Metadata-Version: 2.1
Name: katsdptelstate
Version: 0.12
Summary: Karoo Array Telescope - Telescope State Client
Home-page: https://github.com/ska-sa/katsdptelstate
Author: MeerKAT SDP team
Author-email: sdpdev+katsdptelstate@ska.ac.za
License: Modified BSD
Keywords: meerkat ska
Platform: OS Independent
Classifier: Development Status :: 5 - Production/Stable
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: BSD License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python
Classifier: Programming Language :: Python :: 3.5
Classifier: Programming Language :: Python :: 3.6
Classifier: Programming Language :: Python :: 3.7
Classifier: Programming Language :: Python :: 3.8
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Classifier: Topic :: Scientific/Engineering :: Astronomy
Requires-Python: >=3.5
Provides-Extra: rdb
Provides-Extra: aio
Provides-Extra: test
License-File: LICENSE

MeerKAT Science Data Processor Telescope State
==============================================

This is a client package that allows connection to a database that
stores telescope state information for the Science Data Processor of the
MeerKAT radio telescope. This database is colloquially known as *telstate*.

The telescope state is a key-value store. There are three types of keys:

immutables (aka *attributes*)
  Stores a single value that is not allowed to change once set.

mutables (aka *sensors*)
  Stores multiple timestamped values organised into an ordered set.

indexed
  Stores a dictionary of key-value pairs, each of which behaves like an
  immutable. This is useful to avoid the main key-space becoming too large.
  It also supports some patterns like incrementally storing values but
  fetching all values in a single operation. Furthermore, it allows more
  general keys than just strings.

The keys are strings and the values (and the sub-keys of indexed keys) are
Python objects serialised via MessagePack_, which has been extended to support
tuples, complex numbers and NumPy arrays. Older versions of the database stored
the values as pickles, and the package warns the user if that's the case. Keys
can be retrieved from the telstate object using attribute syntax or dict
syntax.

.. _MessagePack: http://www.msgpack.org/

Databases can be accessed via one of two backends: a Redis client backend
that allows shared access to an actual Redis server over the network (or a
simulated server via fakeredis) and a simplified in-memory backend for
stand-alone access. Both backends support loading and saving a Redis snapshot
in the form of an RDB dump file.

It is possible to have multiple *views* on the same database (one per telstate
instance). A view is defined as a list of *prefixes* acting as namespaces that
group keys. When reading from the database, each prefix is prepended to the key
in turn until a match is found. When writing to the database, the first prefix
is prepended to the key. The first prefix therefore serves as the primary
namespace while the rest are supplementary read-only namespaces.

.. warning::

  **WARNING**: The standard warning about Python pickles applies. Never
  retrieve data from an untrusted telstate database with values encoded as
  pickles, or connect to such a database over an untrusted network. Pickle
  support is disabled by default, but can be enabled for trusted databases
  by setting the environment variable KATSDPTELSTATE_ALLOW_PICKLE=1.

Getting Started
---------------

The simplest way to test out `katsdptelstate` is to use the in-memory backend.
If you want to run a real Redis server you will need to install Redis (version
4.0 or newer) on a suitable machine on the network. For example, do this:

- macOS: ``brew install redis``
- Ubuntu: ``apt-get install redis-server``

Then ``pip install katsdptelstate`` and run a local ``redis-server``. If you
also want to load RDB files, do ``pip install katsdptelstate[rdb]``.

A Simple Example
----------------

.. code:: python

  import time
  import katsdptelstate

  # Connect to an actual Redis server via an endpoint or an URL
  telstate = katsdptelstate.TelescopeState('localhost:6379')
  telstate = katsdptelstate.TelescopeState('redis://localhost')
  # Or use the in-memory backend (useful for testing)
  telstate = katsdptelstate.TelescopeState()
  # Load RDB file into Redis if katsdptelstate is installed with [rdb] option
  telstate.load_from_file('dump.rdb')

  # Attribute / dict style access returns the latest value
  telstate.add('n_chans', 32768)
  print(telstate.n_chans)  # -> 32768
  print(telstate['n_chans'])  # -> 32768

  # List all keys (attributes and sensors)
  print(telstate.keys())  # -> ['n_chans']

  # Sensors are timestamped underneath
  st = time.time()
  telstate.add('n_chans', 4096)
  et = time.time()
  telstate.add('n_chans', 16384)
  # Time ranges can be used and are really fast
  telstate.get_range('n_chans', st=st, et=et)  # -> [(4096, 1556112474.453495)]
  # Add an item 10 seconds back
  telstate.add('n_chans', 1024, ts=time.time() - 10)

  # Attributes cannot be changed (only deleted)
  telstate.add('no_change', 1234, immutable=True)
  # Adding it again is OK as long as the value doesn't change
  telstate.add('no_change', 1234, immutable=True)
  # Simpler notation for setting attributes
  telstate['no_change'] = 1234
  # Will raise katsdptelstate.ImmutableKeyError
  telstate['no_change'] = 456

  # Create a new view with namespace 'ns' and standard underscore separator
  view = telstate.view('ns')
  # Insert a new attribute in this namespace and retrieve it
  view['x'] = 1
  print(view['x'])  # -> 1
  print(view.prefixes)  # -> ('ns_', '')
  print(view.keys())  # -> ['n_chans', 'no_change', 'ns_x']

Asynchronous interface
----------------------
There is also an interface that works with asyncio. Use
``katsdptelstate.aio.TelescopeState`` instead of
``katsdptelstate.TelescopeState``. Functions that interact with the database are now
coroutines. Python 3.6+ is required.

There are a few differences from the synchronous version, partly necessary due
to the nature of asyncio and partly to streamline and modernise the code:

- The constructor only takes a backend, not an endpoint. See below for an
  example of how to construct a redis backend.
- There is currently no support for reading or writing RDB files; you'll need
  to create a synchronous telescope state client that connects to the same
  storage.
- There is no support for attribute-style access.
- Item-style access is supported for read (``await ts.get('key')``), but not
  for write. Use ``await ts.set('key', 'value')`` instead to set immutable
  keys.
- Instead of ``key in ts``, use ``await ts.exists(key)``.
- The ``wait_key`` and ``wait_indexed`` methods do not take a timeout or a
  cancellation future. They can be used with asyncio's cancellation machinery.
  The `async-timeout`_ package is useful for timeouts.
- The backend should be closed when no longer needed to avoid warnings.

.. _async-timeout: https://pypi.org/project/async-timeout/

Example
^^^^^^^

.. code:: python

  from katsdptelstate.aio import TelescopeState
  from katsdptelstate.aio.redis import RedisBackend

  # Create a connection to localhost redis server
  backend = await RedisBackend.from_url('redis://localhost')
  ts = TelescopeState(backend)

  # Store and retrieve some data
  await ts.set('key', 'value')
  print(await ts.get('key'))

  # Close the connections (do not try to use ts after this)
  ts.backend.close()
  await ts.backend.wait_closed()


History
=======

0.12 (2023-03-13)
-----------------
* Switch to aioredis 2.x (#124)
* Add async (and synchronous) `RedisBackend.from_url` constructor (#125, #126)
* Switch from nose to pytest (#129)
* Remove `Endpoint.multicast_subscribe` and `netifaces` dependency (#130)
* Fix Lua script for `set_indexed` so that `wait_indexed` actually works (#127)
* Make aio `wait_key` more robust (#128)
* General cleanup (#122, #123)

0.11 (2021-05-07)
-----------------
* Add asynchronous RDBWriter class (#108)
* Use a transaction for get_range instead of Lua: faster on server side (#110)
* Multiplex aio pubsubs over pool's standard connection (#113)
* Require hiredis for speed and aioredis<2 for compatibility (#114, #118)
* Improve `wait_keys` responsiveness for `MemoryBackend` (#111, #116)
* Avoid blocking the Redis server on telstate.clear (#112)
* Update Docker image to use Redis 6.x instead of 4.x (#109)
* Support older HMSET Redis command in unit tests (#107)
* Requirements cleanup (#115, #117, #119, #120)

0.10 (2020-05-25)
-----------------
* Remove Python 2 support. At least Python 3.5 is required.
* Remove support for old versions of redis-py (#100)
* Use redis-py health checks to improve robustness (#99)
* Add "indexed" keys (#98)
* Add an asyncio interface (#103)
* No longer throw InvalidKeyError when setting a key that shadows a method (#102)
* Add type annotations for mypy (#101)

0.9 (2020-05-25)
----------------
* Deprecate Python 2 support: this is the last release that will support Python 2 (#94)
* Remove ``get_message`` and ``send_message``, which were never used (#89)
* Publish the documentation on https://katsdptelstate.readthedocs.io (#90)
* Disable pickles by default for security (#92)

0.8 (2019-05-06)
----------------
* The default encoding is now msgpack; warn on loading pickles (#75, #79)
* The default backend is now in-memory (#76)
* Add the ability to dump in-memory backend to an RDB file (#77)
* Construct from RDB file-like objects and Redis URLs (#80, #82)
* Report keys and prefixes to the user as strings (#73)
* Add IPython tab completion (#83)
* RDB reader and writer cleanup (#85, #86)

0.7 (2019-02-12)
----------------
* Introduce encodings and add msgpack encoding as alternative to pickle (#64, #65)
* Introduce backends and add in-memory backend as alternative to redis (#71, #72)
* Simplify setting attributes via `__setitem__` (#68)
* Let keys be bytes internally, but allow specification as unicode strings (#63)
* The GitHub repository is now public as well

0.6 (2018-05-10)
----------------
* Initial release of katsdptelstate
