Lazy attributes for Python objects
- A decorator to create lazy attributes.
Lazy attributes are computed attributes that are evaluated only once, the first time they are used. Subsequent uses return the results of the first call. They come handy when code should run
- late, i.e. just before it is needed, and
- once, i.e. not twice, in the lifetime of an object.
You can think of it as deferred initialization. The possibilities are endless.
The class below creates its store resource lazily:
from lazy import lazy class FileUploadTmpStore(object): @lazy def store(self): location = settings.get('fs.filestore') return FileSystemStore(location) def put(self, uid, fp): self.store.put(uid, fp) fp.seek(0) def get(self, uid, default=None): return self.store.get(uid, default)
Another application area is caching:
class PersonView(View): @lazy def person_id(self): return self.request.get('person_id', -1) @lazy def person_data(self): return self.session.query(Person).get(self.person_id)
I first encountered this type of descriptor in the zope.cachedescriptors package, which is part of the Zope Toolkit.
1.3 - 2017-02-05
- Support Python 2.6-3.6 without 2to3. [stefan]
- Add a LICENSE file. [stefan]
1.2 - 2014-04-19
- Remove setuptools from install_requires because it isn’t. [stefan]
1.1 - 2012-10-12
- Use functools.wraps() properly; the list of attributes changes with every version of Python 3. [stefan]
1.0 - 2011-03-24
- Initial release.