node-consumer-pact-interceptor

Node Consumer Pact interceptor

This is a pure nodejs implementation of the PACT system of mocking and testing system integration points. It is presently in an Alpha state and you should expect some bugs. Be sure to read the caveats below.

Quickstart

See the examples/passing-example for the fastest way to get started.

Workflow:

(This assumes understanding of - Consumer driven contracts - The PACT v2 specification)

1. Create pact JSON spec
2. Feed this to the interceptor (As below examples)
3. Send requests as expected to the provider, they will be intercepted as outgoing HTTP requests
4. Throw any failures received by the interceptor, else verify the response is as expected
5. Publish pact to broker

Why?

There is a JS DSL for creating pacts which is already compatible with the Pact v2 Spec. However, this requires the use of a ruby server with which to create assertions with.

We found that it was difficult to induct new developers into using this system and the incidental complexity barrier (often compounted by CI servers and docker-containers) was such that this became a real pain-point.

How does it work

The interceptor is just wrapping the excellent MITM library which is catching outgoing HTTP requests at the NodeJS core level and allowing responses to be injected.

From a high-level, the interceptor waits for requests the URL it's watching and, once it receives them, it will try verify this against the PACT specification. If it fails to do so, it will return an assertion error, if there is no assertion failures it will respond as per the PACT specification.

Caveats:

• As yet this does not fully implement Pact V2. At the time of writing I have been unable to fully meet the requirements of nested type-matching.
• This is Based around a single interaction per pact-file. This was a first-cut design-decsion and will be revisited if it appears necessary.
• Outgoing HTTP or socket requests which are not part of the pact test are going to be blocked. This is unfortunate and less than ideal. I have not yet found a way to use MITM's API to allow HTTP requests on a per-url basis.