add new pipeline component

Rdmp.Dicom/Extraction/DicomTagToCSV.cs

@@ -0,0 +1,115 @@
+using Amazon.SSOOIDC.Internal;
+using CsvHelper;
+using FellowOakDicom;
+using NPOI.HPSF;
+using Org.BouncyCastle.Tls;
+using Rdmp.Core.Curation.Data;
+using Rdmp.Core.DataFlowPipeline;
+using Rdmp.Core.ReusableLibraryCode.Checks;
+using Rdmp.Core.ReusableLibraryCode.Progress;
+using Rdmp.Dicom.Extraction.FoDicomBased;
+using Rdmp.Dicom.Extraction.FoDicomBased.DirectoryDecisions;
+using System;
+using System.Collections.Generic;
+using System.Data;
+using System.IO;
+using System.Linq;
+using System.Text;
+using System.Threading.Tasks;
+
+namespace Rdmp.Dicom.Extraction
+{
+    public class DicomTagToCSV : IPluginDataFlowComponent<DataTable>
+    {
+        [DemandsInitialization("If the path filename contains relative file uris to images then this is the root directory")]
+        public string ArchiveRootIfAny { get; set; }
+
+        [DemandsInitialization("The column name in the extracted dataset which contains the location of the dicom files", Mandatory = true)]
+        public string RelativeArchiveColumnName { get; set; }
+
+        [DemandsInitialization("How many tries to allow for fetching the file. This setting may be useful on network drives or oversubscribed resources", DefaultValue = 0)]
+        public int FileFetchRetryLimit { get; set; }
+
+        [DemandsInitialization("How long to wait between file fetch retries in milliseconds.", DefaultValue = 100)]
+        public int FileFetchRetryTimeout { get; set; }
+
+        [DemandsInitialization("The number of errors (e.g. failed to find/anonymise file) to allow before abandoning the extraction", DefaultValue = 100)]
+        public int ErrorThreshold { get; set; }
+
+        private int _errors = 0;
+        private IPutDicomFilesInExtractionDirectories _putter;
+
+
+        public void Abort(IDataLoadEventListener listener)
+        {
+            //throw new NotImplementedException();
+        }
+
+        public void Check(ICheckNotifier notifier)
+        {
+            //throw new NotImplementedException();
+        }
+
+        public void Dispose(IDataLoadEventListener listener, Exception pipelineFailureExceptionIfAny)
+        {
+            //throw new NotImplementedException();
+        }
+
+        public DataTable ProcessPipelineData(DataTable toProcess, IDataLoadEventListener listener, GracefulCancellationToken cancellationToken)
+        {
+
+            var fileRows = new Dictionary<string, DataRow>();
+            var releaseIDs = new Dictionary<string, string>();
+            var dicomFiles = new List<(string, string)>();
+            foreach (DataRow processRow in toProcess.Rows)
+            {
+                var file = (string)processRow[RelativeArchiveColumnName];
+                fileRows.Add(file, processRow);
+                dicomFiles.Add((file, file));
+            }
+
+            var dicomFilePaths = new AmbiguousFilePath(ArchiveRootIfAny, dicomFiles).GetDataset(FileFetchRetryLimit, FileFetchRetryTimeout, listener);
+            foreach (var dcm in dicomFilePaths)
+            {
+                var filepath = Path.Combine("C:\\temp\\csvtest", dcm.Item1);
+                var sw = new StreamWriter(filepath);
+                using var w = new CsvWriter(sw, System.Globalization.CultureInfo.InvariantCulture);
+                w.WriteRecord(
+                dcm.Item2.Dataset.SelectMany(t => Entry.ProcessTag(dcm.Item1, t))
+                );
+            }
+
+            return toProcess;
+        }
+    }
+
+    internal class Entry
+    {
+        public string Id { get; }
+        public string Name { get; }
+        public string Value { get; }
+
+        public Entry(string id, string name, string value)
+        {
+            Id = id;
+            Name = name;
+            Value = value;
+        }
+
+        public static IEnumerable<Entry> ProcessTag(string id, DicomItem item)
+        {
+            return item switch
+            {
+                DicomAttributeTag aTag => aTag.Values.Select(v => new Entry(id, aTag.Tag.DictionaryEntry.Name, v.DictionaryEntry.Name)),
+                DicomStringElement s => StringEntries(id, s.Tag.DictionaryEntry.Name, s),
+                DicomSequence seq => seq.Items.SelectMany(ds => ds.SelectMany(i => ProcessTag(id, i))),
+                _ => new[] { new Entry(id, item.Tag.DictionaryEntry.Name, item.ToString()) }
+            };
+        }
+        private static IEnumerable<Entry> StringEntries(string id, string tag, DicomStringElement e)
+        {
+            for (int i = 0; i < e.Count; i++)
+                yield return new Entry(id, tag, e.Get<string>(i));
+        }
+    }
+}