Fix for LT-22563

Src/LexText/ParserCore/ParserCoreTests/ParseFilerProcessingTests.cs

@@ -13,6 +13,8 @@
 
 using System;
 using System.Linq;
+using System.Reflection;
+using System.Xml.Linq;
 using NUnit.Framework;
 using SIL.FieldWorks.Common.FwUtils;
 using SIL.LCModel.Core.Text;
@@ -99,6 +101,37 @@ protected void ExecuteIdleQueue()
 			m_idleQueue.Clear();
 		}
 
+		/// <summary>
+		/// Creates a stem entry with two senses, each with its own MoStemMsa, so tests can
+		/// verify that a specific MSA (not just "the first one") gets resolved.
+		/// </summary>
+		private ILexEntry CreateStemEntryWithTwoMsas(string form, out IMoStemMsa firstMsa, out IMoStemMsa secondMsa)
+		{
+			ILexEntry entry = m_entryFactory.Create();
+			IMoStemAllomorph entryForm = m_stemAlloFactory.Create();
+			entry.AlternateFormsOS.Add(entryForm);
+			entryForm.Form.VernacularDefaultWritingSystem = TsStringUtils.MakeString(form, m_vernacularWS.Handle);
+			firstMsa = m_stemMsaFactory.Create();
+			entry.MorphoSyntaxAnalysesOC.Add(firstMsa);
+			ILexSense firstSense = m_senseFactory.Create();
+			entry.SensesOS.Add(firstSense);
+			firstSense.MorphoSyntaxAnalysisRA = firstMsa;
+			secondMsa = m_stemMsaFactory.Create();
+			entry.MorphoSyntaxAnalysesOC.Add(secondMsa);
+			ILexSense secondSense = m_senseFactory.Create();
+			entry.SensesOS.Add(secondSense);
+			secondSense.MorphoSyntaxAnalysisRA = secondMsa;
+			return entry;
+		}
+
+		private static ParseMorph InvokeTryCreateParseMorph(LcmCache cache, XElement morphElem, out bool succeeded)
+		{
+			MethodInfo method = typeof(XAmpleParser).GetMethod("TryCreateParseMorph", BindingFlags.NonPublic | BindingFlags.Static);
+			var args = new object[] { cache, morphElem, null };
+			succeeded = (bool) method.Invoke(null, args);
+			return (ParseMorph) args[2];
+		}
+
 		#endregion // Non-tests
 
 		#region Setup and TearDown
@@ -522,6 +555,118 @@ public void LexEntryInflTypeTwoAnalyses()
 			}
 		}
 
+		/// ------------------------------------------------------------------------------------
+		/// <summary>
+		/// LT-22563: XAmple's MSI DbRef for an irregularly inflected variant is
+		/// "lexEntryHvo.refIndex.msaHvo" (see LT-22422). TryCreateParseMorph must resolve the
+		/// specific msaHvo from that DbRef; it must not fall back to just the variant's
+		/// main/first sense, since that can be a different MSA than XAmple actually chose
+		/// when the underlying entry has more than one MSA.
+		/// </summary>
+		/// ------------------------------------------------------------------------------------
+		[Test]
+		public void TryCreateParseMorph_IrregularVariantWithMultipleMsas_UsesMsaFromDbRef()
+		{
+			ILexEntry crebV = null;
+			IMoStemAllomorph crebVForm = null;
+			IMoStemMsa firstMsa = null;
+			IMoStemMsa secondMsa = null;
+
+			UndoableUnitOfWorkHelper.Do("Undo stuff", "Redo stuff", m_actionHandler, () =>
+			{
+				// 'seek' has two senses/MSAs; 'creb' is an irregularly inflected variant of it.
+				ILexEntryInflType inflType = m_lexEntryInflTypeFactory.Create();
+				Cache.LangProject.LexDbOA.VariantEntryTypesOA.PossibilitiesOS.Add(inflType);
+
+				ILexEntry seekV = CreateStemEntryWithTwoMsas("seekVTEST", out firstMsa, out secondMsa);
+
+				crebV = m_entryFactory.Create();
+				crebVForm = m_stemAlloFactory.Create();
+				crebV.AlternateFormsOS.Add(crebVForm);
+				crebVForm.Form.VernacularDefaultWritingSystem = TsStringUtils.MakeString("crebVTEST", m_vernacularWS.Handle);
+				ILexEntryRef lexEntryRef = m_lexEntryRefFactory.Create();
+				crebV.EntryRefsOS.Add(lexEntryRef);
+				lexEntryRef.ComponentLexemesRS.Add(seekV);
+				lexEntryRef.VariantEntryTypesRS.Add(inflType);
+			});
+
+			// Mimic what XAmple echoes back for this kind of variant: MoForm DbRef is the
+			// allomorph hvo, and MSI DbRef is "lexEntryHvo.refIndex.msaHvo", pointing
+			// specifically at the *second* MSA of 'seek'.
+			var morphElem = new XElement("Morph",
+				new XElement("MoForm", new XAttribute("DbRef", crebVForm.Hvo)),
+				new XElement("MSI", new XAttribute("DbRef", string.Format("{0}.0.{1}", crebV.Hvo, secondMsa.Hvo))));
+
+			bool succeeded;
+			ParseMorph morph = InvokeTryCreateParseMorph(Cache, morphElem, out succeeded);
+
+			Assert.That(succeeded, Is.True);
+			Assert.That(morph, Is.Not.Null);
+			Assert.That(morph.Msa.Hvo, Is.EqualTo(secondMsa.Hvo),
+				"Should use the MSA hvo encoded in the MSI DbRef, not just the variant's first/main sense");
+			Assert.That(morph.Msa.Hvo, Is.Not.EqualTo(firstMsa.Hvo));
+		}
+
+		/// ------------------------------------------------------------------------------------
+		/// <summary>
+		/// LT-22563: hardens the above test by giving the variant *two* EntryRefsOS (pointing
+		/// at two different base entries, each itself with two MSAs) and checking both possible
+		/// refIndex values. This guards both halves of the "lexEntryHvo.refIndex.msaHvo" DbRef:
+		/// picking the right LexEntryRef by index, and then the right MSA within that ref's
+		/// target entry -- not just the case where there is only one ref to pick from.
+		/// </summary>
+		/// ------------------------------------------------------------------------------------
+		[TestCase(0)]
+		[TestCase(1)]
+		public void TryCreateParseMorph_IrregularVariantWithMultipleEntryRefsAndMsas_UsesRefIndexAndMsaFromDbRef(int entryRefIndex)
+		{
+			ILexEntry crebV = null;
+			IMoStemAllomorph crebVForm = null;
+			IMoStemMsa believeSecondMsa = null;
+			IMoStemMsa seekSecondMsa = null;
+
+			UndoableUnitOfWorkHelper.Do("Undo stuff", "Redo stuff", m_actionHandler, () =>
+			{
+				ILexEntryInflType inflType = m_lexEntryInflTypeFactory.Create();
+				Cache.LangProject.LexDbOA.VariantEntryTypesOA.PossibilitiesOS.Add(inflType);
+
+				IMoStemMsa believeFirstMsa;
+				ILexEntry believeV = CreateStemEntryWithTwoMsas("believeVTEST2", out believeFirstMsa, out believeSecondMsa);
+				IMoStemMsa seekFirstMsa;
+				ILexEntry seekV = CreateStemEntryWithTwoMsas("seekVTEST2", out seekFirstMsa, out seekSecondMsa);
+
+				crebV = m_entryFactory.Create();
+				crebVForm = m_stemAlloFactory.Create();
+				crebV.AlternateFormsOS.Add(crebVForm);
+				crebVForm.Form.VernacularDefaultWritingSystem = TsStringUtils.MakeString("crebVTEST2", m_vernacularWS.Handle);
+
+				// EntryRefsOS[0] -> believeV, EntryRefsOS[1] -> seekV.
+				ILexEntryRef believeRef = m_lexEntryRefFactory.Create();
+				crebV.EntryRefsOS.Add(believeRef);
+				believeRef.ComponentLexemesRS.Add(believeV);
+				believeRef.VariantEntryTypesRS.Add(inflType);
+
+				ILexEntryRef seekRef = m_lexEntryRefFactory.Create();
+				crebV.EntryRefsOS.Add(seekRef);
+				seekRef.ComponentLexemesRS.Add(seekV);
+				seekRef.VariantEntryTypesRS.Add(inflType);
+			});
+
+			IMoStemMsa expectedMsa = entryRefIndex == 0 ? believeSecondMsa : seekSecondMsa;
+
+			var morphElem = new XElement("Morph",
+				new XElement("MoForm", new XAttribute("DbRef", crebVForm.Hvo)),
+				new XElement("MSI", new XAttribute("DbRef", string.Format("{0}.{1}.{2}", crebV.Hvo, entryRefIndex, expectedMsa.Hvo))));
+
+			bool succeeded;
+			ParseMorph morph = InvokeTryCreateParseMorph(Cache, morphElem, out succeeded);
+
+			Assert.That(succeeded, Is.True);
+			Assert.That(morph, Is.Not.Null);
+			Assert.That(morph.Msa.Hvo, Is.EqualTo(expectedMsa.Hvo),
+				string.Format("Should resolve EntryRefsOS[{0}] and the specific msa hvo encoded in the DbRef", entryRefIndex));
+		}
+
 		[Test]
 		public void LexEntryInflTypeAnalysisWithNullForSlotFiller()
 		{

Src/LexText/ParserCore/ParserXmlWriterExtensions.cs

@@ -18,7 +18,7 @@ namespace SIL.FieldWorks.WordWorks.Parser
 {
 	internal static class ParserXmlWriterExtensions
 	{
-		private static Tuple<int, int, int> ProcessMsaHvo(string msaHvo)
+		public static Tuple<int, int, int> ProcessMsaHvo(string msaHvo)
 		{
 			string[] msaHvoParts = msaHvo.Split('.');
 			// the msa hvo has one part or three parts separated by a period.

Src/LexText/ParserCore/XAmpleParser.cs

@@ -245,9 +245,11 @@ private static bool TryCreateParseMorph(LcmCache cache, XElement morphElem, out
 			//       required slots in affix templates.  The parser filer can ignore these.
 			// 3. <MSI DbRef="y"... and y is an hvo for a LexEntry.
 			//       The LexEntry is a variant form for the first set of LexEntryRefs.
-			// 4. <MSI DbRef="y"... and y is an hvo for a LexEntry followed by a period and an index digit.
-			//       The LexEntry is a variant form and the (non-zero) index indicates
-			//       which set of LexEntryRefs it is for.
+			// 4. <MSI DbRef="y.i.z"... and y is an hvo for a LexEntry, i is an index to the
+			//       LexEntryRef, and z is the hvo of the specific MSA (of that LexEntryRef's
+			//       target entry) that was used. The LexEntry is a variant form; i indicates
+			//       which set of LexEntryRefs it is for, and z indicates which of that entry's
+			//       MSAs to use, since it can have more than one (see LT-22422/LT-22563).
 			ICmObject objForm;
 			if (!cache.ServiceLocator.GetInstance<ICmObjectRepository>().TryGetObject(int.Parse(formHvo), out objForm))
 			{
@@ -261,8 +263,9 @@ private static bool TryCreateParseMorph(LcmCache cache, XElement morphElem, out
 				return true;
 			}
 
-			// Irregulary inflected forms can have a combination MSA hvo: the LexEntry hvo, a period, and an index to the LexEntryRef
-			Tuple<int, int> msaTuple = ProcessMsaHvo(msaHvo);
+			// Irregulary inflected forms can have a combination MSA hvo: the LexEntry hvo, a period, an index to the LexEntryRef,
+			// another period and the MSA hvo
+			Tuple<int, int, int> msaTuple = ParserXmlWriterExtensions.ProcessMsaHvo(msaHvo);
 			ICmObject objMsa;
 			if (!cache.ServiceLocator.GetInstance<ICmObjectRepository>().TryGetObject(msaTuple.Item1, out objMsa))
 			{
@@ -286,10 +289,19 @@ private static bool TryCreateParseMorph(LcmCache cache, XElement morphElem, out
 					ILexEntryRef lexEntryRef = msaAsLexEntry.EntryRefsOS[msaTuple.Item2];
 					if (lexEntryRef != null && lexEntryRef.ComponentLexemesRS.Count() > 0)
 					{
-						// make sure there is at least one component lexeme (LT-22328)
-						ILexSense sense = MorphServices.GetMainOrFirstSenseOfVariant(lexEntryRef);
 						var inflType = lexEntryRef.VariantEntryTypesRS[0] as ILexEntryInflType;
-						morph = new ParseMorph(form, sense.MorphoSyntaxAnalysisRA, inflType);
+						IMoStemMsa stemMsa = (IMoStemMsa)cache.ServiceLocator.GetInstance<ICmObjectRepository>().GetObject(msaTuple.Item3);
+						if (stemMsa != null)
+						{
+							// use the msa itself
+							morph = new ParseMorph(form, stemMsa, inflType);
+						}
+						else
+						{
+							// make sure there is at least one component lexeme (LT-22328)
+							ILexSense sense = MorphServices.GetMainOrFirstSenseOfVariant(lexEntryRef);
+							morph = new ParseMorph(form, sense.MorphoSyntaxAnalysisRA, inflType);
+						}
 						return true;
 					}
 				}
@@ -300,12 +312,6 @@ private static bool TryCreateParseMorph(LcmCache cache, XElement morphElem, out
 			return true;
 		}
 
-		private static Tuple<int, int> ProcessMsaHvo(string msaHvo)
-		{
-			string[] msaHvoParts = msaHvo.Split('.');
-			return Tuple.Create(int.Parse(msaHvoParts[0]), msaHvoParts.Length == 2 ? int.Parse(msaHvoParts[1]) : 0);
-		}
-
 		public XDocument ParseWordXml(string word)
 		{
 			CheckDisposed();