Is there a way to override the histogram function?

[frosencrantz]

How does one override the histogram function for the FreqSheet?

visidata/visidata/freqtbl.py

Lines 55 to 58 in 25a412a

	def histogram(col, row):
	histogram = col.sheet.options.disp_histogram
	histolen = col.sheet.options.disp_histolen
	return histogram*(histolen*len(row.sourcerows)//col.sheet.largest)

I was wondering how hard it would be to change the function to use the Unicode block characterswe to get a smoother graph.

[andycraig]

I was able to override it by creating a plugin with the following code. When a histogram is created it adds the status message 'hello from histogram'. The code is a copy-pasted visidata.freqtbl.FreqTableSheet.reload(), although I did have to change the super().initCols() line to explicitly reference the parent class PivotSheet.

from visidata import AttrColumn, asyncthread, Column, status, vlen, vd
from visidata.freqtbl import FreqTableSheet
from visidata.pivot import PivotSheet


@asyncthread
def reloadPatch(self):
    'Generate frequency table then reverse-sort by length.'
    PivotSheet.initCols(self) # changed from super().initCols()

    # add default bonus columns
    for c in [
        AttrColumn('count', 'sourcerows', type=vlen),
        Column('percent', type=float, getter=lambda col,row: len(row.sourcerows)*100/col.sheet.source.nRows),
    ]:
        self.addColumn(c)

    if self.options.disp_histolen and self.options.disp_histogram:
        def histogram(col, row):
            vd.status("hello from histogram")
            histogram = col.sheet.options.disp_histogram
            histolen = col.sheet.options.disp_histolen
            return histogram*(histolen*len(row.sourcerows)//col.sheet.largest)

        c = Column('histogram', type=str, getter=histogram, width=self.options.disp_histolen+2)
        self.addColumn(c)

    # two more threads
    vd.sync(self.addAggregateCols(),
            self.groupRows(self.updateLargest))

    if self.nCols > len(self.groupByCols)+3:  # hide percent/histogram if aggregations added
        self.column('percent').hide()
        self.column('histogram').hide()

    if not [c for c in self.groupByCols if vd.isNumeric(c)]:
        self.orderBy(self.column('count'), reverse=True)

FreqTableSheet.reload = reloadPatch

[frosencrantz]

@andycraig I hadn't read your post when I replied other than to see you had a solution and there was code. When I looked at what you did I realized this is what I was trying to avoid by copying the rest of visidata.freqtbl.FreqTableSheet.reload(). Since Saul or Anja are likely to make changes that cause that code to be out-of-date.

It doesn't look like histogram needs to be a sub-function in that method. Is there a way to override that function without copying the rest of the surrounding method . . If not I think perhaps VisiData could change so histogram is easier to override.

[andycraig]

@frosencrantz Yeah, it works but it's fragile.

It looks like it's possible although somewhat complicated to replace a nested function: https://stackoverflow.com/a/27550237/7785832 But I agree that it seems like histogram doesn't need to be a nested function so if it could be extracted outside of reload it feels like that would be more elegant.

[frosencrantz]

@andycraig would you be interested in fixing the code? If you proposed a fix, I feel Saul and Anja would consider including it.

[andycraig]

I'd be happy to do that. I've created an issue to propose it: #1621

[frosencrantz]

Thanks @andycraig. I was thinking of a Pull Request, in case you want to directly contribute to VisiData.

Edit: Saul said that would be useful. Thanks @andycraig! Looking forward to your PR.