Update key parser and worksheet parsers

nfdi4plants · Jul 17, 2023 · 4623a46 · 4623a46
1 parent 2a49bfd
commit 4623a46
Show file tree

Hide file tree

Showing 3 changed files with 73 additions and 23 deletions.
diff --git a/playground.fsx b/playground.fsx
@@ -9,8 +9,8 @@
 
 #r "nuget: DocumentFormat.OpenXml"
 #r "nuget: FSharpAux"
-#r "nuget: FsSpreadsheet, 1.2.0-preview"
-#r "nuget: FsSpreadsheet.ExcelIO, 1.2.0-preview"
+#r "nuget: FsSpreadsheet, 1.3.0-preview"
+#r "nuget: FsSpreadsheet.ExcelIO, 1.3.0-preview"
 #r "nuget: FSharp.FGL"
 #r "nuget: FSharp.FGL.ArrayAdjacencyGraph"
 
@@ -40,16 +40,65 @@ open ArcType
 
 
 
-let inves = FsWorkbook.fromXlsxFile @"C:\Users\revil\OneDrive\CSB-Stuff\NFDI\testARC30\isa.investigation.xlsx"
+//let inves = FsWorkbook.fromXlsxFile @"C:\Users\revil\OneDrive\CSB-Stuff\NFDI\testARC30\isa.investigation.xlsx"
+let inves = FsWorkbook.fromXlsxFile @"C:\Users\olive\OneDrive\CSB-Stuff\NFDI\testARC30\isa.investigation.xlsx"
 
 let invesWs = FsWorkbook.getWorksheets inves |> Seq.head
 invesWs.RescanRows()
 invesWs.CellCollection
 invesWs.Rows
 
-let invesWsParsed = ArcGraphModel.IO.Worksheet.parseRowsAggregated invesWs
+//let invesWsParsed = ArcGraphModel.IO.Worksheet.parseRowsAggregated invesWs
+//let invesWsParsed = ArcGraphModel.IO.Worksheet.parseColumnsAggregated invesWs
 let invesWsParsed = ArcGraphModel.IO.Worksheet.parseRowsFlat invesWs
-
+//invesWsParsed |> Seq.cast<CvParam>
+invesWsParsed |> Seq.cast<IParamBase> |> Seq.cast<CvParam>
+//invesWsParsed |> List.map (fun x -> x :> CvParam)
+invesWsParsed
+|> List.take 10
+|> List.map (
+    fun x ->
+        match x with
+        | :? CvParam as p -> 
+            Param.getValue p
+            p.ToString()
+        | :? UserParam as p -> 
+            Param.getValue p
+            p.ToString()
+)
+
+invesWsParsed
+invesWsParsed |> List.iter (List.iter (printfn "%A"))
+invesWsParsed.Head |> printfn "%A"
+invesWsParsed.Head |> printfn "%O"
+(invesWsParsed.Head :?> CvAttributeCollection)
+invesWsParsed.Head.ToString()
+
+[1 .. 10] |> printfn "%A"
+[1 .. 10] |> printfn "%O"
+
+open System.Text.RegularExpressions
+
+let namePattern = @"(?<=\[).*(?=[\]])"
+let key = "Comment[<Investigation Person ORCID>]"
+Regex.tryParseValue namePattern key
+|> Option.map (fun n ->
+    CvTerm("",n,"")
+)
+|> Option.get
+|> fun term -> CvParam(term,ParamValue.Value "Hallo",[])
+
+let lol = CvParam("lol", "lol", "lol", ParamValue.Value "lol")
+let lol2 = CvParam("lol", "lol", "lol", ParamValue.Value "lol")
+
+lol = lol2
+
+let dict1 = Dictionary<string,string>()
+dict1.Add("kek", "lil")
+let dict2 = Dictionary<string,string>()
+dict2.Add("kek", "lil")
+
+dict1 = dict2
 
 
 // new CvTypes - testin' and foolin' around

diff --git a/src/ArcGraphModel.IO/ISA/KeyParser.fs b/src/ArcGraphModel.IO/ISA/KeyParser.fs
@@ -133,14 +133,14 @@ module KeyParser =
         let namePattern = @"(?<=\[).*(?=[\]])"
         Regex.tryParseValue namePattern key
         |> Option.map (fun n ->
-            "",n,""
+            CvTerm("",n,"")
         )
 
     let (|AnnotationID|_|) (key : string) : CvTerm Option =
         let namePattern = @"(?<=\[).*(?=[\]])"
         Regex.tryParseValue namePattern key
         |> Option.map (fun n ->
-            "",n,""
+            CvTerm("",n,"")
         )
 
     let (|UnMatchable|) (key : string) : string =
@@ -167,21 +167,22 @@ module KeyParser =
         | Token TokenBase.phone term -> 
             fun (pv) -> CvParam(term,pv,attributes)
 
-        | Attribute AttributeBase.person (attribute,key) 
-        | Attribute AttributeBase.investigation (attribute,key) 
-        | Attribute AttributeBase.study (attribute,key) 
-        | Attribute AttributeBase.assay (attribute,key) 
-        | Attribute AttributeBase.publication (attribute,key) 
-        | Attribute AttributeBase.factorType (attribute,key) 
-        | Attribute AttributeBase.designType (attribute,key)
-        | Attribute AttributeBase.parameter (attribute,key) 
-        | Attribute AttributeBase.characteristic (attribute,key) 
-        | Attribute AttributeBase.rawData (attribute,key) 
-        | Attribute AttributeBase.processedData (attribute,key)
-        | Attribute AttributeBase.factor (attribute,key) -> 
-            parseKey (attribute :: attributes) key
+        //| Attribute AttributeBase.person (attribute,key) 
+        //| Attribute AttributeBase.investigation (attribute,key) 
+        //| Attribute AttributeBase.study (attribute,key) 
+        //| Attribute AttributeBase.assay (attribute,key) 
+        //| Attribute AttributeBase.publication (attribute,key) 
+        //| Attribute AttributeBase.factorType (attribute,key) 
+        //| Attribute AttributeBase.designType (attribute,key)
+        //| Attribute AttributeBase.parameter (attribute,key) 
+        //| Attribute AttributeBase.characteristic (attribute,key) 
+        //| Attribute AttributeBase.rawData (attribute,key) 
+        //| Attribute AttributeBase.processedData (attribute,key)
+        //| Attribute AttributeBase.factor (attribute,key) -> 
+        //    parseKey (attribute :: attributes) key
 
         | StructuredName term ->
+            printfn "term is %O\tkey is %s" term key
             fun (pv) -> CvParam(term,pv,attributes)
 
         | UnMatchable name -> 

diff --git a/src/ArcGraphModel.IO/ISA/Worksheet.fs b/src/ArcGraphModel.IO/ISA/Worksheet.fs
@@ -24,7 +24,7 @@ module Worksheet =
 
     /// Parses rows of a given FsWorksheet via a given tokenization function and returns the resulting IAttributeCollection list.
     let parseRows tokenizationFunction (worksheet : FsWorksheet) = 
-        parseCells (worksheet.Rows |> List.map (fun x -> x.Cells)) tokenizationFunction worksheet
+        parseCells (worksheet.Rows) tokenizationFunction worksheet
 
     /// Parses rows of a given FsWorksheet and returns the resulting aggregated ICvBase list.
     let parseRowsAggregated (worksheet : FsWorksheet) = 
@@ -36,7 +36,7 @@ module Worksheet =
 
     /// Parses columns of a given FsWorksheet via a given tokenization function and returns the resulting IAttributeCollection list.
     let parseColumns tokenizationFunction (worksheet : FsWorksheet) = 
-        parseCells (Seq.toList worksheet.Columns |> List.map (fun x -> x.Cells)) tokenizationFunction worksheet
+        parseCells (Seq.toList worksheet.Columns) tokenizationFunction worksheet
 
     /// Parses columns of a given FsWorksheet and returns the resulting aggregated ICvBase list.
     let parseColumnsAggregated (worksheet : FsWorksheet) =
@@ -48,7 +48,7 @@ module Worksheet =
 
     /// Parses the columns of the first FsTable in a given FsWorksheet via a given tokenization function and returns the resulting IAttributeCollection list.
     let parseTableColumns tokenizationFunction (worksheet : FsWorksheet) = 
-        parseCells (worksheet.Tables.Head.Columns(worksheet.CellCollection) |> Seq.toList |> List.map (fun x -> x.Cells)) tokenizationFunction worksheet
+        parseCells (worksheet.Tables.Head.Columns(worksheet.CellCollection) |> Seq.toList) tokenizationFunction worksheet
 
     /// Parses the columns of the first FsTable in a given FsWorksheet and returns the resulting aggregated ICvBase list.
     let parseTableColumnsAggregated (worksheet : FsWorksheet) =