Forum Discussion

davidleal's avatar
davidleal
Frequent Visitor
3 years ago
Solved

Replace duplicated values with null in the same column per a given list of columns

I found several solutions for a similar problem, but it creates an additional column or does find and replace value which is not what I need. Let's say I have a Name column with the following values:

name1
name1
name2
name2
name1

and I would like to have the following output (the same previous column but with some values transformed):

name1
null
name2
null
null

In this question: Testing a query function to replace duplicated values with null, but I am getting errors  wdx223_Daniel provided the following solution that returns a list with the expected output:

let
Source = Excel.CurrentWorkbook(){[Name="TB_source"]}[Content],
#"Changed Type" = Table.TransformColumnTypes(Source,{{"Name", type text}, {"Amount", Int64.Type}}),
RemoveDups = (lst as list)=>List.Accumulate(lst,{},(x,y)=> x&{ if List.Contains(x,y) then null else y}),
replaceValues = RemoveDups(Table.Column(#"Changed Type", "Name"))
in
replaceValues

but I was not able to assign this list to the existing column Name. The question: Find and Replace text string in one table using value from another table does a find and replace, I was trying to customize it, but it doesn't work because for every value in the Find will be replaced wil null. Because it does a search (not by index position).

What I would need is to loop for all values of the a given column and replace with the corresponding value at the same index position of #"uniqueList". In other languages it would be a simple assignment: a=b, but I was not able to find a way in Power Query.

I was trying the following for example:

Table.ReplaceValue(#"Changed Type", each [Name], each replaceValues,Replacer.ReplaceValue,{"Name"})

but it return a the list of value on each row.

 

For the general case I would like to do the following transformation in #"Changed Type": For a list of selected columns replace  per column duplicated values with null

 

Thanks

  • Hi davidleal ,
    utilizing on what you have already, I would suggest the following approach:

    let
        Source = Table.FromRows(
            Json.Document(
                Binary.Decompress(
                    Binary.FromText(
                        "i45W8kvMTTVU0lFyziypBNGGBkqxOpjCRghhI6gwiDbGLmyCKWwMpE2BwrEA",
                        BinaryEncoding.Base64
                    ),
                    Compression.Deflate
                )
            ),
            let
                _t = ((type nullable text) meta [Serialized.Text = true])
            in
                type table [Name = _t, City = _t, Amount = _t]
        ),
        #"Changed Type" = Table.TransformColumnTypes(
            Source,
            {{"Name", type text}, {"City", type text}, {"Amount", Int64.Type}}
        ),
        ListOfColumns = {"Name", "City"},
        RemainingColumns = List.Difference(Table.ColumnNames(#"Changed Type"), ListOfColumns),
        RemoveDups = (lst as list) =>
            List.Accumulate(lst, {}, (x, y) => x & {if List.Contains(x, y) then null else y}),
        replaceValues = List.Transform(ListOfColumns, each RemoveDups(Table.Column(#"Changed Type", _))),
        Custom1 = Table.FromColumns(
            replaceValues & Table.ToColumns(Table.SelectColumns(#"Changed Type", RemainingColumns)),
            ListOfColumns & RemainingColumns
        )
    in
        Custom1



3 Replies

  • ImkeF's avatar
    ImkeF
    Icon for Community Champion rankCommunity Champion

    Hi davidleal ,
    utilizing on what you have already, I would suggest the following approach:

    let
        Source = Table.FromRows(
            Json.Document(
                Binary.Decompress(
                    Binary.FromText(
                        "i45W8kvMTTVU0lFyziypBNGGBkqxOpjCRghhI6gwiDbGLmyCKWwMpE2BwrEA",
                        BinaryEncoding.Base64
                    ),
                    Compression.Deflate
                )
            ),
            let
                _t = ((type nullable text) meta [Serialized.Text = true])
            in
                type table [Name = _t, City = _t, Amount = _t]
        ),
        #"Changed Type" = Table.TransformColumnTypes(
            Source,
            {{"Name", type text}, {"City", type text}, {"Amount", Int64.Type}}
        ),
        ListOfColumns = {"Name", "City"},
        RemainingColumns = List.Difference(Table.ColumnNames(#"Changed Type"), ListOfColumns),
        RemoveDups = (lst as list) =>
            List.Accumulate(lst, {}, (x, y) => x & {if List.Contains(x, y) then null else y}),
        replaceValues = List.Transform(ListOfColumns, each RemoveDups(Table.Column(#"Changed Type", _))),
        Custom1 = Table.FromColumns(
            replaceValues & Table.ToColumns(Table.SelectColumns(#"Changed Type", RemainingColumns)),
            ListOfColumns & RemainingColumns
        )
    in
        Custom1



  • ImkeF's avatar
    ImkeF
    Icon for Community Champion rankCommunity Champion

    Hi davidleal ,

    not with the native Table.ReplaceValue or Table.TransformColumns- functions, as they don't support recursion.

  • davidleal's avatar
    davidleal
    Frequent Visitor

    Thanks, it works, so it looks like that the only we to do it is to recreate a new table again, but not replacing the existing column values with the new value list.