A DynamoDB client for Go
Clone or download
Pull request Compare This branch is 2 commits ahead of underarmour:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
internal
schema
streams
.travis.yml
LICENSE.txt
README.md
client.go
client_env.go
client_test.go
const.go
consumed_capacity.go
consumed_capacity_test.go
doc.go
errors.go
errors_test.go
example_client_test.go
example_test.go
executor.go
functional_test.go
helpers.go
mock_executor.go
mock_executor_test.go
request_batch.go
request_batch_test.go
request_delete_item.go
request_get_item.go
request_item_test.go
request_put_item.go
request_query.go
request_scan.go
request_table.go
request_update_item.go
types.go
types_test.go
wire.go
wire_test.go

README.md

Dynago

Build Status GoDoc

Dynago is a DynamoDB client API for Go.

Key design tenets of Dynago:

  • Most actions are done via chaining to build filters and conditions
  • objects are completely safe for passing between goroutines (even queries and the like)
  • To make understanding easier via docs, we use Amazon's naming wherever possible.

Installation

Install using go get:

go get github.com/almonteb/dynago

Docs are at http://godoc.org/github.com/almonteb/dynago

Example

Run a query:

client := dynago.NewAwsClient(region, accessKey, secretKey)
// or you can use below if you have AWS credential values in ENV,
// in case of running on AWS Lambda, etc.
// client := dynago.NewAwsClientFromEnv()

query := client.Query(table).
	KeyConditionExpression("UserId = :uid", dynago.P(":uid", 42)).
	FilterExpression("NumViews > :views").
	Param(":views", 50).
	Desc()

result, err := query.Execute()
if err != nil {
	// do something
}
for _, row := range result.Items {
	fmt.Printf("Name: %s, Views: %d", row["Name"], row["NumViews"])
}

Type Marshaling

Dynago lets you use go types instead of having to understand a whole lot about dynamo's internal type system.

Example:

doc := dynago.Document{
	"name": "Bob",
	"age": 45,
	"height": 2.1,
	"address": dynago.Document{
		"city": "Boston",
	},
	"tags": dynago.StringSet{"male", "middle_aged"},
}
client.PutItem("person", doc).Execute()
  • Strings use golang string
  • Numbers can be input as int (int64, uint64, etc) or float64 but always are returned as dynago.Number to not lose precision.
  • Maps can be either map[string]interface{} or dynago.Document
  • Opaque binary data can be put in []byte
  • String sets, number sets, binary sets are supported using dynago.StringSet dynago.NumberSet dynago.BinarySet
  • Lists are supported using dynago.List
  • time.Time is only accepted if it's a UTC time, and is marshaled to a dynamo string in iso8601 compact format. It comes back as a string, an can be got back using GetTime() on Document.

Debugging

Dynago can dump request or response information for you to use in debugging. Simply set dynago.Debug with the necessary flags:

dynago.Debug = dynago.DebugRequests | dynago.DebugResponses

If you would like to change how the debugging is printed, please set dynago.DebugFunc to your preference.

Version Compatibility

Dynago follows Semantic Versioning via the gopkg.in interface, and within the v1 chain, we will not break the existing API or behaviour of existing code using Dynago. We will add new methods and features, but it again should not break code.

Additional resources

The past, and the future

Dynago currently implements all of its support for the underlying DynamoDB API encoding, AWS signing/authentication, etc. This happened in part because the existing libraries out there at the time of writing used deprecated API's and complicated methods, and it was actually cleaner at the time to support the API by fresh implementation.

AWS-SDK-Go exists as of June 2015 and has a very up to date API, but the API is via bare structs which minimally wrap protocol-level details of DynamoDB, resulting in it being very verbose for writing applications (dealing with DynamoDB's internal type system is boilerplatey). Once Amazon has brought it out of developer preview, the plan is to have Dynago use it as the underlying protocol and signature implementation, but keep providing Dynago's clean and simple API for building queries and marshaling datatypes in DynamoDB.