Skip to content
kweekly edited this page Jul 27, 2012 · 2 revisions

Synopsis

A source describes to the system a repository of data streams from which it can retrieve data. Sources are distinguished by their source name, which is mapped to a corresponding .src file in the sources/ directory of the python repository. In general, all a source needs to define is a type, such as TIMESERIES, and a driver which knows how to retrieve data from that source.

Each specific driver may require additional fields to tell it where to get data, for instance, the SMAP driver needs to know the base url from which to retrieve the data.

A source ID is a string distinguishing a single data stream to the driver.

Known Sources

Below are a list of the provided sources (hopefully updated with changes). These can also be obtained by running fetcher.py list.

csv

Type Time Series
source ID meaning Filename of CSV file to read from. This file must contain the time points that are requested, i.e. by `--to` and `--from`, otherwise it will be truncated to the data available.

This source is mainly for testing, i.e. if you already have a CSV file, or you want to run a partial flow on an intermediate result.

smapcsv

This is the same as above, but it assumes that time and value units are in millis-. Use this source if pulling from raw-data CSVs from SMAP.

cosm

Type Time Series
source ID meaning The last part of the url to retrieve data via the curl API. e.g. "504/datastreams/energy"

Link

Requires that you obtain a cosm API key and set the apikey variable in the cosm.src file.

openbms

Type Time Series
source ID meaning UUID of the SMAP device.

Link

As of now, this driver only works for some of the streams of openbms, such as the ACME motes. Assumes that time and value units are in millis-

sensezilla

Same as above, but pointing to the local SMAP instance instead of openbms

Clone this wiki locally