Universum 0.19.2

Unifier of Continuous Integration

Project 'Universum'

Universum integrates various CI systems and provides additional features, such as: customized downloading sources from VCS, running tests described in configuration file and reporting the results to code review systems.

Full documentation can be found here: https://universum.readthedocs.io/

Please check out our code of conduct and contribution policy

Project is executed with python3.7 -m universum command. Independent analyzers are executed with their module name, e.g. python3.7 -m universum.analyzers.pylint. Other Universum modes, such as poller or submitter, are called via command line, e.g. python3.7 -m universum poll

Installation

Minimum prerequisites (see documentation for details):

1. OS Linux
2. Python version 3.7 or greater
3. Pip version 9.0 or greater

sudo pip3.7 install -U universum

or

pip3.7 install --user -U universum

Can also be installed with extras for using VCS, but they also require installing respective command-line tools, such as git or p4.

Development

In order to prepare the development environment for the Universum, please fulfill the prerequisites, and then use the commands listed below. Please note we use venv to properly select python interpreter version and to isolate development environment from the system.

Prerequisites:

1. Install all of the VCS extras as described in the Universum installation manual, (including installation of Git and P4 CLI)

2. Install Docker (docker-ce, docker-ce-cli) as described in the official installation manual

   • Also add current user to 'docker' group (use sudo usermod -a -G docker $USER and then relogin)

Further commands:

python3.7 -m venv virtual-environment-python3.7
source ./virtual-environment/bin/activate
git clone https://github.com/Samsung/Universum.git universum-working-dir
cd universum-working-dir
git checkout master
pip install -U .[test]
make images

And after this the pytest or make test commands can be executed (see below).

The [test] extra will install/update the following additional Python modules:

* `sphinx`
* `sphinx-argparse` (extension for `Sphinx`)
* `sphinx_rtd_theme` (extension for `Sphinx`)
* `docker`
* `httpretty`
* `mock`
* `pytest`
* `pylint`
* `pytest-pylint`
* `teamcity-messages` (is not actually used in manual testing, but is there for CI)
* `pytest-cov`
* `coverage`
* `mypy`

Although it is possible to get these modules via pip3.7 install -U universum[test], it might be more convenient to checkout the Universum branch you are currently working on, change working directory to project root and run a pip3.7 install -U .[test] command from there for more flexibility. Using virtual environment (via venv command) allows to separate test environment from system and provides even more control over additional modules.

Uninstalling Universum via pip uninstall universum will not uninstall all the dependencies installed along with it. Simply deleting the directory with virtual environment will leave the system completely cleaned of all changes.

Docker images used in tests can be built manually or using the make images command. Also make rebuild command can be used to update images ignoring cache (e.g. to rerun apt update).

The make test command runs all the tests (including the doctests) and collects coverage. Tests can also be launched manually via pytest command with any required options (such as -k for running tests based on keywords or -s for showing the suppressed output).

Project contents

universum is main project folder, that is being copied to Python libraries location (e.g. dist-packages or site-packages) when installed. It contains __main__.py script, that is the main entry point to the whole project. It also contains the following modules:

• main/poll/submit/api/nonci - managing modules for different Universum modes

• configuration_support - special module for configuring the project

• analyzers directory is not quite a part of Universum itself. It contains independent scripts compatible with Universum for implementing static (and other types of) analysis support.

• lib - utility functions libraries

  • ci_exception - internal exceptions
  • module_arguments - handles command line and other parameters
  • gravity - inter-module communication
  • utils - miscellaneous

• modules - independent packages

  • api_support - 'main' mode module to answer API requests
  • automation_server - drivers for CI systems (e.g. Jenkins)
  • artifact_collector - implements build artifacts
  • code_report_collector - support for external 'code report' modules
  • launcher - executes build scenario, described in project configuration file
  • output - drivers for environment-based logs
  • project_directory - interaction with host file system
  • reporter - interaction with code review systems
  • structure_handler - execution 'blocks' isolation, order, reporting, etc.
  • vcs - CI target sources preparation

Also there are 'base' modules/classes for driver implementation standardization, and 'main' modules/classes for automated driver choosing based on environment and settings.

doc directory contains sources for project documentation. It can be generated locally with running make from root directory using Sphinx.

tests directory contains test system, based on PyTest. Full tests can be started from root directory via make tests command, otherwise use standard PyTest syntax. Commits failing any of project tests should not be merged into 'master' branch!

examples contains various examples of project configuration files. Usage of such files is illustrated in run_basic_example.sh script.

universum_log_collapser contains Universum Jenkins plugin sources. The plugin implements coloring of failed steps and provides collapsing/expansion of log sections.

setup.py is 'setuptools' configuration file, and shouldn't be executed on its own.

Quick architecture overview

1. Project only entry point (except 'analyzers') is universum.py. Based on chosen execution mode (default, submitting, polling, etc.) it calls one of 'main' modules, passing them all parameters

2. Universum is a set of separate modules, each implementing its own piece of functionality. They are connected using special gravity library

3. All classes, inherited from Module (defined in gravity), automatically can:

   • use Dependency mechanism to use other modules
   • describe any module parameters in define_arguments() and receive them parsed via self.settings

4. configuration_support is, in fact, an 'external' module, used not only by Universum, but by user configuration file for generating project configuration

5. 'Base' classes are virtual, not implementing any actual functionality, but describing the structure of inherited classes and ensuring they have all required functions that will be called by modules using them

Contributing

Further versions of this README file should include:

1. Notification on mandatory code review for all commits to master
2. Notification on mandatory documenting of the newly added features
3. Description of CI process, links to configurations/logs/build results/etc.