Fix LT-22008: Do not crash importing flextext with empty word content (…

…#242)

Src/LexText/Interlinear/BIRDInterlinearImporter.cs

@@ -728,6 +728,8 @@ private static IAnalysis CreateWordAnalysisStack(LcmCache cache, Word word)
 
 			foreach (var wordItem in word.Items)
 			{
+				if (wordItem.Value == null)
+					continue;
 				ITsString wordForm = null;
 				switch (wordItem.type)
 				{
@@ -758,7 +760,8 @@ private static IAnalysis CreateWordAnalysisStack(LcmCache cache, Word word)
 			}
 			else
 			{
-				Debug.Assert(analysis != null, "What else could this do?");
+				// There was an invalid analysis in the file. We can't do anything with it.
+				return null;
 			}
 
 			// Fill in morphemes, lex. entries, lex. gloss, and lex.gram.info

Src/LexText/Interlinear/ITextDllTests/BIRDFormatImportTests.cs

@@ -449,6 +449,77 @@ public void UglyEmptyDataShouldNotCrash()
 			}
 		}
 
+		[Test]
+		public void EmptyTxtItemUnderWordShouldNotCrash()
+		{
+			// an interlinear text example xml string
+			const string xml =
+"<?xml version=\"1.0\" encoding=\"utf-8\"?>" +
+"<document version=\"2\">" +
+"<interlinear-text guid=\"6a424526-aa64-4912-b8c7-9e5ae0ab4aae\">" +
+"<item type=\"title\" lang=\"en\">Test</item>" +
+"<paragraphs>" +
+"<paragraph guid=\"ccd62b33-c9f6-4e52-917f-00c67c3638c8\">" +
+"<phrases>" +
+"<phrase begin-time-offset=\"0\" end-time-offset=\"3750\" guid=\"b57a2665-402a-4ab1-af16-655c70df062f\">" +
+"<item lang=\"fr\" type=\"txt\">testing paragraph without words</item>" +
+"<words>" +
+"<word guid=\"093fe3c6-d467-4c28-a03e-d511b94185da\">" +
+"<item lang=\"fr\" type=\"txt\"/>" + // empty txt item
+"</word>" +
+"</words>" +
+"<item lang=\"en\" type=\"gls\">In the country of a Mongol king lived three sisters.</item>" +
+"</phrase>" +
+"<phrase guid=\"441c1171-78a1-481b-9732-b9c558948ce5\">" +
+"<item type=\"txt\" lang=\"fr\">This is a test.</item>" +
+"<item type=\"segnum\" lang=\"en\">1</item>" +
+"<words>" +
+"<word guid=\"d161bf25-b6df-418d-b9c1-a396ff3ec5b1\">" +
+"<item type=\"txt\" lang=\"fr\">This</item>" +
+"</word>" +
+"<word guid=\"ab9e81c7-3157-4011-a8a1-68eb1afc0be1\">" +
+"<item type=\"txt\" lang=\"fr\">is</item>" +
+"</word>" +
+"<word guid=\"0ef6172b-07c3-4c91-b0b2-afbebca5fca0\">" +
+"<item type=\"txt\" lang=\"fr\">a</item>" +
+"</word>" +
+"<word guid=\"48949d94-869c-45d5-9251-b68ce7d66cee\">" +
+"<item type=\"txt\" lang=\"fr\">test</item>" +
+"</word>" +
+"<word>" +
+"<item type=\"punct\" lang=\"fr\">.</item>" +
+"</word>" +
+"</words>" +
+"<item type=\"gls\" lang=\"en\"></item>" +
+"</phrase>" +
+"</phrases>" +
+"</paragraph>" +
+"</paragraphs>" +
+"<languages>" +
+"<language lang=\"en\" font=\"Charis SIL\" />" +
+"<language lang=\"fr\" font=\"Times New Roman\" vernacular=\"true\" />" +
+"</languages>" +
+"</interlinear-text>" +
+"</document>";
+
+			var li = new LinguaLinksImport(Cache, null, null);
+			LCModel.IText text = null;
+			using(var stream = new MemoryStream(Encoding.ASCII.GetBytes(xml.ToCharArray())))
+			{
+				// SUT - Verify that no crash occurs importing this data: see LT-22008
+				Assert.DoesNotThrow(()=> li.ImportInterlinear(new DummyProgressDlg(), stream, 0, ref text));
+				using(var firstEntry = Cache.LanguageProject.Texts.GetEnumerator())
+				{
+					firstEntry.MoveNext();
+					var imported = firstEntry.Current;
+					Assert.That(imported.ContentsOA.ParagraphsOS.Count, Is.EqualTo(1));
+					Assert.That(((IStTxtPara)imported.ContentsOA.ParagraphsOS[0]).SegmentsOS.Count, Is.EqualTo(2));
+					// Verify that the words with non-empty txt were imported
+					Assert.That(((IStTxtPara)imported.ContentsOA.ParagraphsOS[0]).SegmentsOS[1].AnalysesRS.Count, Is.EqualTo(5));
+				}
+			}
+		}
+
 		[Test]
 		public void TestImportMergeFlexTextWithSegnumItem()
 		{