Coverage for python/lsst/pipe/base/graph.py : 35%

Hot-keys on this page
r m x p toggle line displays
j k next/prev highlighted chunk
0 (zero) top of page
1 (one) first highlighted chunk
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"""Module defining quantum graph classes and related methods.
25There could be different representations of the quantum graph depending
26on the client needs. Presently this module contains graph implementation
27which is based on requirements of command-line environment. In the future
28we could add other implementations and methods to convert between those
29representations.
30"""
32# "exported" names
33__all__ = ["QuantumGraph", "QuantumGraphTaskNodes", "QuantumIterData"]
35# -------------------------------
36# Imports of standard modules --
37# -------------------------------
38from itertools import chain
39from dataclasses import dataclass
40import pickle
41from typing import List, FrozenSet, Mapping
43# -----------------------------
44# Imports for other modules --
45# -----------------------------
46from .pipeline import TaskDef
47from .pipeTools import orderPipeline
48from lsst.daf.butler import DatasetRef, DatasetType, NamedKeyDict, Quantum
50# ----------------------------------
51# Local non-exported definitions --
52# ----------------------------------
54# ------------------------
55# Exported definitions --
56# ------------------------
59@dataclass
60class QuantumIterData:
61 """Helper class for iterating over quanta in a graph.
63 The `QuantumGraph.traverse` method needs to return topologically ordered
64 Quanta together with their dependencies. This class is used as a value
65 for the iterator, it contains enumerated Quantum and its dependencies.
66 """
68 __slots__ = ["index", "quantum", "taskDef", "dependencies"]
70 index: int
71 """Index of this Quantum, a unique but arbitrary integer."""
73 quantum: Quantum
74 """Quantum corresponding to a graph node."""
76 taskDef: TaskDef
77 """Task class to be run on this quantum, and corresponding label and
78 config.
79 """
81 dependencies: FrozenSet(int)
82 """Possibly empty set of indices of dependencies for this Quantum.
83 Dependencies include other nodes in the graph; they do not reflect data
84 already in butler (there are no graph nodes for those).
85 """
88@dataclass
89class QuantumGraphTaskNodes:
90 """QuantumGraphTaskNodes represents a bunch of nodes in an quantum graph
91 corresponding to a single task.
93 The node in quantum graph is represented by the `PipelineTask` and a
94 single `~lsst.daf.butler.Quantum` instance. One possible representation
95 of the graph is just a list of nodes without edges (edges can be deduced
96 from nodes' quantum inputs and outputs if needed). That representation can
97 be reduced to the list of PipelineTasks (or their corresponding TaskDefs)
98 and the corresponding list of Quanta. This class is used in this reduced
99 representation for a single task, and full `QuantumGraph` is a sequence of
100 tinstances of this class for one or more tasks.
102 Different frameworks may use different graph representation, this
103 representation was based mostly on requirements of command-line
104 executor which does not need explicit edges information.
105 """
107 taskDef: TaskDef
108 """Task defintion for this set of nodes."""
110 quanta: List[Quantum]
111 """List of quanta corresponding to the task."""
113 initInputs: Mapping[DatasetType, DatasetRef]
114 """Datasets that must be loaded or created to construct this task."""
116 initOutputs: Mapping[DatasetType, DatasetRef]
117 """Datasets that may be written after constructing this task."""
120class QuantumGraph(list):
121 """QuantumGraph is a sequence of `QuantumGraphTaskNodes` objects.
123 Typically the order of the tasks in the list will be the same as the
124 order of tasks in a pipeline (obviously depends on the code which
125 constructs graph).
127 Parameters
128 ----------
129 iterable : iterable of `QuantumGraphTaskNodes`, optional
130 Initial sequence of per-task nodes.
131 """
132 def __init__(self, iterable=None):
133 list.__init__(self, iterable or [])
134 self.initInputs = NamedKeyDict()
135 self.initIntermediates = NamedKeyDict()
136 self.initOutputs = NamedKeyDict()
138 initInputs: NamedKeyDict
139 """Datasets that must be provided to construct one or more Tasks in this
140 graph, and must be obtained from the data repository.
142 This is disjoint with both `initIntermediates` and `initOutputs`.
143 """
145 initIntermediates: NamedKeyDict
146 """Datasets that must be provided to construct one or more Tasks in this
147 graph, but are also produced after constructing a Task in this graph.
149 This is disjoint with both `initInputs` and `initOutputs`.
150 """
152 initOutputs: NamedKeyDict
153 """Datasets that are produced after constructing a Task in this graph,
154 and are not used to construct any other Task in this graph.
156 This is disjoint from both `initInputs` and `initIntermediates`.
157 """
159 @classmethod
160 def load(cls, file, universe):
161 """Read QuantumGraph from a file that was made by `save`.
163 Parameters
164 ----------
165 file : `io.BufferedIOBase`
166 File with pickle data open in binary mode.
167 universe: `~lsst.daf.butler.DimensionUniverse`
168 DimensionUniverse instance, not used by the method itself but
169 needed to ensure that registry data structures are initialized.
171 Returns
172 -------
173 graph : `QuantumGraph`
174 Resulting QuantumGraph instance.
176 Raises
177 ------
178 TypeError
179 Raised if pickle contains instance of a type other than
180 QuantumGraph.
182 Notes
183 -----
184 Reading Quanta from pickle requires existence of singleton
185 DimensionUniverse which is usually instantiated during Registry
186 initializaion. To make sure that DimensionUniverse exists this method
187 accepts dummy DimensionUniverse argument.
188 """
189 qgraph = pickle.load(file)
190 if not isinstance(qgraph, QuantumGraph):
191 raise TypeError(f"QuantumGraph pickle file has contains unexpected object type: {type(qgraph)}")
192 return qgraph
194 def save(self, file):
195 """Save QuantumGraph to a file.
197 Presently we store QuantumGraph in pickle format, this could
198 potentially change in the future if better format is found.
200 Parameters
201 ----------
202 file : `io.BufferedIOBase`
203 File to write pickle data open in binary mode.
204 """
205 pickle.dump(self, file)
207 def quanta(self):
208 """Iterator over quanta in a graph.
210 Quanta are returned in unspecified order.
212 Yields
213 ------
214 taskDef : `TaskDef`
215 Task definition for a Quantum.
216 quantum : `~lsst.daf.butler.Quantum`
217 Single quantum.
218 """
219 for taskNodes in self:
220 taskDef = taskNodes.taskDef
221 for quantum in taskNodes.quanta:
222 yield taskDef, quantum
224 def quantaAsQgraph(self):
225 """Iterator over quanta in a graph.
227 QuantumGraph containing individual quanta are returned.
229 Yields
230 ------
231 graph : `QuantumGraph`
232 """
233 for qdata in self.traverse():
234 node = QuantumGraphTaskNodes(qdata.taskDef, [qdata.quantum],
235 qdata.quantum.initInputs, qdata.quantum.outputs)
236 graph = QuantumGraph([node])
237 yield graph
239 def countQuanta(self):
240 """Return total count of quanta in a graph.
242 Returns
243 -------
244 count : `int`
245 Number of quanta in a graph.
246 """
247 return sum(len(taskNodes.quanta) for taskNodes in self)
249 def traverse(self):
250 """Return topologically ordered Quanta and their dependencies.
252 This method iterates over all Quanta in topological order, enumerating
253 them during iteration. Returned `QuantumIterData` object contains
254 Quantum instance, its ``index`` and the ``index`` of all its
255 prerequsites (Quanta that produce inputs for this Quantum):
257 - the ``index`` values are generated by an iteration of a
258 QuantumGraph, and are not intrinsic to the QuantumGraph
259 - during iteration, each ID will appear in index before it ever
260 appears in dependencies.
262 Yields
263 ------
264 quantumData : `QuantumIterData`
265 """
267 def orderedTaskNodes(graph):
268 """Return topologically ordered task nodes.
270 Yields
271 ------
272 nodes : `QuantumGraphTaskNodes`
273 """
274 # Tasks in a graph are probably topologically sorted already but there
275 # is no guarantee for that. Just re-construct Pipeline and order tasks
276 # in a pipeline using existing method.
277 nodesMap = {id(item.taskDef): item for item in graph}
278 pipeline = orderPipeline([item.taskDef for item in graph])
279 for taskDef in pipeline:
280 yield nodesMap[id(taskDef)]
282 index = 0
283 outputs = {} # maps (DatasetType.name, dataId) to its producing quantum index
284 for nodes in orderedTaskNodes(self):
285 for quantum in nodes.quanta:
287 # Find quantum dependencies (must be in `outputs` already)
288 prereq = []
289 for dataRef in chain.from_iterable(quantum.predictedInputs.values()):
290 # if data exists in butler then `id` is not None
291 if dataRef.id is None:
292 # Get the base name if this is a component
293 name, component = dataRef.datasetType.nameAndComponent()
294 key = (name, dataRef.dataId)
295 try:
296 prereq.append(outputs[key])
297 except KeyError:
298 # The Quantum that makes our inputs is not in the graph,
299 # this could happen if we run on a "split graph" which is
300 # usually just one quantum. Check for number of Quanta
301 # in a graph and ignore error if it's just one.
302 # TODO: This code has to be removed or replaced with
303 # something more generic
304 if not (len(self) == 1 and len(self[0].quanta) == 1):
305 raise
307 # Update `outputs` with this quantum outputs
308 for dataRef in chain.from_iterable(quantum.outputs.values()):
309 key = (dataRef.datasetType.name, dataRef.dataId)
310 outputs[key] = index
312 yield QuantumIterData(index=index, quantum=quantum, taskDef=nodes.taskDef,
313 dependencies=frozenset(prereq))
314 index += 1