wip more restructuring

playground.fsx

@@ -9,6 +9,7 @@
 
 #r "nuget: DocumentFormat.OpenXml"
 #r "nuget: FSharpAux"
+#r "nuget: FsOboParser"
 #r "nuget: FsSpreadsheet, 3.1.1"
 #r "nuget: FsSpreadsheet.ExcelIO, 3.1.1"
 #r "nuget: FSharp.FGL"
@@ -21,7 +22,6 @@ open FSharp.FGL.ArrayAdjacencyGraph
 open System.Collections.Generic
 
 
-
 //#r "c:/repos/csbiology/fsspreadsheet/src/FsSpreadsheet/bin/Debug/netstandard2.0/FsSpreadsheet.dll"
 //#r "c:/repos/csbiology/fsspreadsheet/src/FsSpreadsheet.CsvIO/bin/Debug/netstandard2.0/FsSpreadsheet.CsvIO.dll"
 //#r "c:/repos/csbiology/fsspreadsheet/src/FsSpreadsheet.ExcelIO/bin/Debug/netstandard2.0/FsSpreadsheet.ExcelIO.dll"
@@ -34,20 +34,139 @@ open System.Collections.Generic
 
 open FsSpreadsheet
 open FsSpreadsheet.ExcelIO
+open FsOboParser
 //open FsSpreadsheet.DSL
 open ControlledVocabulary
 open ControlledVocabulary.ParamBase
 open ArcGraphModel
 
+
+let expectedTermValuesSimple = 
+    [
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""; "iid"]
+        [""; "ititle"]
+        [""; "idesc"]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""; "Maus"; "Keider"; "müller"; "oih"]
+        [""; "Oliver"; "andreas"]
+        [""; "L. I."; "C."]
+        [""; "[email protected]"]
+        [""]
+        [""]
+        [""]
+        [""; "Affe"]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""; "sid"]
+        [""; "stitle"]
+        [""; "sdesc"]
+        [""]
+        [""]
+        [""; "sid\isa.study.xlsx"]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""; "aid\isa.assay.xlsx"; "aid2\isa.assay.xlsx"]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""; "weil"]
+        [""; "lukas"]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+        [""]
+    ]
+
+let allExpectedMetadataTermsFull =
+    Terms.InvestigationMetadata.cvTerms
+    |> List.skip 1
+    |> List.zip expectedTermValuesSimple
+    |> List.map (fun (values,term) ->
+        values |> List.map ( fun v ->  CvParam(term, ParamValue.Value v, []))
+    )
+    |> List.concat
+
+let parsedInvestigationMetadataSimple = Investigation.parseMetadataSheetfromXlsxFile (__SOURCE_DIRECTORY__ + "/tests/ArcGraphModel.Tests/Fixtures/correct/investigation_simple.xlsx")
+
+let i_fs = FsWorkbook.fromXlsxFile (__SOURCE_DIRECTORY__ + "/tests/ArcGraphModel.Tests/Fixtures/correct/investigation_simple.xlsx")
+
+(FsWorkbook.getWorksheetByName "isa_investigation" i_fs).CellCollection.GetCells()
+|> Seq.filter(fun c -> c.RowNumber = 1)
+
+parsedInvestigationMetadataSimple
+|> List.skip 20
+|> List.take 20
+
 // Assay annotation table parsing
 
 let assayTokens = Assay.parseAnnotationTablesFromFile (__SOURCE_DIRECTORY__ + "/tests/ArcGraphModel.Tests/Fixtures/correct/assay_with_single_characteristics.xlsx")
 
 // Investigation metadata parsing
 
+let investigationTokens = Investigation.parseMetadataSheetfromXlsxFile (__SOURCE_DIRECTORY__ + "/tests/ArcGraphModel.Tests/Fixtures/correct/full_investigation_mkay.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 inves = FsWorkbook.fromXlsxFile (__SOURCE_DIRECTORY__ + "/tests/ArcGraphModel.Tests/Fixtures/isa.investigation.xlsx")
+let inves = FsWorkbook.fromXlsxFile (__SOURCE_DIRECTORY__ + "/tests/ArcGraphModel.Tests/Fixtures/correct/full_investigation_mkay.xlsx")
 
 let invesWs = FsWorkbook.getWorksheets inves |> Seq.head
 invesWs.RescanRows()

src/ArcGraphModel/ArcGraphModel.fsproj

@@ -16,14 +16,20 @@
     <None Include="structural_ontologies/assay_metadata_structural_ontology.yml" />
 
     <Compile Include="Address.fs" />
-    <Compile Include="ISA\Terms.fs" />
-    <Compile Include="ISA\AnnotationTable.fs" />
-    <Compile Include="ISA\Regex.fs" />
-    <Compile Include="ISA\KeyParser.fs" />
-    <Compile Include="ISA\Tokenization.fs" />
-    <Compile Include="ISA\TokenAggregation.fs" />
-    <Compile Include="ISA\Worksheet.fs" />
-    <Compile Include="ISA\TopLevelParsers.fs" />
+
+    <Compile Include="Terms.fs" />
+
+    <Compile Include="Regex.fs" />
+
+    <Compile Include="AnnotationTable.fs" />
+
+    <Compile Include="MetadataSheet.fs" />
+
+    <Compile Include="Tokenization.fs" />
+
+    <Compile Include="Workbook.fs" />
+    <Compile Include="Worksheet.fs" />
+    <Compile Include="TopLevelParsers.fs" />
   </ItemGroup>
 
 
@@ -38,9 +44,6 @@
     <RepositoryType>git</RepositoryType>
   </PropertyGroup>
 
-
-  <ItemGroup />
-
   <ItemGroup>
     <ProjectReference Include="..\ControlledVocabulary\ControlledVocabulary.fsproj" PackageVersion="[0.0.0, 1.0.0)" />
     <PackageReference Include="FSharp.FGL.ArrayAdjacencyGraph" Version="[0.0.2]" />
@@ -50,6 +53,10 @@
     <PackageReference Include="FsSpreadsheet.ExcelIO" Version="[3.1.1]" />
   </ItemGroup>
 
+  <ItemGroup>
+    <Folder Include="ISA\" />
+  </ItemGroup>
+
   <Target Name="UseExplicitPackageVersions" BeforeTargets="GenerateNuspec">
     <ItemGroup>
       <_ProjectReferenceWithExplicitPackageVersion Include="@(ProjectReference->'%(FullPath)')" Condition="'%(ProjectReference.PackageVersion)' != ''" />