Doxygen custom pages not linked correctly.

[crud89]

Hey there! :)
I am using your framework with doxygen 1.8.13 and got everything working well so far, except for the Pages page. I want to include custom page files, generated from markdown text. Basically I want to organize my tutorials that way. So say I have a file ./docs/tutorials/helloworld.md. I then include the file by adding it to the INPUT list:

INPUT = ../ \
        tutorials/helloworld.md

Doxygen links the resulting md_tutorials__helloworld.html file correctly, however when I run it through dox2html.py, the created link is invalid, since it points to md_tutorials_helloworld.html. Note that there is one underscore missing. The debug output does not give much meaningful information either, since it just tells me that it does not find a brief or detailed description. However, the correct file is created, just the link within the Pages page is invalid.

Any ideas?

[nephatrine]

I'm having a similar issue. I have a LICENSE.md that Doxygen is putting into its own page. The XML generated is:

<compound refid="md__l_i_c_e_n_s_e" kind="page"><name>md_LICENSE</name></compound>

In the HTML that m.css creates for that, it generates a file md__l_i_c_e_n_s_e.html. I can get to it without issues from the search bar, but the hyperlink in the Pages view links to md_license.html which doesn't exist and returns an error. For some reason the Pages page is eating the underscores?

[mosra]

Hi all, sorry for the late reply, I am extremely busy right now...

I think both of these issues are related to the CASE_SENSE_NAMES option in the Doxyfile. It defaults to NO on Windows and causes the extra underscores to be added ... in some cases, but not all, which means I get a very inconsistent and sometimes irrepairable data in the XML files :/

Can you try setting this option to YES to see if it fixes anything?

[nephatrine]

@mosra I can confirm that setting CASE_SENSE_NAMES to YES corrected the issue for me, though that doesn't seem like an ideal solution.

Looking at dox2html5 though, I noticed this:

# Compound URL is ID, except for index page
compound.url = (compounddef.find('compoundname').text if compound.kind == 'page' else compound.id) + '.html'

I'm not much of a python guy so the order of that if statement is a bit confusing to me, but it looks like it's using the compoundname instead of compound.id for all pages. In at least my case though, that's what is causing the mismatch. If I change that line of code to always use compound.id, then the generated HTML is correct and the Pages links all work as does the index page and nothing seems to break, but this is just a small test project so idk.

[mosra]

Uh, sorry, I forgot that I already implemented some support for CASE_SENSE_NAMES = NO as a fix for #42. Turns out I missed one thing, thank you @nephatrine for pointing it out. Should be fixed in 03b69fd.

@Aschratt I was not able to reproduce the problem you're having with the exact casing you mentioned, but trying out tutorials/Helloworld.md (uppercase H) together with CASE_SENSE_NAMES = NO resulted in the same double underscore you're having. I think this is the same case and so the above commit should fix that too.

Closing as resolved, please complain if something is still wrong ;)

[crud89]

Hi! I tried again with your changes and can confim, that it now works with CASE_SENSE_NAMES = NO. Thank you! 👍

[mosra]

Awesome, thanks for the confirmation! :)