.inputs.values():
1315 _update_input_refs_in_place(refs, run)
1317 for refs in self._initInputRefs.values():
1318 _update_input_refs_in_place(refs, run)
1320 if update_graph_id:
1321 self._buildId = BuildId(f"{time.time()}-{os.getpid()}")
1323 # Update metadata if present.
1324 if self._metadata is not None and metadata_key is not None:
1325 metadata = dict(self._metadata)
1326 metadata[metadata_key] = run
1327 self._metadata = metadata
1329 @property
1330 def graphID(self) -> BuildId:
1331 """The ID generated by the graph at construction time (`str`)."""
1332 return self._buildId
1334 @property
1335 def universe(self) -> DimensionUniverse:
1336 """Dimension universe associated with this graph
1337 (`~lsst.daf.butler.DimensionUniverse`).
1338 """
1339 return self._universe
1341 def __iter__(self) -> Generator[QuantumNode, None, None]:
1342 yield from nx.topological_sort(self._connectedQuanta)
1344 def __len__(self) -> int:
1345 return self._count
1347 def __contains__(self, node: QuantumNode) -> bool:
1348 return self._connectedQuanta.has_node(node)
1350 def __getstate__(self) -> dict:
1351 """Stores a compact form of the graph as a list of graph nodes, and a
1352 tuple of task labels and task configs. The full graph can be
1353 reconstructed with this information, and it preserves the ordering of
1354 the graph nodes.
1355 """
1356 universe: Optional[DimensionUniverse] = None
1357 for node in self:
1358 dId = node.quantum.dataId
1359 if dId is None:
1360 continue
1361 universe = dId.graph.universe
1362 return {"reduced": self._buildSaveObject(), "graphId": self._buildId, "universe": universe}
1364 def __setstate__(self, state: dict) -> None:
1365 """Reconstructs the state of the graph from the information persisted
1366 in getstate.
1367 """
1368 buffer = io.BytesIO(state["reduced"])
1369 with LoadHelper(buffer, minimumVersion=3) as loader:
1370 qgraph = loader.load(state["universe"], graphID=state["graphId"])
1372 self._metadata = qgraph._metadata
1373 self._buildId = qgraph._buildId
1374 self._datasetDict = qgraph._datasetDict
1375 self._nodeIdMap = qgraph._nodeIdMap
1376 self._count = len(qgraph)
1377 self._taskToQuantumNode = qgraph._taskToQuantumNode
1378 self._taskGraph = qgraph._taskGraph
1379 self._connectedQuanta = qgraph._connectedQuanta
1380 self._initInputRefs = qgraph._initInputRefs
1381 self._initOutputRefs = qgraph._initOutputRefs
1383 def __eq__(self, other: object) -> bool:
1384 if not isinstance(other, QuantumGraph):
1385 return False
1386 if len(self) != len(other):
1387 return False
1388 for node in self:
1389 if node not in other:
1390 return False
1391 if self.determineInputsToQuantumNode(node) != other.determineInputsToQuantumNode(node):
1392 return False
1393 if self.determineOutputsOfQuantumNode(node) != other.determineOutputsOfQuantumNode(node):
1394 return False
1395 if set(self.allDatasetTypes) != set(other.allDatasetTypes):
1396 return False
1397 return set(self.taskGraph) == set(other.taskGraph)