How can I have set the type of some columns when opening a CSV file?

[halloleo]

I have a CSV file with lots of columns, a few of them (monthly_price,lease_duration and total) I always want to be of type int or float when I open this sheet.

I know about the SheetsSheet.knownCols attribute. But with it I set the type for all sheets I open in future. Can Ito specify column types just for this sheet?

Is there a way to specify column types on the command line? ( something like like vd my_table.csv --column-type=monthly_price,float)

Or can I specify colum types in a file-specific visidatarc file, maybe in the same directory as the CSV file? (e.g. something like my_table.visidatarc`)

I am curious what's possible in VisiData!

[saulpw]

I think knownCols is your best bet. Do you really view different sheets with the same column names but different types?

[midichef]

How about setting the sheet's knownCols in an afterLoad() function?

@Sheet.before
def afterLoad(sheet):
#    override the Sheet.knownCols with an empty one, for this specific sheet object
#    to avoid changing the type of the Date column in every sheet
     sheet.knownCols = DefaultAttrDict()
     if sheet.name == 'benchmark':
         sheet.knownCols.Date.type = date
     elif sheet.name == 'benchmark2':
         sheet.knownCols.Date.width = 0  #hide the column

[halloleo]

@midichef Cool idea to set knownCols only for this specific file! Where would your snippet go? .visidatarc?

[halloleo]

@saulpw You have a point that this columns are probably not used for other types in other sheets... However, I'm uncomfortable to set these for all sheets: I am worried in a year's time I load another sheet with the same column name, and even if the type is fittingI will ask myself: Why does it has auto-set the type? I then will wonder: Is this a new feature of the newly auto-updated homebrew version of vd? And I will start trawling once more through the doco before I realise: Ah, a year ago I did this fancy thing for this column name...

[midichef]

Yes, the snippet goes in .visidatarc. You may want to change @Sheet.before to @CsvSheet.before to limit it to only run for CsvSheet, not for all sheets.

@midichef Cool idea to set knownCols only for this specific file! Where would your snippet go? .visidatarc?

Based on your earlier use case comments, some things you may want to try inside the afterLoad(): if you want to examine the sheet data file path, look at sheet.source.given and parse it to look for an absolute paths. If there isn't an absolute path, i.e. if there's no '/' in the given, then maybe you'd want to use os.getcwd() to change behavior based on the directory visidata was started in.

[midichef]

Another option is to create a custom file type that assigns types to its columns:

## set columns by using -f with a custom sheet and custom file type:
#    vd -f csv_price datacsv
class CsvPriceSheet(CsvSheet):
    pass
@VisiData.api
def open_csv_price(vd, p):
    return CsvPriceSheet(p.base_stem, source=p)
# the sheet has to get its own empty .knownCols, otherwise all sheets will use the new column types
CsvPriceSheet.knownCols = DefaultAttrDict()
CsvPriceSheet.knownCols.monthly_price.type = float