def write_custom_annotation(h5file,
annotation_name,
annotation_value,
annotations_sub_group=None):
"""
Write a custom annotation to an HDF5 database.
The annotation is written to the `h5file` under the `annotation_name` such
that it holds the `annotation_value`.
Parameters
----------
h5file : tables.file.File
The handle to the HDF5 database to which the annotation is written.
annotation_name : str
The name of the annotation to write.
annotation_value : a JSON serialisable object
The annotation value to write.
annotations_sub_group : str
The group node under "/annotations" to which the custom annotation is
written (the default is None which implies that the custom annotation
is written directly under "/annotations").
Notes
-----
The `annotation_value` must be a JSON seriablisable object.
Examples
--------
Write a custom annotation to an HDF5 database.
>>> import magni
>>> from magni.reproducibility.io import write_custom_annotation
>>> annotation_name = 'custom_annotation'
>>> annotation_value = 'the value'
>>> with magni.utils.multiprocessing.File('db.hdf5', mode='a') as h5file:
... write_custom_annotation(h5file, annotation_name, annotation_value)
... annotations = magni.reproducibility.io.read_annotations(h5file)
>>> str(annotations['custom_annotation'])
'the value'
"""
@_decorate_validation
def validate_input():
_generic('h5file', tables.file.File)
_generic('annotation_name', 'string')
_generic('annotations_sub_group', 'string', ignore_none=True)
validate_input()
if annotations_sub_group is not None:
annotations_group = '/'.join(['/annotations', annotations_sub_group])
else:
annotations_group = '/annotations'
try:
ann_val = json.dumps(annotation_value)
except TypeError:
raise TypeError('The annotation value does not have a valid JSON ' +
'representation. It may not be used as an annotation.')
try:
h5file.create_array(annotations_group,
annotation_name,
obj=ann_val.encode(),
createparents=True)
h5file.flush()
except tables.NodeError:
raise tables.NodeError(
'The annotation "{!r}" already exists '.format(annotation_name) +
'in the database. Remove the old annotation before placing a ' +
'new one.')
def annotate_database(h5file):
"""
Annotate an HDF5 database with information about Magni and the platform.
The annotation consists of a group in the root of the `h5file` having nodes
that each provide information about Magni or the platform on which this
function is run.
Parameters
----------
h5file : tables.file.File
The handle to the HDF5 database that should be annotated.
See Also
--------
magni.reproducibility._annotation.get_conda_info : Conda annotation
magni.reproducibility._annotation.get_git_revision : Git annotation
magni.reproducibility._annotation.get_platform_info : Platform annotation
magni.reproducibility._annotation.get_datetime : Date and time annotation
magni.reproducibility._annotation.get_magni_config : Magni config
annotation
magni.reproducibility._annotation.get_magni_info : Magni info annotation
Notes
-----
The annotations of the database includes the following:
* conda_info - Information about Continuum Anacononda install
* git_revision - Git revision and tag of Magni
* platform_info - Information about the current platform (system)
* datetime - The current date and time
* magni_config - Infomation about the current configuration of Magni
* magni_info - Information from `help(magni)`
Examples
--------
Annotate the database named 'db.hdf5':
>>> import magni
>>> from magni.reproducibility.io import annotate_database
>>> with magni.utils.multiprocessing.File('db.hdf5', mode='a') as h5file:
... annotate_database(h5file)
"""
@_decorate_validation
def validate_input():
_generic('h5file', tables.file.File)
validate_input()
annotations = {
'conda_info': json.dumps(_annotation.get_conda_info()),
'git_revision': json.dumps(_annotation.get_git_revision()),
'platform_info': json.dumps(_annotation.get_platform_info()),
'datetime': json.dumps(_annotation.get_datetime()),
'magni_config': json.dumps(_annotation.get_magni_config()),
'magni_info': json.dumps(_annotation.get_magni_info())
}
try:
annotations_group = h5file.create_group('/', 'annotations')
for annotation in annotations:
h5file.create_array(annotations_group,
annotation,
obj=annotations[annotation].encode())
h5file.flush()
except tables.NodeError:
raise tables.NodeError('The database has already been annotated. ' +
'Remove the existing annotation prior to ' +
'(re)annotating the database.')
def chase_database(h5file):
"""
Chase an HDF5 database to track information about stack and source code.
The chase consist of a group in the root of the `h5file` having nodes that
each profide information about the program execution that led to this chase
of the database.
Parameters
----------
h5file : tables.file.File
The handle to the HDF5 database that should be chased.
See Also
--------
magni.reproducibility._chase.get_main_file_name : Name of main file
magni.reproducibility._chase.get_main_file_source : Main file source code
magni.reproducibility._chase.get_main_source : Source code around main
magni.reproducibility._chase.get_stack_trace : Complete stack trace
Notes
-----
The chase include the following information:
* main_file_name - Name of the main file/script that called this function
* main_file_source - Full source code of the main file/script
* main_source - Extract of main file source code that called this function
* stack_trace - Complete stack trace up until the call to this function
Examples
--------
Chase the database named 'db.hdf5':
>>> import magni
>>> from magni.reproducibility.io import chase_database
>>> with magni.utils.multiprocessing.File('db.hdf5', mode='a') as h5file:
... chase_database(h5file)
"""
@_decorate_validation
def validate_input():
_generic('h5file', tables.file.File)
validate_input()
chases = {
'main_file_name': json.dumps(_chase.get_main_file_name()),
'main_file_source': json.dumps(_chase.get_main_file_source()),
'main_source': json.dumps(_chase.get_main_source()),
'stack_trace': json.dumps(_chase.get_stack_trace())
}
try:
chase_group = h5file.create_group('/', 'chases')
for chase in chases:
h5file.create_array(chase_group, chase, obj=chases[chase].encode())
h5file.flush()
except tables.NodeError:
raise tables.NodeError('The database has already been chased. ' +
'Remove the existing chase prior to ' +
'(re)chasing the database.')