Yourls changes link

[mike14-2]

Hello,
I have the problem that I have a link which also contains capital letters. These must remain capitalized.
However, Yourls changes this link to be resolved.
Can anyone tell me what I need to do differently?
Thank you very much

[dgw]

Based on additional details posted in another thread, I believe the problem is that an application called DEVONthink misapplies the URI format in its own links as follows:

x-devonthink-item://930FBCF1-CB42-45B8-9092-904030947FBF
\________________/^^\__________________________________/
        |         |                  |
  custom scheme   |              path (UUID)
                  |
        authority prefix in URI
        that defines only a path

// immediately after the scheme: is reserved for an optional "authority" component—commonly, a hostname. In the next URI component, the "path", // can appear if a path segment is empty—but the first path segment may not be empty if no authority is present in the URI. See more details in RFC 3986 or in slightly plainer English at Wikipedia.

I think this 5-year-old thread about DEVONthink links in Anki means that Anki is parsing URIs correctly, and DEVONthink needs to use zero, one, or three slashes, but not two for its links. If it is going to continue using URIs as they currently are, where the item URL is constructed as if the item UUID is a hostname, the item UUID should be treated as case-insensitive when the user clicks an item link.

Further detail of my investigation in the collapsible section below.

Extra analysis

Like YOURLS' normalization function written in PHP, the Python standard-library URI parser urllib.parse.urlparse() also thinks this item UUID is a network location instead of a path:

>>> from urllib.parse import urlparse
>>> res = urlparse('x-devonthink-item://930FBCF1-CB42-45B8-9092-904030947FBF')
>>> res
ParseResult(scheme='x-devonthink-item', netloc='930FBCF1-CB42-45B8-9092-904030947FBF', path='', params='', query='', fragment='')

As I suggested above, using any of the possible slash counts other than the two that DEVONthink currently uses will work:

>>> noslash = urlparse('x-devonthink-item:930FBCF1-CB42-45B8-9092-904030947FBF')
>>> noslash
ParseResult(scheme='x-devonthink-item', netloc='', path='930FBCF1-CB42-45B8-9092-904030947FBF', params='', query='', fragment='')
>>> oneslash = urlparse('x-devonthink-item:/930FBCF1-CB42-45B8-9092-904030947FBF')
>>> oneslash
ParseResult(scheme='x-devonthink-item', netloc='', path='/930FBCF1-CB42-45B8-9092-904030947FBF', params='', query='', fragment='')
>>> threeslash = urlparse('x-devonthink-item:///930FBCF1-CB42-45B8-9092-904030947FBF')
>>> threeslash
ParseResult(scheme='x-devonthink-item', netloc='', path='/930FBCF1-CB42-45B8-9092-904030947FBF', params='', query='', fragment='')

Zero slashes is the best choice if they do not want to handle the leading / of the resulting path, but three slashes is likely the most "normal-looking" to most users who have never paid attention to the exact number of slashes in the links they click every day.

[mike14-2]

Hello,
sorry for posting my answer incorrectly.
I have understood it, I think.
I now use “///” and it works.
Brilliant.

Thank you very much