From cdc4687db56950fa8e7a572fd93717aa66fadeb3 Mon Sep 17 00:00:00 2001 From: Samuel Sperling Date: Mon, 27 Oct 2025 11:20:25 +1100 Subject: [PATCH] Fixing #408 Exponential Time Complexity Bug Causing Hangs with Nested

Tags --- ..._RegressionWithVaryingNewLines.verified.md | 2 +- ...aragraphs_ThenConvertCorrectly.verified.md | 9 ++ ...HTML_ThenConvertWithoutHanging.verified.md | 1 + ...aragraphs_ThenConvertCorrectly.verified.md | 1 + ...sAndSpans_ThenConvertCorrectly.verified.md | 4 + ...aragraphs_ThenConvertCorrectly.verified.md | 15 +++ ...evelsDeep_ThenConvertCorrectly.verified.md | 9 ++ ...evelsDeep_ThenConvertCorrectly.verified.md | 19 +++ ...evelsDeep_ThenConvertCorrectly.verified.md | 1 + ...TextNodes_ThenConvertCorrectly.verified.md | 5 + src/ReverseMarkdown.Test/ConverterTests.cs | 117 ++++++++++++++++++ src/ReverseMarkdown/Converters/P.cs | 8 +- 12 files changed, 188 insertions(+), 3 deletions(-) create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_AlternatingEmptyAndFilledNestedParagraphs_ThenConvertCorrectly.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_DeeplyNestedParagraphs_WithMalformedHTML_ThenConvertWithoutHanging.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_EmptyNestedParagraphs_ThenConvertCorrectly.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_InterleavedParagraphsAndSpans_ThenConvertCorrectly.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_TenLevelsDeep_ThenConvertCorrectly.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly.verified.md create mode 100644 src/ReverseMarkdown.Test/ConverterTests.When_UnclosedParagraphsWithSpansAndTextNodes_ThenConvertCorrectly.verified.md diff --git a/src/ReverseMarkdown.Test/ConverterTests.Bug393_RegressionWithVaryingNewLines.verified.md b/src/ReverseMarkdown.Test/ConverterTests.Bug393_RegressionWithVaryingNewLines.verified.md index fa57f5f..90b692d 100644 --- 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 new file mode 100644 index 0000000..58a9f34 --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_AlternatingEmptyAndFilledNestedParagraphs_ThenConvertCorrectly.verified.md @@ -0,0 +1,9 @@ +A + +B + +C + +D + +E \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_DeeplyNestedParagraphs_WithMalformedHTML_ThenConvertWithoutHanging.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_DeeplyNestedParagraphs_WithMalformedHTML_ThenConvertWithoutHanging.verified.md new file mode 100644 index 0000000..5f28270 --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_DeeplyNestedParagraphs_WithMalformedHTML_ThenConvertWithoutHanging.verified.md @@ -0,0 +1 @@ + \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_EmptyNestedParagraphs_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_EmptyNestedParagraphs_ThenConvertCorrectly.verified.md new file mode 100644 index 0000000..5f28270 --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_EmptyNestedParagraphs_ThenConvertCorrectly.verified.md @@ -0,0 +1 @@ + \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_InterleavedParagraphsAndSpans_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_InterleavedParagraphsAndSpans_ThenConvertCorrectly.verified.md new file mode 100644 index 0000000..b9b0568 --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_InterleavedParagraphsAndSpans_ThenConvertCorrectly.verified.md @@ -0,0 +1,4 @@ +Text1 +Text2 +Text3 +Text4 \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly.verified.md new file mode 100644 index 0000000..7caf2ad --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly.verified.md @@ -0,0 +1,15 @@ +Part1 + +Part2 + +Part3 + +Part4 + +Part5 + +Part6 + +Part7 + +Part8 \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly.verified.md new file mode 100644 index 0000000..286680b --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly.verified.md @@ -0,0 +1,9 @@ +Level1 + +Level2 + +Level3 + +Level4 + +Level5 \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_TenLevelsDeep_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_NestedParagraphs_TenLevelsDeep_ThenConvertCorrectly.verified.md new file mode 100644 index 0000000..a78352c --- /dev/null +++ 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 \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly.verified.md new file mode 100644 index 0000000..c6e7929 --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly.verified.md @@ -0,0 +1 @@ +Level1Level2Level3Level4Level5 \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.When_UnclosedParagraphsWithSpansAndTextNodes_ThenConvertCorrectly.verified.md b/src/ReverseMarkdown.Test/ConverterTests.When_UnclosedParagraphsWithSpansAndTextNodes_ThenConvertCorrectly.verified.md new file mode 100644 index 0000000..747b6ad --- /dev/null +++ b/src/ReverseMarkdown.Test/ConverterTests.When_UnclosedParagraphsWithSpansAndTextNodes_ThenConvertCorrectly.verified.md @@ -0,0 +1,5 @@ +Intro + Filler text here. +Section1 +Section2 +Section3 \ No newline at end of file diff --git a/src/ReverseMarkdown.Test/ConverterTests.cs b/src/ReverseMarkdown.Test/ConverterTests.cs index fc4c971..803cf41 100644 --- a/src/ReverseMarkdown.Test/ConverterTests.cs +++ b/src/ReverseMarkdown.Test/ConverterTests.cs @@ -1518,5 +1518,122 @@ public Task Bug400_MissingSpanSpaceWithItalics() var html = "

What we thought: When we built Pages, we assumed that customers would use them like newsletters to share relevant, continually-updated information with field teams.

"; return CheckConversion(html); } + + [Fact] + public Task When_NestedParagraphs_FiveLevelsDeep_ThenConvertCorrectly() + { + // Tests moderately nested

tags where HtmlAgilityPack creates nested structure + var html = "

Level1

Level2

Level3

Level4

Level5

"; + + var config = new Config + { + GithubFlavored = true, + UnknownTags = Config.UnknownTagsOption.Bypass + }; + + return CheckConversion(html, config); + } + + [Fact] + public Task When_NestedSpans_FiveLevelsDeep_ThenConvertCorrectly() + { + // Tests moderately nested tags to ensure span bypass converter handles nesting + var html = "Level1Level2Level3Level4Level5"; + + var config = new Config + { + UnknownTags = Config.UnknownTagsOption.Bypass + }; + + return CheckConversion(html, config); + } + + [Fact] + public Task When_InterleavedParagraphsAndSpans_ThenConvertCorrectly() + { + // Tests the interleaved

pattern common in malformed HTML + var html = "

Text1

Text2

Text3

Text4

"; + + var config = new Config + { + GithubFlavored = true, + UnknownTags = Config.UnknownTagsOption.Bypass + }; + + return CheckConversion(html, config); + } + + [Fact] + public Task When_ManySequentialUnclosedParagraphs_ThenConvertCorrectly() + { + // Tests sequential unclosed

tags as found in user-generated content + var html = "

Part1

Part2

Part3

Part4

Part5

Part6

Part7

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 = @"

Intro

Filler text here.

Section1

Section2

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

tags + var html = "

"; + + 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

tags + var html = "

A

B

C

D

E

"; + + 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 = "

L1

L2

L3

L4

L5

L6

L7

L8

L9

L10

"; + + 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 index b90bc99..d6cfe20 100644 --- 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)