new: Split and Filter by Longest UO

Author: ninmonkey

forumQuestions/Select Distinct Longest OUs.pbix

@@ -0,0 +1,47 @@
+original question: <https://www.reddit.com/r/PowerBI/comments/1brmxet/split_lines_in_excel_and_delimiters_in_power_query/>
+
+The problem causing duplicates is from splitting and expanding in multiple steps. 
+
+There's a different function you can call named `Table.TransformRows`
+That lets you modify multiple columns in one pass. Here's the full code:
+
+- [Final results screenshot.png](https://raw.githubusercontent.com/ninmonkey/ninMonkQuery-examples/main/forumQuestions/img/2024-03_SplitExcelNewlines.png)
+- [SplitExcelNewlines.pq](https://github.com/ninmonkey/ninMonkQuery-examples/blob/5e0b338b6d211c4e712b0819588f83c90f014256/forumQuestions/pq/2024-03_SplitExcelNewlines.pq#L19-L56)
+- [SplitExcelNewlines.pbix](https://github.com/ninmonkey/ninMonkQuery-examples/blob/5e0b338b6d211c4e712b0819588f83c90f014256/forumQuestions/2024-03_SplitExcelNewlines.pbix)
+
+### Importing multiple date formats
+
+As a bonus: I wrote a function that imports both date formats in your example
+
+    try Something catch (e) => // ... handler function
+
+It's better than `try otherwise` because you can optionally return any value, including the original error record. Like `null meta [ info = e ]` 
+
+### The Main Code
+
+- Split each column by a newline
+- then generate a list of N records, using position and expanding the PartId
+
+    ExpandRows = (row as record) as any => [
+            Delim = "#(lf)", // some apps use #(cf,lf) instead
+            invoiceList = Text.Split(row[Invoice], Delim ),
+            qtyList     = Text.Split( row[Qty], Delim ),
+            datesList   = Text.Split( row[Expected], Delim ),
+
+            totalRecords = List.Max({
+                List.Count(invoiceList),
+                List.Count(qtyList),
+                List.Count(datesList)
+            }),
+            
+            return = List.Transform(
+                { 0..(totalRecords - 1) },
+                (num) => [
+                    Part        = partId,
+                    Invoice     = invoiceList{ num }?,
+                    Qty         = qtyList{ num }?,
+                    Expected    = datesList{ num }?,
+                    LineNumber  = num
+                ]
+            )
+        ][return]

forumQuestions/img/Select Distinct Longest OUs.part2.png

@@ -0,0 +1,74 @@
+// DevicesOU_MultiRow
+let
+    EnterData = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45Wckkty0xONbRSqKis0ssvSo+JgYgUK8Xq4JGNiQkoyk8hRk1MjH9aGpCLpNYIU60CWFoBwjPG6xosshiuMcFrAhbZmJiQ1OISoCtiAQ==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [Text = _t]),
+    Source = Table.TransformColumnTypes(EnterData,{{"Text", type text}}),
+
+    #"Added Custom" = Table.AddColumn(Source, 
+        "SplitToRecords",         
+        (row) => [
+            Segments = Text.Split( row[Text], ": "),
+            DeviceName = Segments{0},
+            OU = Segments{1}
+        ], Record.Type
+    ),
+    // returns a single item, the longest string
+    List.SelectLongest = (source as list) as any => 
+        List.First( List.SortByLongest( source ) ),
+
+    // returns original list, sorted as longest first 
+    List.SortByLongest = (source as list) as list => 
+        List.Sort( source, (item) => -Text.Length( item ) ),
+
+
+    #"Expanded SplitToRecords" = Table.ExpandRecordColumn(#"Added Custom", "SplitToRecords", {"DeviceName", "OU"}, {"Device", "OU"}),
+    #"Changed Type" = Table.TransformColumnTypes(#"Expanded SplitToRecords",{{"Device", type text}, {"OU", type text}}),
+    #"Cleaned Text" = Table.TransformColumns(#"Changed Type",{ {"Device", Text.Trim, type text}, {"OU", Text.Trim, type text}} ),
+    #"Removed Other Columns" = Table.SelectColumns(#"Cleaned Text",{"Device", "OU"}),
+    #"Grouped Rows" =  Table.Group(
+            #"Removed Other Columns", {"Device"},
+            { 
+                {
+                    "OU_Longest",
+                    (row) => List.SelectLongest( row[OU] ), 
+                    Text.Type
+                } 
+            }
+            
+        )
+in
+    #"Grouped Rows"
+
+// DevicesOU_SingleLine
+let
+    
+    EnterData = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45Wckkty0xONbRSqKis0ssvSo+JgYgUW6MLxMQEFOWn4BCOifFPSwNylWJ1YGYaYZqpgCRtTLSVSJpMiNIUklpcArQqFgA=", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [Text = _t]),
+    Source = Table.TransformColumnTypes(EnterData,{{"Text", type text}}),
+
+    // this is overkill. I kept it verbose so you can inspect each stage 
+    // it splits a single line, multiple times, as a single "step"
+    SplitToRecords = Table.AddColumn(Source, 
+        "SplitToRecords",         
+        (row) => [
+            // Segments = Text.Split( row[Text], ": "),
+            line        = row[Text],
+            DeviceName  = Text.BeforeDelimiter( line, ": ", 0 ),
+            RestOfLine  = Text.AfterDelimiter( line, ": ", 0 ),
+            OU_list     = Text.Split( RestOfLine, ";"),
+            sortByLongest = List.Sort( OU_list, each -Text.Length(_) ),
+            OU_Longest    = List.First( sortByLongest, null )
+        ], Record.Type
+    ),
+    
+    // for testing, this lets you see one record at a time
+    InspectOneRow = SplitToRecords{0}[SplitToRecords],
+
+    // else grab the only two columns that we care about
+    #"Expanded SplitToRecords" = Table.ExpandRecordColumn(
+        SplitToRecords, "SplitToRecords", 
+        {"DeviceName", "OU_Longest"}, {"DeviceName", "OU_Longest"} ),
+
+    #"Changed Type" = Table.TransformColumnTypes(
+        #"Expanded SplitToRecords",
+        { {"DeviceName", type text}, {"OU_Longest", type text}} )
+in
+    #"Changed Type"