[Fix] parseMarkdown startDelimiter

[NicolasRoehm]

I was not able to build the pages on Windows 10 using unibit build or unibit develop.
It always outputs [Unibit] Generating 0 pages...
I removed the \n on each startDelimiter and it now works.
Surprisingly, endDelimiter works fine as is, for example, \n---.

[NicolasRoehm]

I have found other issues related to Windows 10 :

• In unibit.js -> getPages(), the split by '/' result is irrelevant because Windows provides '\\'
  • A trick can be to add some urlPath = urlPath.replace('\\', '/'); here and there
• In browser routing with base/index pages does not work and sends 404, Windows users must add /index manually at the end of the URL

Because Unibit works great on Linux, I'll continue to use it in the right way.
You can close this issue and/or ping me with guidelines to fix these issues properly.

Thank you for sharing this library, it's very useful and working well (on real OS) ! :)

[NicolasRoehm]

Updated in 49108ea

[smnh]

@NicolasRoehm thanks for the PR.

Good catch with the EOL.

Regarding path.sep, please see comments inline

[smnh]

Here we are handling page URL paths. URLs always use forward slash disregards to the underlying system /.
On windows system, the path.sep will be \ so the logic will not work for URL paths

[NicolasRoehm]

Here the path.sep is not about URL but rather about file routing/parsing.

On windows system, the page.url value prints \\blog\\author\\dianne-ameter.

• When we try to remove the leading and trailing / character(s) with _.trim, nothing happens :
  • pageUrl contains \\blog\\author\\dianne-ameter instead of blog\\author\\dianne-ameter.
• Then we split the url parts by the hardcoded /, so we get :
  • pageUrlParts = [ '\\blog\\author\\dianne-ameter' ] instead of ['blog', 'author', 'dianne-ameter'].

This causes the blog page to not list any blog posts and all subdirectory routes are skipped/broken.

[smnh]

Interesting,
Then I guess the root cause of this bug is somewhere before the execution of this code.
The page.url should never have OS specific separators and should always have standard URL forward slash - /.
Maybe the code that constructs the original url property?