Add chart series support for point shapes and sizes

[mofojed]

• Add point shape and size to series info, so markers can be plotted with the correct shapes and sizes
• Tested with Web UI branch https://github.com/mofojed/web-client-ui/tree/figure-point-shapes
• Used the following code in Python to test shapes/sizes, and plot-by series:

from deephaven import empty_table
from deephaven.plot import Figure
import math

math_funcs = ["sin", "cos"]
shapes = ["SQUARE", "CIRCLE", "UP_TRIANGLE", "DOWN_TRIANGLE", "RIGHT_TRIANGLE", "LEFT_TRIANGLE", "DIAMOND"]
size = 50
# Unsupported shapes: ["ELLIPSE", "HORIZONTAL_RECTANGLE", "VERTICAL_RECTANGLE"]

# Create a generic table that has enough columns to display all the shapes
# Re-uses some of the funcs
t = empty_table(size).update([f"name=`` + shapes[(int)Math.floor(i / Math.ceil(size / {len(shapes)}))]", "x=i*0.1"])
for i in range(len(shapes)):
    t = t.update([f"y{i}=Math.{math_funcs[i % len(math_funcs)]}(x+{math.floor(i / len(math_funcs))})"])

# Create a figure showing all the different shapes one for each series
p = Figure()
for i in range(len(shapes)):
    p = p.plot_xy(series_name=shapes[i], t=t, x="x", y=f"y{i}").point(shape=shapes[i], size=len(shapes) - i, visible=True)
p = p.show();

# Create a plot by figure
p2 = Figure().plot_xy(series_name="Multi", t=t, by=["name"], x="x", y="y1").line(visible=True)
for i in range(len(shapes)):
    p2 = p2.point(shape=shapes[i], multi_series_key=[shapes[i]], size=len(shapes) - i, visible=True)
p2 = p2.show()

• Used the following code in Groovy to test with plotBy as well:

names = ["A", "B", "C", "D", "E"]
mathFuncs = ["sin", "cos"]
shapes = ["SQUARE", "CIRCLE", "UP_TRIANGLE", "DOWN_TRIANGLE", "RIGHT_TRIANGLE", "LEFT_TRIANGLE", "DIAMOND"]
// Unsupported shapes: ["ELLIPSE", "HORIZONTAL_RECTANGLE", "VERTICAL_RECTANGLE"]

// Create a generic table that has enough columns to display all the shapes
// Re-uses some of the funcs
t = emptyTable(50).update("name=shapes[i % 7]", "x=i")
for (i = 0; i < shapes.size; i += 1) {
    t = t.update("y" + i + "=Math." + mathFuncs[i % mathFuncs.size] + "(x+" + Math.floor(i / mathFuncs.size) + ")")
}

// # Generate the table and figure based on the shapes created
p = figure()
for (i = 0; i < shapes.size; i += 1) {
    p = p.plot(shapes[i], t, "x", "y" + i).pointShape(shapes[i]).pointSize(shapes.size - i).pointsVisible(true)
}
p = p.show()

p2 = plotBy("plotBy", t, "x", "y1", ["name"] as String[]).pointsVisible(true)
for (i = 0; i < shapes.size; i += 1) {
  p2 = p2.pointShape(shapes[i], [shapes[i]] as String[]).linesVisible(true)
}
p2 = p2.show()

[rcaudy]

What about MultiSeries?

[mofojed]

MultiSeries does set the point shapes (line 465, line 502) but I'm not sure how to test it. I think I need plot_by functionality which hasn't been plumbed through yet. Or do you know of a way to trigger it?

[rcaudy]

We can merge like this, but we should figure out about multi-series before merging.

[rcaudy]

MultiSeries does set the point shapes (line 465, line 502) but I'm not sure how to test it. I think I need plot_by functionality which hasn't been plumbed through yet. Or do you know of a way to trigger it?

It looks to me like plotBy, catPlotBy, and ohlcPlotBy will make a MultiSeries.

[mofojed]

plot_by doesn't seem to be available in python, so I converted the code to Groovy and add a plotBy to get a multi-series. It looks like the point shapes are appearing as expected (see the screenshot and code snippet in my original comment).

[mofojed]

@niloc132 pointed out there is actually a plot-by-key in python: https://deephaven.io/core/docs/reference/cheat-sheets/cheat-sheet/#plot-by-some-key

I tried to map the same groovy snippet above to using plotBy in Python, but server is throwing an NPE:

from deephaven import empty_table
from deephaven.plot import Figure
import math

math_funcs = ["sin", "cos"]
shapes = ["SQUARE", "CIRCLE", "UP_TRIANGLE", "DOWN_TRIANGLE", "RIGHT_TRIANGLE", "LEFT_TRIANGLE", "DIAMOND"]
# Unsupported shapes: ["ELLIPSE", "HORIZONTAL_RECTANGLE", "VERTICAL_RECTANGLE"]

# Create a generic table that has enough columns to display all the shapes
# Re-uses some of the funcs
t = empty_table(50).update([f"name=shapes[i % 5]", "x=i*0.1"])
for i in range(len(shapes)):
    t = t.update([f"y{i}=Math.{math_funcs[i % len(math_funcs)]}(x+{math.floor(i / len(math_funcs))})"])

# Generate the table and figure based on the shapes created
p = Figure()
for i in range(len(shapes)):
    p = p.plot_xy(series_name=shapes[i], t=t, x="x", y=f"y{i}").point(shape=shapes[i], size=len(shapes) - i, visible=True)

p = p.show();

p2 = Figure().plot_xy(series_name="Multi", t=t, x="x", y="y1", by=["name"]).point(visible=True)
for i in range(len(shapes)):
    p2 = p2.point(shape=shapes[i], multi_series_key=[shapes[i]])

p2 = p2.show()

Throws:

heduler-Concurrent-1 | i.d.s.s.SessionState      | Internal Error '2d7baeac-6803-4c68-bef1-e1b31c0be75d' java.lang.NullPointerException: null
        at io.deephaven.extensions.barrage.util.ExportUtil.buildTableCreationResponse(ExportUtil.java:20)
        at io.deephaven.extensions.barrage.util.ExportUtil.buildTableCreationResponse(ExportUtil.java:13)
        at io.deephaven.server.partitionedtable.PartitionedTableServiceGrpcImpl.lambda$getTable$8(PartitionedTableServiceGrpcImpl.java:166)
        at io.deephaven.server.session.SessionState$ExportObject.doExport(SessionState.java:886)
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:515)
        at java.util.concurrent.FutureTask.run(FutureTask.java:264)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:304)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1128)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:628)
        at io.deephaven.server.runner.DeephavenApiServerModule$ThreadFactory.lambda$newThread$0(DeephavenApiServerModule.java:165)
        at java.lang.Thread.run(Thread.java:829)

Not sure what I'm doing wrong there, but I tried a different example in python and it worked:

from deephaven.plot import Figure

table_3_keys = static_source_1.head(2000).where("String1 in `A`, `B`, `C`")\
    .sort("Int1").update(["New_Row = i", "Y = Int1 * (String1 = `A` ? 2 : String1 = `B` ? 0.5 : 1.0)"])
p = Figure().plot_xy(series_name="partitioned", t=table_3_keys, by=["String1"], x="New_Row", y="Y").point(shape="SQUARE", visible=True, multi_series_key=["A"]).point(shape="DIAMOND", visible=True, multi_series_key=["B"]).point(shape="UP_TRIANGLE", visible=True, multi_series_key=["C"]).show()

[mofojed]

Figured out what the issue was in my snippet above. The problem was on this line:

t = empty_table(50).update([f"name=shapes[i % 5]", "x=i*0.1"])

This was creating a table where name was of type PyObject. I needed to change it to a native String, I did it with string concatenation:

t = empty_table(50).update([f"name=`` + shapes[i % 5]", "x=i"])

Not sure if there's anything we can do about that, pretty annoying that it can't handle the first syntax correctly.

[deephaven-internal]

Labels indicate documentation is required. Issues for documentation have been opened:

How-to: https://github.com/deephaven/deephaven.io/issues/2112
Conceptual: https://github.com/deephaven/deephaven.io/issues/2114
Reference: https://github.com/deephaven/deephaven.io/issues/2113
Blog: https://github.com/deephaven/deephaven.io/issues/2111