Package dimer :: Module archive

Module archive

utilities and wrappers for working with HDF archives.

* an archive contains one of more datasets

* a dataset contains

an array X: the data
an optional array Y: the raw response
an optional array T: labels (targets) associated with the response

a dataset specification (dset_path) is the concatenation (`:` is used as a separator) of the path to the HDF file and the dataset name. All datasets hang on the root.

Functions

[hide private]

__parse(path, PATT=re.compile(r'(.+\.h5):(.+)')) source code

basename(path)
a dataset has the form <archname>:<basename>

source code

archname(path)
a dataset has the form <archname>:<basename>

source code

split(path)
a dataset has the form <archname>:<basename>

source code

join(path, dsname)
join archname with basename

source code

save_object(file_path, key, obj)
atomic save operation

source code

load_object(file_path, key)
atomic load operation

source code

dset_path(path)
a checker for correct specification of an archive path

source code

Variables

[hide private]

log = logging.getLogger(__name__)

__HDF_SUFFIX__ = 'h5'

__SPEC_SEP__ = ':'

DSPEC_MSG = 'dataset specification; of the form <hdf_archive_p...

__package__ = 'dimer'

Function Details

[hide private]

basename(path)

source code

a dataset has the form <archname>:<basename>

Returns:: the archive (inside the file) part of the path

archname(path)

source code

a dataset has the form <archname>:<basename>

Returns:: the archname part of the path

split(path)

source code

a dataset has the form <archname>:<basename>

Returns:: (archname, basename)

Variables Details

[hide private]

DSPEC_MSG

Value:

'dataset specification; of the form <hdf_archive_path>:<dataset_name>'