Refresh pipeline directed graph colors

The current pipeline directed graph nodes use dark gray as their
background color. It has been reported that it is difficult to read the
black text on the dark gray background. This ticket updates the directed
graph color scheme with an aim towards making node text easier to read.
In the process of exploring what color schemes would be optimal to
satisfy the aim of this ticket, it emerged that making use of the Rubin
visual identity colors may be desirable. This will help to make LSST
pipeline graphs more instantly recognizable as Rubin-associated
products. Colors: https://rubin.canto.com/g/RubinVisualIdentity

doc/changes/DM-39294.misc.md

@@ -0,0 +1,6 @@
+The current pipeline directed graph nodes use dark gray as their background color.
+It has been reported that it is difficult to read the black text on the dark gray background.
+This ticket updates the directed graph color scheme with an aim towards making node text easier to read.
+In the process of exploring what color schemes would be optimal to satisfy the aim of this ticket, it emerged that making use of the Rubin visual identity colors may be desirable.
+This will help to make LSST pipeline graphs more instantly recognizable as Rubin-associated products.
+Colors: https://rubin.canto.com/g/RubinVisualIdentity

python/lsst/ctrl/mpexec/dotTools.py

@@ -30,6 +30,7 @@
 # -------------------------------
 #  Imports of standard modules --
 # -------------------------------
+import html
 import io
 import re
 from collections.abc import Iterable
@@ -49,32 +50,52 @@
 #  Local non-exported definitions --
 # ----------------------------------
 
-# Node styles indexed by node type.
-_STYLES = dict(
-    task=dict(shape="box", style="filled,bold", fillcolor="gray70"),
-    quantum=dict(shape="box", style="filled,bold", fillcolor="gray70"),
-    dsType=dict(shape="box", style="rounded,filled", fillcolor="gray90"),
-    dataset=dict(shape="box", style="rounded,filled", fillcolor="gray90"),
+# Attributes applied to directed graph objects.
+_NODELABELPOINTSIZE = 18
+_ATTRIBS = dict(
+    defaultGraph=dict(splines="ortho", nodesep="0.5", ranksep="0.75", pad="0.5"),
+    defaultNode=dict(shape="box", fontname="Monospace", fontsize="14", margin="0.2,0.1", penwidth="3"),
+    defaultEdge=dict(color="black", arrowsize="1.5", penwidth="1.5"),
+    task=dict(style="filled", color="black", fillcolor="#B1F2EF"),
+    quantum=dict(style="filled", color="black", fillcolor="#B1F2EF"),
+    dsType=dict(style="rounded,filled,bold", color="#00BABC", fillcolor="#F5F5F5"),
+    dataset=dict(style="rounded,filled,bold", color="#00BABC", fillcolor="#F5F5F5"),
 )
 
 
+def _renderDefault(type: str, attribs: dict[str, str], file: io.TextIOBase) -> None:
+    """Set default attributes for a given type."""
+    default_attribs = ", ".join([f'{key}="{val}"' for key, val in attribs.items()])
+    print(f"{type} [{default_attribs}];", file=file)
+
+
 def _renderNode(file: io.TextIOBase, nodeName: str, style: str, labels: list[str]) -> None:
     """Render GV node"""
-    label = r"\n".join(labels)
-    attrib_dict = dict(_STYLES[style], label=label)
-    attrib = ", ".join([f'{key}="{val}"' for key, val in attrib_dict.items()])
+    label = r"</TD></TR><TR><TD>".join(labels)
+    attrib_dict = dict(_ATTRIBS[style], label=label)
+    pre = '<<TABLE BORDER="0" CELLPADDING="5"><TR><TD>'
+    post = "</TD></TR></TABLE>>"
+    attrib = ", ".join(
+        [
+            f'{key}="{html.escape(val)}"' if key != "label" else f"{key}={pre}{val}{post}"
+            for key, val in attrib_dict.items()
+        ]
+    )
     print(f'"{nodeName}" [{attrib}];', file=file)
 
 
 def _renderTaskNode(nodeName: str, taskDef: TaskDef, file: io.TextIOBase, idx: Any = None) -> None:
     """Render GV node for a task"""
-    labels = [taskDef.label, taskDef.taskName]
+    labels = [
+        f'<B><FONT POINT-SIZE="{_NODELABELPOINTSIZE}">' + taskDef.label + "</FONT></B>",
+        taskDef.taskName,
+    ]
     if idx is not None:
         labels.append(f"index: {idx}")
     if taskDef.connections:
         # don't print collection of str directly to avoid visually noisy quotes
         dimensions_str = ", ".join(sorted(taskDef.connections.dimensions))
-        labels.append(f"dimensions: {dimensions_str}")
+        labels.append(f"<I>dimensions:</I>&nbsp;{dimensions_str}")
     _renderNode(file, nodeName, "task", labels)
 
 
@@ -91,9 +112,9 @@ def _renderQuantumNode(
 
 def _renderDSTypeNode(name: str, dimensions: list[str], file: io.TextIOBase) -> None:
     """Render GV node for a dataset type"""
-    labels = [name]
+    labels = [f'<B><FONT POINT-SIZE="{_NODELABELPOINTSIZE}">' + name + "</FONT></B>"]
     if dimensions:
-        labels.append("Dimensions: " + ", ".join(sorted(dimensions)))
+        labels.append("<I>dimensions:</I>&nbsp;" + ", ".join(sorted(dimensions)))
     _renderNode(file, name, "dsType", labels)
 
 
@@ -165,6 +186,9 @@ def graph2dot(qgraph: QuantumGraph, file: Any) -> None:
         close = True
 
     print("digraph QuantumGraph {", file=file)
+    _renderDefault("graph", _ATTRIBS["defaultGraph"], file)
+    _renderDefault("node", _ATTRIBS["defaultNode"], file)
+    _renderDefault("edge", _ATTRIBS["defaultEdge"], file)
 
     allDatasetRefs: dict[str, str] = {}
     for taskId, taskDef in enumerate(qgraph.taskGraph):
@@ -242,6 +266,9 @@ def expand_dimensions(connection: connectionTypes.BaseConnection) -> list[str]:
         close = True
 
     print("digraph Pipeline {", file=file)
+    _renderDefault("graph", _ATTRIBS["defaultGraph"], file)
+    _renderDefault("node", _ATTRIBS["defaultNode"], file)
+    _renderDefault("edge", _ATTRIBS["defaultEdge"], file)
 
     allDatasets: set[str | tuple[str, str]] = set()
     if isinstance(pipeline, Pipeline):

tests/test_dotTools.py

@@ -129,11 +129,12 @@ def testPipeline2dot(self):
         # It's hard to validate complete output, just checking few basic
         # things, even that is not terribly stable.
         lines = file.getvalue().strip().split("\n")
+        nglobals = 3
         ndatasets = 10
         ntasks = 6
         nedges = 16
         nextra = 2  # graph header and closing
-        self.assertEqual(len(lines), ndatasets + ntasks + nedges + nextra)
+        self.assertEqual(len(lines), nglobals + ndatasets + ntasks + nedges + nextra)
 
         # make sure that all node names are quoted
         nodeRe = re.compile(r"^([^ ]+) \[.+\];$")
@@ -142,8 +143,9 @@ def testPipeline2dot(self):
             match = nodeRe.match(line)
             if match:
                 node = match.group(1)
-                self.assertEqual(node[0] + node[-1], '""')
-                continue
+                if node not in ["graph", "node", "edge"]:
+                    self.assertEqual(node[0] + node[-1], '""')
+                    continue
             match = edgeRe.match(line)
             if match:
                 for group in (1, 2):
@@ -152,7 +154,7 @@ def testPipeline2dot(self):
                 continue
 
         # make sure components are connected appropriately
-        self.assertIn('"D" -> "D.C";', file.getvalue())
+        self.assertIn('"D" -> "D.C"', file.getvalue())
 
         # make sure there is a connection created for metadata if someone
         # tries to read it in