This package is deprecated!!! Please use
bob.pipelines for generic database interfaces instead.
A Bob db package (or database) contains a set of interfaces (API) to programmatically query and access samples and metadata from raw data (usually) stored on disk. Because of the growing number of packages of this nature, we decided to centralise common functionality and management routines in this package, to avoid constant re-writing of basic functionality. This guide explains basic concepts of db packages, the common functionality that is available, how to create your own and how to connect db packages to real-world applications.
A Bob db package is normally named
corresponds to the original database name or an acronym. So, for example, for
the Iris Flower Dataset, the corresponding db package is named
bob.db.iris. You should choose the name of your db package carefully so it
is relatively easy to figure out its relationship with the original raw data it
The raw data of a database is, normally, not shipped with the equivalent Bob db package. The reasons for these are twofold:
More often than not, raw data is very voluminous and cannot be stored on the Python Package Index (PyPI), where we post our packages.
Occasionally, data is subject to end-user license agreements, which must be undertaken between processors and data controllers directly. In this case, we’re simply not entitled to distribute these raw data files.
In cases where both reasons 1 and 2, above, are non-issues, the Bob db package may include the raw data. This decision is made by the package developer and varies from case to case.
A Bob db package is a normal Bob package and should, without exceptions, be developed by following the guidelines for any other package available at Bob Package Development. To simplify maintenance and improve homogeneity between different db packages, we further suggest you follow our Development Guide.
- Using Bob Databases
- Development Guide
- Python API