id of Anchor is wrong

[yuki-kimoto]

From metacpan/metacpan-web#2668

Grinnz said "This can be discussed in https://metacpan.org/dist/MetaCPAN-Pod-XHTML if necessary, as this behavior is controlled by those roles."

I post this issue.

The id of anchor is wrong when there are multiple head names that start with the same string.

=head1 Method Names

Method Names.

=head1 Method

Method.

=head2 Method Definition

Method Definition.

L<"Method">

L<"Method"> want to jump to "=head1 Method", but it jumps to "=head1 Method Names".

Pod Renderer renders the following HTML.

<h1 id="Method-Names"><a href="#Method-Names" class="anchor"><span class="fa fa-bookmark black"></span></a><a id="Method"></a>Method Names</h1>

<p>Method Names.</p>

<h1 id="Method1"><a href="#Method1" class="anchor"><span class="fa fa-bookmark black"></span></a>Method</h1>

<p>Method.</p>

<h2 id="Method-Definition"><a href="#Method-Definition" class="anchor"><span class="fa fa-bookmark black"></span></a><a id="Method2"></a>Method Definition</h2>

<p>Method Definition.</p>

<p><a href="#Method">"Method"</a></p></div>

Is this behavior changed?

[haarg]

This is definitely wrong, but fixing it is not easy.

The ids are generated as the Pod is parsed. Extra targets are generated for the first word of the heading, as it's very common for people to include things like parameter specifications in headings but try to link with just the function or method name. Since they are generated as the Pod is parsed, the abbreviated ids can conflict with future headings that consist of only the first word of a previous heading. To avoid this, the additional ids would need to be generated after parsing the entire document and inserted.

[yuki-kimoto]

I understand now the following heading create additional "foo" id .

=head1 foo ($nums)

This is definitely wrong, but fixing it is not easy.

To avoid this, the additional ids would need to be generated after parsing the entire document and inserted.

I will read the source codes of MetaCPAN-Pod-XHTML.

Could I try to fix this?

The fixing is "If there is an exact matching heading, the correct id is given to the heading"

=head1 Method Names -> Method1

Method Names.

=head1 Method -> Method

Method.

=head2 Method Definition -> Method2

Method Definition.

L<"Method">

What could be the problem with the fix?

[haarg]

The problem is that given the pod:

=head1 Method Names

Method Names.

=head1 Method

Method.

By the time the parser sees =head1 Method it has already generated the HTML for =head1 Method Names, which includes the Method target.

You are certainly welcome to try to fix this.

[yuki-kimoto]

Thank you to allow me to trying fix this.

I will try to fix this in few weeks.

[yuki-kimoto]

I try to fix this from today.

[yuki-kimoto]

@haarg

I'm reading source codes.

I want to get original POD that is passed to "parse_string_document" to get heading information.

https://github.com/metacpan/MetaCPAN-Pod-XHTML/blob/master/t/accurate-targets.t#L33

but Pod::Simple doesn't save original POD.

https://metacpan.org/module/Pod::Simple/source#L418

I want to hook "parse_string_document" to get heading information.

How do you think this?

[haarg]

I made an initial implementation of this in #5, but I'm going to hold off on releasing it for now.

I've worked on getting some fixes and refactoring merged into Pod::Simple itself. Once there is a new release of that, I'll be updating this dist to work with it, and then merging in something like I have in the PR I prepared for this issue.

[yuki-kimoto]

Thank you very much! pre_process method seems to allow us to parse whole html before parsing.

I wait for the merge.

[haarg]

This should be fixed in 0.003000.

[yuki-kimoto]

Thank you very much!

[yuki-kimoto]

This bug does not appear to be fixed in metaCPAN or Pod Renderer.

[oalders]

We may need to update the dep in metacpan-api