Package mekano :: Package atoms :: Module atomfactory :: Class AtomFactory

Class AtomFactory

A single AtomFactory makes unique atoms for the given objects. By atoms, we just mean numbers. Objects just have to be hashable.

To get the Pipeline's (type, object) atoms, just use tuples!

>>> af = AtomFactory("mytokens")
>>> a1 = af["apples"]
>>> a2 = af["oranges"]
>>> assert(a1 == 1)
>>> assert(a2 == 2)
>>> assert(af(1) == "apples")
>>> a.lock()                        # Do not allow changes.

Loading/saving:

>>> a = AtomFactory.fromfile(filename)
>>> a.save(filename)

Note: af(1) is candy for af.get_object(1)

Instance Methods

[hide private]

__init__(self, name='noname') source code

__repr__(self)

source code

__getitem__(self, obj)

source code

get_object(self, a)

source code

__call__(self, a)

source code

__len__(self)

source code

__contains__(self, obj)

source code

lock(self)
Lock the AtomFactory.

source code

remove(self, objects)
Returns a new AtomFactory with the given objects removed.

source code

save(self, filename)

source code

savetxt(self, filename)
Save each object on a line.

source code

Static Methods

[hide private]

fromfile(filename)

source code

Method Details

[hide private]

lock(self)

source code

Lock the AtomFactory.

No new atoms can be added; Only old ones can be retrieved.

savetxt(self, filename)

source code

Save each object on a line.

This should be enough to reconstruct the AtomFactory, and is also useful for things like LDA's vocabulary file.