Repository navigation
help migrating to CSV v1.0 #1207
Description
Hello,
If i understand correctly, CSV v1 does not accept functions for the drop argument anymore, in favor of Tables.Scan. one problem with this is that Tables.Scan only works on the actual names of the columns, whereas the function i am using performs a transformation on the column name before the selection procedure.
The function is the following:
function _drop(i,name)
norm_header = normalisestring(name)
normalisestring(norm_header; tofilter=r"[ \-\_\d]") ∈ ignorelist #ignorelist is an user-supplied list
end
function normalisestring(str; tofilter = ' ')
res = Base.Unicode.normalize(str,casefold=true,stripmark=true)
return replace(res, tofilter => "")
endwhat can be done to migrate this workflow to CSV v1?
Activity
quinnj commentedon Sep 23, 2026
[PR link description generated by AI]
Thanks for the report. #1210 brings back function-valued
selectanddrop, and releases them as CSV 1.2.0. After that release, your 0.10 code works again:As in 0.10, the function receives the column position and the column name as a
Symbol. CallString(name)beforeUnicode.normalize, becausenormalizedoes not accept aSymbol.Until 1.2.0 is registered, you can compute the mask from the header names first on CSV 1.0 or 1.1:
CSV.File(lf)reuses the index fromCSV.lazy, so this costs about the same as one read. ForCSV.RowsorCSV.Chunks, get the names withTables.columnnames(CSV.File(path; limit=0))and pass the samedrop=mask.