Coverage for python/lsst/pipe/base/pipeTools.py: 11%
Shortcuts 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
Shortcuts 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/>.
22"""Module defining few methods to manipulate or query pipelines.
23"""
25# No one should do import * from this module
26__all__ = ["isPipelineOrdered", "orderPipeline"]
28# -------------------------------
29# Imports of standard modules --
30# -------------------------------
31import itertools
33# -----------------------------
34# Imports for other modules --
35# -----------------------------
36from .connections import iterConnections
38# ----------------------------------
39# Local non-exported definitions --
40# ----------------------------------
42# ------------------------
43# Exported definitions --
44# ------------------------
47class MissingTaskFactoryError(Exception):
48 """Exception raised when client fails to provide TaskFactory instance.
49 """
50 pass
53class DuplicateOutputError(Exception):
54 """Exception raised when Pipeline has more than one task for the same
55 output.
56 """
57 pass
60class PipelineDataCycleError(Exception):
61 """Exception raised when Pipeline has data dependency cycle.
62 """
63 pass
66def isPipelineOrdered(pipeline, taskFactory=None):
67 """Checks whether tasks in pipeline are correctly ordered.
69 Pipeline is correctly ordered if for any DatasetType produced by a task
70 in a pipeline all its consumer tasks are located after producer.
72 Parameters
73 ----------
74 pipeline : `pipe.base.Pipeline`
75 Pipeline description.
76 taskFactory: `pipe.base.TaskFactory`, optional
77 Instance of an object which knows how to import task classes. It is
78 only used if pipeline task definitions do not define task classes.
80 Returns
81 -------
82 True for correctly ordered pipeline, False otherwise.
84 Raises
85 ------
86 `ImportError` is raised when task class cannot be imported.
87 `DuplicateOutputError` is raised when there is more than one producer for a
88 dataset type.
89 `MissingTaskFactoryError` is raised when TaskFactory is needed but not
90 provided.
91 """
92 # Build a map of DatasetType name to producer's index in a pipeline
93 producerIndex = {}
94 for idx, taskDef in enumerate(pipeline):
96 for attr in iterConnections(taskDef.connections, 'outputs'):
97 if attr.name in producerIndex:
98 raise DuplicateOutputError("DatasetType `{}' appears more than "
99 "once as output".format(attr.name))
100 producerIndex[attr.name] = idx
102 # check all inputs that are also someone's outputs
103 for idx, taskDef in enumerate(pipeline):
105 # get task input DatasetTypes, this can only be done via class method
106 inputs = {name: getattr(taskDef.connections, name) for name in taskDef.connections.inputs}
107 for dsTypeDescr in inputs.values():
108 # all pre-existing datasets have effective index -1
109 prodIdx = producerIndex.get(dsTypeDescr.name, -1)
110 if prodIdx >= idx:
111 # not good, producer is downstream
112 return False
114 return True
117def orderPipeline(pipeline):
118 """Re-order tasks in pipeline to satisfy data dependencies.
120 When possible new ordering keeps original relative order of the tasks.
122 Parameters
123 ----------
124 pipeline : `list` of `pipe.base.TaskDef`
125 Pipeline description.
127 Returns
128 -------
129 Correctly ordered pipeline (`list` of `pipe.base.TaskDef` objects).
131 Raises
132 ------
133 `DuplicateOutputError` is raised when there is more than one producer for a
134 dataset type.
135 `PipelineDataCycleError` is also raised when pipeline has dependency
136 cycles. `MissingTaskFactoryError` is raised when TaskFactory is needed but
137 not provided.
138 """
140 # This is a modified version of Kahn's algorithm that preserves order
142 # build mapping of the tasks to their inputs and outputs
143 inputs = {} # maps task index to its input DatasetType names
144 outputs = {} # maps task index to its output DatasetType names
145 allInputs = set() # all inputs of all tasks
146 allOutputs = set() # all outputs of all tasks
147 for idx, taskDef in enumerate(pipeline):
148 # task outputs
149 dsMap = {name: getattr(taskDef.connections, name) for name in taskDef.connections.outputs}
150 for dsTypeDescr in dsMap.values():
151 if dsTypeDescr.name in allOutputs:
152 raise DuplicateOutputError("DatasetType `{}' appears more than "
153 "once as output".format(dsTypeDescr.name))
154 outputs[idx] = set(dsTypeDescr.name for dsTypeDescr in dsMap.values())
155 allOutputs.update(outputs[idx])
157 # task inputs
158 connectionInputs = itertools.chain(taskDef.connections.inputs, taskDef.connections.prerequisiteInputs)
159 dsMap = [getattr(taskDef.connections, name).name for name in connectionInputs]
160 inputs[idx] = set(dsMap)
161 allInputs.update(inputs[idx])
163 # for simplicity add pseudo-node which is a producer for all pre-existing
164 # inputs, its index is -1
165 preExisting = allInputs - allOutputs
166 outputs[-1] = preExisting
168 # Set of nodes with no incoming edges, initially set to pseudo-node
169 queue = [-1]
170 result = []
171 while queue:
173 # move to final list, drop -1
174 idx = queue.pop(0)
175 if idx >= 0:
176 result.append(idx)
178 # remove task outputs from other tasks inputs
179 thisTaskOutputs = outputs.get(idx, set())
180 for taskInputs in inputs.values():
181 taskInputs -= thisTaskOutputs
183 # find all nodes with no incoming edges and move them to the queue
184 topNodes = [key for key, value in inputs.items() if not value]
185 queue += topNodes
186 for key in topNodes:
187 del inputs[key]
189 # keep queue ordered
190 queue.sort()
192 # if there is something left it means cycles
193 if inputs:
194 # format it in usable way
195 loops = []
196 for idx, inputNames in inputs.items():
197 taskName = pipeline[idx].label
198 outputNames = outputs[idx]
199 edge = " {} -> {} -> {}".format(inputNames, taskName, outputNames)
200 loops.append(edge)
201 raise PipelineDataCycleError("Pipeline has data cycles:\n" + "\n".join(loops))
203 return [pipeline[idx] for idx in result]