Merge pull request #5 from nfdi4plants/ro-crate

Include ro-crate and rework output format selection and handling

README.md

@@ -1,6 +1,8 @@
 # arc-export
 Automatic building of a Docker container for exporting ARCs to Arc.json
 
+# Setup
+
 ## local build
 ```shell
 docker build . -t arc-export
@@ -14,3 +16,38 @@ docker pull ghcr.io/nfdi4plants/arc-export:main
 
 docker run -v C:\Repos\ArcRepo:/arc ghcr.io/nfdi4plants/arc-export:main /arc-export -p arc
 ```
+
+# Basic usage
+
+## Output format Selection
+
+Output format can be specified with the `-f` flag, followed by one of the following:
+
+- `isa-json` will produce a `arc-isa.json` file
+- `rocrate-metadata` will produce a `arc-ro-crate-metadata.json` file
+- `summary-markdown` will produce a `arc-summary.md` file
+
+E.g. 
+
+```shell
+docker run -v C:\Repos\ArcRepo:/arc arc-export:latest /arc-export -p arc -f rocrate-metadata -f isa-json -f summary-markdown
+```
+
+will produce all three output files.
+
+## Help
+
+```cli
+USAGE: arc-export [--help] --arc-directory <path> [--out-directory <path>]
+                  [--output-format <isa-json|rocrate-metadata|summary-markdown>]
+
+OPTIONS:
+
+    --arc-directory, -p <path>
+                          Specify a directory that contains the arc to convert.
+    --out-directory, -o <path>
+                          Optional. Specify a output directory for the invenio metadata record.
+    --output-format, -f <isa-json|rocrate-metadata|summary-markdown>
+                          Optional. Specify the output format. Default is ISA-JSON.
+    --help                display this list of options.
+```

src/arc-export/CLIArgs.fs

@@ -3,9 +3,15 @@
 open Argu
 open System.IO
 
+type OutputFormat =
+    | ISA_Json
+    | ROCrate_Metadata
+    | Summary_Markdown
+
 type CliArguments =
     | [<Mandatory>][<AltCommandLine("-p")>] ARC_Directory of path:string
-    | [<AltCommandLine("-o")>] Out_Directory of path:string
+    | [<Unique>][<AltCommandLine("-o")>] Out_Directory of path:string
+    | [<AltCommandLine("-f")>] Output_Format of OutputFormat
 
     // to-do once we have a common schema, implement as optional flag to validate all generated json via schema
     //| [<AltCommandLine("-val")>] Validate 
@@ -15,6 +21,7 @@ type CliArguments =
             match s with
             | ARC_Directory _ -> "Specify a directory that contains the arc to convert."
             | Out_Directory _ -> "Optional. Specify a output directory for the invenio metadata record."
+            | Output_Format _ -> "Optional. Specify the output format. Default is ISA-JSON."
             // to-do once we have a common schema, implement as optional flag to validate all generated json via schema
             //| Validate -> "Optional. Validate the output against the metadata record schema"
 

src/arc-export/Program.fs

@@ -1,90 +1,44 @@
 open System.IO
 open ARCtrl
 open ARCtrl.FileSystem
-open ARCtrl.ISA
 open ARCtrl.NET
 open Argu
-open FileSystemTreeExtension
-open ARCExtension
-open ArcSummaryMarkdown
-open ARCtrl.NET.Contract
 
-let getAllFilePaths (directoryPath : string) =
-    let directoryPath = System.IO.Path.GetFullPath(directoryPath)
-    let rec allFiles dirs =
-        if Seq.isEmpty dirs then Seq.empty else
-            seq { yield! dirs |> Seq.collect Directory.EnumerateFiles
-                  yield! dirs |> Seq.collect Directory.EnumerateDirectories |> allFiles }
-
-    allFiles [directoryPath] 
-    |> Seq.toArray
-    |> Array.map System.IO.Path.GetFullPath
-    |> Array.map (fun p -> p.Replace(directoryPath, "").Replace("\\","/"))
-
-let loadARCCustom (arcPath : string) =
-
-    //// EINFACH DIESE ZEIELE AUSTAUSCHEN
-
-    let paths = getAllFilePaths arcPath
-
-    let arc = ARC.fromFilePaths paths
-
-    let contracts = arc.GetReadContracts()
-
-    let fulFilledContracts = 
-        contracts 
-        |> Array.map (fulfillReadContract arcPath)
-
-    arc.SetISAFromContracts(fulFilledContracts,true)
-    arc
 
 try
     let args = CLIArgs.cliArgParser.ParseCommandLine()
 
     let arcPath = args.GetResult(CLIArgs.ARC_Directory)
 
-    let outPath = 
+    let outDir = 
         args.TryGetResult(CLIArgs.Out_Directory)
         |>Option.defaultValue arcPath
 
-    let jsonFile = Path.Combine(outPath,"arc.json")
-
-    let mdfile = Path.Combine(outPath,"arc-summary.md")
-
-
-    let inv, mdContent = 
-        try 
-            let arc = loadARCCustom arcPath
-            let registeredPayload = 
-                try 
-                    arc.GetRegisteredPayload(IgnoreHidden = true)
-                with err ->
-                    printfn "Could not get payload content, defaulting to all filesystem entries."
-                    printfn "error: %s" err.Message
-                    arc.FileSystem.Tree
-
-            let inv = arc.ISA |> Option.get
-
-            getAllFilePaths arcPath |> Seq.iter (printfn "%s")
-
-            inv,
-            MARKDOWN_TEMPLATE
-                .Replace("[[ARC_TITLE]]", inv.Title |> Option.defaultValue "Untitled ARC")
-                .Replace("[[FILE_TREE]]", FileSystemTree.toMarkdownTOC registeredPayload)
-        with 
-        | err ->
+    let arc = 
+
+        try ARC.load arcPath with
+        | err -> 
             printfn "Could not read investigation, writing empty arc json."
-            let comment1 = Comment.fromString "Status" "Could not parse ARC"
-            let comment2 = Comment.fromString "ErrorMessage" $"Could not parse ARC:\n{err.Message}"
-            ArcInvestigation(Identifier.createMissingIdentifier() , comments = [|comment1;comment2|]),
-            "Could not read investigation, unable to display registered file content."
+            let comment1 = Comment("Status","Could not parse ARC")
+            let comment2 = Comment("ErrorMessage",$"Could not parse ARC:\n{err.Message}")
+            let fs = ARCtrl.NET.Path.getAllFilePaths arcPath |> FileSystemTree.fromFilePaths |> FileSystem.create
+            let inv = 
+                ArcInvestigation(Helper.Identifier.createMissingIdentifier() , comments = ResizeArray [|comment1;comment2|])
+            let arc = ARC(inv,fs = fs)
+            arc
+
+    let outputFormats = args.GetResults(CLIArgs.Output_Format)
 
+    //args.Contains(CLIArgs.Output_Format)
+
+    if outputFormats |> List.contains CLIArgs.OutputFormat.ISA_Json || List.isEmpty outputFormats then
+        Writers.write_isa_json outDir arc
 
-    File.WriteAllText(mdfile, mdContent)
+    if outputFormats |> List.contains CLIArgs.OutputFormat.ROCrate_Metadata then
+        Writers.write_ro_crate_metadata outDir arc
 
-    inv
-    |> ARCtrl.ISA.Json.ArcInvestigation.toString
-    |> fun json -> File.WriteAllText(jsonFile, json)
+    if outputFormats |> List.contains CLIArgs.OutputFormat.Summary_Markdown then
+        Writers.write_arc_summary_markdown outDir arc
 
 with
     | :? ArguParseException as ex ->

src/arc-export/Writers.fs

@@ -0,0 +1,37 @@
+module Writers
+
+open ARCtrl
+open ARCtrl.Json
+open System.IO
+open ARCtrl.FileSystem
+open ArcSummaryMarkdown
+
+[<Literal>]
+let ro_crate_metadata_filename = "arc-ro-crate-metadata.json"
+
+[<Literal>]
+let isa_json_filename = "arc-isa.json"
+
+[<Literal>]
+let arc_summary_markdown_filename = "arc-summary.md"
+
+let write_ro_crate_metadata (outDir: string) (arc: ARC) =
+    let ro_crate_metadata = ARC.toROCrateJsonString(2) arc
+    let ro_crate_metadata_path = Path.Combine(outDir, ro_crate_metadata_filename)
+    File.WriteAllText(ro_crate_metadata_path, ro_crate_metadata)
+
+let write_isa_json (outDir: string) (arc: ARC) =
+    let inv = arc.ISA |> Option.get
+    let isa_json = inv.ToROCrateJsonString(2)
+    let isa_json_path = Path.Combine(outDir, isa_json_filename)
+    File.WriteAllText(isa_json_path, isa_json)
+
+let write_arc_summary_markdown (outDir: string) (arc: ARC) =
+    let inv = arc.ISA |> Option.get
+    let registeredPayload = arc.GetRegisteredPayload(IgnoreHidden = true)
+    let markdownContent =
+        MARKDOWN_TEMPLATE
+            .Replace("[[ARC_TITLE]]", inv.Title |> Option.defaultValue "Untitled ARC")
+            .Replace("[[FILE_TREE]]", FileSystemTree.toMarkdownTOC registeredPayload)
+    let arc_summary_markdown_path = Path.Combine(outDir, arc_summary_markdown_filename)
+    File.WriteAllText(arc_summary_markdown_path, markdownContent)

src/arc-export/arc-export.fsproj

@@ -10,15 +10,14 @@
     <Compile Include="Defaults.fs" />
     <Compile Include="CLIArgs.fs" />
     <Compile Include="FileSystemTreeExtension.fs" />
-    <Compile Include="ARCExtension.fs" />
     <Compile Include="ArcSummaryMarkdown.fs" />
+    <Compile Include="Writers.fs" />
     <Compile Include="Program.fs" />
   </ItemGroup>
 
   <ItemGroup>
-    <PackageReference Include="ARCtrl" Version="[1.0.0-alpha3]" />
-    <PackageReference Include="ARCtrl.NET" Version="[1.0.0-alpha1]" />
-    <PackageReference Include="Argu" Version="[6.1.1]" />
+    <PackageReference Include="ARCtrl.NET" Version="2.0.0-beta.1" />
+    <PackageReference Include="Argu" Version="6.2.4" />
     <PackageReference Include="JsonDSL" Version="[0.1.0]" />
     <PackageReference Update="FSharp.Core" Version="7.*" />
   </ItemGroup>