Merge pull request #9 from CSBiology/feature-minimalParser-#5

#5 Minimal parser

playground.fsx

@@ -2,6 +2,7 @@
 #r "FsOboParser.dll"
 
 #r "nuget: IsaDotNet"
+//#r "nuget: FsOboParser"
 
 open FsOboParser
 
@@ -10,8 +11,29 @@ open System.IO
 
 let testPath = Path.Combine(__SOURCE_DIRECTORY__,  "./../../nfdi4plants/arc-validate/ErrorClassOntology.obo")
 
+//OboEntries.fromFile true testPath
 let testOntology = OboOntology.fromFile true testPath
 
+let testTerms = [
+    OboTerm.Create("test:000", Name = "test0")
+    OboTerm.Create("test:001", Name = "test1a", IsA = ["test:000"])
+    OboTerm.Create("test:002", Name = "test2", IsA = ["test:001"])
+    OboTerm.Create("test:003", Name = "test1b", IsA = ["test:000"])
+]
+
+let testOntology = OboOntology.create testTerms []
+
+testOntology.GetChildOntologyAnnotations(testTerms.Head.Id)
+testOntology.GetChildOntologyAnnotations(testTerms.Head.Id, Depth = 1)
+testOntology.GetChildOntologyAnnotations(testTerms.Head.Id, Depth = 2)
+
+let performanceTerms = List.init 7000000 (fun i -> OboTerm.Create($"lol:{i}"))
+let performanceOboOntology = OboOntology.create performanceTerms []
+OboOntology.toFile @"C:\Repos\CSBiology\FsOboParser\performanceOntology.obo" performanceOboOntology
+
+let x = OboOntology.fromFile false @"C:\Repos\CSBiology\FsOboParser\performanceOntology.obo"
+
+
 //let fileLines = File.ReadAllLines testPath
 
 //OboTerm.fromLines true ((fileLines |> Seq.ofArray).GetEnumerator()) 0

src/FsOboParser/FastOboGraph.fs

@@ -1,6 +1,7 @@
 namespace FsOboParser
 
-//open OboTerm
+
+//open type OboTerm
 
 
 ////########################################

src/FsOboParser/FsOboParser.fsproj

@@ -17,6 +17,8 @@
     <Compile Include="TermSynonym.fs" />
     <Compile Include="OboTerm.fs" />
     <Compile Include="OboTypeDef.fs" />
+    <Compile Include="OboEntry.fs" />
+    <Compile Include="OboEntries.fs" />
     <Compile Include="OboOntology.fs" />
     <Compile Include="FastOboGraph.fs" />
   </ItemGroup>
@@ -35,7 +37,7 @@
   </PropertyGroup>
 
   <ItemGroup>
-    <PackageReference Include="Microsoft.SourceLink.GitHub" Version="1.1.1" PrivateAssets="All"/>
+    <PackageReference Include="Microsoft.SourceLink.GitHub" Version="1.1.1" PrivateAssets="All" />
     <PackageReference Include="ISADotNet" Version="0.6.1" />
   </ItemGroup>
 

src/FsOboParser/OboEntries.fs

@@ -0,0 +1,30 @@
+namespace FsOboParser
+
+
+/// Functions for working with OboEntries.
+module OboEntries =
+
+    /// Reads a collection of strings and parses them into a list of OboEntries.
+    let fromLines verbose (input : seq<string>) =
+
+        let en = input.GetEnumerator()
+        let rec loop (en : System.Collections.Generic.IEnumerator<string>) entries lineNumber =
+
+            match en.MoveNext() with
+            | true ->
+                match (en.Current |> DBXref.trimComment) with
+                | "[Term]" -> 
+                    let lineNumber, parsedTerm = OboTerm.fromLines verbose en lineNumber "" "" false [] "" "" [] [] [] [] [] [] [] [] false [] [] [] false "" ""
+                    loop en (Term parsedTerm :: entries) lineNumber
+                | "[Typedef]" -> 
+                    let lineNumber, parsedTypeDef = OboTypeDef.fromLines verbose en lineNumber "" "" "" "" [] [] false false false false false false false
+                    loop en (TypeDef parsedTypeDef :: entries) lineNumber
+                | _ -> loop en entries (lineNumber + 1)
+            | false -> entries
+
+        loop en [] 1
+
+    /// Reads an OBO file and returns a list of OboEntries.
+    let fromFile verbose filepath =
+        System.IO.File.ReadAllLines filepath
+        |> fromLines verbose

src/FsOboParser/OboEntry.fs

@@ -0,0 +1,7 @@
+namespace FsOboParser
+
+
+/// Model of raw OboEntries, divided into Terms (as `OboTerm`s) and TypeDefs (as `OboTypeDef`s).
+type OboEntry =
+    | Term of OboTerm
+    | TypeDef of OboTypeDef

src/FsOboParser/OboOntology.fs

@@ -33,10 +33,10 @@ type OboOntology =
             | true ->             
                 match (en.Current |> trimComment) with
                 | "[Term]"    -> 
-                    let lineNumber,parsedTerm = (OboTerm.fromLines verbose en lineNumber "" "" false [] "" "" [] [] [] [] [] [] [] [] false [] [] [] false "" "")
+                    let lineNumber,parsedTerm = OboTerm.fromLines verbose en lineNumber "" "" false [] "" "" [] [] [] [] [] [] [] [] false [] [] [] false "" ""
                     loop en (parsedTerm :: terms) typedefs lineNumber
                 | "[Typedef]" -> 
-                    let lineNumber,parsedTypeDef = (OboTypeDef.fromLines verbose en lineNumber "" "" "" "" [] [] false false false false false false false)
+                    let lineNumber,parsedTypeDef = OboTypeDef.fromLines verbose en lineNumber "" "" "" "" [] [] false false false false false false false
                     loop en terms (parsedTypeDef :: typedefs) lineNumber
                 | _ -> loop en terms typedefs (lineNumber + 1)
             | false -> OboOntology.create (List.rev terms) (List.rev typedefs)
@@ -48,6 +48,21 @@ type OboOntology =
         System.IO.File.ReadAllLines path
         |> OboOntology.fromLines verbose
 
+    /// Takes a list of OboEntries and returns the OboOntology based on it.
+    static member fromOboEntries entries =
+
+        let rec loop terms typedefs entries =
+            match entries with
+            | h :: t ->
+                match h with
+                | Term term         -> loop (term :: terms) typedefs t
+                | TypeDef typedef   -> loop terms (typedef :: typedefs) t
+            | [] -> terms, typedefs
+
+        let terms, typedefs = loop [] [] entries
+
+        OboOntology.create terms typedefs
+
     /// Writes an OBO Ontology to term and type def stanzas in line form.
     static member toLines (oboOntology : OboOntology) =
         seq {

src/FsOboParser/OboTypeDef.fs

@@ -5,7 +5,7 @@ open DBXref
 open System
 
 
-/// Models the relationship between OBO Terms 
+/// Models the relationship between OBO Terms.
 type OboTypeDef = 
     {
         ///The unique id of the current term.