Forum Discussion

RobRayborn's avatar
RobRayborn
Helper IV
2 years ago
Solved

Cleaning Text field columns from incoming SQL database

I'm trying to pull in SQL data in to a Dataflow from our server.  This incoming data has Date/Time, Number, and Text data columns.  The problem is the SQL Text columns have always have have {Space} c...
Need Help
Show and Tell
Tutorial Requests
  • m_dekorte's avatar
    m_dekorte
    2 years ago

    Hi RobRayborn,

     

    Let's try something else...

    Create a new blank query and replace everything inside with the following code:

    let
  TrimAllTxtCols = (t as table, optional n as number) as table =>
    Table.TransformColumns( t, 
      List.Transform(
        List.Select( Table.ColumnNames(t), 
          each if n = null 
            then List.MatchesAll(
              Table.Column(t, _), 
              (x) => Value.Is(x, type nullable text)
            )
            else List.MatchesAll(
              List.FirstN(List.RemoveNulls(Table.Column(t, _)), n), 
              (x) => Value.Is(x, type text)
            )
        ), each {_, Text.Trim, type text}
      )
    )
in
    TrimAllTxtCols

    This returns a function value, you can give this query a more suitable name if you like.

    Now for "t" pass it query with your incoming SQL data
    When you have many rows in your table, I'd encourage you to also pass a number for "n" that tells the function how many rows to check, to see if your column contains text values.

     

    Hope that works for you.

    Cheers. 

m_dekorte's avatar
m_dekorte
Resident Rockstar
2 years ago

Hi RobRayborn 

 

Give something like this a go. If your columns aren't typed instead of Table.ColumnsOfType you can pass a list with all relevant column names. Note that you have to replace "Source" twice with the previous step name in your Query.

 

Table.TransformColumns( Source, 
    List.Transform(
        Table.ColumnsOfType( Source, {type text}), 
        each {_, Text.Trim, type text}
    )
)

 

To illustrate, you can copy this code into a new blank query.

 

let
    TypedSource = Table.FromRows(
        {
            {"   text", "text   ", #date(2024,2,15)},
            {"   text", "text   ", #date(2024,2,15)},
            {"   text", "text   ", #date(2024,2,15)}
        }, type table[ Col1=text, Col2=text, Col3=date]
    ),
    ChType = Table.TransformColumns( TypedSource, 
        List.Transform(
            Table.ColumnsOfType(TypedSource, {type text}), 
            each {_, Text.Trim, type text})
    ),
    UntypedSource = Table.FromRows(
        {
            {"   text", "text   ", #date(2024,2,15)},
            {"   text", "text   ", #date(2024,2,15)},
            {"   text", "text   ", #date(2024,2,15)}
        }
    ),
    ChType2 = Table.TransformColumns( UntypedSource, 
        List.Transform(
            List.FirstN( Table.ColumnNames(UntypedSource), 2),
            each {_, Text.Trim, type text}
        )
    )
in
    ChType2

 

I hope this is helpful

RobRayborn's avatar
RobRayborn
Helper IV
2 years ago

Should this come after the "Navigation" step?
It also keeps generating an error, - "Token '=' expected", which is apparently supposed to come after the Table.TransformColumns.

  • m_dekorte's avatar
    m_dekorte
    Resident Rockstar
    2 years ago

    See the second example, where this code has been implemented.

     

    It needs to come after the Navigation step BUT your columns have to be typed so when you see an "ABC123" beside your column names, you need to set column type first before applying it or use the second method illustrated in that example.

     

    I hope this is helpful