fix(bluebubbles): preserve intra-word underscores in stripMarkdown by sudabg · Pull Request #46292 · openclaw/openclaw

sudabg · 2026-03-14T15:34:22Z

Summary

The stripMarkdown function used in BlueBubbles outbound text processing strips single underscores inside words (e.g., here_is_a_message_with_underscores → hereisamessagewithunderscores).

Root Cause

The underscore-based italic regex (?<!_)_(?!_)(.+?)(?<!_)_(?!_) matches _text_ anywhere, even when underscores are clearly word-internal separators rather than markdown italic delimiters.

Fix

Updated the regex to require whitespace or start/end position around underscore delimiters:

// Before
result.replace(/(?<!_)_(?!_)(.+?)(?<!_)_(?!_)/g, "$1");

// After (CommonMark-compliant)
result.replace(/(?<=\s|^)_(?!_)(.+?)(?<!_)_(?=\s|$|[.,;:!?)/\]])/g, "$1");

This matches CommonMark spec for flanking delimiter runs — only _text_ with proper word boundaries is treated as italic formatting. Underscores inside words (like snake_case identifiers) are preserved.

Test Cases

Input	Before	After
`here_is_a_message`	`hereisamessage`	`here_is_a_message` ✅
`hello _world_ there`	`hello world there`	`hello world there` ✅
`_italic text_`	`italic text`	`italic text` ✅
`snake_case_var`	`snakecasevar`	`snake_case_var` ✅

The underscore-based italic regex was too aggressive and stripped underscores inside words like `here_is_a_message_with_underscores`. Fix: Require whitespace or start/end position around underscore delimiters, matching CommonMark spec for flanking delimiter runs. Only underscores that look like actual markdown italic formatting (`_text_` with proper boundaries) are now stripped. Fixes openclaw#46185

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7f8184f075

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

chatgpt-codex-connector · 2026-03-14T15:36:45Z

src/line/markdown-to-line.ts

  result = result.replace(/(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)/g, "$1");
-  result = result.replace(/(?<!_)_(?!_)(.+?)(?<!_)_(?!_)/g, "$1");
+  // Only match _text_ when underscores are at word boundaries (CommonMark-compliant)
+  result = result.replace(/(?<=\s|^)_(?!_)(.+?)(?<!_)_(?=\s|$|[.,;:!?)/\]])/g, "$1");


Allow punctuation-flanked _italic_ delimiters

The new underscore-italic regex only opens on (?<=\s|^), so valid markdown like (_italic_) or "_italic_" is no longer stripped and leaks literal underscores into plain-text output. In stripMarkdown, this is a regression because the function is intended to remove markdown markers generally, and CommonMark permits underscore emphasis when delimiters are adjacent to punctuation (not only whitespace/start).

Useful? React with 👍 / 👎.

greptile-apps · 2026-03-14T15:36:56Z

Greptile Summary

This PR fixes a bug in stripMarkdown where the underscore-italic regex was incorrectly stripping underscores inside snake_case identifiers (e.g. here_is_a_message → hereisamessage). The fix works for the reported case, but the updated regex introduces a regression for italic text that appears after opening punctuation characters.

Fix is correct for the reported case: snake_case identifiers like here_is_a_message are now preserved since none of their underscores are preceded by whitespace or start-of-string.
Regression for italic after punctuation: The (?<=\s|^) lookbehind is too strict — valid italic like (_italic_), "_italic_", or [_see this_] will no longer be stripped, leaving raw underscores in the plain-text output.
Closing lookahead is asymmetric: the closing _ requires (?=\s|$|[.,;:!?)/\]]) (specific punctuation), but opening punctuation characters ((, [, ", etc.) are not included in the lookbehind.

Confidence Score: 3/5

The fix resolves the reported snake_case bug but introduces a regression for italic text after opening punctuation characters.
The change correctly fixes the snake_case identifier stripping issue, but the (?<=\s|^) lookbehind is overly restrictive. Italic-marked text following (, [, ", or ' will no longer be stripped, leaving raw underscores in plain-text output where none were expected before.
src/line/markdown-to-line.ts — the updated italic regex needs a broader lookbehind to cover opening punctuation characters.

Prompt To Fix All With AI

This is a comment left during a code review.
Path: src/line/markdown-to-line.ts
Line: 354

Comment:
**Opening delimiter lookbehind too restrictive — regression for italic after punctuation**

The `(?<=\s|^)` lookbehind only allows whitespace or start-of-string before the opening `_`. This means valid italic markdown that immediately follows opening punctuation — such as `(`, `[`, `"`, or `'` — will **no longer be stripped**, leaving raw underscores in the plain-text output.

For example:
| Input | Before fix | After this fix |
|---|---|---|
| `(_italic text_)` | `(italic text)` ✅ | `(_italic text_)` ❌ |
| `"_italic_"` | `"italic"` ✅ | `"_italic_"` ❌ |
| `[_see this_]` | `[see this]` ✅ | `[_see this_]` ❌ |

A more complete fix would also allow opening punctuation characters in the lookbehind, matching closer to the CommonMark spec (which says a left-flanking delimiter run may be preceded by Unicode punctuation):

```suggestion
  result = result.replace(/(?<=\s|^|[([{"'`])_(?!_)(.+?)(?<!_)_(?=\s|$|[.,;:!?)/\]}"'`])/g, "$1");
```

This extends both the lookbehind and the closing lookahead to cover matching punctuation pairs, preserving the snake_case fix while not regressing on italic-inside-punctuation cases.

How can I resolve this? If you propose a fix, please make it concise.

_{Last reviewed commit: 7f8184f}

greptile-apps · 2026-03-14T15:36:59Z

src/line/markdown-to-line.ts

  result = result.replace(/(?<!\*)\*(?!\*)(.+?)(?<!\*)\*(?!\*)/g, "$1");
-  result = result.replace(/(?<!_)_(?!_)(.+?)(?<!_)_(?!_)/g, "$1");
+  // Only match _text_ when underscores are at word boundaries (CommonMark-compliant)
+  result = result.replace(/(?<=\s|^)_(?!_)(.+?)(?<!_)_(?=\s|$|[.,;:!?)/\]])/g, "$1");


Opening delimiter lookbehind too restrictive — regression for italic after punctuation

The (?<=\s|^) lookbehind only allows whitespace or start-of-string before the opening _. This means valid italic markdown that immediately follows opening punctuation — such as (, [, ", or ' — will no longer be stripped, leaving raw underscores in the plain-text output.

For example:

Input Before fix After this fix

(_italic text_) (italic text) ✅ (_italic text_) ❌

"_italic_" "italic" ✅ "_italic_" ❌

[_see this_] [see this] ✅ [_see this_] ❌

A more complete fix would also allow opening punctuation characters in the lookbehind, matching closer to the CommonMark spec (which says a left-flanking delimiter run may be preceded by Unicode punctuation):

Suggested change

result = result.replace(/(?<=\s|^)_(?!_)(.+?)(?<!_)_(?=\s|$|[.,;:!?)/\]])/g, "$1");

result = result.replace(/(?<=\s|^|[([{"'`])_(?!_)(.+?)(?<!_)_(?=\s|$|[.,;:!?)/\]}"'`])/g, "$1");

This extends both the lookbehind and the closing lookahead to cover matching punctuation pairs, preserving the snake_case fix while not regressing on italic-inside-punctuation cases.

Prompt To Fix With AI

This is a comment left during a code review. Path: src/line/markdown-to-line.ts Line: 354 Comment: **Opening delimiter lookbehind too restrictive — regression for italic after punctuation** The `(?<=\s|^)` lookbehind only allows whitespace or start-of-string before the opening `_`. This means valid italic markdown that immediately follows opening punctuation — such as `(`, `[`, `"`, or `'` — will **no longer be stripped**, leaving raw underscores in the plain-text output. For example: | Input | Before fix | After this fix | |---|---|---| | `(_italic text_)` | `(italic text)` ✅ | `(_italic text_)` ❌ | | `"_italic_"` | `"italic"` ✅ | `"_italic_"` ❌ | | `[_see this_]` | `[see this]` ✅ | `[_see this_]` ❌ | A more complete fix would also allow opening punctuation characters in the lookbehind, matching closer to the CommonMark spec (which says a left-flanking delimiter run may be preceded by Unicode punctuation): ```suggestion result = result.replace(/(?<=\s|^|[([{"'`])_(?!_)(.+?)(?<!_)_(?=\s|$|[.,;:!?)/\]}"'`])/g, "$1"); ``` This extends both the lookbehind and the closing lookahead to cover matching punctuation pairs, preserving the snake_case fix while not regressing on italic-inside-punctuation cases. How can I resolve this? If you propose a fix, please make it concise.

openclaw-barnacle bot added the size: XS label Mar 14, 2026

chatgpt-codex-connector bot reviewed Mar 14, 2026

View reviewed changes

greptile-apps bot reviewed Mar 14, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(bluebubbles): preserve intra-word underscores in stripMarkdown#46292

fix(bluebubbles): preserve intra-word underscores in stripMarkdown#46292
sudabg wants to merge 1 commit intoopenclaw:mainfrom
sudabg:fix/stripmarkdown-intra-word

sudabg commented Mar 14, 2026

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 14, 2026

Uh oh!

greptile-apps bot commented Mar 14, 2026

Uh oh!

greptile-apps bot Mar 14, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Input	Before fix	After this fix
`(_italic text_)`	`(italic text)` ✅	`(_italic text_)` ❌
`"_italic_"`	`"italic"` ✅	`"_italic_"` ❌
`[_see this_]`	`[see this]` ✅	`[_see this_]` ❌

	result = result.replace(/(?<=\s\|^)_(?!_)(.+?)(?<!_)_(?=\s\|$\|[.,;:!?)/\]])/g, "$1");
	result = result.replace(/(?<=\s\|^\|[([{"'`])_(?!_)(.+?)(?<!_)_(?=\s\|$\|[.,;:!?)/\]}"'`])/g, "$1");

Uh oh!

Conversation

sudabg commented Mar 14, 2026

Summary

Root Cause

Fix

Test Cases

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 14, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps bot commented Mar 14, 2026

Greptile Summary

Confidence Score: 3/5

Uh oh!

greptile-apps bot Mar 14, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant