Merge pull request #48 from nfdi4plants/ARCtrl

Add process graph tokenization
nfdi4plants · Jan 9, 2024 · 5116014 · 5116014
2 parents cd23da6 + 5798f6a
commit 5116014
Show file tree

Hide file tree

Showing 32 changed files with 1,707 additions and 789 deletions.
diff --git a/build/ProjectInfo.fs b/build/ProjectInfo.fs
@@ -76,6 +76,7 @@ let structuralOntologySources = [
     "src/ARCTokenization/structural_ontologies/investigation_metadata_structural_ontology.yml"
     "src/ARCTokenization/structural_ontologies/study_metadata_structural_ontology.yml"
     "src/ARCTokenization/structural_ontologies/assay_metadata_structural_ontology.yml"
+    "src/ARCTokenization/structural_ontologies/arc_process_graph_structural_ontology.yml"
     "src/ARCTokenization/structural_ontologies/arc_file_structure_ontology.yml"
 ]
 

diff --git a/src/ARCTokenization/ARCMock.fs b/src/ARCTokenization/ARCMock.fs
@@ -1,6 +1,8 @@
 namespace ARCTokenization
 
 open ControlledVocabulary
+open ARCtrl
+open ARCtrl.ISA
 
 type ARCMock =
 
@@ -418,4 +420,26 @@ type ARCMock =
                 else
                     CvParam(term, ParamValue.Value v, [])
             )
-        )
+        )
+
+    static member ProcessGraphColumn(
+        header: ARCtrl.ISA.CompositeHeader,
+        cells: seq<ARCtrl.ISA.CompositeCell>
+    ) =
+        CompositeColumn.create(header, cells |> Array.ofSeq)
+        |> Tokenization.ARCtrl.CompositeColumn.tokenize
+
+    static member ProcessGraph(
+        columns: seq<ARCtrl.ISA.CompositeHeader *  seq<ARCtrl.ISA.CompositeCell>>
+    ) =
+        let table = ArcTable.create("", new ResizeArray<_>(), new System.Collections.Generic.Dictionary<_,_>())
+
+        columns
+        |> Seq.map (fun (headerTerm, cells) -> 
+            CompositeColumn.create(headerTerm, cells |> Array.ofSeq)
+        )
+        |> Array.ofSeq
+        |> table.AddColumns
+
+        table
+        |> Tokenization.ARCtrl.ARCTable.tokenizeColumns
diff --git a/src/ARCTokenization/ARCTokenization.fsproj b/src/ARCTokenization/ARCTokenization.fsproj
@@ -19,14 +19,17 @@
   <ItemGroup>
     <None Include="RELEASE_NOTES.md" />
     <Compile Include="structural_ontologies\AFSO.fs" />
+    <Compile Include="structural_ontologies\APGSO.fs" />
     <Compile Include="structural_ontologies\INVMSO.fs" />
     <Compile Include="structural_ontologies\STDMSO.fs" />
     <Compile Include="structural_ontologies\ASSMSO.fs" />
     <EmbeddedResource Include="structural_ontologies/arc_file_structure_ontology.obo" CopyToOutputDirectory="Always" Pack="true" PackagePath="structural_ontologies" />
+    <EmbeddedResource Include="structural_ontologies/arc_process_graph_structural_ontology.obo" CopyToOutputDirectory="Always" Pack="true" PackagePath="structural_ontologies" />
     <EmbeddedResource Include="structural_ontologies/investigation_metadata_structural_ontology.obo" CopyToOutputDirectory="Always" Pack="true" PackagePath="structural_ontologies" />
     <EmbeddedResource Include="structural_ontologies/study_metadata_structural_ontology.obo" CopyToOutputDirectory="Always" Pack="true" PackagePath="structural_ontologies" />
     <EmbeddedResource Include="structural_ontologies/assay_metadata_structural_ontology.obo" CopyToOutputDirectory="Always" Pack="true" PackagePath="structural_ontologies" />
     <None Include="structural_ontologies/arc_file_structure_ontology.yml" />
+    <None Include="structural_ontologies/arc_process_graph_structural_ontology.yml" />
     <None Include="structural_ontologies/investigation_metadata_structural_ontology.yml" />
     <None Include="structural_ontologies/study_metadata_structural_ontology.yml" />
     <None Include="structural_ontologies/assay_metadata_structural_ontology.yml" />
@@ -60,9 +63,10 @@
   <ItemGroup>
     <ProjectReference Include="..\ControlledVocabulary\ControlledVocabulary.fsproj" PackageVersion="[1.0.0, 2.0.0)" />
     <PackageReference Include="FSharpAux.Core" Version="[2.0.0]" />
-    <PackageReference Include="FsOboParser" Version="[0.3.0]" />
-    <PackageReference Include="FsSpreadsheet" Version="[4.1.0]" />
-    <PackageReference Include="FsSpreadsheet.ExcelIO" Version="[4.1.0]" />
+    <PackageReference Include="OBO.NET" Version="[0.4.2]" />
+    <PackageReference Include="FsSpreadsheet" Version="[5.0.2]" />
+    <PackageReference Include="FsSpreadsheet.ExcelIO" Version="[5.0.2]" />
+    <PackageReference Include="ARCtrl" Version="[1.0.1]" />
     <PackageReference Include="Microsoft.SourceLink.GitHub" Version="1.1.1" PrivateAssets="All" />
   </ItemGroup>