TitleIndex Macro should also take relative prefixes

[glinders@…]

The TitleIndex macro only takes the full path of a Wiki page.

For instance, if you are on Wiki page WikiStart:Top/Deeper/Deepest and want to include a list of all subpages of Deepest, you have to include [[TitleIndex(Top/Deeper/Deepest/)]] in your page.

It would be great if you could also use [[TitleIndex(./)]] to achieve the same.

Similarly, [[TitleIndex(../)]] should give the same result as [[TitleIndex(Top/Deeper/)]]

[Peter Suter]

The attached patch​ implements this functionality that I've been missing myself.

[Jun Omae]

I tried the patch. It is inconsistent between [[TitleIndex(./)]] and [[TitleIndex(SandBox/)]].

Also, we should add unit tests for this feature.

[Peter Suter]

Thanks for testing. You are of course correct on both accounts. I will attach an updated patch later.

[Peter Suter]

Replying to jomae:

It is inconsistent between [[TitleIndex(./)]] and [[TitleIndex(SandBox/)]]. Also, we should add unit tests for this feature.

This updated patch​ fixes the inconsistency ([[TitleIndex(./)]] does not list the current wiki page anymore) and adds unit tests.

(The fix is changing _resolve_relative_name:

-elif comp and comp != '.': 
+elif comp != '.': 

It is not immediately obvious that this breaks no other callers, but I can not come up with a scenario where this change would affect the only other caller​ after pagename.rstrip('/') and pagename.startswith(('./', '../')) or pagename in ('.', '..').)

[Jun Omae]

(The fix is changing _resolve_relative_name:

-elif comp and comp != '.': 
+elif comp != '.': 

It is not immediately obvious that this breaks no other callers, but I can not come up with a scenario where this change would affect the only other caller​ after pagename.rstrip('/') and pagename.startswith(('./', '../')) or pagename in ('.', '..').)

After the changes, in SandBox page,

• ..//WikiStart will be WikiStart? text
• ../WikiStart will link to WikiStart page

Before the changes, both will link to WikiStart page.

But I think the patch is okay because page name doesn't have repeated-slashes.

Unit and functional tests pass on Python 2.5 - 2.7 with SQLite, PostgreSQL and MySQL on Linux.

[Peter Suter]

Committed in [12584]. Thanks for reviewing and testing. I had not considered such "invalid" page names with repeated slashes.