Reference: https://trello.com/c/Ohh7Q6qjSigned-off-by: NCaio Carrara <ccarrara@redhat.com>

There was an else on while without break. That has no meaning because
since while has no break, the else block will be always executed. This
change removes the else statement and keeps the block.
Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

Reference: https://trello.com/c/Ohh7Q6qjSigned-off-by: NCaio Carrara <ccarrara@redhat.com>

Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

This started as a hunt for "foo" doesn't exist in module "bar" errors
in "inspekt lint" checks.  It looks like it fixes those issues, but
even if it doesn't, it's a valid change in itself.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

Sometimes the URL used is a link to the content that gets updated
between jobs.  When comparing jobs, it's unclear if we're downloading
the same file or another one.

Let's print the available information we have, such as the content
length, and the creation/modification dates.

Reference: https://trello.com/c/wQ8ir7xASigned-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

Which is used for the (usually temporary) snapshot image.  By default,
it's kept in the on the same directory as the base image.  When
downloading the base image into a test's cache_dir, it makes sense to
keep the snapshot out of the cache_dir, usually in the test's workdir.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

This check is based on a number of packages that are either Python 2
only and/or are not available on most distros.  And, pylint itself
supports this type of check.

Reference: https://trello.com/c/rfSAXdyiSigned-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

There's some duplicated code dealing with the lzma module presence and
consequently lzma handling capabilities on avocado.utils.archive.

Also, it makes a lot more sense to have that functionality there.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Sometimes multiple links to the same OS image file version will exist
on an index web page.  For instance, on openSUSE download server,
a folder icon, and the name of the version itself, appear side
by side and point to the same version:

               Name                    Last Modified  Size  Metadata
 [parent_icon] Parent Directory                       -
 [folder_icon] Leap_15.0/              08-Feb-2018    -
 ...

Let's make the set of found patterns unique, and not one per link.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Unfortunately, the vmimage module doesn't have any unittests, so the
compatibility with Python 3 has not been tested.  Let's fix the
obvious compatiblity issues and add the most simple test possible
to at least load the utility module on Python 3.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

While the readability aspect is debatable, this is IMO easier to
grasp than a multiline condition.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Let's give a little more info, and split the parameter type into its
own docstring line.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Another simple refactor, because it makes sense to have related
attributes set together instead of scattered all over the place.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Let's move the logging map definitions to a class scope, and simplify
the setting of the loglevel attribute.

In theory, even though it's not supported now, this should also allow
code that define their own jobs to easily customize the logging level.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

When `make develop` is run egg-info directories are created. We
shouldn't add them to version control, so this commit adds the proper
entry to gitignore file.
Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

E1111 error happens when a function which doesn't return is used to
assign to something. In this particular case the unittest.TextTestResut
stopTest method doesn't return and its result is being assigned to a
variable. This behaviour makes pylint fails with E1111 error. This
commit just don't assign the result of super() call and also keeps the
super class method behavior (not returing).
Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Python 3.7 introduced a new way to add breakpoint on the code. It was
added the built-in breackpoint function. There was couple of places
where the word breakpoint was being redefined making pylint execution
fail with W0622 error (Redefining built-in). This commit replaces
'breakpoint' with 'break_point' in these places.
Signed-off-by: NCaio Carrara <ccarrara@redhat.com>

While abusing the term tests, this is an initial set of files that are
intended to test how Avocado can be deployed on various platforms,
using different installation methods.

At this point, a single combination of installation method (pip) and
destination (virtual environment).
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Which will return the first member of the archive, either a file or
directory.  If the archive is empty, it'll return None.
Signed-off-by: NCleber Rosa <crosa@redhat.com>

Signed-off-by: NCleber Rosa <crosa@redhat.com>

Currently we compare the full TreeNodeEnvironment object to make sure
they come from the same origin, which only works for connected tree. But
Avocado supports (and uses in json-load) passing leaves of unconnected
trees, but then the TreeNodeEnvironments are not matching.

Let's rely only on "TreeNodeEnvironment.path" and declare these matching
environments, no matter what values are stored (see selftest for
details)
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

On python3 we can get "TypeError: can't pickle zipimport.zipimporter
objects" error when trying to "copy.deepcopy" the arguments and I
haven't found any real need for it on "initialize" call (plugins are not
overriding arguments, or they copy it themselves). Let's avoid this by
not copying args.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

* https://github.com/avocado-framework/avocado:
  avocado.utils.network: introduce default port values for find_free_port()

The "paths" is supposed to be a list of paths and not just string.
Fortunatelly this was why this plugin actually worked, because it turned
paths into `['/', 'r', 'u', 'n']` making the root node available. Anyway
let's fix the value and use the only location this plugin puts variables
in, which is `/`.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>

This plugin uses flat structure, there is no need to set the default
path.
Signed-off-by: NLukáš Doktor <ldoktor@redhat.com>