Return sequences from set-backed readers

Author: kMutagene

src/BioFSharp.IO.INSDC/Analysis.fs

@@ -6,13 +6,13 @@ open BioFSharp.IO.INSDC.Internal
 /// Read and write INSDC Analysis records.
 module Analysis =
 
-    /// Read an INSDC Analysis XML record from the file at `filePath`.
-    let read (filePath: string) : Analysis =
-        XmlSerializer.read<Analysis> filePath
+    /// Read INSDC Analysis XML records from the file at `filePath`.
+    let read (filePath: string) : seq<Analysis> =
+        XmlSerializer.readOrSet<Analysis, AnalysisSet> "ANALYSIS_SET" (fun set -> set.Analysis) filePath
 
-    /// Parse an INSDC Analysis XML record from an in-memory string.
-    let readString (xml: string) : Analysis =
-        XmlSerializer.readString<Analysis> xml
+    /// Parse INSDC Analysis XML records from an in-memory string.
+    let readString (xml: string) : seq<Analysis> =
+        XmlSerializer.readStringOrSet<Analysis, AnalysisSet> "ANALYSIS_SET" (fun set -> set.Analysis) xml
 
     /// Write an INSDC Analysis `analysis` to the file at `filePath` as XML.
     let write (filePath: string) (analysis: Analysis) : unit =

src/BioFSharp.IO.INSDC/BioProject.fs

@@ -8,13 +8,13 @@ open BioFSharp.IO.INSDC.Internal
 /// this library).
 module BioProject =
 
-    /// Read an INSDC BioProject XML record from the file at `filePath`.
-    let read (filePath: string) : BioProject =
-        XmlSerializer.read<BioProject> filePath
+    /// Read INSDC BioProject XML records from the file at `filePath`.
+    let read (filePath: string) : seq<BioProject> =
+        XmlSerializer.readOrSet<BioProject, BioProjectSet> "PROJECT_SET" (fun set -> set.Project) filePath
 
-    /// Parse an INSDC BioProject XML record from an in-memory string.
-    let readString (xml: string) : BioProject =
-        XmlSerializer.readString<BioProject> xml
+    /// Parse INSDC BioProject XML records from an in-memory string.
+    let readString (xml: string) : seq<BioProject> =
+        XmlSerializer.readStringOrSet<BioProject, BioProjectSet> "PROJECT_SET" (fun set -> set.Project) xml
 
     /// Write an INSDC BioProject `project` to the file at `filePath` as XML.
     let write (filePath: string) (project: BioProject) : unit =

src/BioFSharp.IO.INSDC/BioSample.fs

@@ -8,13 +8,13 @@ open BioFSharp.IO.INSDC.Internal
 /// this library).
 module BioSample =
 
-    /// Read an INSDC BioSample XML record from the file at `filePath`.
-    let read (filePath: string) : BioSample =
-        XmlSerializer.read<BioSample> filePath
+    /// Read INSDC BioSample XML records from the file at `filePath`.
+    let read (filePath: string) : seq<BioSample> =
+        XmlSerializer.readOrSet<BioSample, BioSampleSet> "SAMPLE_SET" (fun set -> set.Sample) filePath
 
-    /// Parse an INSDC BioSample XML record from an in-memory string.
-    let readString (xml: string) : BioSample =
-        XmlSerializer.readString<BioSample> xml
+    /// Parse INSDC BioSample XML records from an in-memory string.
+    let readString (xml: string) : seq<BioSample> =
+        XmlSerializer.readStringOrSet<BioSample, BioSampleSet> "SAMPLE_SET" (fun set -> set.Sample) xml
 
     /// Write an INSDC BioSample `sample` to the file at `filePath` as XML.
     let write (filePath: string) (sample: BioSample) : unit =

src/BioFSharp.IO.INSDC/Experiment.fs

@@ -6,13 +6,13 @@ open BioFSharp.IO.INSDC.Internal
 /// Read and write INSDC Experiment records.
 module Experiment =
 
-    /// Read an INSDC Experiment XML record from the file at `filePath`.
-    let read (filePath: string) : Experiment =
-        XmlSerializer.read<Experiment> filePath
+    /// Read INSDC Experiment XML records from the file at `filePath`.
+    let read (filePath: string) : seq<Experiment> =
+        XmlSerializer.readOrSet<Experiment, ExperimentSet> "EXPERIMENT_SET" (fun set -> set.Experiment) filePath
 
-    /// Parse an INSDC Experiment XML record from an in-memory string.
-    let readString (xml: string) : Experiment =
-        XmlSerializer.readString<Experiment> xml
+    /// Parse INSDC Experiment XML records from an in-memory string.
+    let readString (xml: string) : seq<Experiment> =
+        XmlSerializer.readStringOrSet<Experiment, ExperimentSet> "EXPERIMENT_SET" (fun set -> set.Experiment) xml
 
     /// Write an INSDC Experiment `experiment` to the file at `filePath` as XML.
     let write (filePath: string) (experiment: Experiment) : unit =

src/BioFSharp.IO.INSDC/Internal/XmlSerializer.fs

@@ -3,6 +3,7 @@ namespace BioFSharp.IO.INSDC.Internal
 open System
 open System.Collections.Concurrent
 open System.IO
+open System.Xml.Linq
 open System.Xml.Serialization
 
 /// Internal helpers wrapping `System.Xml.Serialization.XmlSerializer` for the
@@ -27,6 +28,24 @@ module XmlSerializer =
         use reader = new StringReader(xml)
         (getSerializer typeof<'T>).Deserialize(reader) :?> 'T
 
+    /// Deserialize INSDC records from XML that may either use the single
+    /// record root (`RUN`, `STUDY`, ...) or the ENA API's set root
+    /// (`RUN_SET`, `STUDY_SET`, ...).
+    let readStringOrSet<'T, 'TSet> (setRootName: string) (select: 'TSet -> seq<'T>) (xml: string) : seq<'T> =
+        let rootName = XDocument.Parse(xml).Root.Name.LocalName
+
+        if rootName = setRootName then
+            xml |> readString<'TSet> |> select
+        else
+            xml |> readString<'T> |> Seq.singleton
+
+    /// Deserialize INSDC records from a file that may either use the single
+    /// record root (`RUN`, `STUDY`, ...) or the ENA API's set root
+    /// (`RUN_SET`, `STUDY_SET`, ...).
+    let readOrSet<'T, 'TSet> (setRootName: string) (select: 'TSet -> seq<'T>) (filePath: string) : seq<'T> =
+        File.ReadAllText(filePath)
+        |> readStringOrSet<'T, 'TSet> setRootName select
+
     /// Serialize an INSDC record `value` to the file at `filePath`.
     let write<'T> (filePath: string) (value: 'T) : unit =
         use stream = File.Create(filePath)

src/BioFSharp.IO.INSDC/Run.fs

@@ -6,13 +6,13 @@ open BioFSharp.IO.INSDC.Internal
 /// Read and write INSDC Run records.
 module Run =
 
-    /// Read an INSDC Run XML record from the file at `filePath`.
-    let read (filePath: string) : Run =
-        XmlSerializer.read<Run> filePath
+    /// Read INSDC Run XML records from the file at `filePath`.
+    let read (filePath: string) : seq<Run> =
+        XmlSerializer.readOrSet<Run, RunSet> "RUN_SET" (fun set -> set.Run) filePath
 
-    /// Parse an INSDC Run XML record from an in-memory string.
-    let readString (xml: string) : Run =
-        XmlSerializer.readString<Run> xml
+    /// Parse INSDC Run XML records from an in-memory string.
+    let readString (xml: string) : seq<Run> =
+        XmlSerializer.readStringOrSet<Run, RunSet> "RUN_SET" (fun set -> set.Run) xml
 
     /// Write an INSDC Run `run` to the file at `filePath` as XML.
     let write (filePath: string) (run: Run) : unit =

src/BioFSharp.IO.INSDC/Study.fs

@@ -6,13 +6,13 @@ open BioFSharp.IO.INSDC.Internal
 /// Read and write INSDC Study records.
 module Study =
 
-    /// Read an INSDC Study XML record from the file at `filePath`.
-    let read (filePath: string) : Study =
-        XmlSerializer.read<Study> filePath
+    /// Read INSDC Study XML records from the file at `filePath`.
+    let read (filePath: string) : seq<Study> =
+        XmlSerializer.readOrSet<Study, StudySet> "STUDY_SET" (fun set -> set.Study) filePath
 
-    /// Parse an INSDC Study XML record from an in-memory string.
-    let readString (xml: string) : Study =
-        XmlSerializer.readString<Study> xml
+    /// Parse INSDC Study XML records from an in-memory string.
+    let readString (xml: string) : seq<Study> =
+        XmlSerializer.readStringOrSet<Study, StudySet> "STUDY_SET" (fun set -> set.Study) xml
 
     /// Write an INSDC Study `study` to the file at `filePath` as XML.
     let write (filePath: string) (study: Study) : unit =

src/BioFSharp.IO.INSDC/Submission.fs

@@ -6,13 +6,13 @@ open BioFSharp.IO.INSDC.Internal
 /// Read and write INSDC Submission records.
 module Submission =
 
-    /// Read an INSDC Submission XML record from the file at `filePath`.
-    let read (filePath: string) : Submission =
-        XmlSerializer.read<Submission> filePath
+    /// Read INSDC Submission XML records from the file at `filePath`.
+    let read (filePath: string) : seq<Submission> =
+        XmlSerializer.readOrSet<Submission, SubmissionSet> "SUBMISSION_SET" (fun set -> set.Submission) filePath
 
-    /// Parse an INSDC Submission XML record from an in-memory string.
-    let readString (xml: string) : Submission =
-        XmlSerializer.readString<Submission> xml
+    /// Parse INSDC Submission XML records from an in-memory string.
+    let readString (xml: string) : seq<Submission> =
+        XmlSerializer.readStringOrSet<Submission, SubmissionSet> "SUBMISSION_SET" (fun set -> set.Submission) xml
 
     /// Write an INSDC Submission `submission` to the file at `filePath` as XML.
     let write (filePath: string) (submission: Submission) : unit =