Ensure tag operations are case insensitive on insert across database types (#19439)

* Ensure tag operations are case insensitve on insert across database types.

* Ensure tags provided in a single property are case insensitively distinct when saving the tags and relationships.

* Update src/Umbraco.Infrastructure/Persistence/Repositories/Implement/TagRepository.cs

Co-authored-by: Copilot <[email protected]>

* Handle case sensitivity on insert with tag groups too.

---------

Co-authored-by: Copilot <[email protected]>

Author: AndyButland

src/Umbraco.Infrastructure/Persistence/Repositories/Implement/TagRepository.cs

@@ -128,10 +128,13 @@ public void Assign(int contentId, int propertyTypeId, IEnumerable<ITag> tags, bo
         var group = SqlSyntax.GetQuotedColumnName("group");
 
         // insert tags
+        // - Note we are checking in the subquery for the existence of the tag, so we don't insert duplicates, using a case-insensitive comparison (the
+        //   LOWER keyword is consistent across SQLite and SQLServer). This ensures consistent behavior across databases as by default, SQLServer will
+        //   perform a case-insensitive comparison, while SQLite will not.
         var sql1 = $@"INSERT INTO cmsTags (tag, {group}, languageId)
 SELECT tagSet.tag, tagSet.{group}, tagSet.languageId
 FROM {tagSetSql}
-LEFT OUTER JOIN cmsTags ON (tagSet.tag = cmsTags.tag AND tagSet.{group} = cmsTags.{group} AND COALESCE(tagSet.languageId, -1) = COALESCE(cmsTags.languageId, -1))
+LEFT OUTER JOIN cmsTags ON (LOWER(tagSet.tag) = LOWER(cmsTags.tag) AND LOWER(tagSet.{group}) = LOWER(cmsTags.{group}) AND COALESCE(tagSet.languageId, -1) = COALESCE(cmsTags.languageId, -1))
 WHERE cmsTags.id IS NULL";
 
         Database.Execute(sql1);
@@ -142,7 +145,7 @@ LEFT OUTER JOIN cmsTags ON (tagSet.tag = cmsTags.tag AND tagSet.{group} = cmsTag
 FROM (
     SELECT t.Id
     FROM {tagSetSql}
-    INNER JOIN cmsTags as t ON (tagSet.tag = t.tag AND tagSet.{group} = t.{group} AND COALESCE(tagSet.languageId, -1) = COALESCE(t.languageId, -1))
+    INNER JOIN cmsTags as t ON (LOWER(tagSet.tag) = LOWER(t.tag) AND LOWER(tagSet.{group}) = LOWER(t.{group}) AND COALESCE(tagSet.languageId, -1) = COALESCE(t.languageId, -1))
 ) AS tagSet2
 LEFT OUTER JOIN cmsTagRelationship r ON (tagSet2.id = r.tagId AND r.nodeId = {contentId} AND r.propertyTypeID = {propertyTypeId})
 WHERE r.tagId IS NULL";
@@ -245,14 +248,18 @@ private class TagComparer : IEqualityComparer<ITag>
     {
         public bool Equals(ITag? x, ITag? y) =>
             ReferenceEquals(x, y) // takes care of both being null
-            || (x != null && y != null && x.Text == y.Text && x.Group == y.Group && x.LanguageId == y.LanguageId);
+            || (x != null &&
+                y != null &&
+                string.Equals(x.Text, y.Text, StringComparison.OrdinalIgnoreCase) &&
+                string.Equals(x.Group, y.Group, StringComparison.OrdinalIgnoreCase) &&
+            x.LanguageId == y.LanguageId);
 
         public int GetHashCode(ITag obj)
         {
             unchecked
             {
-                var h = obj.Text.GetHashCode();
-                h = (h * 397) ^ obj.Group.GetHashCode();
+                var h = StringComparer.OrdinalIgnoreCase.GetHashCode(obj.Text);
+                h = (h * 397) ^ StringComparer.OrdinalIgnoreCase.GetHashCode(obj.Group);
                 h = (h * 397) ^ (obj.LanguageId?.GetHashCode() ?? 0);
                 return h;
             }

tests/Umbraco.Tests.Integration/Umbraco.Infrastructure/Persistence/Repositories/TagRepositoryTest.cs

@@ -1,7 +1,6 @@
 // Copyright (c) Umbraco.
 // See LICENSE for more details.
 
-using System.Linq;
 using Microsoft.Extensions.Logging;
 using NUnit.Framework;
 using Umbraco.Cms.Core.Cache;
@@ -1047,6 +1046,98 @@ public void Can_Get_Tagged_Entities_For_Tag()
         }
     }
 
+    [Test]
+    public void Can_Create_Tag_Relations_With_Mixed_Casing_For_Tag()
+    {
+        var provider = ScopeProvider;
+        using (var scope = ScopeProvider.CreateScope())
+        {
+            (IContentType contentType, IContent content1, IContent content2) = CreateContentForCreateTagTests();
+
+            var repository = CreateRepository(provider);
+
+            // Note two tags are applied, but they differ only in case for the tag.
+            Tag[] tags1 = { new() { Text = "tag1", Group = "test" }, new() { Text = "Tag1", Group = "test" } };
+            repository.Assign(
+                content1.Id,
+                contentType.PropertyTypes.First().Id,
+                tags1,
+                false);
+
+            // Note the casing is different from the tag in tags1, but both should be considered equivalent.
+            Tag[] tags2 = { new() { Text = "TAG1", Group = "test" } };
+            repository.Assign(
+                content2.Id,
+                contentType.PropertyTypes.First().Id,
+                tags2,
+                false);
+
+            // Only one tag should have been saved.
+            var tagCount = scope.Database.ExecuteScalar<int>(
+                "SELECT COUNT(*) FROM cmsTags WHERE [group] = 'test'");
+            Assert.AreEqual(1, tagCount);
+
+            // Both content items should be found as tagged by the tag, even though one was assigned with the tag differing in case.
+            Assert.AreEqual(2, repository.GetTaggedEntitiesByTag(TaggableObjectTypes.Content, "tag1").Count());
+        }
+    }
+
+    [Test]
+    public void Can_Create_Tag_Relations_With_Mixed_Casing_For_Group()
+    {
+        var provider = ScopeProvider;
+        using (var scope = ScopeProvider.CreateScope())
+        {
+            (IContentType contentType, IContent content1, IContent content2) = CreateContentForCreateTagTests();
+
+            var repository = CreateRepository(provider);
+
+            // Note two tags are applied, but they differ only in case for the group.
+            Tag[] tags1 = { new() { Text = "tag1", Group = "group1" }, new() { Text = "tag1", Group = "Group1" } };
+            repository.Assign(
+                content1.Id,
+                contentType.PropertyTypes.First().Id,
+                tags1,
+                false);
+
+            // Note the casing is different from the group in tags1, but both should be considered equivalent.
+            Tag[] tags2 = { new() { Text = "tag1", Group = "GROUP1" } };
+            repository.Assign(
+                content2.Id,
+                contentType.PropertyTypes.First().Id,
+                tags2,
+                false);
+
+            // Only one tag/group should have been saved.
+            var tagCount = scope.Database.ExecuteScalar<int>(
+                "SELECT COUNT(*) FROM cmsTags WHERE [tag] = 'tag1'");
+            Assert.AreEqual(1, tagCount);
+
+            var groupCount = scope.Database.ExecuteScalar<int>(
+                "SELECT COUNT(*) FROM cmsTags WHERE [group] = 'group1'");
+            Assert.AreEqual(1, groupCount);
+
+            // Both content items should be found as tagged by the tag, even though one was assigned with the group differing in case.
+            Assert.AreEqual(2, repository.GetTaggedEntitiesByTagGroup(TaggableObjectTypes.Content, "group1").Count());
+        }
+    }
+
+    private (IContentType ContentType, IContent Content1, IContent Content2) CreateContentForCreateTagTests()
+    {
+        var template = TemplateBuilder.CreateTextPageTemplate();
+        FileService.SaveTemplate(template);
+
+        var contentType = ContentTypeBuilder.CreateSimpleContentType("test", "Test", defaultTemplateId: template.Id);
+        ContentTypeRepository.Save(contentType);
+
+        var content1 = ContentBuilder.CreateSimpleContent(contentType);
+        var content2 = ContentBuilder.CreateSimpleContent(contentType);
+        DocumentRepository.Save(content1);
+        DocumentRepository.Save(content2);
+
+        return (contentType, content1, content2);
+    }
+
     private TagRepository CreateRepository(IScopeProvider provider) =>
         new((IScopeAccessor)provider, AppCaches.Disabled, LoggerFactory.CreateLogger<TagRepository>());
 }