rework ontologyannotation localID field

src/ISA/ISA.Json/Ontology.fs

@@ -109,27 +109,16 @@ module OntologyAnnotation =
         |> GEncode.choose
         |> Encode.object
 
-    let localIDDecoder : Decoder<string> =
-        fun s json ->
-            match Decode.string s json with
-            | Ok (Regex.ActivePatterns.TermAnnotation tan) -> 
-                Ok (tan.TermSourceREF)
-            | _ -> Ok ""
-            //| Ok s -> Error (DecoderError(s,ErrorReason.FailMessage "Could not parse local ID from string"))
-            //| Error e -> Error e
-
 
     let decoder (options : ConverterOptions) : Decoder<OntologyAnnotation> =
         Decode.object (fun get ->
-            {
-                ID = get.Optional.Field "@id" GDecode.uri
-                Name = get.Optional.Field "annotationValue" (AnnotationValue.decoder options)
-                TermSourceREF = get.Optional.Field "termSource" Decode.string
-                //LocalID = try get.Optional.Field "termAccession" localIDDecoder with | _ -> None
-                LocalID = get.Optional.Field "termAccession" localIDDecoder |> Option.bind (fun s -> if s = "" then None else Some s)
-                TermAccessionNumber = get.Optional.Field "termAccession" Decode.string
-                Comments = get.Optional.Field "comments" (Decode.array (Comment.decoder options))               
-            }
+            OntologyAnnotation.create(
+                ?Id = get.Optional.Field "@id" GDecode.uri,
+                ?Name = get.Optional.Field "annotationValue" (AnnotationValue.decoder options),
+                ?TermSourceREF = get.Optional.Field "termSource" Decode.string,
+                ?TermAccessionNumber = get.Optional.Field "termAccession" Decode.string,
+                ?Comments = get.Optional.Field "comments" (Decode.array (Comment.decoder options))               
+            )
         )
 
     let fromString (s:string) = 

src/ISA/ISA.Spreadsheet/CompositeHeader.fs

@@ -7,10 +7,10 @@ module ActivePattern =
 
     open ARCtrl.ISA.Regex.ActivePatterns
 
-    let mergeTerms tsr1 tan1 tsr2 tan2 =
-        if tsr1 <> tsr2 then failwithf "TermSourceRef %s and %s do not match" tsr1 tsr2
-        if tan1 <> tan2 then failwithf "TermAccessionNumber %s and %s do not match" tan1 tan2
-        {|TermSourceRef = tsr1; TermAccessionNumber = tan1|}
+    let mergeIDInfo idSpace1 localID1 idSpace2 localID2 =
+        if idSpace1 <> idSpace2 then failwithf "TermSourceRef %s and %s do not match" idSpace1 idSpace2
+        if localID1 <> localID2 then failwithf "LocalID %s and %s do not match" localID1 localID2
+        {|TermSourceRef = idSpace1; TermAccessionNumber = $"{idSpace1}:{localID1}"|}
 
     let (|Term|_|) (categoryParser : string -> string option) (f : OntologyAnnotation -> CompositeHeader) (cells : FsCell list) =
         let (|AC|_|) s =
@@ -25,7 +25,7 @@ module ActivePattern =
         //| [AC name; TermAccessionNumber term1; TermSourceREF term2] 
         //| [AC name; Unit; TermAccessionNumber term1; TermSourceREF term2] 
         | [AC name; UnitColumnHeader; TSRColumnHeader term1; TANColumnHeader term2] ->
-            let term = mergeTerms term1.TermSourceREF term1.TermAccessionNumber term2.TermSourceREF term2.TermAccessionNumber
+            let term = mergeIDInfo term1.IDSpace term1.LocalID term2.IDSpace term2.LocalID
             let ont = OntologyAnnotation.fromString(name, term.TermSourceRef, term.TermAccessionNumber)
             f ont
             |> Some

src/ISA/ISA/ArcTypes/CompositeCell.fs

@@ -75,10 +75,10 @@ type CompositeCell =
     // TODO: i would really love to have an overload here accepting string input
     static member createTerm (oa:OntologyAnnotation) = Term oa
     static member createTermFromString (?name: string, ?tsr: string, ?tan: string) =
-        Term <| OntologyAnnotation.fromString(?term = name, ?tsr = tsr, ?tan = tan)
+        Term <| OntologyAnnotation.fromString(?termName = name, ?tsr = tsr, ?tan = tan)
     static member createUnitized (value: string, ?oa:OntologyAnnotation) = Unitized (value, Option.defaultValue (OntologyAnnotation.empty) oa)
     static member createUnitizedFromString (value: string, ?name: string, ?tsr: string, ?tan: string) = 
-        Unitized <| (value, OntologyAnnotation.fromString(?term = name, ?tsr = tsr, ?tan = tan))
+        Unitized <| (value, OntologyAnnotation.fromString(?termName = name, ?tsr = tsr, ?tan = tan))
     static member createFreeText (value: string) = FreeText value
 
     static member emptyTerm = Term OntologyAnnotation.empty

src/ISA/ISA/ArcTypes/CompositeHeader.fs

@@ -140,10 +140,10 @@ type CompositeHeader =
         // Input/Output have similiar naming as Term, but are more specific. 
         // So they have to be called first.
         | Regex.ActivePatterns.Regex Regex.Pattern.InputPattern r ->
-            let iotype = r.Groups.["iotype"].Value
+            let iotype = r.Groups.[Regex.Pattern.MatchGroups.iotype].Value
             Input <| IOType.ofString (iotype)
         | Regex.ActivePatterns.Regex Regex.Pattern.OutputPattern r ->
-            let iotype = r.Groups.["iotype"].Value
+            let iotype = r.Groups.[Regex.Pattern.MatchGroups.iotype].Value
             Output <| IOType.ofString (iotype)
         // Is term column
         | Regex.ActivePatterns.TermColumn r ->

src/ISA/ISA/JsonTypes/Component.fs

@@ -69,7 +69,7 @@ type Component =
 
     /// Create a ISAJson Component from ISATab string entries
     static member fromString (?name: string, ?term:string, ?source:string, ?accession:string, ?comments : Comment []) = 
-        let cType = OntologyAnnotation.fromString (?term = term, ?tsr=source, ?tan=accession, ?comments = comments) |> Option.fromValueWithDefault OntologyAnnotation.empty
+        let cType = OntologyAnnotation.fromString (?termName = term, ?tsr=source, ?tan=accession, ?comments = comments) |> Option.fromValueWithDefault OntologyAnnotation.empty
         match name with
         | Some n -> 
             let v,u = Component.decomposeName n

src/ISA/ISA/JsonTypes/OntologyAnnotation.fs

@@ -11,28 +11,31 @@ type OntologyAnnotation =
         ID : URI option
         Name : AnnotationValue option
         TermSourceREF : string option
-        LocalID : string option
         TermAccessionNumber : URI option
         Comments : Comment [] option
     }
 
-    static member make id name termSourceREF localID termAccessionNumber comments= 
+    static member make id name termSourceREF termAccessionNumber comments= 
         {
             ID = id
             Name = name 
             TermSourceREF = termSourceREF
-            LocalID = localID
             TermAccessionNumber = termAccessionNumber  
             Comments = comments
         }
-
+    
     /// This function creates the type exactly as given. If you want a more streamlined approach use `OntologyAnnotation.fromString`.
-    static member create(?Id,?Name,?TermSourceREF,?LocalID,?TermAccessionNumber,?Comments) : OntologyAnnotation =
-        OntologyAnnotation.make Id Name TermSourceREF LocalID TermAccessionNumber Comments
+    static member create(?Id,?Name,?TermSourceREF,?TermAccessionNumber,?Comments) : OntologyAnnotation =
+        OntologyAnnotation.make Id Name TermSourceREF TermAccessionNumber Comments
 
     static member empty =
         OntologyAnnotation.create()
 
+
+    member this.IDInfo = 
+        this.TermAccessionNumber
+        |> Option.bind Regex.tryParseTermAnnotation 
+
     /// Returns the name of the ontology as string
     // TODO: Why is this called Text, while everything else is called string?
     member this.NameText =
@@ -76,20 +79,12 @@ type OntologyAnnotation =
     ///</summary>
     ///<param name="tsr">Term source reference</param>
     ///<param name="tan">Term accession number</param>
-    static member fromString (?term:string, ?tsr:string, ?tan:string, ?comments : Comment []) =
-
-        let tsr,localID = 
-            match tan with
-            | Some (Regex.ActivePatterns.TermAnnotation tan) -> 
-                (if tsr.IsSome then tsr else Some tan.TermSourceREF),
-                Some tan.LocalTAN
-            | _ -> tsr,None
+    static member fromString (?termName:string, ?tsr:string, ?tan:string, ?comments : Comment []) =
 
         OntologyAnnotation.make 
             None 
-            (term |> Option.map AnnotationValue.fromString)
+            (termName |> Option.map AnnotationValue.fromString)
             tsr
-            localID
             tan
             comments
 
@@ -99,20 +94,20 @@ type OntologyAnnotation =
         |> Regex.tryParseTermAnnotation
         |> Option.get 
         |> fun r ->
-            let accession = r.TermSourceREF + ":" + r.LocalTAN
-            OntologyAnnotation.fromString ("", r.TermSourceREF, accession)
+            let accession = r.IDSpace + ":" + r.LocalID
+            OntologyAnnotation.fromString ("", r.IDSpace, accession)
 
     /// Parses any value in `TermAccessionString` to term accession format "termsourceref:localtan". Exmp.: "MS:000001".
     ///
     /// If `TermAccessionString` cannot be parsed to this format, returns empty string!
     member this.TermAccessionShort = 
-        match this.TermSourceREF, this.LocalID with
-        | Some tsr, Some id -> $"{tsr}:{id}"
+        match this.IDInfo with
+        | Some id -> $"{id.IDSpace}:{id.LocalID}"
         | _ -> ""
 
     member this.TermAccessionOntobeeUrl = 
-        match this.TermSourceREF, this.LocalID with
-        | Some tsr, Some id -> OntologyAnnotation.createUriAnnotation tsr id
+        match this.IDInfo with
+        | Some id -> OntologyAnnotation.createUriAnnotation id.IDSpace id.LocalID
         | _ -> ""
 
     member this.TermAccessionAndOntobeeUrlIfShort = 
@@ -234,4 +229,4 @@ type OntologyAnnotation =
 
     member this.Copy() =
         let nextComments = this.Comments |> Option.map (Array.map (fun c -> c.Copy()))
-        OntologyAnnotation.make this.ID this.Name this.TermSourceREF this.LocalID this.TermAccessionNumber nextComments
+        OntologyAnnotation.make this.ID this.Name this.TermSourceREF this.TermAccessionNumber nextComments

src/ISA/ISA/Regex.fs

@@ -7,6 +7,20 @@ open System
 
 module Pattern =
 
+    module MatchGroups =
+
+        [<Literal>]
+        let numberFormat = "numberFormat"
+
+        [<Literal>]
+        let localID = "localid"
+
+        [<Literal>]
+        let idspace = "idspace"
+
+        [<Literal>]
+        let iotype = "iotype"
+
     /// This pattern is only used to remove any leftover #id attributes from previous Swate version. 
     /// `"Parameter [biological replicate#2]"` This #id is deprecated but the pattern can still be used to remove any from files. 
     /// Was deprecated before 2023.
@@ -21,8 +35,7 @@ module Pattern =
     let SquaredBracketsTermNamePattern = "\[.*\]" //  @"(?<= \[)[^#\]]*(?=[\]#])" <- Cannot be used in IE11
 
     /// Used to get unit name from Excel numberFormat: 0.00 "degree Celsius" --> degree Celsius
-    [<LiteralAttribute>]
-    let ExcelNumberFormat = "\"(?<numberFormat>(.*?))\""
+    let ExcelNumberFormat = $"\"(?<{MatchGroups.numberFormat}>(.*?))\""
 
     /// Hits Unit column header
     [<LiteralAttribute>]
@@ -55,29 +68,23 @@ module Pattern =
     let TermAccessionNumberColumnPattern = @"Term Accession Number\s\((?<id>.*)\)" 
 
     /// Hits term accession, without id: ENVO:01001831
-    [<LiteralAttribute>]
-    let TermAnnotationShortPattern = @"(?<termsourceref>\w+?):(?<localtan>\w+)" //prev: @"[\w]+?:[\d]+"
+    let TermAnnotationShortPattern = $@"(?<{MatchGroups.idspace}>\w+?):(?<{MatchGroups.localID}>\w+)" //prev: @"[\w]+?:[\d]+"
 
     // https://obofoundry.org/id-policy.html#mapping-of-owl-ids-to-obo-format-ids
     /// <summary>Regex pattern is designed to hit only Foundry-compliant URIs.</summary>
-    [<LiteralAttribute>]
-    let TermAnnotationURIPattern = @"http://purl.obolibrary.org/obo/(?<termsourceref>\w+?)_(?<localtan>\w+)"
+    let TermAnnotationURIPattern = $@"http://purl.obolibrary.org/obo/(?<{MatchGroups.idspace}>\w+?)_(?<{MatchGroups.localID}>\w+)"
 
     /// Watch this closely, this could hit some edge cases we do not want to cover.
-    [<LiteralAttribute>]
-    let TermAnnotationURIPattern_lessRestrictive = @".*\/(?<termsourceref>\w+?)[:_](?<localtan>\w+)"
+    let TermAnnotationURIPattern_lessRestrictive = $@".*\/(?<{MatchGroups.idspace}>\w+?)[:_](?<{MatchGroups.localID}>\w+)"
 
     /// This pattern is used to match both Input and Output columns and capture the IOType as `iotype` group.
-    [<LiteralAttribute>]
-    let IOTypePattern = @"(Input|Output)\s\[(?<iotype>.+)\]"
+    let IOTypePattern = $@"(Input|Output)\s\[(?<{MatchGroups.iotype}>.+)\]"
 
     /// This pattern is used to match Input column and capture the IOType as `iotype` group.
-    [<LiteralAttribute>]
-    let InputPattern = @"Input\s\[(?<iotype>.+)\]"
+    let InputPattern = $@"Input\s\[(?<{MatchGroups.iotype}>.+)\]"
 
     /// This pattern is used to match Output column and capture the IOType as `iotype` group.
-    [<LiteralAttribute>]
-    let OutputPattern = @"Output\s\[(?<iotype>.+)\]"
+    let OutputPattern = $@"Output\s\[(?<{MatchGroups.iotype}>.+)\]"
 
     /// This pattern matches any column header starting with some text, followed by one whitespace and a term name inside squared brackets.
     ///
@@ -176,9 +183,9 @@ module ActivePatterns =
     let (|TermAnnotationShort|_|) input =
         match input with
         | Regex Pattern.TermAnnotationShortPattern value ->
-            let termsourceref = value.Groups.["termsourceref"].Value
-            let localtan = value.Groups.["localtan"].Value
-            {|TermSourceREF = termsourceref; LocalTAN = localtan|}
+            let idspace = value.Groups.[Pattern.MatchGroups.idspace].Value
+            let localID = value.Groups.[Pattern.MatchGroups.localID].Value
+            {|IDSpace = idspace; LocalID = localID|}
             |> Some
         | _ ->
             None
@@ -193,9 +200,9 @@ module ActivePatterns =
         | Regex Pattern.TermAnnotationShortPattern value 
         | Regex Pattern.TermAnnotationURIPattern value 
         | Regex Pattern.TermAnnotationURIPattern_lessRestrictive value ->
-            let termsourceref = value.Groups.["termsourceref"].Value
-            let localtan = value.Groups.["localtan"].Value
-            {|TermSourceREF = termsourceref; LocalTAN = localtan; TermAccessionNumber = input|}
+            let idspace = value.Groups.[Pattern.MatchGroups.idspace].Value
+            let localID = value.Groups.[Pattern.MatchGroups.localID].Value
+            {|IDSpace = idspace; LocalID = localID|}
             |> Some
         | _ ->
             None
@@ -208,7 +215,7 @@ module ActivePatterns =
         | Regex Pattern.TermSourceREFColumnPattern r ->
             match r.Groups.["id"].Value with
             | TermAnnotation r -> Some r 
-            | _ -> Some {|LocalTAN = ""; TermAccessionNumber = ""; TermSourceREF = ""|}
+            | _ -> Some {|IDSpace = ""; LocalID = ""|}
          | _ -> None
 
     /// Matches a "Term Accession Number (ShortTerm)" column header and returns the ShortTerm as Term Source Ref and Annotation Number.
@@ -219,21 +226,21 @@ module ActivePatterns =
         | Regex Pattern.TermAccessionNumberColumnPattern r ->
             match r.Groups.["id"].Value with
             | TermAnnotation r -> Some r 
-            | _ -> Some {|LocalTAN = ""; TermAccessionNumber = ""; TermSourceREF = ""|}
+            | _ -> Some {|IDSpace = ""; LocalID = ""|}
          | _ -> None
 
     /// Matches a "Input [InputType]" column header and returns the InputType as string.
     let (|InputColumnHeader|_|) input = 
         match input with
         | Regex Pattern.InputPattern r ->
-            Some r.Groups.["iotype"].Value          
+            Some r.Groups.[Pattern.MatchGroups.iotype].Value          
          | _ -> None
 
     /// Matches a "Output [OutputType]" column header and returns the OutputType as string.
     let (|OutputColumnHeader|_|) input =
         match input with
         | Regex Pattern.OutputPattern r ->
-            Some r.Groups.["iotype"].Value
+            Some r.Groups.[Pattern.MatchGroups.iotype].Value
         | _ -> None
 
 
@@ -261,9 +268,9 @@ let tryParseReferenceColumnHeader (str : string) =
 let tryParseTermAnnotationShort (str:string) =
     match str.Trim() with
     | Regex TermAnnotationShortPattern value ->
-        let termsourceref = value.Groups.["termsourceref"].Value
-        let localtan = value.Groups.["localtan"].Value
-        {|TermSourceREF = termsourceref; LocalTAN = localtan|} 
+        let idspace = value.Groups.[Pattern.MatchGroups.idspace].Value
+        let localid = value.Groups.[Pattern.MatchGroups.localID].Value
+        {|IDSpace = idspace; LocalID = localid|} 
         |> Some
     | _ -> None
 
@@ -279,17 +286,17 @@ let tryParseTermAnnotation (str:string) =
     | Regex TermAnnotationShortPattern value 
     | Regex TermAnnotationURIPattern value 
     | Regex TermAnnotationURIPattern_lessRestrictive value ->
-        let termsourceref = value.Groups.["termsourceref"].Value
-        let localtan = value.Groups.["localtan"].Value
-        {|TermSourceREF = termsourceref; LocalTAN = localtan|}
+        let idspace = value.Groups.[Pattern.MatchGroups.idspace].Value
+        let localid = value.Groups.[Pattern.MatchGroups.localID].Value
+        {|IDSpace = idspace; LocalID = localid|}
         |> Some
     | _ ->
         None
 
 /// Tries to parse 'str' to term accession and returns it in the format `Some "termsourceref:localtan"`. Exmp.: `Some "MS:000001"`
 let tryGetTermAnnotationShortString (str:string) = 
     tryParseTermAnnotation str
-    |> Option.map (fun r -> r.TermSourceREF + ":" + r.LocalTAN)
+    |> Option.map (fun r -> r.IDSpace + ":" + r.LocalID)
 
 /// Parses 'str' to term accession and returns it in the format "termsourceref:localtan". Exmp.: "MS:000001"
 let getTermAnnotationShortString (str:string) =
@@ -320,7 +327,7 @@ let tryParseIOTypeHeader (headerStr: string) =
     match headerStr.Trim() with
     | Regex IOTypePattern value ->
         // remove quotes at beginning and end of matched string
-        let numberFormat = value.Groups.["iotype"].Value
+        let numberFormat = value.Groups.[Pattern.MatchGroups.iotype].Value
         Some numberFormat
     | _ ->
         None