Extracting a Specific string format from a string of text

[agc1234]

Good Day Everyone,

I am trying to extract a specific format of: ##a# or (number)(number)(letter)(number)

I need first 4 parts of the serial number. I've attached examples of source data and what I'd like to extract:

Does anyone have any ideas? Really stuck trying to get a reliable formula for this.

Thanks Everyone!

[AlienSx]

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("jY4xCgIxEEWvMqQ2kMxMJoldomAhpLERYgqFLRa0Wdn7q5scwPLNPPivVmVNEiaPqu2quhSDhyCGqeO0zPcnlPX1mBaY37AHS+cYkXjo30MsJOy8BLitxqCAxZML/ip+c6xAAgkkfQFRZ+8Cx05WZybNTrprtpR8HPQrSyMtjecfw6q1Dw==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [#"Serial #" = _t]),
    pattern = {false, false, true, false},
    chars = List.Buffer({"0".."9", "A".."Z"}),    
    finder = (t as text) =>
        [lst = List.Buffer(List.Select(Text.ToList(t), (x) => List.Contains(chars, x))),
        gen = List.Generate(
            () => [ch = lst, okay = false, serial = {}],
            (x) => not List.IsEmpty(x[ch]) or x[okay], 
            (x) => 
                [okay = List.Transform(List.FirstN(x[ch], 4), (w) => (try Number.From(w))[HasError]) = pattern,
                serial =  if okay then List.FirstN(x[ch], 4) else {},
                ch = List.Skip(x[ch], if okay then 4 else 1)],
            (x) => Text.Combine(x[serial])
        ),
        z = Text.Combine(List.Select(gen, each _ <> ""), ", ")][z],
    out = Table.AddColumn(Source, "Desired Outcome", (x) => finder(x[#"Serial #"]))
in
    out

[dufoq3]

Hello @agc1234,
same result as @Vijay_A_Verma but different approach.

See line four. This one also matches the criteria.

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("LY4xCwIxDEb/SuhsoU3StHVrFRyELi5C7aBww8G5nNz/V68dX/Lge7Uqa5IweVTtUNWtGDwFMUwdp3V+LlC292taYf7AESxdY0Tiof8OsZCw8xLgsRmDAhYvLvi7+N2xAgkkkPQFRJ29Cxw7WZ2ZNDvprtlT8nnQvyyNtDSeqrUv", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [#"Serial #" = _t]),
    UppercasedText = Table.TransformColumns(Source,{{"Serial #", Text.Upper, type text}}),
    Ad_Cleaned = Table.AddColumn(UppercasedText , "Cleaned", each Text.Select([#"Serial #"], List.Buffer({"0".."9", "A".."Z"}))),
    Ad_EachFourCombination = Table.AddColumn(Ad_Cleaned, "Each 4 Combination", each List.Accumulate( 
    {0..Text.Length([Cleaned])-1},
    {},
    (s,c)=> List.RemoveNulls(s & { try Text.Range([Cleaned], c, 4) otherwise null })
), type list),
    Ad_OnlyMatching = Table.AddColumn(Ad_EachFourCombination, "Only Matching", each List.RemoveNulls(List.Transform([Each 4 Combination], each 
      if  (try Value.Is(Number.From(Text.ToList(_){0}), Number.Type) otherwise false) //1st digit check if is number
      and (try Value.Is(Number.From(Text.ToList(_){1}), Number.Type) otherwise false) //2nd digit check if is number
      and (try not Value.Is(Number.From(Text.ToList(_){2}), Number.Type) otherwise true) //3rd digit check if is text
      and (try Value.Is(Number.From(Text.ToList(_){3}), Number.Type) otherwise false) //4th digit check if is number
      then _ else null)), type list),
    Ad_FourDigits = Table.AddColumn(Ad_OnlyMatching, "4 digits", each Text.Combine([Only Matching], ", "), type text),
    RemovedOtherColumns = Table.SelectColumns(Ad_FourDigits,{"Serial #", "4 digits"})
in
    RemovedOtherColumns

[latimeria]

Hi @agc1234 ,

Another approach

let
    Source = Excel.CurrentWorkbook(){[Name="Table1"]}[Content],
    #"Changed Type" = Table.TransformColumnTypes(Source,{{"Serial#", type text}}),
    #"Added Desired Output" = Table.AddColumn(
        #"Changed Type", 
        "Desired Output", 
        each 
            List.Transform(
                Text.Split(Text.Remove([#"Serial#"],{" ", "-"}),"&"), 
                each Text.Start(
                    Text.RemoveRange(_,
                    0,
                    Text.PositionOfAny( 
                        _,
                        {"0".."9"},Occurrence.First
                    )
                ),
                4
            )
        )
    ),
    #"Extracted Values" = Table.TransformColumns(
        #"Added Desired Output", 
        {"Desired Output", each Text.Combine(List.Transform(_, Text.From), ", "), type text}
        )
in
    #"Extracted Values"

[dufoq3]

Hi @latimeria,

These should be excluded.

[Vijay_A_Verma]

Use below code

Note - In row 4, 87X6 is also a pattern matching your requirement.

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("LY4xCwIxDEb/SuhsoU3StHVrFRyELi5C7aBww8G5nNz/V68dX/Lge7Uqa5IweVTtUNWtGDwFMUwdp3V+LlC292taYf7AESxdY0Tiof8OsZCw8xLgsRmDAhYvLvi7+N2xAgkkkPQFRJ29Cxw7WZ2ZNDvprtlT8nnQvyyNtDSeqrUv", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [#"Serial #" = _t]),
    #"Added Custom2" = Table.AddColumn(Source, "Custom.1", (x)=> 
    let
        p = Text.Select(x[#"Serial #"], {"0".."9", "A".."Z"}),
        c = Text.Length(p)-3,
        r = List.Generate(()=>[i=0, a=Text.Start(p,4)], each [i]<c, each [i=[i]+1, a=Text.Middle(p,i,4)], each [a])
    in
        Text.Combine(List.Select(r, (x)=> List.ContainsAll({"0".."9"}, Text.ToList(Text.Start(x, 2) & Text.End(x, 1))) and Text.Select(x, {"A".."Z"}) = Text.At(x,2)), "; ")    )
in
    #"Added Custom2"

[v-stephen-msft]

Hi @agc1234 ,

In Power BI, when we want to process complex and unordered data, adding many steps using power query may not get the desired results. As we know, power query cannot use regular expressions. If we call Python script and use regular expressions to process the data this time, it will become simple. Please refer to

Using regular expressions in power bi desktop - Microsoft Fabric Community

Best Regards,

Stephen Tao

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.