Collapse nested markup with surrounding text; leave nested <sup> alone (BL-16387)

StephenMcConnel · claude · StephenMcConnel · commit 6da195441c51 · 2026-07-08T16:12:15.000-06:00
Responds to review feedback on PR #8019: the collapse regex previously only removed markup doubled exactly at the boundary (<b><b>...</b></b>). It now also collapses cases with text inside the outer element but outside the inner one, e.g. <b>abc<b>def</b>ghi</b> -> <b>abcdefghi</b>, since the inner tag adds no formatting and can cause surprises when toggling the style. Also drop <sup> from the collapse set: nested superscript is semantically meaningful, so those are left as-is (attributes are still stripped from the inner tag). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
diff --git a/src/BloomExe/Book/Book.cs b/src/BloomExe/Book/Book.cs
@@ -2698,8 +2698,8 @@ public static void UpdateCharacterStyleMarkup(HtmlDom bookDOM)
                 // Only one level of nesting is handled, but that should (almost always) be enough.
                 inner = Regex.Replace(
                     inner,
-                    @"<(strong|em|sup|u)><\1>(.*?)</\1></\1>",
-                    "<$1>$2</$1>",
+                    @"<(strong|em|u)>([^<>]*)<\1>([^<>]*)</\1>([^<>]*)</\1>",
+                    "<$1>$2$3$4</$1>",
                     RegexOptions.CultureInvariant | RegexOptions.IgnoreCase
                 );
                 // Remove empty (or essentially empty) character markup tags.  (BL-16387)
diff --git a/src/BloomTests/Book/BookTests.cs b/src/BloomTests/Book/BookTests.cs
@@ -3183,12 +3183,12 @@ public void UpdateCharacterStyleMarkup_LeavesPlainTextUnchanged()
         public void UpdateCharacterStyleMarkup_ReducesNestingOfSameTags()
         {
             var dom = new HtmlDom(
-                "<html><body><div class='bloom-editable'><p><b><b style='color:red'>text</b></b></p></div></body></html>"
+                "<html><body><div class='bloom-editable'><p><b>some <b style='color:red'>text</b> here</b></p></div></body></html>"
             );
             Bloom.Book.Book.UpdateCharacterStyleMarkup(dom);
             Assert.That(
                 GetFirstEditableParagraph(dom).InnerXml,
-                Is.EqualTo("<strong>text</strong>")
+                Is.EqualTo("<strong>some text here</strong>")
             );
             var dom1 = new HtmlDom(
                 "<html><body><div class='bloom-editable'><p><i><i style='color:red'>text</i></i></p></div></body></html>"
@@ -3199,7 +3199,11 @@ public void UpdateCharacterStyleMarkup_ReducesNestingOfSameTags()
                 "<html><body><div class='bloom-editable'><p><sup><sup style='color:red'>text</sup></sup></p></div></body></html>"
             );
             Bloom.Book.Book.UpdateCharacterStyleMarkup(dom2);
-            Assert.That(GetFirstEditableParagraph(dom2).InnerXml, Is.EqualTo("<sup>text</sup>"));
+            // <sup> does have a nesting effect, so we don't collapse them. We just remove the attributes from the inner tag.
+            Assert.That(
+                GetFirstEditableParagraph(dom2).InnerXml,
+                Is.EqualTo("<sup><sup>text</sup></sup>")
+            );
             var dom3 = new HtmlDom(
                 "<html><body><div class='bloom-editable'><p>This is <u><u style='color:red'>some text.</u></u></p></div></body></html>"
             );
