GitLab

Now the --attachments option of the test command will fetch all
test run and test case attachments files.

These files will be saved in a directory hierarchy of the form
<attachments-dir>/<test-run-dir>/<test-case-dir>, and each of the files
will be saved in the specific category for each of the test runs and
test cases.

Also the output message is much simpler now, and it will only print
a logger info message after all the files have been saved to the
location.

This commit adds new sections to cover all the possible job states
in the analyse report.

It will now show the following main categories if there are jobs
in such a states: Failed jobs, Pending jobs, Canceled jobs and
Successful jobs.

This commit will make the output command to wait for the job to start
when using the --live option, or will immediately exit and print a
message showing the current job status when not using such an option.

This commit also introduces some code refactoring and cleaning, specially
in the way the errors are handled to fit better the above changes.

Now the job command will only show the job name by default if no
other option is passed, and it will require explicitly the new
--tests option to list the job tests.

This commit also makes the command output cleaner.

These changes are intended to make the job command resemble how
the test command behaves.

This will show the job failure comment if it is available in the
analyse report. This comment contains information about
infrastructure failures for the job.

This commit wraps all the XMLRPC methods inside the Connection class
for easy and direct access from the lqa API, which will allow (among
other things) to abstract such a details from the external code and
commands.

One of the important changes in this commit is that now the Connection
object will create a server proxy at once and will be referenced by
the property self._conn which will be later used by the rest of its
methods to access the server API.

Only the XMLRPC methods currently used by the lqa tool have been added
to the Connection class, others should be added later here if required.

This option will allow to fetch all the available attachments for
the specified tests, and it takes as an argument the directory path
where these attachments will be saved.

This commit implements the following changes to the test command
line interface:

- It now won't show the test results by default, and instead the
  command line option --no-results is replaced by the option
  -r/--results to explicitly tell the command to show the results.
- It makes the displayed output cleaner for the --info option.

This commit adds the flag -a/--all-results to the analyse command
so it expands and show all test results (not only failed results).

It sets a limit to show only 500 test results for each status, and
it prints a message to check the bundle link for further results if
that's the case. This limit can be ignored with the --no-limit
option.

It also refactors some of the code to be cleaner for printing the
different test results.

This new method will allow to fetch the test job attachments
for the Test objects.

The status command will now show the full name and id for the job status.

The image field of a test can be None, so check that it exists
beforing removing white spaces.

This job API property will allow to fetch the failure comment for
the job. This can be useful for reporting and debugging job issues.

This command will list all the submitted and running jobs with their
submit or start time respectively.

These properties will help to get the whole value for the bundle
and details structures of the Job API.

This commit changes the module name to the new name waitqueue for
importing WAIT_DEFAULT_TIMEOUT.

This offers a more concise output only showing the info relevant
to each job status and also avoids the error during the calculation
of the job duration if the end_time is not yet available.

This commit reports a cleaner error message and will exit with code
APPLICATION_ERROR if the job bundle is not found for the job and test
commands.

This commit makes lot of changes to the current wait API, that
it is now renamed waitqueue since this represents more clearly what
this API is intended to be: A queue to wait for jobs completion.

The API now inherits from the list object too, since it behaves the
same to a list except from some new methods specific to wait for
the jobs in the queue.

These changes will now primarily allow (among other things) to use
the waitqueue API independently from the lqa_tool since it also
properly abstracts the API behaviour from such a tool.

This commit changes the special argument name from 'job_id' to
'job_ids' to build the job id list. This is a more appropriate
name for such an argument and fixes the case where the command
only takes a single job id (named job_id) and doesn't need to
build a list.

This commit adds a range notation to specify the JOB ID in those
commands requirng such an argument.

The notation for JOB ID's can now be a range like: '0-N' to
process jobs with ID from 0 to N. The notation keeps also working
fine specifying a single JOB ID, or even a combination of both,
for example: '0 1-6 20-23 99' will work as expected
processing each of the specific JOB ID's.

This command will show information for the tests executed
by the specified job id. It offers several options to display the
test results and test metadata.

This command can be seen like a front-end to the test API lqa_api.test.

This command will list all the bundle streams the logged in user
has access to.

This commit adds the new job command at lqa_tool.commands.job.

This command is used to fetch and display the information of the
specified job ids. It is basically a command line front-end that
directly maps to methods in the job API to show job metadata.

Remove old/deprecated links and adapt some parts of the instructions
to the current project status.

This commit renames some of the module names to match the exact
command name as available from the command line.

The standard os.path functions basename and splitext are now used
to properly get the test name from the testdef field when finding
the missing tests in the analyse command.

This commit fix the _find_missing_tests method to properly find
missing tests with the same name. It also removes unused code.

This will catch such a signal for all the commands. This commit also
removes catching this signal specifically from the API lqa_api.wait.

Signed-off-by: Héctor Orón Martínez <hector.oron@collabora.co.uk>

Change the fetch_results code to use the new Test object API.

This commit also changes the argument variable of the command to the
more standard name job_id.

This includes all the values for the results of a test, since the
result hash can contain different fields depending on the test, and
it also makes the code simpler, since the test results will
directly map to the test_run structure.

This commit also removes the unused self._packages variable.

Handle XMLRPC protocol errors at the top level for all the commands,
and exit with APPLICATION_ERROR code if such an error is raised.