LT-22697: Normalize the range-element ids written to LIFT ranges (#1063)

FieldWorks holds strings as NFD and normalizes them to NFC on LIFT
export, but several writes in the ranges file bypass the normalizing
helper. The part-of-speech and lexical-relation range-element ids went
out through MakeSafeXmlAttribute while the parent attribute of the same
element, two lines away, was normalized -- one name in two encodings
inside one element, which a consumer resolving a range value by string
equality fails to match. The morph-type id was not escaped at all, so a
name holding an ampersand left the document unparseable.

Route those six writes through MakeSafeAndNormalizedAttribute, which
escapes as well as normalizes. The two writes left alone are writing
system tags, where normalizing is a no-op. Two regression tests cover
the two defects.

Fixes LT-22697.

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: imnasnainaec

Src/LexText/LexTextControls/LexTextControlsTests/LiftExportTests.cs

@@ -2064,6 +2064,86 @@ public void LiftExportRanges_PartOfSpeechCatalogIdIsExported()
 			AssertThatXmlIn.Dom(xdocRangeFile).HasAtLeastOneMatchForXpath("//range[@id='grammatical-info']/range-element/trait[@name='catalog-source-id']");
 		}
 
+		///--------------------------------------------------------------------------------------
+		/// <summary>
+		/// LT-22697: FLEx holds names decomposed and normalizes them on export. A
+		/// range-element id must be normalized like the parent attribute naming it and like
+		/// its own label, or a consumer comparing raw strings cannot resolve the reference.
+		/// </summary>
+		///--------------------------------------------------------------------------------------
+		[Test]
+		public void LiftExportRanges_PartOfSpeechIdIsNormalizedLikeItsParentAndLabel()
+		{
+			const string ksDecomposed = "Comple\u0301ments";	// e + U+0301
+			const string ksComposed = "Compl\u00e9ments";	// U+00E9
+			IPartOfSpeech parentPos = null;
+			IPartOfSpeech childPos = null;
+			NonUndoableUnitOfWorkHelper.Do(m_cache.ActionHandlerAccessor, () =>
+			{
+				var posFactory = m_cache.ServiceLocator.GetInstance<IPartOfSpeechFactory>();
+				parentPos = posFactory.Create();
+				m_cache.LangProject.PartsOfSpeechOA.PossibilitiesOS.Add(parentPos);
+				parentPos.Name.set_String(m_cache.DefaultAnalWs, ksDecomposed);
+				childPos = posFactory.Create();
+				parentPos.SubPossibilitiesOS.Add(childPos);
+				childPos.Name.set_String(m_cache.DefaultAnalWs, "Comple\u0301ment du lieu");
+			});
+			var xdocRangeFile = new XmlDocument();
+			using (var w = new StringWriter())
+			{
+				// SUT
+				new LiftExporter(m_cache).ExportLiftRanges(w);
+				xdocRangeFile.LoadXml(w.ToString());
+			}
+
+			var parentElement = xdocRangeFile.SelectSingleNode(string.Format(
+				"//range[@id='grammatical-info']/range-element[@guid='{0}']", parentPos.Guid));
+			var childElement = xdocRangeFile.SelectSingleNode(string.Format(
+				"//range[@id='grammatical-info']/range-element[@guid='{0}']", childPos.Guid));
+			Assert.That(parentElement, Is.Not.Null);
+			Assert.That(childElement, Is.Not.Null);
+			var sId = parentElement.Attributes["id"].Value;
+			Assert.That(sId, Is.EqualTo(ksComposed),
+				"the id must be normalized, not the decomposed form held in memory");
+			Assert.That(childElement.Attributes["parent"].Value, Is.EqualTo(sId),
+				"the parent attribute must match the id it names");
+			Assert.That(parentElement.SelectSingleNode("label/form/text").InnerText, Is.EqualTo(sId),
+				"the label must match the id of its own element");
+		}
+
+		///--------------------------------------------------------------------------------------
+		/// <summary>
+		/// LT-22697: the morph-type id was written raw, so a name holding a markup character
+		/// left the whole ranges document unparseable.
+		/// </summary>
+		///--------------------------------------------------------------------------------------
+		[Test]
+		public void LiftExportRanges_MorphTypeIdWithMarkupCharacterIsEscaped()
+		{
+			const string ksName = "prefix & suffix";
+			IMoMorphType morphType = null;
+			NonUndoableUnitOfWorkHelper.Do(m_cache.ActionHandlerAccessor, () =>
+			{
+				morphType = m_cache.ServiceLocator.GetInstance<IMoMorphTypeFactory>().Create();
+				m_cache.LangProject.LexDbOA.MorphTypesOA.PossibilitiesOS.Add(morphType);
+				morphType.Name.set_String(m_cache.DefaultAnalWs, ksName);
+			});
+			var xdocRangeFile = new XmlDocument();
+			using (var w = new StringWriter())
+			{
+				// SUT
+				new LiftExporter(m_cache).ExportLiftRanges(w);
+				Assert.That(() => xdocRangeFile.LoadXml(w.ToString()), Throws.Nothing,
+					"an unescaped id leaves the whole ranges document unparseable");
+			}
+
+			var element = xdocRangeFile.SelectSingleNode(string.Format(
+				"//range[@id='morph-type']/range-element[@guid='{0}']", morphType.Guid));
+			Assert.That(element, Is.Not.Null);
+			Assert.That(element.Attributes["id"].Value, Is.EqualTo(ksName),
+				"the parsed id must be the name as stored, escaping undone");
+		}
+
 		private int m_flidLongText;
 
 		private void AddStTextCustomFieldAndData()

Src/LexText/LexTextControls/LiftExporter.cs

@@ -1741,13 +1741,13 @@ private void WritePartOfSpeechRangeElement(TextWriter w, IPartOfSpeech pos)
 			{
 				var liftIdOwner = ((IPartOfSpeech)(pos.Owner)).Name.BestAnalysisVernacularAlternative.Text;
 				w.WriteLine("<range-element id=\"{0}\" guid=\"{1}\" parent=\"{2}\">",
-					XmlUtils.MakeSafeXmlAttribute(liftId), pos.Guid,
+					MakeSafeAndNormalizedAttribute(liftId), pos.Guid,
 					MakeSafeAndNormalizedAttribute(liftIdOwner));
 			}
 			else
 			{
 				w.WriteLine("<range-element id=\"{0}\" guid=\"{1}\">",
-					XmlUtils.MakeSafeXmlAttribute(liftId), pos.Guid);
+					MakeSafeAndNormalizedAttribute(liftId), pos.Guid);
 			}
 			WriteAllForms(w, "label", null, "form", pos.Name);
 			WriteAllForms(w, "abbrev", null, "form", pos.Abbreviation);
@@ -1778,13 +1778,13 @@ private void WriteLexRefType(TextWriter w, ILexRefType refer)
 			{
 				var liftIdOwner = ((ILexRefType)refer.Owner).Name.BestAnalysisVernacularAlternative.Text;
 				w.WriteLine("<range-element id=\"{0}\" guid=\"{1}\" parent=\"{2}\">",
-					XmlUtils.MakeSafeXmlAttribute(liftId), refer.Guid,
+					MakeSafeAndNormalizedAttribute(liftId), refer.Guid,
 					MakeSafeAndNormalizedAttribute(liftIdOwner));
 			}
 			else
 			{
 				w.WriteLine("<range-element id=\"{0}\" guid=\"{1}\">",
-					XmlUtils.MakeSafeXmlAttribute(liftId), refer.Guid);
+					MakeSafeAndNormalizedAttribute(liftId), refer.Guid);
 			}
 			WriteAllForms(w, "label", null, "form", refer.Name);
 			WriteAllForms(w, "abbrev", null, "form", refer.Abbreviation);
@@ -2177,19 +2177,19 @@ void WriteMorphTypeRange(TextWriter w)
 				var liftId = type.Name.get_String(m_wsEn).Text;
 				if (String.IsNullOrEmpty(liftId))
 					liftId = type.Name.BestAnalysisVernacularAlternative.Text;
-				w.WriteLine("<range-element id=\"{0}\" guid=\"{1}\">", liftId, type.Guid);
+				w.WriteLine("<range-element id=\"{0}\" guid=\"{1}\">", MakeSafeAndNormalizedAttribute(liftId), type.Guid);
 				WriteAllForms(w, "label", null, "form", type.Name);
 				WriteAllForms(w, "abbrev", null, "form", type.Abbreviation);
 				WriteAllForms(w, "description", null, "form", type.Description);
 				if (type.Prefix != null)
 				{
 					w.WriteLine("<trait name=\"leading-symbol\" value=\"{0}\"/>",
-						XmlUtils.MakeSafeXmlAttribute(type.Prefix));
+						MakeSafeAndNormalizedAttribute(type.Prefix));
 				}
 				if (type.Postfix != null)
 				{
 					w.WriteLine("<trait name=\"trailing-symbol\" value=\"{0}\"/>",
-						XmlUtils.MakeSafeXmlAttribute(type.Postfix));
+						MakeSafeAndNormalizedAttribute(type.Postfix));
 				}
 				w.WriteLine("</range-element>");
 			}
@@ -2298,7 +2298,7 @@ void WriteStemNameRanges(TextWriter w)
 					foreach (var region in stem.RegionsOC)
 					{
 						w.WriteLine("<trait name=\"feature-set\" value=\"{0}\"/>",
-							XmlUtils.MakeSafeXmlAttribute(region.LiftName));
+							MakeSafeAndNormalizedAttribute(region.LiftName));
 					}
 					w.WriteLine("</range-element>");
 				}