Add a new filter - splitDataBySegments (#344)

src/Microsoft.Health.Fhir.Liquid.Converter.UnitTests/Filters/SegmentFiltersTests.cs

@@ -4,6 +4,7 @@
 // -------------------------------------------------------------------------------------------------
 
 using System;
+using System.Collections.Generic;
 using System.Linq;
 using Microsoft.Health.Fhir.Liquid.Converter.Models.Hl7v2;
 using Microsoft.Health.Fhir.Liquid.Converter.Parsers;
@@ -116,6 +117,65 @@ public void GivenAnHl7v2Data_WhenHasSegments_CorrectResultShouldBeReturned()
             Assert.Throws<NullReferenceException>(() => Filters.HasSegments(new Hl7v2Data(), null));
         }
 
+        [Fact]
+        public void GivenAnHl7v2Data_WhenSplitDataBySegments_CorrectResultShouldBeReturned()
+        {
+            List<Hl7v2Data> splitDataByOrc = Filters.SplitDataBySegments(TestData, "ORC");
+            Assert.Equal(4, splitDataByOrc.Count);
+            Assert.Equal(3, splitDataByOrc[0].Meta.Count);
+            Assert.Equal(7, splitDataByOrc[1].Meta.Count);
+            Assert.Equal(2, splitDataByOrc[2].Meta.Count);
+            Assert.Equal(2, splitDataByOrc[3].Meta.Count);
+            Assert.Equal(3, splitDataByOrc[0].Data.Count);
+            Assert.Equal(7, splitDataByOrc[1].Data.Count);
+            Assert.Equal(2, splitDataByOrc[2].Data.Count);
+            Assert.Equal(2, splitDataByOrc[3].Data.Count);
+            Assert.Equal("MSH", splitDataByOrc[0].Meta[0]);
+            Assert.Equal("ORC", splitDataByOrc[1].Meta[0]);
+            string expectedOrcValue = @"ORC|RE|4422^NIST-AA-IZ-2|13696^NIST-AA-IZ-2|||||||7824^Jackson^Lily^Suzanne^^^^^NIST-PI-1^L^^^PRN||654^Thomas^Wilma^Elizabeth^^^^^NIST-PI-1^L^^^MD|||||NISTEHRFAC^NISTEHRFacility^HL70362|";
+            Assert.Equal(expectedOrcValue, splitDataByOrc[1].Data[0].Value);
+
+            List<Hl7v2Data> splitDataByMsh = Filters.SplitDataBySegments(TestData, "MSH");
+            Assert.Equal(2, splitDataByMsh.Count);
+            Assert.Empty(splitDataByMsh[0].Meta);
+            Assert.Empty(splitDataByMsh[0].Data);
+            Assert.Equal(14, splitDataByMsh[1].Meta.Count);
+            Assert.Equal(14, splitDataByMsh[1].Data.Count);
+
+            List<Hl7v2Data> splitData = Filters.SplitDataBySegments(TestData, "PID|OBX");
+            Assert.Equal(6, splitData.Count);
+            Assert.Single(splitData[0].Meta);
+            Assert.Equal(5, splitData[1].Meta.Count);
+            Assert.Single(splitData[2].Meta);
+            Assert.Single(splitData[3].Meta);
+            Assert.Single(splitData[4].Meta);
+            Assert.Equal(5, splitData[5].Meta.Count);
+            Assert.Single(splitData[0].Data);
+            Assert.Equal(5, splitData[1].Data.Count);
+            Assert.Single(splitData[2].Data);
+            Assert.Single(splitData[3].Data);
+            Assert.Single(splitData[4].Data);
+            Assert.Equal(5, splitData[5].Data.Count);
+            Assert.Equal("MSH", splitData[0].Meta[0]);
+            Assert.Equal("PID", splitData[1].Meta[0]);
+            Assert.Equal("OBX", splitData[2].Meta[0]);
+            Assert.Equal(@"OBX|1|CE|30963-3^Vaccine Funding Source^LN|1|PHC70^Private^CDCPHINVS||||||F|||20150624", splitData[2].Data[0].Value);
+
+            splitData = Filters.SplitDataBySegments(TestData, "PID|||OBX|KKK|");
+            Assert.Equal(6, splitData.Count);
+            Assert.Equal("MSH", splitData[0].Meta[0]);
+            Assert.Equal("PID", splitData[1].Meta[0]);
+            Assert.Equal("OBX", splitData[2].Meta[0]);
+
+            // If separators are not in the data or empty, a data list containing only the original data will be returned.
+            Assert.StrictEqual(TestData, Filters.SplitDataBySegments(TestData, string.Empty)[0]);
+            Assert.StrictEqual(TestData, Filters.SplitDataBySegments(TestData, "PV1")[0]);
+
+            // Hl7v2Data and separators could not be null. If one of them is null, NullReferenceException will be thrown.
+            Assert.Throws<NullReferenceException>(() => Filters.SplitDataBySegments(null, "ORC"));
+            Assert.Throws<NullReferenceException>(() => Filters.SplitDataBySegments(TestData, null));
+        }
+
         private static Hl7v2Data LoadTestData()
         {
             var parser = new Hl7v2DataParser();

src/Microsoft.Health.Fhir.Liquid.Converter.UnitTests/Utilities/Hl7v2DataUtilityTests.cs

@@ -0,0 +1,41 @@
+// -------------------------------------------------------------------------------------------------
+// Copyright (c) Microsoft Corporation. All rights reserved.
+// Licensed under the MIT License (MIT). See LICENSE in the repo root for license information.
+// -------------------------------------------------------------------------------------------------
+
+using System;
+using Microsoft.Health.Fhir.Liquid.Converter.Utilities;
+using Xunit;
+
+namespace Microsoft.Health.Fhir.Liquid.Converter.UnitTests.Utilities
+{
+    public class Hl7v2DataUtilityTests
+    {
+        [Fact]
+        public void GivenValidMessage_WhenSplitMessageToSegments_CorrectResultShouldBeReturned()
+        {
+            string[] testMessage =
+            {
+                "\r\nMSH|^~\\&|test\r\nPID|test\r\n\r\nPD1|test\r\nPD1|test\r\n",
+                "\rMSH|^~\\&|test\rPID|test\r\rPD1|test\rPD1|test\r",
+                "\nMSH|^~\\&|test\nPID|test\n\nPD1|test\nPD1|test\n",
+                "\nMSH|^~\\&|test\r\nPID|test\r\r\nPD1|test\r\n\nPD1|test\r\n",
+            };
+
+            foreach (string message in testMessage)
+            {
+                string[] segments = Hl7v2DataUtility.SplitMessageToSegments(message);
+                Assert.Equal(4, segments.Length);
+                Assert.Equal("MSH|^~\\&|test", segments[0]);
+                Assert.Equal("PID|test", segments[1]);
+                Assert.Equal("PD1|test", segments[2]);
+                Assert.Equal("PD1|test", segments[3]);
+            }
+
+            Assert.Empty(Hl7v2DataUtility.SplitMessageToSegments(string.Empty));
+
+            // message could not be null
+            Assert.Throws<NullReferenceException>(() => Hl7v2DataUtility.SplitMessageToSegments(null));
+        }
+    }
+}

src/Microsoft.Health.Fhir.Liquid.Converter/Filters/SegmentFilters.cs

@@ -128,5 +128,37 @@ private static Dictionary<string, List<Hl7v2Segment>> GetSegmentListsInternal(Hl
 
             return result;
         }
+
+        public static List<Hl7v2Data> SplitDataBySegments(Hl7v2Data hl7v2Data, string segmentIdSeparators)
+        {
+            var results = new List<Hl7v2Data>();
+            var result = new Hl7v2Data();
+            var segmentIds = new HashSet<string>(segmentIdSeparators.Split(@"|", StringSplitOptions.RemoveEmptyEntries));
+
+            if (segmentIdSeparators == string.Empty || !segmentIds.Intersect(hl7v2Data.Meta).Any())
+            {
+                results.Add(hl7v2Data);
+                return results;
+            }
+
+            for (var i = 0; i < hl7v2Data.Meta.Count; ++i)
+            {
+                if (segmentIds.Contains(hl7v2Data.Meta[i]))
+                {
+                    results.Add(result);
+                    result = new Hl7v2Data();
+                }
+
+                result.Meta.Add(hl7v2Data.Meta[i]);
+                result.Data.Add(hl7v2Data.Data[i]);
+            }
+
+            if (result.Meta.Count > 0)
+            {
+                results.Add(result);
+            }
+
+            return results;
+        }
     }
 }

src/Microsoft.Health.Fhir.Liquid.Converter/Parsers/Hl7v2DataParser.cs

@@ -10,14 +10,14 @@
 using Microsoft.Health.Fhir.Liquid.Converter.InputProcessors;
 using Microsoft.Health.Fhir.Liquid.Converter.Models;
 using Microsoft.Health.Fhir.Liquid.Converter.Models.Hl7v2;
+using Microsoft.Health.Fhir.Liquid.Converter.Utilities;
 using Microsoft.Health.Fhir.Liquid.Converter.Validators;
 
 namespace Microsoft.Health.Fhir.Liquid.Converter.Parsers
 {
     public class Hl7v2DataParser : IDataParser
     {
         private static readonly Hl7v2DataValidator Validator = new Hl7v2DataValidator();
-        private static readonly string[] SegmentSeparators = { "\r\n", "\r", "\n" };
 
         public object Parse(string message)
         {
@@ -30,7 +30,7 @@ public object Parse(string message)
             {
                 var result = new Hl7v2Data(message);
 
-                var segments = message.Split(SegmentSeparators, StringSplitOptions.RemoveEmptyEntries);
+                var segments = Hl7v2DataUtility.SplitMessageToSegments(message);
                 Validator.ValidateMessageHeader(segments[0]);
                 var encodingCharacters = ParseHl7v2EncodingCharacters(segments[0]);
                 result.EncodingCharacters = encodingCharacters;

src/Microsoft.Health.Fhir.Liquid.Converter/Utilities/Hl7v2DataUtility.cs

@@ -0,0 +1,22 @@
+// -------------------------------------------------------------------------------------------------
+// Copyright (c) Microsoft Corporation. All rights reserved.
+// Licensed under the MIT License (MIT). See LICENSE in the repo root for license information.
+// -------------------------------------------------------------------------------------------------
+
+using System;
+using System.Collections.Generic;
+using System.Text;
+
+namespace Microsoft.Health.Fhir.Liquid.Converter.Utilities
+{
+    public static class Hl7v2DataUtility
+    {
+        private static readonly string[] SegmentSeparators = { "\r\n", "\r", "\n" };
+
+        public static string[] SplitMessageToSegments(string message)
+        {
+            var segments = message.Split(SegmentSeparators, StringSplitOptions.RemoveEmptyEntries);
+            return segments;
+        }
+    }
+}