Python 的 Redis 对象映射器.zip

#----------------------------------- 1.1.1 ----------------------------------- [added] IDENTITY_STRING and IDENTITY_STRING_CI keygens for [fixed] exception edge case when saving / deleting index-only columns with a unique constraint and no relevant data. #----------------------------------- 1.1.0 ----------------------------------- [added] UnsafeColumns(); used for direct access to Redis structures per entity, which are deleted on entity deletion. #----------------------------------- 1.0.7 ----------------------------------- [fixed] searching by unique indexonly columns; previously would try to use the useless "to_redis" on the column type, which returns none. #----------------------------------- 1.0.6 ----------------------------------- [added] IndexOnly columns can now have unique indexes generated via keygen. #----------------------------------- 1.0.5 ----------------------------------- Re-release for git tags #----------------------------------- 1.0.4 ----------------------------------- [added] dockerfiles + docker-compose yamls for testing. [updated] Python version compatability. #----------------------------------- 1.0.3 ----------------------------------- [fix] explicitly added support for Python 3.7, 3.8, and 3.9 that were already being tested in travis, but which we missed. Thanks to github user https://github.com/Frank5000 for the bug report. Also removed extra-old and generally unsupported Python 2.6 and 3.3 from the list. If you *need* support for these, please contact me about a support license. #----------------------------------- 1.0.2 ----------------------------------- [fix] thanks to bug report by github user https://github.com/Goury, fixed old bug with rebuilding indexes via scan. #----------------------------------- 1.0.1 ----------------------------------- [fix] thanks to bug report and instructions from github user https://github.com/mshalom-meta, I have fixed rom to be compatible with Redis connections having `decode_responses=True` set. #----------------------------------- 1.0.0 ----------------------------------- [WARNING] data indexed in this version is NO LONGER COMPATIBLE with data indexed in versions < 1.0.0, THIS VERSION WILL AUTOMATICALLY CONVERT YOUR INDEXES on save, and older clients saving may not update indexes properly. [future change] At some point, we'll stop supporting Python 2.7. I don't know when that is, but it will be soon-ish. [changed] data storage for "old" indexed data (so we can delete items) has changed to reduce the number of big keys. [changed] If you previously used no-filter model.query.count() without a primary index, that previously worked. That will now return a lie, in order to be fast. Sorry. [added] You can now transfer value from one entity to another, marking success in a 3rd "transfer" entity in a single Redis round-trip. Added from an inbound email + self-answered StackOverflow answer. Check the new test for how one might use that + indexes + json in order to handle the 1) payment, and 2) item transfer for purchaseable items in a market. Thanks to Roman V for the question / feature request. [added] You can now iterate over all entities that do not match a given set of prefixes or suffixes on a known prefix/suffix column. Thanks to github user https://github.com/bgervan for the feature request. [fixed] upstream travis config; Redis 3.2 and 4.0 were removed from Docker repo [fixed] Thanks to github user https://github.com/cdessez for the bug report and PR to not pass None to a Lua script. Had to fix upstream Travis CI [added] Utility function to refresh all indexes, while showing progress. #---------------------------------- 0.42.6 ----------------------------------- [added] get related model from relationship columns thanks to PR from github user https://github.com/Aimage ; MyModel.column_name.get_related_model() #---------------------------------- 0.42.5 ----------------------------------- [fixed] Odd indexing and query edge cases that would make some 3.6 String (bytes) indexes not usable when Text (unicode) was passed. Reindexing should fix the issue. #---------------------------------- 0.42.4 ----------------------------------- [fixed] Thanks to github user https://github.com/stacywsmith for the bug report and PR for fixing the Redis client compatability edge case due to the Redis Python client upgrades [changed] Upgraded machines, so upgraded sphinx, so docs got mangled, huge changelog for what amounts to a few lines of library code change #---------------------------------- 0.42.3 ----------------------------------- [fixed] IndexOnly() columns no longer accept the unique parameter (was ignored in effect, anyway) [added] IndexOnly() columns now take a keygen2 parameter in lieu of the column + keygen argument. [fixed] Some links in the docs #---------------------------------- 0.42.2 ----------------------------------- [fixed] Model.query.select().limit() will now honor the passed-in limit. Thanks to Github user https://github.com/healiseu for the bug report [fixed] Model.query.select() will now no longer pollute non-null session objects with partial / broken objects. Thanks to Github user https://github.com/healiseu for the bug report #---------------------------------- 0.42.1 ----------------------------------- [fixed] Session object wasn't always initialized correctly. Bug report thanks to github user https://github.com/cdessez . #---------------------------------- 0.42.0 ----------------------------------- [changed] Attributes passed on object creation that don't exist will now raise an exception, unless you explicitly pass _extra_ok=True. [fixed] a fix from bugs 108/109 where objects wouldn't be added to the session when they should be, and would when they shouldn't be. [fixed] potential issue when using select and order_by; could have returned model entities instead of columns selected. [fixed] issue with iterators over select queries possibly returning bad data silently (columns would all be 'False'). [fixed] issue with entities going into the session during arbitrary queries where they shouldn't have. [fixed] hopefully got sphinx docs to generate consistent order, so subsequent commits don't result in rewriting it all for every commit :/ #---------------------------------- 0.41.2 ----------------------------------- [added] Model.update(*args, **kwargs) as a convenience method per requset from github user https://github.com/61t [fixed] multi-subclassed models may have raised a duplicate PrimaryKey error on initialization. [fixed] small issue with over-writing values on IndexOnly() columns not raising an exception. #---------------------------------- 0.41.1 ----------------------------------- [fixed] pythonhosted.org is not going to be hosting docs forever, this moves the docs to github pages and updates references. #---------------------------------- 0.41.0 ----------------------------------- [fixed] in some cases, bad column metadata could get left in the foreign key column metadata in rom internals. Shouldn't have affected anything, except some extra memory being used unnecessarily. [added] Model.query.delete(blocksize=100) - for deleting models that match a query. Requires that the model not have any ManyToOne, OneToMany, or OneToOne columns or other foreign-key relationships. This should speed up common deletion/cleanup tasks. [added] util.session.delete(...) - for deleting a list of arbitrary models in one round trip to Redis, if possible (if all entites are of one type, and you haven't done anything strange with per-entity connections). Same no foreign-key relationships limitation exists here. [fixed] a few misspellings in the docs (there are probably more). [added] columns.IndexOnly() - allowing multiple indexes and index types over columns, without explic