Skip to content

Latest commit

 

History

History
197 lines (138 loc) · 8.26 KB

README.md

File metadata and controls

197 lines (138 loc) · 8.26 KB
Raw

DynamoDB configuration and table namespacing

This repository is a collection of small functions that build on top of boto3's dynamodb to encourage better configuration and per-environment tables.

Here's a bullet-point summary:

  • provides a convenient model interface to query, scan, create, update and delete items
  • boto3 integration with conversion between dynamodb and pythonic data
  • parses table configuration as defined in a YAML file (see tests/dynamodb.yml)
  • namespaces tables so you can share the same configuration between different environments
  • gives you Table objects that have the schema and indexes loaded locally so you avoid extra lookups
  • direct calls to create or update tables by name as the configuration changes
  • optional ability to define non-indexed columns and types of data you expect to store

TODO: add back to Solano. Old cc_dynamodb was running.

Example usage:

Models:

from cc_dynamodb3.models import DynamoDBModel

from schematics import types as fields

class TestModel(DynamoDBModel):
    TABLE_NAME = 'test'

    agency_subdomain = fields.StringType(required=True)
    external_id = fields.IntType()
    name = fields.StringType()
    is_enabled = fields.BooleanType()


cc_dynamodb3.set_config(
    config_file_path='path/to/yaml/file.yml',
    aws_access_key_id='<KEY>',
    aws_secret_access_key='<SECRET>',
    namespace='dev_')

obj = TestModel.create(agency_subdomain='test')  # calls PutItem
obj.is_enabled = True
obj.save()                                       # calls UpdateItem

for obj in TestModel.all():
    print(obj.agency_subdomain)  # prints 'test'

And configuration:

schemas:
    test:  # note: no namespacing here
        -
            type: HashKey
            name: agency_subdomain
            data_type: STRING

Plain:

import cc_dynamodb3

cc_dynamodb3.set_config(
    config_file_path='path/to/yaml/file.yml',
    aws_access_key_id='<KEY>',
    aws_secret_access_key='<SECRET>',
    namespace='dev_')

table = cc_dynamodb3.table.get_table('employment_screening_reports')
    # Returns the boto Table object
    # after figuring out the DynamoDB table name (via namespace)
    # and loading the schema and indexes from the config.
table.scan()  # Table uses boto3 interface

API

Configuration

Configuration is currently stored globally as a variable and otherwise not cached. So you can think of it as per-thread level caching. There are two functions:

get_config(**kwargs)

Returns the cached config. Calls set_config first if no cached config was found.

set_config(**kwargs)

Loads up the YAML configuration file and validates dynamodb connection details. The following are required, either set through the environment, or passed in as kwargs (to overwrite):

  • namespace, determines the table name prefix. Each repository using this library should have a unique namespace.
  • aws_access_key_id and aws_secret_access_key, the AWS connection credentials for boto's connection. Examples shown in the tutorial
  • table_config, a path to the YAML file for table configuration.

dynamodb.yml

This file contains the table schema for each table (required), and optional secondary indexes (global_indexes or indexes (local secondary indexes).

set_redis_config(host='localhost', port=6379, db=3)

The headline is an example call. Redis caching is optional, but may greatly speed up your server performance.

Redis caching is used to avoid parsing the YAML file every time set_config() is called.

Usage

The following are all at the cc_dynamodb3 top level. With the exception of get_reverse_table_name, you should always use the unprefixed table name (exactly as from the configuration file).

|------------------------------------------------------------------------------------------|
| Function name            | Docstring                                                     |
|------------------------------------------------------------------------------------------|
| get_table_name           | Given a table name, return the prefixed version.              |
|------------------------------------------------------------------------------------------|
| get_reverse_table_name   | Given a namespaced table name, return the unprefixed version. |
|------------------------------------------------------------------------------------------|
| get_table_index          | Given a table name and an index name, return the index.       |
|------------------------------------------------------------------------------------------|
| get_connection           | Returns a DynamoDBConnection even if credentials are invalid. |
|------------------------------------------------------------------------------------------|
| get_table_columns        | Return known columns for a table and their data type.         |
|------------------------------------------------------------------------------------------|
| query_table              | Provides a nicer interface to query a table than boto3        |
|                          | default.                                                      |
|------------------------------------------------------------------------------------------|
| get_table                | Returns a dict with table and preloaded schema, plus columns. |
|------------------------------------------------------------------------------------------|
| list_table_names         | List known table names from configuration, without namespace. |
|------------------------------------------------------------------------------------------|
| create_table             | Create table. Throws an error if table already exists.        |
|------------------------------------------------------------------------------------------|
| update_table             | Handles updating primary index and global secondary indexes.  |
|                          | Updates throughput and creates/deletes indexes.               |
|------------------------------------------------------------------------------------------|

Mocks: cc_dynamodb3.mocks

This file provides convenient functions for testing with boto3's dynamodb.

You can do most of the testing using the moto library directly, but some things are not supported (mainly, global secondary indexes and querying via them).

Paul's fork of moto supports querying, and there is a Pull Request open to merge it upstream.

mock_table_with_data

Create a table and populate it with array of items from data.

Example:

data = [{'key_1': 'value 1'}, {'key_1': 'value 2'}]
table = mock_table_with_data('some_table', data)
# Expect 2 results
assert len(table.scan()) == 2

Quickstart

In your configuration file, e.g. config.py:

DYNAMODB_CONFIG_PATH = os.path.join(os.path.dirname(__file__), 'dynamodb.yml')
DATABASE = dict(
    namespace='test_',
    table_config=DYNAMODB_CONFIG_PATH,
    aws_access_key_id='test',
    aws_secret_access_key='secret',
)

If you want to use DynamoDB Local, just pass host as a parameter in the connection, e.g.:

DATABASE = dict(
    namespace='test_',
    host='localhost',
    table_config=DYNAMODB_CONFIG_PATH,
    aws_access_key_id='test',
    aws_secret_access_key='secret',
)

This uses AWS's provided jar file to run DynamoDB locally. Read more here.

In your database file, e.g. db.py:

import cc_dynamodb3
dynamodb_config = config.DATABASE
cc_dynamodb3.set_config(**dynamodb_config)

Then you can use the library directly:

import cc_dynamodb3
TABLE_NAME = 'some_table'

table = cc_dynamodb3.get_table(TABLE_NAME)
item = table.get_item(Key={'some_key': 'value'})

Dynamodb Tutorial

For more on boto3's dynamodb interface, please see their guide.