pkginfo picks random package if multiple package with the same prefix are installes

[Gerhard Schmidt]

I'm trying to use pkginfo to retrive infos about some installed Packages.

e.g. these packages are installed

pyramid 2.0
pyramid-chameleon 0.3
pyramid-debugtoolbar 4.9
pyramid-deform 0.2
pyramid-layout 1.0.1
pyramid-mailer 0.15.1.1
pyramid-mako 1.1.0
pyramid-mytum 0.17.5
pyramid-retry 2.1.1
pyramid-session-cookie 1.0.2
pyramid-tm 2.5
pyramid-utilities 0.9.2

when in try to get information about pyramid-chameleon everything works fine.

When in want some information about pyramid, pkg returns one of the other packages.

the problem is the read method. It builds a list of all entries matching pyramid*.egg-info and pyramid*.dist-info and pick the first one that exists.

I changed the lines 34-35 to
            egg_pattern = '%s-*.egg-info' % package
            dist_pattern = '%s-*.dist-info' % package

fixes the problem for me. I don't think this is the best solution.

Regards
  Estartu

[Tres Seaver (tseaver)]

Thanks very much for the report!

I'm curious why you think your fix is not ideal. It looks to me as though any installed '.egg-info' or '.dist-info' directory *should* be prefixed by '<package_name>-'.

[Gerhard Schmidt (estartu71)]

I wasn't sure if _ and - are consistently exchanged. As you see in the
list the the package name are with - while the directory on the disks
are with _. So there might be packages that have a - in the directory
name which will break my fix again.

The version string is quite fixed in it's format. So it might be
possible to define a regular expression for the version information and
matching against <projectname>-<versionexpression>.{egg|dist)-info.
This whould be the premium solution. But I hadn't had the time to look
into it.

Am 19.05.22 um 18:31 schrieb Tres Seaver:
> Thanks very much for the report!
>
> I'm curious why you think your fix is not ideal. It looks to me as
> though any installed '.egg-info' or '.dist-info' directory *should* be
> prefixed by '<package_name>-'.
>

[Gerhard Schmidt (estartu71)]

the problem is still in 1.8.3

[Gerhard Schmidt (estartu71)]

the problem ist still in 1.9.6