Install:
pip install whisper-pandas
Use as Python package:
# Simple to read any Whisper file
>>> from whisper_pandas import WhisperFile
>>> wsp = WhisperFile.read("example.wsp")
# Simple to work with metadata as objects
>>> wsp.meta.archives
[WhisperArchiveMeta(index=0, offset=52, seconds_per_point=10, points=1555200, retention=15552000),
WhisperArchiveMeta(index=1, offset=18662452, seconds_per_point=60, points=5256000, retention=315360000),
WhisperArchiveMeta(index=2, offset=81734452, seconds_per_point=3600, points=87601, retention=315363600)]
# Simple to work with data as `pandas.Series`
>>> wsp.data[1]
2017-02-10 07:07:00+00:00 0.000000
2017-02-10 07:08:00+00:00 0.000000
2017-02-10 07:09:00+00:00 0.000000
2017-02-10 07:10:00+00:00 0.000000
2017-02-10 07:11:00+00:00 0.000000
...
2021-07-20 13:35:00+00:00 4.099915
2021-07-20 13:36:00+00:00 4.104024
2021-07-20 13:37:00+00:00 4.099772
2021-07-20 13:38:00+00:00 4.101358
2021-07-20 13:39:00+00:00 4.099854
Length: 2331015, dtype: float32
Use as command line tool:
whisper-pandas example.wsp
WhisperDB is a fixed-size time series format (see docs).
The official Python package is here: whisper
You should use it, except if you like Pandas and only need
to read (not write) Whisper files, then you should use whisper-pandas.
Why?
- Mucho simpler to use
- Mucho less likely you will shoot yourself in the foot
- Mucho speedy
Currently we use whisper.info internally to read the metadata,
and then Numpy & Pandas to read the data,
using objects for the metadata and some conveniences like quickly
showing what data is available in a given file or reading zipped files.
Partial reading is not implemented, a given file is always read completely
and the data directly converted to columnar pandas.Series format with
a datetime index.
Contributions are welcome via Github pull requests.
You can run the tests via pytest.
The packaging follows the recommendation in
Packaging Python Projects.
To make a release increase the version number in setup.cfg and run the commands there.