nanshe.io.hdf5.record module

The module record provides the equivalent of a logger for array data.

Overview

Applying the decorator static_subgrouping_array_recorders with the names of desired recorders as strings or keyword arguments with a given recorder instance initializes the recorders attribute, which contains each recorder as a dictionary would. Results are stored under a group named after decorated function. Nested function calls can be paralleled by nesting results within groups. All that is required is that the recorder be assigned from outer scope to inner (see example below).

@static_subgrouping_array_recorders("r")
def a(x):
    a.r = x
    return x * x.T

@static_subgrouping_array_recorders("r")
def b(x, y):
    a.r = b.r
    return a(x + y)

All recorders default to an instance of EmptyArrayRecorder unless otherwise specified. All operations work on them exactly the same, but no recording occurs. This guarantees that code using recorders will not break if they don’t have them. A similar strategy can be applied for classes and their methods by using class_static_array_debug_recorder.

Todo

Complete documentation with more examples.

API

class nanshe.io.hdf5.record.EmptyArrayRecorder(**kwargs)

Bases: object

get(**kwargs)

class nanshe.io.hdf5.record.HDF5ArrayRecorder(**kwargs)

Bases: object

get(**kwargs)

class nanshe.io.hdf5.record.HDF5EnumeratedArrayRecorder(**kwargs)

Bases: object

get(**kwargs)

nanshe.io.hdf5.record.class_static_array_debug_recorder(*args, **kwargs)

Creates a decorator that adds a static variable recorders that contains the variable array_debug_recorder to the function it decorates. By default, array_debug_recorder is set to an EmptyArrayRecorder instance. Also, on assignment it automatically creates a subgroup with the same name as the function.

Parameters:a_class (class) – All variables to be named (set to EmptyArrayRecorder()). Returns:

A decorator that adds the static variable

array_debug_recorder to the given function.

Return type:(class)

nanshe.io.hdf5.record.class_static_subgrouping_array_recorders(*args, **kwargs)

Creates a decorator that adds a static variable, recorders, that holds as many recorders as are supplied.

Parameters:args (tuple of strs) – All variables to be named (set to EmptyArrayRecorder()). Keyword Arguments:  kwargs (dict of strs and ArrayRecorders) – All variables to be named with values of type ArrayRecorder. Returns:

A decorator that adds

the static variable, recorders, to the given function.

Return type:(callable)

nanshe.io.hdf5.record.generate_HDF5_array_recorder(*args, **kwargs)

Generates a function used for writing arrays (structured or otherwise) to a group in an HDF5 file.

Parameters:

• hdf5_handle – The HDF5 file group to place the debug contents into.
• group_name – The name of the group within hdf5_handle to save the contents to. (If set to the empty string, data will be saved to hdf5_handle directly)
• enable – Whether to generate a real recorder or a fake one.
• overwrite_group – Whether to overwrite the group where data is stored.
• recorder_constructor – Type of recorder to use if enable is True.
• **kwargs – Other arguments to pass through to the recorder_constructor (won’t pass through if enable is false).

Returns:

A function, which will take a given array

name and value and write them out.

Return type:

ArrayRecorder

nanshe.io.hdf5.record.static_array_debug_recorder(*args, **kwargs)

Creates a decorator that adds a static variable recorders that contains the variable array_debug_recorder to the function it decorates. By default, array_debug_recorder is set to an EmptyArrayRecorder instance. Also, on assignment it automatically creates a subgroup with the same name as the function.

Parameters:callable (callable) – All variables to be named (set to EmptyArrayRecorder()). Returns:

A decorator that adds the static variable

array_debug_recorder to the given function.

Return type:(callable)

nanshe.io.hdf5.record.static_subgrouping_array_recorders(*args, **kwargs)

Creates a decorator that adds a static variable, recorders, that holds as many recorders as are supplied.

Parameters:args (tuple of strs) – All variables to be named (set to EmptyArrayRecorder()). Keyword Arguments:  kwargs (dict of strs and ArrayRecorders) – All variables to be named with values of type ArrayRecorder. Returns:

A decorator that adds

the static variable, recorders, to the given function.

Return type:(callable)