Add support for custom tabulator options

[droumis]

Fixes #125

Use tabulator_kwargs and refer to the options https://panel.holoviz.org/reference/widgets/Tabulator.html

TODO:

• allow for show_indices to be overwritten (another PR)

For instance:

AnnotatorTable(annotator, tabulator_kwargs=dict(pagination='local', page_size=7))

Code

import panel as pn
import holoviews as hv
hv.extension('bokeh')
import pandas as pd
from holonote.annotate import Annotator
from holonote.annotate.connector import SQLiteDB
from holonote.app import PanelWidgets, AnnotatorTable

data = {
    'description': ['T0', 'T1', 'T2'] * 10,
    'start': range(30),
    'end': [i + .8 for i in range(30)]
}

annotations_df = pd.DataFrame(data)

curve_data = {
    'time': range(30),
    'val': [1, 3, 2] * 10,
}

curve_df = pd.DataFrame(curve_data)

plot = hv.Curve(curve_df, 'time', 'val')

annotator = Annotator({'time': float}, fields=["description"],
                      connector=SQLiteDB(filename=':memory:')
                     )

annotator.define_annotations(annotations_df, time=("start", "end"))

annotator.groupby = "description"
annotator.visible = ["T1"]

annotator_widgets = pn.Column(
    PanelWidgets(annotator),
    AnnotatorTable(
        annotator,
        tabulator_kwargs=dict(
            pagination='local',
            page_size=7,
        )
    ),
    width=400
)

# Display the widgets and plot
pn.Row(annotator_widgets, annotator * plot.opts(responsive=True)).servable()

[codspeed-hq]

CodSpeed Performance Report

Merging #126 will not alter performance

_{Comparing tabulator_kwargs (ffe1ce7) with main (ceb60de)}

Summary

✅ 6 untouched benchmarks

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.34%. Comparing base (ceb60de) to head (ffe1ce7).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #126   +/-   ##
=======================================
  Coverage   86.33%   86.34%           
=======================================
  Files          26       26           
  Lines        2774     2775    +1     
=======================================
+ Hits         2395     2396    +1     
  Misses        379      379           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ahuang11]

If it's a param should this be wrapped in a function decorated with pn.depends('tabulator_kwargs'), running self.tabulator.param.update(**tabulator_kwargs)?

[droumis]

I'm guessing not, since tabulator_kwargs (and _create_tabulator) are intended to only be used once. But it's a good question for someone with a greater understanding to jump in here

[philippjfr]

I agree that the tabulator_kwargs should be synced with the Tabulator. There actually is an API on Reactive objects to do this, specifically, could you try:

Suggested change

	**self.tabulator_kwargs,
	refs=self.param.tabulator_kwargs

This will update keep the tabulator_kwargs dictionary synced with the Tabulator parameters.

[droumis]

So my understanding (which I just pushed changes for) is that it is appropriate to make tabulator_kwargs into a param (I'm guessing now requiring allow_refs=True?), and that it will be synced solely by using the refs= API, without the dependence on the depends decorator.

Just curious, what is the reasoning that tabulator_kwargs should remain synced with Tabulator beyond initialization? Is it just the general design inclination to allow for future flexibility in case some property about the Tabulator will be internally modified?

Also, am I correct in assuming the allow_refs=True is now required or am I confusing concepts?

Thanks for the guidance!

[droumis]

also, if my recent changes look good, please approve and merge! thank you

[droumis]

Note to self, I was thinking about adding input validation warnings for dict items like this:

class _TabulatorOpts(param.Dict):
    def __init__(self, default={}, doc="kwargs to pass to tabulator", allow_refs=True):
        super().__init__(default=default, doc=doc, allow_refs=allow_refs)

    def _validate(self, val) -> None:
        super()._validate(val)
        warn_opts = ("show_index", "selectable", "buttons")
        for k in val:
            if k in warn_opts:
                warn(f"`{k}` is set directly and should not be passed through `tabulator_opts`")

But then the doc entry doesn't appear in the docstring, so I think I'll leave it out and just let the validation be a side effect of passing in a duplicate kwarg of one that is set directly

[ahuang11]

You could create your own param type with validation, e.g.

tabulator_kwargs = param.TabulatorKwargs()

As an example, Panel does it for a few like Margin.
https://github.com/holoviz/panel/blob/9f412a0cb6a3093c0a0f34489ae24db85bea60c6/panel/_param.py#L51

[droumis]

I think that's pretty much what I proposed in the comment above, no? I think I'll avoid for now because I prefer the docs to show up, but thanks. How does this PR look currently?

[ahuang11]

I think the PR looks fine; I'm just not sure if allow_refs=True is necessary or not since I don't have any experience with it so I was hoping for Philipp to review :P Are you able to confirm it doesn't work when allow_refs=False?

I think that's pretty much what I proposed in the comment above,

Oops yeah, I didn't recognize it since I'm used to seeing inheriting from param.Parameter

the docs to show up

I'm a bit confused about this one.

from param import Parameter, Parameterized


class Aspect(Parameter):
    """
    A Parameter type to validate aspect ratios. Supports numeric values
    and auto.
    """

    def __init__(self, default=None, allow_None=True, **params):
        super().__init__(default=default, allow_None=allow_None, **params)
        self._validate(default)

    def _validate(self, val):
        self._validate_value(val, self.allow_None)

    def _validate_value(self, val, allow_None):
        ...


class Test(Parameterized):

    tabulator_aspect = Aspect(doc="Just testing")


Test?

results in

Init signature: Test(*, tabulator_aspect, name)
Docstring:     
Parameters of 'Test'
====================

Parameters changed from their default values are marked in red.
Soft bound values are marked in cyan.
C/V= Constant/Variable, RO/RW = ReadOnly/ReadWrite, AN=Allow None

Name              Value   Type     Mode  

tabulator_aspect   None  Aspect  V RW AN 

Parameter docstrings:
=====================

tabulator_aspect: Just testing
Type:           ParameterizedMetaclass
Subclasses:     

[philippjfr]

There's no reason not to enable allow_refs but it isn't strictly needed. Tabulator allows refs so that's all that's necessary, enabling it here would make it possible to bind references to these objects which is also nice.

[philippjfr]

Could do something like this to validate the values:

@param.depends('tabulator_kwargs')
def _validated_tabulator_kwargs(self):
    ... 
    return self.tabulator_kwargs

[philippjfr]

I'm happy with this as is but if you want to validate the kwargs I'd be happy with that too.