Markdown Tree Parser Truncates Sections At # Comments Within Embedded Bash Code Blocks

[halso]

Describe the bug
The md-tree parser truncates section prematurely if they have code blocks in with comments starting with # because the tools thinks it has hit a Level 1 Header.

Steps to Reproduce

Run:

md-tree explode architectureBugTestVersion.md

on the following file:

architectureBugTestVersion.md

PR

If this can be fixed by the md-tree devs, that's ideal. I've raised this as a bug at:

ksylvan/markdown-tree-parser#7

Expected behavior

Expected:

architectureBugTestVersion/development-workflow.md

should contain the full "Development Workflow" section.

Actual behavior

architectureBugTestVersion/development-workflow.md

only contain the "Development Workflow" section up to the # comment in the bash code excerpt.

Here's a screenshot showing the truncated section:

Additional context

This is causing the agents to get confused as they can't find sections that should be in the sharded document - for example "Code Quality and Standards" which is a subsection of the "Development Workflow" section.

[manjaroblack]

When I get this error, I just tell my ai agent to fix the markdown file. the agent will reduce the needed sections down to ## level headings so that the md-tree can parse it correctly.

[halso]

When I get this error, I just tell my ai agent to fix the markdown file. the agent will reduce the needed sections down to ## level headings so that the md-tree can parse it correctly.

@manjaroblack thanks for the reply. I got Claude to fix the bug in the md-tree parser project. First time round it over engineered the solution. Second time I asked it for a minimal solution and it did a good job. It just needed to turn off document sharding while it was inside a code section, so any comments would not be misinterpreted as headings. I asked it to create a pull request, but I hadn't forked the repo, so have to follow the proper process (never done that before...). It's funny that I raised this bug without thinking of getting AI to fix it. Shows me that it's going to be a while before we realise the full power of this technology...

[bmadcode]

@ksylvan wonder if you have some ideas on if it will be doable to make it so these code blocks do not get including in calculation of the doc levels. maybe a token counter for start and end ```n so we know if we are in a code block or nexted code blocks with n number of `'s

[ksylvan]

I'll take a look.

Please file a bug at my repo for md-tree

[halso]

@ksylvan - I asked Claude Code to fix this and it seems to have done a good job. The bug issue and the pull request are at:
ksylvan/markdown-tree-parser#7
cc @bmadcode

[halso]

@bmadcode good news. Looks like @ksylvan has fixed this in ksylvan/markdown-tree-parser#9 and I think now all you have to do is update markdown-tree-parser to 1.6.1 to get the fix? Thanks @ksylvan for this, and is the version right?

[ksylvan]

Yes, that is correct @halso

[ksylvan]

@bmadcode Brian, I created a PR to update this and a few other packages.