Merge pull request #424 from nfdi4plants/fixes

Various fixes and small additions

src/ARCtrl/ARC.fs

@@ -329,7 +329,7 @@ type ARC(?isa : ArcInvestigation, ?cwl : CWL.CWL, ?fs : FileSystem.FileSystem) =
         let workbooks = System.Collections.Generic.Dictionary<string, DTOType*FsWorkbook>()
         match this.ISA with
         | Some inv -> 
-            let investigationConverter = Spreadsheet.ArcInvestigation.toFsWorkbook
+            let investigationConverter = ArcInvestigation.toFsWorkbook
             workbooks.Add (InvestigationFileName, (DTOType.ISA_Investigation, investigationConverter inv))
             inv.StaticHash <- inv.GetLightHashCode()
             inv.Studies
@@ -365,7 +365,7 @@ type ARC(?isa : ArcInvestigation, ?cwl : CWL.CWL, ?fs : FileSystem.FileSystem) =
 
         | None -> 
             //printfn "ARC contains no ISA part."
-            workbooks.Add (InvestigationFileName, (DTOType.ISA_Investigation, Spreadsheet.ArcInvestigation.toFsWorkbook (ArcInvestigation.create(Identifier.MISSING_IDENTIFIER))))
+            workbooks.Add (InvestigationFileName, (DTOType.ISA_Investigation, ArcInvestigation.toFsWorkbook (ArcInvestigation.create(Identifier.MISSING_IDENTIFIER))))
 
         // Iterates over filesystem and creates a write contract for every file. If possible, include DTO.       
         _fs.Tree.ToFilePaths(true)
@@ -566,11 +566,14 @@ type ARC(?isa : ArcInvestigation, ?cwl : CWL.CWL, ?fs : FileSystem.FileSystem) =
         let isa = ARCtrl.Json.Decode.fromJsonString ARCtrl.Json.ARC.ROCrate.decoder s
         ARC(?isa = isa)
 
+    member this.ToROCrateJsonString(?spaces) =
+        ARCtrl.Json.ARC.ROCrate.encoder (Option.get _isa)
+        |> ARCtrl.Json.Encode.toJsonString (ARCtrl.Json.Encode.defaultSpaces spaces)
+
         /// exports in json-ld format
     static member toROCrateJsonString(?spaces) =
         fun (obj:ARC) ->
-            ARCtrl.Json.ARC.ROCrate.encoder (Option.get obj.ISA)
-            |> ARCtrl.Json.Encode.toJsonString (ARCtrl.Json.Encode.defaultSpaces spaces)
+            obj.ToROCrateJsonString(?spaces = spaces)
 
     /// <summary>
     /// Returns the write contract for the input ValidationPackagesConfig object.

src/Core/ArcTypes.fs

@@ -103,10 +103,13 @@ module ArcTypesAux =
 [<AttachMembers>]
 type ArcAssay(identifier: string, ?measurementType : OntologyAnnotation, ?technologyType : OntologyAnnotation, ?technologyPlatform : OntologyAnnotation, ?tables: ResizeArray<ArcTable>, ?datamap : DataMap, ?performers : ResizeArray<Person>, ?comments : ResizeArray<Comment>) = 
     inherit ArcTables(defaultArg tables <| ResizeArray())
-    
+
     let performers = defaultArg performers <| ResizeArray()
     let comments = defaultArg comments <| ResizeArray()
-    let mutable identifier : string = identifier
+    let mutable identifier : string =
+        let identifier = identifier.Trim()
+        Helper.Identifier.checkValidCharacters identifier
+        identifier
     let mutable investigation : ArcInvestigation option = None
     let mutable measurementType : OntologyAnnotation option = measurementType
     let mutable technologyType : OntologyAnnotation option = technologyType
@@ -530,7 +533,10 @@ type ArcStudy(identifier : string, ?title, ?description, ?submissionDate, ?publi
     let registeredAssayIdentifiers = defaultArg registeredAssayIdentifiers <| ResizeArray()
     let comments = defaultArg comments <| ResizeArray()
 
-    let mutable identifier = identifier
+    let mutable identifier : string =
+        let identifier = identifier.Trim()
+        Helper.Identifier.checkValidCharacters identifier
+        identifier
     let mutable investigation : ArcInvestigation option = None
     let mutable title : string option = title
     let mutable description : string option = description 

src/Core/Table/CompositeColumn.fs

@@ -1,4 +1,4 @@
-namespace ARCtrl
+namespace ARCtrl
 
 open Fable.Core
 
@@ -75,4 +75,7 @@ type CompositeColumn = {
                 if termCellCount >= unitCellCount then
                     CompositeCell.emptyTerm
                 else
-                    CompositeCell.emptyUnitized
+                    CompositeCell.emptyUnitized
+
+    member this.IsUnique =
+        this.Header.IsUnique

src/Core/Table/CompositeHeader.fs

@@ -1,4 +1,4 @@
-namespace ARCtrl
+namespace ARCtrl
 
 open Fable.Core
 open Fable.Core.JsInterop
@@ -92,25 +92,6 @@ type IOType =
         | Some s -> IOType.ofString s |> Some
         | None -> None
 
-    member this.GetUITooltip() = IOType.getUITooltip(U2.Case1 this)
-
-    static member getUITooltip(iotype: U2<IOType,string>) =
-        match iotype with
-        | U2.Case1 Source | U2.Case2 "Source" -> 
-            "The source value must be a unique identifier for an organism or a sample." 
-        | U2.Case1 Sample | U2.Case2 "Sample" -> 
-            "The Sample Name column describes specifc laboratory samples with a unique identifier." 
-        | U2.Case1 Data | U2.Case2 "RawDataFile" -> 
-            "The Raw Data File column defines untransformed and unprocessed data files."
-        | U2.Case1 Data | U2.Case2 "DerivedDataFile" -> 
-            "The Derived Data File column defines transformed and/or processed data files."
-        | U2.Case1 Data | U2.Case2 "ImageFile" -> 
-            "Placeholder"
-        | U2.Case1 Material | U2.Case2 "Material" -> 
-            "Placeholder" 
-        | U2.Case1 (FreeText _) | U2.Case2 "FreeText" -> 
-            "Placeholder"
-        | _ -> failwith $"Unable to parse combination to existing IOType: `{iotype}`"
 #if FABLE_COMPILER
 
     static member source() = IOType.Source
@@ -446,50 +427,10 @@ type CompositeHeader =
         | Output io | Input io -> Some io
         | _ -> None
 
-    member this.GetUITooltip() =
-        // https://fable.io/docs/javascript/features.html#u2-u3--u9
-        CompositeHeader.getUITooltip (U2.Case1
-        this)
-
-    // https://fable.io/docs/javascript/features.html#u2-u3--u9
-    // U2 is an erased union type, allowing seemless integration into js syntax
-    /// <summary>
-    /// Can pass header as `U2.Case1 compositeHeader` or `U2.Case2 string` or (requires `open Fable.Core.JsInterop`) `!^compositeHeader` or `!^string`
-    /// </summary>
-    /// <param name="header"></param>
-    static member getUITooltip(header:U2<CompositeHeader,string>) =
-        match header with
-        | U2.Case1 (Component _) | U2.Case2 "Component" -> 
-            "Component columns are used to describe physical components of a experiment, e.g. instrument names, software names, and reagents names."
-        | U2.Case1 (Characteristic _) | U2.Case2 "Characteristic" ->
-            "Characteristic columns are used for study descriptions and describe inherent properties of the source material, e.g. a certain strain or organism part."
-        | U2.Case1 (Factor _)| U2.Case2 "Factor" ->
-            "Use Factor columns to describe independent variables that result in a specific output of your experiment, e.g. the light intensity under which an organism was grown."
-        | U2.Case1 (Parameter _) | U2.Case2 "Parameter" ->
-            "Parameter columns describe steps in your experimental workflow, e.g. the centrifugation time or the temperature used for your assay."
-        | U2.Case1 (ProtocolType) | U2.Case2 "ProtocolType" ->
-            "Defines the protocol type according to your preferred endpoint repository." 
-        | U2.Case1 (ProtocolDescription) | U2.Case2 "ProtocolDescription" ->
-            "Describe the protocol in free text."
-        | U2.Case1 (ProtocolUri) | U2.Case2 "ProtocolUri" ->
-            "Web or local address where the in-depth protocol is stored."
-        | U2.Case1 (ProtocolVersion) | U2.Case2 "ProtocolVersion" ->
-            "Defines the protocol version."
-        | U2.Case1 (ProtocolREF) | U2.Case2 "ProtocolREF" ->
-            "Defines the protocol name."
-        | U2.Case1 (Performer) | U2.Case2 "Performer" ->
-            "Defines the protocol performer."
-        | U2.Case1 (Date) | U2.Case2 "Date" ->
-            "Defines the date the protocol was performed."
-        | U2.Case1 (Input _) | U2.Case2 "Input" ->
-            "Only one input column per table. E.g. experimental samples or files."
-        | U2.Case1 (Output _) |U2.Case2 "Output" ->
-            "Only one output column per table. E.g. experimental samples or files."
-        | U2.Case1 (FreeText _) | U2.Case2 "FreeText" ->
-            "Placeholder"
-        | U2.Case1 (Comment _) | U2.Case2 "Comment" ->
-            "Comment"
-        | _ -> failwith $"Unable to parse combination to existing CompositeHeader: `{header}`"
+    member this.IsUnique =
+        match this with
+        | ProtocolType | ProtocolREF | ProtocolDescription | ProtocolUri | ProtocolVersion | Performer | Date | Input _ | Output _ -> true
+        | _ -> false
 
 #if FABLE_COMPILER
 

src/Spreadsheet/AnnotationTable/ArcTable.fs

@@ -1,4 +1,4 @@
-module ARCtrl.Spreadsheet.ArcTable
+module ARCtrl.Spreadsheet.ArcTable
 
 open ARCtrl
 open ARCtrl.Helper
@@ -121,7 +121,7 @@ let tryFromFsWorksheet (sheet : FsWorksheet) =
     | err -> failwithf "Could not parse table with name \"%s\":\n%s" sheet.Name err.Message
 
 
-let toFsWorksheet (table : ArcTable) =
+let toFsWorksheet (index : int option) (table : ArcTable) =
     /// This dictionary is used to add spaces at the end of duplicate headers.
     let stringCount = System.Collections.Generic.Dictionary<string,string>()
     let ws = FsWorksheet(table.Name)
@@ -137,7 +137,11 @@ let toFsWorksheet (table : ArcTable) =
         |> List.collect CompositeColumn.toStringCellColumns
     let maxRow = columns.Head.Length
     let maxCol = columns.Length
-    let fsTable = ws.Table("annotationTable",FsRangeAddress(FsAddress(1,1),FsAddress(maxRow,maxCol)))
+    let name =
+        match index with
+        | Some i -> $"{annotationTablePrefix}{i}"
+        | None -> annotationTablePrefix
+    let fsTable = ws.Table(name,FsRangeAddress(FsAddress(1,1),FsAddress(maxRow,maxCol)))
     columns
     |> List.iteri (fun colI col ->         
         col

src/Spreadsheet/ArcAssay.fs

@@ -13,8 +13,8 @@ module ArcAssay =
     let [<Literal>] assaysPrefix = "Assay"
     let [<Literal>] contactsPrefix = "Assay Person"
 
-    let [<Literal>] obsoleteMetaDataSheetName = "Assay"
-    let [<Literal>] metaDataSheetName = "isa_assay"
+    let [<Literal>] obsoleteMetadataSheetName = "Assay"
+    let [<Literal>] metadataSheetName = "isa_assay"
 
     let toMetadataSheet (assay : ArcAssay) : FsWorksheet =
         let toRows (assay:ArcAssay) =
@@ -25,7 +25,7 @@ module ArcAssay =
                 yield SparseRow.fromValues [contactsLabel]
                 yield! Contacts.toRows (Some contactsPrefix) (List.ofSeq assay.Performers)
             }
-        let sheet = FsWorksheet(metaDataSheetName)
+        let sheet = FsWorksheet(metadataSheetName)
         assay
         |> toRows
         |> Seq.iteri (fun rowI r -> SparseRow.writeToSheet (rowI + 1) r sheet)    
@@ -76,6 +76,15 @@ module ArcAssay =
         with 
         | err -> failwithf "Failed while parsing metadatasheet: %s" err.Message
 
+    let isMetadataSheetName (name:string) =
+        name = metadataSheetName || name = obsoleteMetadataSheetName
+
+    let isMetadataSheet (sheet : FsWorksheet) =
+        isMetadataSheetName sheet.Name
+
+    let tryGetMetadataSheet (doc:FsWorkbook) =
+        doc.GetWorksheets()
+        |> Seq.tryFind isMetadataSheet
 
 [<AutoOpen>]
 module ArcAssayExtensions =
@@ -86,27 +95,22 @@ module ArcAssayExtensions =
         static member fromFsWorkbook (doc:FsWorkbook) : ArcAssay = 
             try
                 // Reading the "Assay" metadata sheet. Here metadata 
-                let assayMetaData = 
-
-                    match doc.TryGetWorksheetByName ArcAssay.metaDataSheetName with 
+                let assayMetadata = 
+                    match ArcAssay.tryGetMetadataSheet doc with
                     | Option.Some sheet ->
                         ArcAssay.fromMetadataSheet sheet
                     | None -> 
-                        match doc.TryGetWorksheetByName ArcAssay.obsoleteMetaDataSheetName with 
-                        | Option.Some sheet ->
-                            ArcAssay.fromMetadataSheet sheet
-                        | None -> 
-                            printfn "Cannot retrieve metadata: Assay file does not contain \"%s\" or \"%s\" sheet." ArcAssay.metaDataSheetName ArcAssay.obsoleteMetaDataSheetName
-                            ArcAssay.create(Identifier.createMissingIdentifier())
+                        printfn "Cannot retrieve metadata: Assay file does not contain \"%s\" or \"%s\" sheet." ArcAssay.metadataSheetName ArcAssay.obsoleteMetadataSheetName
+                        ArcAssay.create(Identifier.createMissingIdentifier())
                 let sheets = doc.GetWorksheets()
                 let annotationTables = 
                     sheets |> Seq.choose ArcTable.tryFromFsWorksheet
                 let datamapSheet =
                     sheets |> Seq.tryPick DataMapTable.tryFromFsWorksheet
                 if annotationTables |> Seq.isEmpty |> not then
-                    assayMetaData.Tables <- ResizeArray annotationTables
-                assayMetaData.DataMap <- datamapSheet
-                assayMetaData
+                    assayMetadata.Tables <- ResizeArray annotationTables
+                assayMetadata.DataMap <- datamapSheet
+                assayMetadata
             with
             | err -> failwithf "Could not parse assay: \n%s" err.Message
 
@@ -121,14 +125,20 @@ module ArcAssayExtensions =
         static member toFsWorkbook (assay : ArcAssay, ?datamapSheet: bool) =
             let datamapSheet = defaultArg datamapSheet true
             let doc = new FsWorkbook()
-            let metaDataSheet = ArcAssay.toMetadataSheet (assay)
-            doc.AddWorksheet metaDataSheet
+            let metadataSheet = ArcAssay.toMetadataSheet (assay)
+            doc.AddWorksheet metadataSheet
 
             if datamapSheet then
                 assay.DataMap
                 |> Option.iter (DataMapTable.toFsWorksheet >> doc.AddWorksheet)
 
             assay.Tables
-            |> Seq.iter (ArcTable.toFsWorksheet >> doc.AddWorksheet)
+            |> Seq.iteri (fun i -> ArcTable.toFsWorksheet (Some i) >> doc.AddWorksheet)
+
+            doc
 
-            doc
+        /// Write an assay to a spreadsheet
+        ///
+        /// If datamapSheet is true, the datamap will be written to a worksheet inside assay workbook.
+        member this.ToFsWorkbook (?datamapSheet: bool) =
+            ArcAssay.toFsWorkbook (this, ?datamapSheet = datamapSheet)

src/Spreadsheet/ArcInvestigation.fs

@@ -24,8 +24,8 @@ module ArcInvestigation =
     let [<Literal>] publicationsLabelPrefix = "Investigation Publication"
     let [<Literal>] contactsLabelPrefix = "Investigation Person"
 
-    let [<Literal>] metaDataSheetName = "isa_investigation"
-    let [<Literal>] metaDataSheetName_deprecated = "Investigation"
+    let [<Literal>] metadataSheetName = "isa_investigation"
+    let [<Literal>] obsoleteMetadataSheetName = "Investigation"
 
 
     type InvestigationInfo =
@@ -202,28 +202,45 @@ module ArcInvestigation =
         |> insertRemarks (List.ofSeq investigation.Remarks)
         |> seq
 
-    let fromFsWorkbook (doc:FsWorkbook) =  
-        try
-            match doc.TryGetWorksheetByName metaDataSheetName with
-            | Some sheet -> sheet
-            | None -> 
-                match doc.TryGetWorksheetByName metaDataSheetName_deprecated with
+    let isMetadataSheetName (name : string) =
+        name = metadataSheetName || name = obsoleteMetadataSheetName
+
+    let isMetadataSheet (sheet : FsWorksheet) =
+        isMetadataSheetName sheet.Name
+
+    let tryGetMetadataSheet (doc:FsWorkbook) =
+        doc.GetWorksheets()
+        |> Seq.tryFind isMetadataSheet
+
+
+[<AutoOpen>]
+module ArcInvestigationExtensions =
+
+    open ArcInvestigation
+
+    type ArcInvestigation with
+
+        static member fromFsWorkbook (doc:FsWorkbook) =  
+            try
+                match ArcInvestigation.tryGetMetadataSheet doc with
                 | Some sheet -> sheet
                 | None -> failwith "Could not find metadata sheet with sheetname \"isa_investigation\" or deprecated sheetname \"Investigation\""
-            |> FsWorksheet.getRows
-            |> Seq.map SparseRow.fromFsRow
-            |> fromRows 
-        with
-        | err -> failwithf "Could not read investigation from spreadsheet: %s" err.Message
-
-    let toFsWorkbook (investigation:ArcInvestigation) : FsWorkbook =           
-        try
-            let wb = new FsWorkbook()
-            let sheet = FsWorksheet(metaDataSheetName)
-            investigation
-            |> toRows
-            |> Seq.iteri (fun rowI r -> SparseRow.writeToSheet (rowI + 1) r sheet)                     
-            wb.AddWorksheet(sheet)
-            wb
-        with
-        | err -> failwithf "Could not write investigation to spreadsheet: %s" err.Message
+                |> FsWorksheet.getRows
+                |> Seq.map SparseRow.fromFsRow
+                |> fromRows 
+            with
+            | err -> failwithf "Could not read investigation from spreadsheet: %s" err.Message
+
+        static member toFsWorkbook (investigation:ArcInvestigation) : FsWorkbook =           
+            try
+                let wb = new FsWorkbook()
+                let sheet = FsWorksheet(metadataSheetName)
+                investigation
+                |> toRows
+                |> Seq.iteri (fun rowI r -> SparseRow.writeToSheet (rowI + 1) r sheet)                     
+                wb.AddWorksheet(sheet)
+                wb
+            with
+            | err -> failwithf "Could not write investigation to spreadsheet: %s" err.Message
+
+        member this.ToFsWorkbook() = ArcInvestigation.toFsWorkbook this