Skip to main content

Time Series storage in flat files.

Project description

Time Series Peristence

This Python package provides Time Series storage in flat files according to the TeaFile file format.

In Use

>>> tf = TeaFile.create("acme.tea", "Time Price Volume", "qdq", "ACME at NYSE", {"decimals": 2, "url": "" })
>>> tf.write(DateTime(2011, 3, 4,  9, 0), 45.11, 4500)
>>> tf.write(DateTime(2011, 3, 4, 10, 0), 46.33, 1100)
>>> tf.close()
>>> tf = TeaFile.openread("acme.tea")
TPV(Time=2011-03-04 09:00:00:000, Price=45.11, Volume=4500)
TPV(Time=2011-03-04 10:00:00:000, Price=46.33, Volume=1100)
>>> tf.close()

Exchange Time Series between Apps / OS

You can create, read and write TeaFiles with

  • R,

  • C++,

  • C# or

  • other applications


  • Linux, Unix,

  • Mac OS

  • Windows

Python API Examples


TeaFiles are a very simple, yet highly efficient, way to store time series data providing data exchange between programs written in C++, C# or applications like R, Octave, Matlab, running on Linux, Unix, Mac OS X or Windows.

  • Binary data composed from elementary data types signed and unsigned integers, double and float in IEEE 754 format is prefixed by a header holding a description of the item structure and the content.

  • Data can be directly accessed via memory mapping.

  • TeaFiles are self describing: Containing a description of the item structure they relieve opaqueness of straight binary files. Knowing that a file is a TeaFile is enough to access its data.

link to spec http://tbd

Scope of the Python API

The Python API makes TeaFiles accessible everywhere. It just needs a python installation on any OS to inspect the description and data of a TeaFile:

>>> # Show the decimals and displayname for all files in a folder:
>>> def showdecimals():
    ...     for filename in os.listdir('.'):
    ...         with TeaFile.openread(filename) as tf:
    ...             nvs = tf.description.namevalues
    ...             print('{} {} {}'.format(filename, nvs.get('Decimals'), nvs.get('DisplayName')))
    >>> showdecimals() 2 Alcoa, Inc.
    AA.tick.tea 2 Alcoa, Inc. 2 American Express Co.

Data download from web services is also a good fit. See the file in the package source for a Yahoo finance download function in about 30 lines.


When it comes to high performance processing of very large time series files, this API is currently not as fast as the C++ and C# APIs (Numbers coming soon on http://tbd). There are numerous ways to improve this if necessary, but no current plans at discretelogics to do so. Using the other languages/APIs is recommended. If you wish the Python API to be faster or want to work on that contact us.


$ pip install teafiles

package source with at


Run the unit tests from the package root by

$ python -m pytest .test

Python 2.7 / 3.2

Package tested under CPython 2.7. Python 3.2 planned


This API brought to you by discretelogics, company specialicing in time series analysis and event processing. http://tbd

Version 0.7

The current version is reasonably tested by doctests and some pytests. Better test coverage with unit tests (currently pytest is used) is desirable.

tbd towards version 1.0
  • enhance pytest coverage

  • consider api feedack

  • cleaner test runs, cleanup test files

  • enhance performance after measuring it in python 3 (maybe struct module plays a minor role there)


This package is released under the GNU GENERAL PUBLIC LICENSE, see License.txt.


Welcome at:

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution (44.9 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page