Databases fixtures plugin for py.test.
py.test clean fixtures for: postgresql, mysql, redis, mongo, elasticsearch, dynamodb and rabbitmq.
Starts specific database deamon and cleanup all data produced during tests.
$ pip install --upgrade pytest-dbfixtures
How to use
Please see documentation
- [enhancements] Postgresql client fixture closes other postgresql connection to the database before droping database - prevents the fixture/tests from hanging in some cases
- [enhancements] mysql to use unique tmpdir option for mysql_* commands
- [enhancements] use semicolon to terminate postgresql CREATE/DROP DATABASE statements
- [bugfix] removed unneded dependency
- [enhancement] split out elasticsearch fixtures into separate package. See pytest-elasticsearch
- [feature] use tmpfile.gettempdir instead of hardcoded /tmp directory
- [fix] support for rabbitpy 0.27.x
- [feature] Random port selection ports accept tuples and sets. replace string representation [backward incompatible]
- [feature] Add support for delayTransientStatuses flag (DynamoDB)
- [fix] Update DynamoDB in howtouse
- [fix] packaging/changelog fix
- [feature] add fixture for DynamoDB
- [bugfix] fix dbfixtures packaging
[feature] make it easier to support future postgresql out of the box
[feature] support for postgresql 9.5
[docs] add contribute guidelines
[feature] add info about minimal version supported - closes #132
[enhancements] - moved source code into src folder
[feature] python3.5 compatibility
[docs] enhance docs - refs #126
[feature] PGsql connection error on FreeBSD jailed environments
In FreeBSD jailed environments the loopback interface can not be used to connect to pgsql, because it points to the loopback interface of the host and not the jail.
Thus, it seems to pgsql like the connection is comming from the ip address assigned to the jail.
This ensures that pgsql can be connected to from any host, when run on a FreeBSD systems. This this package can be used in FreeBSD jailed environments
[feature] Use log destination param for pgsql
This commit ensures that stderr is used for logging, by specifying the command line parameter.
On FreeBSD this is very important otherwise syslog will be used and the db-fixtures will hang as it looks in the expected log file and loops forever waiting for a “database is ready” entry to appear… log_destination=stderr is default on most systems and can be set in postgresql.conf or given as an command line argument.
- [bugfix] mongodb fixture no longer removes any of system.* collections
- [bugfix] configured pytest-dbfixtures to work with pymlconf versions compatible with Python 3
- make pytest-dbfixtures compatible with mirakuru 0.5.0
- fix W503 pep8 errors, that appeared after new pylama got released
- update to be able to use pytest-dbfixtures with pymongo 3
- ability to properly use also beta versions of postgresql releases. (previously only stable versions were targeted)
- add ability to set custom location for logs (might be useful for analysing logs from tests)
- added postgresql 9.4 to supported versions.
Use port number in default RabbitMQ node name.
This allows using just the port=’?’ argument to rabbitmq_proc to run multiple independent RabbitMQ instances for use with xdist to parallelize tests on a single machine.
Old versions of Redis notification
displays a message about old version of redis
- Feature: random ports selection Adds a possibility of passing ‘?’ in port= argument of process fixtures.
- Removes ‘port’ from db fixtures
- Fixes postgres missing host - previously using config.postgresql.unixsocketdir
- redisdb fixture allows to specify client connection class
- redisdb fixture uses now StrictRedis by default (backward incompatible)
- replaced MySQLdb with it’s fork mysqlclient - compatibility with python3
- renamed mysqldb fixture to mysql to keep it consistent with other client fixtures.
- replaced pika with python 3 compatible rabbitpy
- removed deprecated mysqldb_session/mysql_session
- bugfix of rabbitmq fixture: cast rabbitmq queues and exchanges to str due to pamq having problems
- internal changes: removed GentleKillingExecutor as mirakuru already implements same functionality
- syntax fixes for python 3
- moved postgresql starting code into it’s own executor
- xdist distributed hosts testing bugfix (but xdist parallelization on one host won’t work)
- make rabbitmq logs persistent
- update mirakuru to min 0.2
- os.killpg to terminate process
- add tests coverage on coveralls
- remove creating old RABBITMQ env variables
- remove elasticsearch home folder at process teardown
- set elasticsearch index.store.type to memory by default
- localized elasticsearch instance by default
- replaced summon_process with mirakuru
- RabbitMQ process fixture is session scoped
- rabbitmq fixture factory accepts custom teardown
- StopRunningExecutor to simply return if process had been already killed.
- MongoDB fixtures can be now initialized by factories
- Bugfix: params in redis fixture
- Add params to mysql fixture
- Client fixtures now check if process (process fixture) is running before every test and starts process if it was terminated after previous test.
- Bugfix: Now we don’t overwrite postgresql config in postgres_proc
- Bugfix of rabbitmq_proc fixture - now it works with scope=function
- Overrides SimpleExecutor’s behavior with a try of more gentle terminating subprocess before killing it.
- Deprecate scope for mysqldb fixture and change it to function by default.
- RabbitMQ factories support (multiple rabbit fixtures).
- Postgresql multiple versions proper support
- Default timeouts and waits for process executors
- introduced Elasticsearch fixture
- mysql fixture now uses factories
- postgresql fixtures and fixture factories
- small code quality improvements
- pylama code check
- splits rabbitmq fixture into process/client fixtures