Skip to content
/ eel Public

A simple proxy service to forward JSON events and transform or filter them along the way.

License

Notifications You must be signed in to change notification settings

Comcast/eel

EEL - A simple Proxy Service for JSON Event Transformation and Forwarding

It's simple - a single JSON event comes in and one (or maybe a few) transformed events get out. Events are arbitrary JSON encoded payloads and they are typically sent around as HTTP POSTs. EEL is stateless and therefore scales easily.

In this example we removed the unwanted "detail" field from the event and adjusted the schema slightly.

Converting JSON payload A from service S1 into JSON payload B for service S2 is a very common operation we perform all the time in service oriented architectures. Here are some good reasons for event transformation:

  1. REDUCTION: Remove everything that's not needed and only pass the data actually required by the downstream service. This can provide performance benefits for the downstream service if processing is expensive and events are bloated.
  2. CANONICALIZATION / CLEANUP: Adjust the structure of the JSON event to validate against a given schema. Typically this can be achieved by rearranging elements in the JSON event, renaming keys and possibly filtering out unwanted subsets of data.
  3. META-TAGGING: Inject additional metadata into the JSON event. Examples are injecting UUIDs or timestamps (for example for log tracing), tagging or labeling events with type fields, calling external auxiliary services and injecting their output into the downstream payload (provided it is JSON).
  4. MAPPING / SUBSTITUTION: Map one ID type to another via look up using an external lookup service.
  5. FILTERING: Filter out unwanted events to reduce load for downstream services.
  6. FANOUT: Forward a single incoming JSON event to multiple downstream services, either by applying the same transformation for all services or by performing different transformations.

You could write each of these transformations in just a few lines of go code, and we often do. The downside is that transformations become engraved in code and cannot be changed easily.

EEL is offering a simple JSON template language combined with a proxy micro service to relay and transform events between upstream and downstream services. The goals of the EEL template language are to be simple and yet powerful. EEL can be the glue in a JSON based service oriented eco-system.

The syntax of the EEL transformation language is inspired by a few of the core concepts from XPath and XSLT applied to JSON.

Installation

go get -u github.com/comcast/eel

Usage

./eel [options...]

Options:

-config  path to config.json (default is ./config-eel/config.json)
-handlers  path to handlers (default is ./config-handlers)
-loglevel  log level (default is "info")
-env  environment name such as qa, prod for logging (default is "default")

No Nonsense

go build -o bin/eel
./bin/starteel.sh

Docker Alternative

Docker 1.13+ verified.

Build a dev image

docker build -t eel:dev .

Run an instance

docker run --rm -p 8080:8080 --name eel-dev eel:dev

The command above will utilize port 8080 of your host. You can change it to any other port via -p ANYOTHERPORT:8080

To pass parameters to eel you can use EEL_PARAMS env variable, e.g.

docker run --rm -e "EEL_PARAMS=-loglevel error" -p 8080:8080 --name eel-dev eel:dev

A Simple Example

Transformation handlers are used to tell EEL if and how to process JSON events it receives and where to forward them to. Each transformation handler is stored in a single JSON file and there may be several such handler files each of them taking care of a certain class of events. See here for more details.

Edit the default transformation handler in config-handlers/tenant1/default.json so it looks like this (there's only a small change needed in the Transformation section).

{
    "Version": "1.0",
    "Name": "Default",
    "Info": "",
    "Active": true,
    "Match": null,
    "IsMatchByExample": false,
    "TerminateOnMatch": true,
    "Transformation": {
        "{{/event}}": "{{/}}"
    },
    "IsTransformationByExample": false,
    "Path": "",
    "Verb": "POST",
    "Endpoint": "http://localhost:8082",
    "HttpHeaders": {
      "X-B3-TraceId": "{{traceid()}}",
      "X-Tenant-Id": "{{tenant()}}"
    }
}

The center piece of the handler configuration is the Transformation section which uses JPath expressions to describe the structural transformations to be performed on the incoming event before forwarding it to the downstream service. In this example we are telling EEL to take the entire payload unmodified and wrap it inside a new element called event.

Compile EEL.

EEL is implemented in go and you will need a golang environment to compile the code. However, you don't need any go coding skills to understand and author EEL templates!

The make file and shell scripts have been tested on Mac and Linux environments.

make all

Launch EEL as a service and start listening to incoming events.

./bin/starteel.sh

Send a test event to EEL.

curl -X POST --data '{ "message" : "hello world!!!" }' http://localhost:8080/v1/events

Output:

{"status":"processed"}

You can check EEL's log output eel.log to see if and how the event is processed. You can also ask for a detailed debug response using the X-Debug header.

curl -X POST -H 'X-Debug: true' --data '{ "message" : "hello world!!!" }' http://localhost:8080/v1/events

Output:

[
	{
		"api": "http",
		"handler": "Default",
		"tenant.id": "tenant1",
		"trace.in.data": {
			"message": "hello world!!!"
		},
		"trace.out.data": {
			"event": {
				"message": "hello world!!!"
			}
		},
		"trace.out.endpoint": "http://localhost:8088",
		"trace.out.headers": {
			"X-B3-TraceId": "20073ee4-d681-4ab5-a973-50c978cd1111",
			"X-Tenant-Id": "tenant1"
		},
		"trace.out.path": "",
		"trace.out.protocol": "http",
		"trace.out.url": "http://localhost:8088",
		"trace.out.verb": "POST",
		"tx.id": "20073ee4-d681-4ab5-a973-50c978cd1111",
		"tx.traceId": "20073ee4-d681-4ab5-a973-50c978cd1111"
	}
]

Or, you can use the /v1/sync/events instead of the /v1/events endpoint to get an immediate response containing the transformed event as it would be forwarded to the downstream service.

curl -X POST --data '{ "message" : "hello world!!!" }' http://localhost:8080/v1/sync/events

Output:

{
	"event": {
		"message": "hello world!!!"
	}
}

To review the current set of active transformation handlers call the health check API.

curl http://localhost:8080/v1/health

Stop EEL.

./bin/stopeel.sh

EEL as Command Line Tool

You can also start experimenting with EEL by using the command line parameters. Example:

./eel -in='{"foo":"bar"}' -tf='{"Foo":"{{/foo}}"}' -istbe=true

More examples can be found here .

Exploring EEL Features

The unit tests are a good starting point to learn more about EEL features and look at some examples in detail.

Each test is contained in its own configuration folder, for example eel/test/data/test01. Each test folder contains a complete set of handler configurations for EEL in the handlers subfolder (usually just one handler), an example input event in.json and one or more expected output events out.json.

The tests can be executed using go test:

cd eel/test
go test -v

Or, you can launch EEL with the handler configurations for a specific test and send the in.json event manually.

./eel -handlers=test/data/test01/handlers > eel.log &
curl -X POST --data @test/data/test01/in.json http://localhost:8080/v1/sync/events
No Name Test Name Description
0 Identity Transformation TestDontTouchEvent Doesn't apply any transformation and forwards everything unchanged.
1 Canonicalize TestCanonicalizeEvent Simple structural changes and array path selectors.
2 External Lookup TestInjectExternalServiceResponse Get JSON data from external service and inject into payload.
3 Transformation By Example TestTransformationByExample Describe transformation using by-example syntax rather than using by-path syntax.
4 Named Transformations TestNamedTransformations Choose from several named transformations based on input data.
5 Conditional Message Generation TestMessageGeneration Assemble message string with ifte() and equals().
6 Handler Matching 1 TestTerminateOnMatchTrue Pick best matching handler and forward single event.
7 Handler Matching 2 TestTerminateOnMatchFalse Pick all matching handlers and forward multiple events.
8 Multi Tenancy TestMultiTenancy Handlers for different tenants or apps.
9 Cascade TestSequentialHandlerCascade Cascade of multiple handlers which will be executed sequentially by using EEL recursively.
10 Java Script For Everything Else TestJavaScript If you really can't avoid it, resort to Java Script.
11 Handler Matching 3 TestMatchByExample Matching handlers using by-example syntax.
12 Convert Headers To Payload TestHeaders Inject HTTP headers from upstream service into JSON event for downstream service.
13 Custom Properties TestCustomProperties Custom properties in handlers for sharing data.
14 Fan Out TestFanOut Send incoming event to several downstream services.
15 Basic String Operations TestStringOps Uppercase, lowercase, substring.
16 Named Transformations 2 TestNamedTransformations2 Perform named transformation on external document.
17 Contains TestContains Check if one JSON event is contained in another.
18 Tenant Id Header TestMultiTenency2 Pass in tenant id as HTTP header.
19 Conditional Message Generation 2 TestMessageGeneration2 Use case() function to simplify conditional string generation.
20 Regex TestRegex Use regex() to evaluate regular expressions.
22 Filter By Path TestFilterByPath Filter event after transformation using by-path syntax.
23 Filter By Example TestFilterByExample Filter event after transformation using by-example syntax.
25 Array Path Selector TestArrayPathSelector Select elements from arrays by index or by path.
27 Iterate Over Array NamedTransformationsAndArrays Iterate over array and apply named transformation.
32 Simple Types NamedTransformationsAndSimpleTypes Apply named transformation to a simple type.
34 Join Join Merge two JSON documents.
41 Partial Matching MatchPartialArrays3 Match event against partial pattern.
46 Filter Cascade FilterCascade Apply multiple filters.
47 Choose From Array ChooseFromArray Choose elements from array by pattern.
48 Named Transformation With Array And Pattern NamedTransformationWithArrayAndPattern Apply named transformation by pattern.
49 Named Transformation With Array And Join NamedTransformationWithArrayAndJoin Apply named transformation with join.
51 Complex Example ComplexExample A real world example.
53 Crush Crush Flatten a deeply nested array.

Further Reading

About

A simple proxy service to forward JSON events and transform or filter them along the way.

Topics

Resources

License

Code of conduct

Security policy

Stars

Watchers

Forks

Packages

No packages published

Languages