mymusictaste / inpynamodb Goto Github PK

Test failed on test_model.test_index_*

        # on base.AsyncConnection.query
        tbl = self.get_meta_table(table_name)
        if tbl is None:
            raise TableError("No such table: {0}".format(table_name))
        if index_name:
            hash_keyname = tbl.get_index_hash_keyname(index_name)
            if not hash_keyname:
>               raise ValueError("No hash key attribute for index: {0}".format(index_name))
E               ValueError: No hash key attribute for index: uid_index

maybe related to: pynamodb/PynamoDB#316

Models

• Model.create_table()
• Model.delete_table()
• Model.save()
• Model.query()
• Model.scan()
• Model.delete()

Indexes

• class GlobalSecondaryIndex
• class LocalSecondaryIndex
• Model.view_index.query()

Batch Operations

• Model.batch_write()
• Model.batch_get()
• Model.scan()

Update

• Model.update()
• Model.update_item()

order of overridden methods in classes inherited from PynamoDB should be same order with parent class.

Add DynamoDB's TTL feature

• UUIDAttribute for automatically generate UUID (low priority)
• PasswordAttribute as a serializer for auto encrypting, decrypting (high priority) -> EncryptedAttribute()

• Documentation
  • readme.md
  • wiki (how-to-use, for developers)

Trying to update a ListAttribute with a MapAttribute value. Some of the values in the MapAttribute have UnicodeAttribute with null=True, default="".
When initializing MapAttribute, the UnicodeAttributes mentioned above have string type "" in them, but when I try to add it to the ListAttribute through update, it raises the following error:

I think the values degenerate between the process from calling the update function to the dispatch. Not sure if this is Inpynamodb problem or pynamodb problem

• Table backup
• Signals
  • post
  • pre
• convert model to Dictionary. maybe Model.as_dict() -> should convert to dict all MapAttributes too. (Use get_attribute() if attr is MapAttribute then attr.as_dict()?)
• Override Model._set_defaults() and make it public method, to be able to change model attributes if default is specified, otherwise no change.

Situation:
I have a global secondary index with a hash key and range key.
When I query the index using a certain hash key with a last evaluated key from a previous query,
I get the list of items that STARTS with the item of last evaluated key.
For example,
let the following list of items [a,b,c,d] be a hypothetical table that I have.
When I do limit = 1, then I get a.
When I do limit =1 and last_evaluated_key=last_evaluated_key of previous query(which is = to a as key), then I still get a.
When I do limit = 2, then I get [a,b].
When I do limit = 2 and last_evaluated_key=last_evaluated_key of previous query(which is = to b as key), then I get [b,c].

The issue does not occur on local secondary index. I am not sure about regular queries on the model itself.

0.0.14 requires following dependency:

aiobotocore version 0.8.0 from 0.6.0
aiohttp version >= 3.1.0 from 3.0.0

but it causes this dependency issue:

in my case, solved with this:

boto3==1.7.12
botocore==1.10.12
aiohttp==3.1.0

Hi, I'd like to use it but want to be sure it is production-ready. I see no commits for 2 years. Does it work?

Currently, in the _get_connection(), new connection is established if there isn't a cached instance of TableConnection. However, this means that even when a new event loop is generated, the Model instances refer to the old connection which may have closed and therefore unable to perform any actions.

A hotfix would be to change the logic in the _get_connection() method to account for this.
But I think this might really have to do with the "session" denoted by the session_cls keyword argument when creating TableConnection. Something more elaborate / smooth can be done with this.

Now get_settings_value from settings.py PynamoDB version was edited to pass tests.

Should find a way to solve this.

PROBLEM:

• since __repr__() calls _get_meta_data(), __repr__() should be written in asyncio way.

  /Users/minjeongkim/.pyenv/versions/mmt-server-event/lib/python3.6/site-packages/pynamodb/attributes.py:232: UserWarning: `Model._get_attributes` is deprecated in favor of `Model.get_attributes` now
    warnings.warn("`Model._get_attributes` is deprecated in favor of `Model.get_attributes` now")

apparently, used in /inpynamodb/indexes.py:94, etc, ...

Currently, when we save an object with a UTCDatetimeAttribute, if the tzinfo is not set, it is assumed that the timezone is UTC 00:00. However, this results in incorrect time data, since if I have the default equal to datetime.datetime.now, then it automatically assumes UTC 00:00 when the location may not necessarily be where UTC is 00:00.

I see two solutions.
One: a hard requirement to give the tzinfo as the parameter - this would break some of the existing code but it will put responsibility on individual developers to specify the tzinfo.
Two: the UTCDatetimeAttribute figuring out the tzinfo of the current location instead of simply defaulting to 00:00. This would be a cleaner solution I believe.

I get the following warning at the end of executing the code

Unclosed client session
client_session: <aiohttp.client.ClientSession object at 0x1193fa0d0>
Unclosed connector
connections: ['[(<aiohttp.client_proto.ResponseHandler object at 0x119419980>, 3.3025964)]']
connector: <aiohttp.connector.TCPConnector object at 0x1193fa050>

Should closing the aiobotocore client still be implemented or am I using it wrong?

1. on test_base_connection.py

with patch(PATCH_METHOD) as req:
    req.side_effect = BotoCoreError . # This doesn't raise BotoCoreError exception.
    await self.assertAsyncRaises(TableError, conn.create_table(self.test_table_name, **kwargs))

• using tox
• should pass PynamoDB's test
• should pass asynchronous operation test

Situation:
Attempting to execute query on a global secondary index of a model.
Attempted query :

query = Promotion.artist_index.query(artist_id, Promotion.progress != "D"))

The query works with

Promotion.progress.startswith("D")

I am not sure what the problem might be. Please take a look into it when you have time. Thank you.