header-ids contain HTML entities

[justanotheranonymoususer]

Version: 9a88ce1

import markdown2

# header-ids contain HTML entities
# Expected second id to be "othertext", not "ltothertext"
# Also weird result in safe mode
markdown_text = R"""
# <sometext>
# &lt;othertext
"""
print(markdown2.markdown(markdown_text, #safe_mode='escape',
                         extras={'header-ids': None}))

[Crozzers]

I'm not sure I see anything wrong with this. If I compare to two other python markdown processors:
Markdown

<h1 id="_1"><sometext></h1>
<h1 id="othertext">&lt;othertext</h1>

Marko

<h1 id=""><sometext></h1>
<h1 id="othertext">&lt;othertext</h1>

Both of them strip the entire < snippet, but neither of them handle <sometext> like we do.

I would argue as well that if your header contains a < symbol, then leaving lt in the header ID is desired. It's a better representation of the actual header and more descriptive of its contents.

Take this example:

# <othertext
# othertext

If we completely strip the < segment we get:

<h1 id="othertext">&lt;othertext</h1>
<h1 id="othertext-2">othertext</h1>

Two near identical header IDs, and its much harder to tell which one belongs to which header

[Crozzers]

Regarding the weird safe mode result, I get this:

<h1 id="&lt;sometext&gt;">&lt;sometext&gt;</h1>

<h1 id="ltothertext">&lt;othertext</h1>

But the HTML spec for ID attrs says:

the id attribute value must be unique amongst all the IDs in the element's tree and must contain at least one character. The value must not contain any ASCII whitespace.

At least one char, no whitespace. These IDs are compliant

[justanotheranonymoususer]

I would argue as well that if your header contains a < symbol, then leaving lt in the header ID is desired

OK, if that's by design I guess it's fine.

Regarding the weird safe mode result [...]

That's weird if having an id contain < is by design, even if it's compliant. But at the very least, there's an inconsistency that can be seen in the output:

<h1 id="&lt;sometext&gt;">&lt;sometext&gt;</h1>

<h1 id="ltothertext">&lt;othertext</h1>

I'd expect the initial < to either stay as < in the id, or (IMO preferably) turn into juts lt, but not to have one or the other depending on how it was escaped.