Fix LT-22121: Analysis guesser should not guess ras for rAs (#326)

* Fix LT-22121: Analysis guesser should not guess ras for rAs

* Add test case for mixed case words

Author: jtmaxwell3

src/SIL.LCModel/DomainServices/AnalysisGuessServices.cs

@@ -729,18 +729,9 @@ public IAnalysis GetBestGuess(AnalysisOccurrence occurrence, bool onlyIndexZeroL
 		private IWfiWordform GetLowercaseWordform(AnalysisOccurrence occurrence, int ws,
 			bool onlyIndexZeroLowercaseMatching, IWfiWordform wordform)
 		{
-			// TODO: make it look for the first word in the sentence...may not be at Index 0!
-			if (occurrence == null)
-				return null;
-			if (onlyIndexZeroLowercaseMatching && occurrence.Index != 0)
-				return null;
-			if (occurrence.Segment == null || !occurrence.Segment.IsValidObject)
-				return null;
-			ITsString tssWfBaseline = occurrence.BaselineText;
-			var cf = new CaseFunctions(Cache.ServiceLocator.WritingSystemManager.Get(ws));
-			string sLower = cf.ToLower(tssWfBaseline.Text);
+			string sLower = GetLowercaseOfTitleCase(occurrence, ws, onlyIndexZeroLowercaseMatching);
 			// don't bother looking up the lowercased wordform if the instanceOf is already in lowercase form.
-			if (sLower != wordform.ShortName)
+			if (sLower != null && sLower != wordform.ShortName)
 			{
 				return GetWordformIfNeeded(sLower, ws);
 			}
@@ -752,18 +743,37 @@ private IWfiWordform GetLowercaseWordform(AnalysisOccurrence occurrence, int ws,
 		// if it lowercases to the given wordform.
 		// Otherwise, return null.
 		// </summary>
-		private IWfiWordform GetOriginalCaseWordform(AnalysisOccurrence occurrence, IWfiWordform wordform, int ws)
+		private IWfiWordform GetOriginalCaseWordform(AnalysisOccurrence occurrence, IWfiWordform wordform,
+			int ws, bool onlyIndexZeroLowercaseMatching)
 		{
-			ITsString tssWfBaseline = occurrence.BaselineText;
-			var cf = new CaseFunctions(Cache.ServiceLocator.WritingSystemManager.Get(ws));
-			string sLower = cf.ToLower(tssWfBaseline.Text);
+			string sLower = GetLowercaseOfTitleCase(occurrence, ws, onlyIndexZeroLowercaseMatching);
 			if (sLower == wordform.GetForm(ws).Text)
 			{
+				ITsString tssWfBaseline = occurrence.BaselineText;
 				return GetWordformIfNeeded(tssWfBaseline.Text, ws);
 			}
 			return null;
 		}
 
+		// <summary>
+		// Get the lowercase form of occurrence if it is Title case.
+		// </summary>
+		private string GetLowercaseOfTitleCase(AnalysisOccurrence occurrence, int ws, bool onlyIndexZeroLowercaseMatching)
+		{
+			// TODO: make it look for the first word in the sentence...may not be at Index 0!
+			if (occurrence == null)
+				return null;
+			if (onlyIndexZeroLowercaseMatching && occurrence.Index != 0)
+				return null;
+			if (occurrence.Segment == null || !occurrence.Segment.IsValidObject)
+				return null;
+			ITsString tssWfBaseline = occurrence.BaselineText;
+			var cf = new CaseFunctions(Cache.ServiceLocator.WritingSystemManager.Get(ws));
+			if (cf.StringCase(tssWfBaseline.Text) == StringCaseStatus.title)
+				return cf.ToLower(tssWfBaseline.Text);
+			return null;
+		}
+
 		/// <summary>
 		/// Get a wordform for word if it already exists or
 		/// if it has an entry in the lexicon.
@@ -862,7 +872,7 @@ private List<IWfiAnalysis> GetSortedAnalysisGuesses(IWfiWordform wordform, int w
 			{
 				// Sometimes the user selects a lowercase wordform for an uppercase word.
 				// Get the original case so that we can include uppercase analyses.
-				var originalCaseWf = GetOriginalCaseWordform(occurrence, wordform, ws);
+				var originalCaseWf = GetOriginalCaseWordform(occurrence, wordform, ws, onlyIndexZeroLowercaseMatching);
 				if (originalCaseWf != null)
 					wordform = originalCaseWf;
 			}

tests/SIL.LCModel.Tests/DomainServices/AnalysisGuessServicesTests.cs

@@ -170,6 +170,17 @@ internal void DoDataSetup()
 					" " + Words_para0[4].Form.BestVernacularAlternative.Text +
 					".", wsVern));
 				Para0.Contents = bldr4.GetString();
+				/* rAs ras */
+				IWfiWordform rAs = wfFactory.Create(TsStringUtils.MakeString("rAs", wsVern));
+				Words_para0.Add(rAs);
+				IWfiWordform ras = wfFactory.Create(TsStringUtils.MakeString("ras", wsVern));
+				Words_para0.Add(ras);
+				var bldr5 = Para0.Contents.GetIncBldr();
+				bldr5.AppendTsString(TsStringUtils.MakeString(
+					" " + Words_para0[20].Form.BestVernacularAlternative.Text +
+					" " + Words_para0[21].Form.BestVernacularAlternative.Text +
+					".", wsVern));
+				Para0.Contents = bldr5.GetString();
 				using (ParagraphParser pp = new ParagraphParser(Cache))
 				{
 					foreach (IStTxtPara para in StText.ParagraphsOS)
@@ -921,6 +932,21 @@ public void ExpectedAnalysisGuess_ForSentenceInitialOnlyLowercase()
 			}
 		}
 
+		/// <summary>
+		/// if a wordform is mixed case, don't look for lower case.
+		/// </summary>
+		[Test]
+		public void ExpectedAnalysisGuess_ForSentenceInitialMixedCase()
+		{
+			using (var setup = new AnalysisGuessBaseSetup(Cache))
+			{
+				WordAnalysisOrGlossServices.CreateNewAnalysisWAG(setup.Words_para0[21]); // ras
+				var wagLowercaseB = new AnalysisOccurrence(setup.Para0.SegmentsOS[5], 0); // rAs
+				var guessActual = setup.GuessServices.GetBestGuess(wagLowercaseB);
+				Assert.AreEqual(new NullWAG(), guessActual);
+			}
+		}
+
 		/// <summary>
 		/// This class allows us to fake out the guesser by passing an analysis occurrence with the analyis we want,
 		/// even though it isn't the analysis recorded in the paragraph.