Coverage for python/lsst/ctrl/mpexec/dotTools.py: 9%

140 statements  

« prev     ^ index     » next       coverage.py v7.2.7, created at 2023-06-11 09:04 +0000

1# This file is part of ctrl_mpexec. 

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/>. 

21 

22"""Module defining few methods to generate GraphViz diagrams from pipelines 

23or quantum graphs. 

24""" 

25 

26from __future__ import annotations 

27 

28__all__ = ["graph2dot", "pipeline2dot"] 

29 

30# ------------------------------- 

31# Imports of standard modules -- 

32# ------------------------------- 

33import io 

34import re 

35from collections.abc import Iterable 

36from typing import TYPE_CHECKING, Any 

37 

38# ----------------------------- 

39# Imports for other modules -- 

40# ----------------------------- 

41from lsst.daf.butler import DatasetType, DimensionUniverse 

42from lsst.pipe.base import Pipeline, connectionTypes, iterConnections 

43 

44if TYPE_CHECKING: 

45 from lsst.daf.butler import DatasetRef 

46 from lsst.pipe.base import QuantumGraph, QuantumNode, TaskDef 

47 

48# ---------------------------------- 

49# Local non-exported definitions -- 

50# ---------------------------------- 

51 

52# Node styles indexed by node type. 

53_STYLES = dict( 

54 task=dict(shape="box", style="filled,bold", fillcolor="gray70"), 

55 quantum=dict(shape="box", style="filled,bold", fillcolor="gray70"), 

56 dsType=dict(shape="box", style="rounded,filled", fillcolor="gray90"), 

57 dataset=dict(shape="box", style="rounded,filled", fillcolor="gray90"), 

58) 

59 

60 

61def _renderNode(file: io.TextIOBase, nodeName: str, style: str, labels: list[str]) -> None: 

62 """Render GV node""" 

63 label = r"\n".join(labels) 

64 attrib_dict = dict(_STYLES[style], label=label) 

65 attrib = ", ".join([f'{key}="{val}"' for key, val in attrib_dict.items()]) 

66 print(f'"{nodeName}" [{attrib}];', file=file) 

67 

68 

69def _renderTaskNode(nodeName: str, taskDef: TaskDef, file: io.TextIOBase, idx: Any = None) -> None: 

70 """Render GV node for a task""" 

71 labels = [taskDef.label, taskDef.taskName] 

72 if idx is not None: 

73 labels.append(f"index: {idx}") 

74 if taskDef.connections: 

75 # don't print collection of str directly to avoid visually noisy quotes 

76 dimensions_str = ", ".join(sorted(taskDef.connections.dimensions)) 

77 labels.append(f"dimensions: {dimensions_str}") 

78 _renderNode(file, nodeName, "task", labels) 

79 

80 

81def _renderQuantumNode( 

82 nodeName: str, taskDef: TaskDef, quantumNode: QuantumNode, file: io.TextIOBase 

83) -> None: 

84 """Render GV node for a quantum""" 

85 labels = [f"{quantumNode.nodeId}", taskDef.label] 

86 dataId = quantumNode.quantum.dataId 

87 assert dataId is not None, "Quantum DataId cannot be None" 

88 labels.extend(f"{key} = {dataId[key]}" for key in sorted(dataId.keys())) 

89 _renderNode(file, nodeName, "quantum", labels) 

90 

91 

92def _renderDSTypeNode(name: str, dimensions: list[str], file: io.TextIOBase) -> None: 

93 """Render GV node for a dataset type""" 

94 labels = [name] 

95 if dimensions: 

96 labels.append("Dimensions: " + ", ".join(sorted(dimensions))) 

97 _renderNode(file, name, "dsType", labels) 

98 

99 

100def _renderDSNode(nodeName: str, dsRef: DatasetRef, file: io.TextIOBase) -> None: 

101 """Render GV node for a dataset""" 

102 labels = [dsRef.datasetType.name, f"run: {dsRef.run!r}"] 

103 labels.extend(f"{key} = {dsRef.dataId[key]}" for key in sorted(dsRef.dataId.keys())) 

104 _renderNode(file, nodeName, "dataset", labels) 

105 

106 

107def _renderEdge(fromName: str, toName: str, file: io.TextIOBase, **kwargs: Any) -> None: 

108 """Render GV edge""" 

109 if kwargs: 

110 attrib = ", ".join([f'{key}="{val}"' for key, val in kwargs.items()]) 

111 print(f'"{fromName}" -> "{toName}" [{attrib}];', file=file) 

112 else: 

113 print(f'"{fromName}" -> "{toName}";', file=file) 

114 

115 

116def _datasetRefId(dsRef: DatasetRef) -> str: 

117 """Make an identifying string for given ref""" 

118 dsId = [dsRef.datasetType.name] 

119 dsId.extend(f"{key} = {dsRef.dataId[key]}" for key in sorted(dsRef.dataId.keys())) 

120 return ":".join(dsId) 

121 

122 

123def _makeDSNode(dsRef: DatasetRef, allDatasetRefs: dict[str, str], file: io.TextIOBase) -> str: 

124 """Make new node for dataset if it does not exist. 

125 

126 Returns node name. 

127 """ 

128 dsRefId = _datasetRefId(dsRef) 

129 nodeName = allDatasetRefs.get(dsRefId) 

130 if nodeName is None: 

131 idx = len(allDatasetRefs) 

132 nodeName = "dsref_{}".format(idx) 

133 allDatasetRefs[dsRefId] = nodeName 

134 _renderDSNode(nodeName, dsRef, file) 

135 return nodeName 

136 

137 

138# ------------------------ 

139# Exported definitions -- 

140# ------------------------ 

141 

142 

143def graph2dot(qgraph: QuantumGraph, file: Any) -> None: 

144 """Convert QuantumGraph into GraphViz digraph. 

145 

146 This method is mostly for documentation/presentation purposes. 

147 

148 Parameters 

149 ---------- 

150 qgraph: `pipe.base.QuantumGraph` 

151 QuantumGraph instance. 

152 file : str or file object 

153 File where GraphViz graph (DOT language) is written, can be a file name 

154 or file object. 

155 

156 Raises 

157 ------ 

158 `OSError` is raised when output file cannot be open. 

159 `ImportError` is raised when task class cannot be imported. 

160 """ 

161 # open a file if needed 

162 close = False 

163 if not hasattr(file, "write"): 

164 file = open(file, "w") 

165 close = True 

166 

167 print("digraph QuantumGraph {", file=file) 

168 

169 allDatasetRefs: dict[str, str] = {} 

170 for taskId, taskDef in enumerate(qgraph.taskGraph): 

171 quanta = qgraph.getNodesForTask(taskDef) 

172 for qId, quantumNode in enumerate(quanta): 

173 # node for a task 

174 taskNodeName = "task_{}_{}".format(taskId, qId) 

175 _renderQuantumNode(taskNodeName, taskDef, quantumNode, file) 

176 

177 # quantum inputs 

178 for dsRefs in quantumNode.quantum.inputs.values(): 

179 for dsRef in dsRefs: 

180 nodeName = _makeDSNode(dsRef, allDatasetRefs, file) 

181 _renderEdge(nodeName, taskNodeName, file) 

182 

183 # quantum outputs 

184 for dsRefs in quantumNode.quantum.outputs.values(): 

185 for dsRef in dsRefs: 

186 nodeName = _makeDSNode(dsRef, allDatasetRefs, file) 

187 _renderEdge(taskNodeName, nodeName, file) 

188 

189 print("}", file=file) 

190 if close: 

191 file.close() 

192 

193 

194def pipeline2dot(pipeline: Pipeline | Iterable[TaskDef], file: Any) -> None: 

195 """Convert Pipeline into GraphViz digraph. 

196 

197 This method is mostly for documentation/presentation purposes. 

198 Unlike other methods this method does not validate graph consistency. 

199 

200 Parameters 

201 ---------- 

202 pipeline : `pipe.base.Pipeline` 

203 Pipeline description. 

204 file : str or file object 

205 File where GraphViz graph (DOT language) is written, can be a file name 

206 or file object. 

207 

208 Raises 

209 ------ 

210 `OSError` is raised when output file cannot be open. 

211 `ImportError` is raised when task class cannot be imported. 

212 `MissingTaskFactoryError` is raised when TaskFactory is needed but not 

213 provided. 

214 """ 

215 universe = DimensionUniverse() 

216 

217 def expand_dimensions(connection: connectionTypes.BaseConnection) -> list[str]: 

218 """Return expanded list of dimensions, with special skypix treatment. 

219 

220 Parameters 

221 ---------- 

222 dimensions : `list` [`str`] 

223 

224 Returns 

225 ------- 

226 dimensions : `list` [`str`] 

227 """ 

228 dimension_set = set() 

229 if isinstance(connection, connectionTypes.DimensionedConnection): 

230 dimension_set = set(connection.dimensions) 

231 skypix_dim = [] 

232 if "skypix" in dimension_set: 

233 dimension_set.remove("skypix") 

234 skypix_dim = ["skypix"] 

235 dimension_graph = universe.extract(dimension_set) 

236 return list(dimension_graph.names) + skypix_dim 

237 

238 # open a file if needed 

239 close = False 

240 if not hasattr(file, "write"): 

241 file = open(file, "w") 

242 close = True 

243 

244 print("digraph Pipeline {", file=file) 

245 

246 allDatasets: set[str | tuple[str, str]] = set() 

247 if isinstance(pipeline, Pipeline): 

248 pipeline = pipeline.toExpandedPipeline() 

249 

250 # The next two lines are a workaround until DM-29658 at which time metadata 

251 # connections should start working with the above code 

252 labelToTaskName = {} 

253 metadataNodesToLink = set() 

254 

255 for idx, taskDef in enumerate(sorted(pipeline, key=lambda x: x.label)): 

256 # node for a task 

257 taskNodeName = "task{}".format(idx) 

258 

259 # next line is workaround until DM-29658 

260 labelToTaskName[taskDef.label] = taskNodeName 

261 

262 _renderTaskNode(taskNodeName, taskDef, file, None) 

263 

264 metadataRePattern = re.compile("^(.*)_metadata$") 

265 for attr in sorted(iterConnections(taskDef.connections, "inputs"), key=lambda x: x.name): 

266 if attr.name not in allDatasets: 

267 dimensions = expand_dimensions(attr) 

268 _renderDSTypeNode(attr.name, dimensions, file) 

269 allDatasets.add(attr.name) 

270 nodeName, component = DatasetType.splitDatasetTypeName(attr.name) 

271 _renderEdge(attr.name, taskNodeName, file) 

272 # connect component dataset types to the composite type that 

273 # produced it 

274 if component is not None and (nodeName, attr.name) not in allDatasets: 

275 _renderEdge(nodeName, attr.name, file) 

276 allDatasets.add((nodeName, attr.name)) 

277 if nodeName not in allDatasets: 

278 dimensions = expand_dimensions(attr) 

279 _renderDSTypeNode(nodeName, dimensions, file) 

280 # The next if block is a workaround until DM-29658 at which time 

281 # metadata connections should start working with the above code 

282 if (match := metadataRePattern.match(attr.name)) is not None: 

283 matchTaskLabel = match.group(1) 

284 metadataNodesToLink.add((matchTaskLabel, attr.name)) 

285 

286 for attr in sorted(iterConnections(taskDef.connections, "prerequisiteInputs"), key=lambda x: x.name): 

287 if attr.name not in allDatasets: 

288 dimensions = expand_dimensions(attr) 

289 _renderDSTypeNode(attr.name, dimensions, file) 

290 allDatasets.add(attr.name) 

291 # use dashed line for prerequisite edges to distinguish them 

292 _renderEdge(attr.name, taskNodeName, file, style="dashed") 

293 

294 for attr in sorted(iterConnections(taskDef.connections, "outputs"), key=lambda x: x.name): 

295 if attr.name not in allDatasets: 

296 dimensions = expand_dimensions(attr) 

297 _renderDSTypeNode(attr.name, dimensions, file) 

298 allDatasets.add(attr.name) 

299 _renderEdge(taskNodeName, attr.name, file) 

300 

301 # This for loop is a workaround until DM-29658 at which time metadata 

302 # connections should start working with the above code 

303 for matchLabel, dsTypeName in metadataNodesToLink: 

304 # only render an edge to metadata if the label is part of the current 

305 # graph 

306 if (result := labelToTaskName.get(matchLabel)) is not None: 

307 _renderEdge(result, dsTypeName, file) 

308 

309 print("}", file=file) 

310 if close: 

311 file.close()