Skip to content

.Net: fix: respect token limit when merging text chunks #14071

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
2830500285 wants to merge 2 commits into
base: main
Choose a base branch
from
+14 −7
Open

.Net: fix: respect token limit when merging text chunks #14071

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d9888b3
fix: respect token limit when merging text chunks
2830500285 Jun 13, 2026
b42ee54
test: clarify text chunker token counter
2830500285 Jun 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 3 additions & 7 deletions dotnet/src/SemanticKernel.Core/Text/TextChunker.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -195,15 +195,11 @@ private static List<string> ProcessParagraphs(List<string> paragraphs, int adjus
var lastParagraphTokens = lastParagraph.Split(s_spaceChar, StringSplitOptions.RemoveEmptyEntries);
var secondLastParagraphTokens = secondLastParagraph.Split(s_spaceChar, StringSplitOptions.RemoveEmptyEntries);

var lastParagraphTokensCount = lastParagraphTokens.Length;
var secondLastParagraphTokensCount = secondLastParagraphTokens.Length;
var mergedParagraph = $"{string.Join(" ", secondLastParagraphTokens)} {string.Join(" ", lastParagraphTokens)}";

if (lastParagraphTokensCount + secondLastParagraphTokensCount <= adjustedMaxTokensPerParagraph)
if (GetTokenCount(mergedParagraph, tokenCounter) <= adjustedMaxTokensPerParagraph)
{
var newSecondLastParagraph = string.Join(" ", secondLastParagraphTokens);
var newLastParagraph = string.Join(" ", lastParagraphTokens);

paragraphs[paragraphs.Count - 2] = $"{newSecondLastParagraph} {newLastParagraph}";
paragraphs[paragraphs.Count - 2] = mergedParagraph;
paragraphs.RemoveAt(paragraphs.Count - 1);
}
Comment on lines 195 to 204

@2830500285 2830500285 Jun 13, 2026

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks. I kept the merge decision on the configured token counter because the whitespace-delimited token array length is the mismatch that caused the bug. This block only runs after the last paragraph is already known to be short, and using word-count short-circuiting for the default path could still disagree with the token counter semantics.

}
Expand Down
11 changes: 11 additions & 0 deletions dotnet/src/SemanticKernel.UnitTests/Text/TextChunkerTests.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -558,6 +558,17 @@ public void CanSplitTextParagraphsWithCustomTokenCounter()
Assert.Equal(expected, result);
}

[Fact]
public void SplitTextParagraphsDoesNotMergeShortLastParagraphPastTokenLimit()
{
var input = new[] { "123456789", "x" };

var result = TextChunker.SplitPlainTextParagraphs(input, 10, tokenCounter: text => text.Length);

Assert.Equal(["123456789", "x"], result);
Assert.All(result, paragraph => Assert.True(paragraph.Length <= 10, $"Paragraph exceeded token limit: {paragraph}"));
}

[Fact]
public void CanSplitTextParagraphsWithOverlapAndCustomTokenCounter()
{
Expand Down
Loading