Fixing #408 Exponential Time Complexity Bug Causing Hangs with Nested <p> Tags (#409)

ssperling-wmp · web-flow · commit 79cb821b0ec4 · 2025-10-27T07:30:16.000+05:30
diff --git a/src/ReverseMarkdown.Test/ConverterTests.Bug393_RegressionWithVaryingNewLines.verified.md b/src/ReverseMarkdown.Test/ConverterTests.Bug393_RegressionWithVaryingNewLines.verified.md
@@ -1,6 +1,6 @@
 ﻿This is regular text
 
-This is HTML: 
+This is HTML:
 
 * Line 1
 * Line 2
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_AlternatingEmptyAndFilledNestedParagraphs_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_AlternatingEmptyAndFilledNestedParagraphs_ThenConvertCorrectly.verified.md
@@ -0,0 +1,9 @@
+﻿A
+
+B
+
+C
+
+D
+
+E
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_DeeplyNestedParagraphs_WithMalformedHTML_ThenConvertWithoutHanging.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_DeeplyNestedParagraphs_WithMalformedHTML_ThenConvertWithoutHanging.verified.md
@@ -0,0 +1 @@
+﻿
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_EmptyNestedParagraphs_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_EmptyNestedParagraphs_ThenConvertCorrectly.verified.md
@@ -0,0 +1 @@
+﻿
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_InterleavedParagraphsAndSpans_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_InterleavedParagraphsAndSpans_ThenConvertCorrectly.verified.md
@@ -0,0 +1,4 @@
+﻿Text1
+Text2
+Text3
+Text4
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly.verified.md
@@ -0,0 +1,15 @@
+﻿Part1
+
+Part2
+
+Part3
+
+Part4
+
+Part5
+
+Part6
+
+Part7
+
+Part8
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly.verified.md
@@ -0,0 +1,9 @@
+﻿Level1
+
+Level2
+
+Level3
+
+Level4
+
+Level5
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_TenLevelsDeep_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_TenLevelsDeep_ThenConvertCorrectly.verified.md
@@ -0,0 +1,19 @@
+﻿L1
+
+L2
+
+L3
+
+L4
+
+L5
+
+L6
+
+L7
+
+L8
+
+L9
+
+L10
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly.verified.md
@@ -0,0 +1 @@
+﻿Level1Level2Level3Level4Level5
diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_UnclosedParagraphsWithSpansAndTextNodes_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_UnclosedParagraphsWithSpansAndTextNodes_ThenConvertCorrectly.verified.md
@@ -0,0 +1,5 @@
+﻿Intro
+ Filler text here. 
+Section1
+Section2
+Section3
diff --git a/src/ReverseMarkdown.Test/ConverterTests.cs b/src/ReverseMarkdown.Test/ConverterTests.cs
@@ -1518,5 +1518,122 @@ public Task Bug400_MissingSpanSpaceWithItalics()
             var html = "<h3 data-reset-style=\"true\" data-anchor-id=\"8b5e184d-26f7-4d9a-80e0-bab2cd825457\"><i style=\"font-size: 14pt;\">What we thought:<span>&nbsp;</span></i><span style=\"color: rgb(41, 63, 77); font-size: 14pt; font-weight: normal;\">When we built Pages, we assumed that customers would use them like newsletters to share relevant, continually-updated information with field teams.</span><div style=\"text-align: left;\"><span style=\"line-height: 16px;\"><span><span height=\"18\" width=\"18\"><span></span></span><span></span></span></span></div></h3>";
             return CheckConversion(html);
         }
+
+        [Fact]
+        public Task When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly()
+        {
+            // Tests moderately nested <p> tags where HtmlAgilityPack creates nested structure
+            var html = "<p>Level1<p>Level2<p>Level3<p>Level4<p>Level5</p></p></p></p></p>";
+            
+            var config = new Config
+            {
+                GithubFlavored = true,
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
+
+        [Fact]
+        public Task When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly()
+        {
+            // Tests moderately nested <span> tags to ensure span bypass converter handles nesting
+            var html = "<span>Level1<span>Level2<span>Level3<span>Level4<span>Level5</span></span></span></span></span>";
+            
+            var config = new Config
+            {
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
+
+        [Fact]
+        public Task When_InterleavedParagraphsAndSpans_ThenConvertCorrectly()
+        {
+            // Tests the interleaved <p><span> pattern common in malformed HTML
+            var html = "<p><span>Text1<p><span>Text2<p><span>Text3<p><span>Text4</span></p></span></p></span></p>";
+            
+            var config = new Config
+            {
+                GithubFlavored = true,
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
+
+        [Fact]
+        public Task When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly()
+        {
+            // Tests sequential unclosed <p> tags as found in user-generated content
+            var html = "<p>Part1<p>Part2<p>Part3<p>Part4<p>Part5<p>Part6<p>Part7<p>Part8";
+            
+            var config = new Config
+            {
+                GithubFlavored = true,
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
+
+        [Fact]
+        public Task When_UnclosedParagraphsWithSpansAndTextNodes_ThenConvertCorrectly()
+        {
+            // Tests mixed content: properly closed tags, text nodes, and unclosed nested tags
+            var html = @"<p><span>Intro</span></p> Filler text here. <p><span>Section1<p><span>Section2<p>Section3";
+            
+            var config = new Config
+            {
+                GithubFlavored = true,
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
+
+        [Fact]
+        public Task When_EmptyNestedParagraphs_ThenConvertCorrectly()
+        {
+            // Tests deeply nested empty <p> tags
+            var html = "<p><p><p><p><p></p></p></p></p></p>";
+            
+            var config = new Config
+            {
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
+
+        [Fact]
+        public Task When_AlternatingEmptyAndFilledNestedParagraphs_ThenConvertCorrectly()
+        {
+            // Tests combination of empty and filled nested <p> tags
+            var html = "<p>A<p><p>B<p><p>C<p><p>D<p><p>E</p></p></p></p></p></p></p></p></p>";
+            
+            var config = new Config
+            {
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
+
+        [Fact]
+        public Task When_NestedParagraphs_TenLevelsDeep_ThenConvertCorrectly()
+        {
+            // Tests deeper nesting (10 levels) to ensure performance remains linear after fix
+            var html = "<p>L1<p>L2<p>L3<p>L4<p>L5<p>L6<p>L7<p>L8<p>L9<p>L10</p></p></p></p></p></p></p></p></p></p>";
+            
+            var config = new Config
+            {
+                GithubFlavored = true,
+                UnknownTags = Config.UnknownTagsOption.Bypass
+            };
+            
+            return CheckConversion(html, config);
+        }
     }
 }
diff --git a/src/ReverseMarkdown/Converters/P.cs b/src/ReverseMarkdown/Converters/P.cs
@@ -16,9 +16,13 @@ public override string Convert(HtmlNode node)
             var indentation = IndentationFor(node);
             var newlineAfter = NewlineAfter(node);
 
-            var content = Converter.Config.CleanupUnnecessarySpaces ? TreatChildren(node).Trim() : TreatChildren(node); 
+            var content = TreatChildren(node);
+            if (Converter.Config.CleanupUnnecessarySpaces)
+            {
+                content = content.Trim();
+            }
 
-            return $"{indentation}{TreatChildren(node)}{newlineAfter}";
+            return $"{indentation}{content}{newlineAfter}";
         }
 
         private static string IndentationFor(HtmlNode node)

-Original file line number
+Diff line change
@@ @@ -0,0 +1,9 @@ @@
 +A
++
 +B
++
 +C
++
 +D
++
 +E