The Markdown Lexer does not handle HTML comments

[jeffreyforan]

This causes the markdown lexer to be wrong when an html comment starts with specific characters.

For example, the following text does not get lexed correctly (bar and baz are italicized):

<!--* foo *-->

bar
baz

[nyamatongwe]

Seems OK to me.

Issue348.md

[zufuliu]

I guess OP want basic tag highlighting:

<!-- *foo* -->

*foo*
<a href=" *foo* "></a>

[nyamatongwe]

It looks to me that the reported behaviour is from a modified Markdown lexer or the Notepad++ UDL (User Defined Language) lexer for Markdown. Notepad++ doesn't show the described result for me although it does recognize the comment.

[jeffreyforan]

No: not "" but rather "".

Thank you for the quick response!

[jeffreyforan]

Per Claude:

I ran into this in the wxPython-bundled Scintilla (which carries an older copy of the Markdown lexer) and fixed it
by adding an SCE_MARKDOWN_COMMENT state.

The problem: <!-- is not recognized, so any Markdown syntax characters inside a comment (e.g. *, _, ~~) are parsed
normally and affect highlighting of subsequent text.

The fix: Add state SCE_MARKDOWN_COMMENT = 22. Detect . This suppresses all inline markup
parsing inside the comment and correctly handles multi-line comments.

Relevant changes to LexMarkdown.cxx:

// In the state machine — exit comment on -->
else if (sc.state == SCE_MARKDOWN_COMMENT) {
if (sc.Match("-->")) {
sc.Forward(2);
sc.SetState(SCE_MARKDOWN_DEFAULT);
}
}

// In LINE_BEGIN handling — detect opening comment
else if (sc.Match("<!--")) {
sc.SetState(SCE_MARKDOWN_COMMENT);
sc.Forward(3);
}

// In DEFAULT handling — detect inline comment
if (sc.Match("<!--")) {
sc.SetState(SCE_MARKDOWN_COMMENT);
sc.Forward(3);
}

And in SciLexer.h:
#define SCE_MARKDOWN_COMMENT 22

[nyamatongwe]

The original problem statement is incorrect as demonstrated.

Per Claude

No. I will not try to make sense of AI-generated text.

[jeffreyforan]

My apologies, the text entry in this bug requries strange escaping to be human readable:

"<!--* foo *--> bar" (spaces around foo, not around the asterisks) italicizes "bar".

Your example was "<!-- *foo* --> bar" which works correctly in the current case.

Feel free to ignore Claude's fixes as appropriate for the code base.

[nyamatongwe]

The current version of LexMarkdown from wxWidgets appears the same as this repository (except for some minor changes to includes) so will be OK.

The reason that the * (SCE_MARKDOWN_EM1) italics mode doesn't leak onto the following bar text in the current code is because SCE_MARKDOWN_EM1 is only entered when a candidate * is followed by a terminating * on the same line. Therefore SCE_MARKDOWN_EM1 is local to a line.