Merge tables using wildcards

[Rudz]

I have a small fact table of about 10,000 records. Based on 5 columns in each row, I need to add 3 extra columns from a small dimension table of around 70 records.

The problem is that the rules in the dimension table include wildcards. The Reference column in the dim table uses prefixes to match. So Fact References of 20123 and 20999 need to match the Dim Reference of 20*. Other columns need a wildcard to match any value, so Fact assignment of A, B, or Z would match a Dim Assignment of *.

In SQL, I can do a join with wildcards:

WHERE fact.company LIKE dim.company + '%'

AND fact.reference LIKE dim.reference+'%'

AND fact.assignment LIKE dim.assignment+'%'

AND fact.plant LIKE dim.plant+'%'

AND fact.customer LIKE dim.customer+'%'

I haven't figured out any way to do this in Power Query. Chris Webb has a Power Query LIKE function, but it looks like it would only work to filter a single column. I need to join on 5 columns. I also found a post that matched on multiple columns using Table.MatchesAnyRows, but it simply indicated where there was a match without returning any columns.

Here is sample data:

FACT

Company	Assignment	Reference	Plant	Customer
A	null	20123	J	5
A	null	20999	J	5
A	Z	M123	J	null
B	CMIX	CMIX	G	null
Z	null	99999	J	5

DIM

Company	Reference	Assignment	Plant	Customer	Result1	Result2	Result3
A	20*	null	J	5	G	P	G
A	21*	G	J	null	G	G	P
A	C M*	*	J	null	Z	P	O
A	M*	*	J	null	Z	O	P
B	10*	null	D	8	E	L	I
*	ADJ*	*	*	*	null	null	null
*	CMI*	*	*	*	null	null	null

The goal is to match records here to the fact table and add the 3 result columns. If there is no match, I need to identify those rows as grief, either in the output or in a separate query. It would be acceptable to return nulls if no match.

The * character in the Dim table indicates match of 0 or more characters (needs to match null).

All values in the reference column are prefixes with the wildcard at the end.

The * character could be changed to % or anything else as needed or removed.

Desired output

Company	Assignment	Reference	Plant	Customer	Result1	Result2	Result3
A	null	20123	J	5	G	P	G
A	null	20999	J	5	G	P	G
A	Z	M123	J	null	Z	O	P
B	CMIX	CMIX	G	null	null	null	null
Z	null	99999	J	5	grief	grief	grief

I have a sample file I could attach if I can figure out how!

Thanks for any help!

[wdx223_Daniel · Accepted solution]

@Rudz 

let
    Fact = Excel.CurrentWorkbook(){[Name="Fact"]}[Content],
    Dim = Excel.CurrentWorkbook(){[Name="Dim"]}[Content],
    fx=(t1,t2)=>
        let
            a=Text.From(t2),
            b=Text.From(t1),
            c=if Text.End(a,1)="*" then Text.StartsWith(b??"",Text.Start(a,Text.Length(a)-1)) else a=b
        in if a="*" then true else c,
    Custom1 = Table.AddColumn(
                              Fact,
                              "n",
                              each let
                                      col=Table.ColumnNames(Fact),
                                      b=Table.SelectRows(
                                                         Dim,
                                                         (x)=>List.AllTrue(List.Transform(col,(y)=>fx(Record.Field(_,y),Record.Field(x,y))))
                                                        )
                                   in if b{0}?=null 
                                      then {null,null,null}
                                      else Record.ToList(Record.SelectFields(b{0},{"Result1","Result2","Result3"}))
                             ),
    Custom2 = Table.SplitColumn(Custom1,"n",each _,{"Result1","Result2","Result3"})
in
    Custom2