Requiring the exact body for an expectation is not scalable

[uglyog]

Having to provide the exact body for a request is not scalable. Have a look at the resulting ruby pact issue #1

My current project has json bodies that include UUIDs and timestamps, which will never be known beforehand.

By adding a path to a matcher, it would mean I could only match on the things that are important in the body.

[uglyog]

As an example, I currently need to match a body of the following form:

    {
        "event_id": "26c7343d-a19c-4bb7-88a2-107079037b17",
        "timestamp": "2014-04-14T07:02:59.154Z",
        "host": "rholshausen.engr.acx",
        "ip_address": "0:0:0:0:0:0:0:1%0",
        "event_type": "DOCUMENT:EVENT_VIEW_VIEWED",
        "user": {
            "id": 1879049107,
            "username": "poleary",
            "organisation": 1879048492
        }, "document": {
            "id": 1879099218
        }
    }

But I'm only interested in the values for event_type, user.id and document.id. And I don't care about the structure, which can change based on some context.

I would like to say something like (in ruby like pseudocode):

.uponReceiving("a PUT request with an Event in JSON format", "PUT",
    "/events/[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}", headers, 
    aBodyThatMatches(
        '/eventType' => "DOCUMENT:EVENT_VIEW_VIEWED",
        '/user/id' => 1879049107,
        '/document/id' => 1879099218
    )
)

This code can match both a JSON and XML response.

[bethesque]

This assumes that you're using path matchers to pull the values out of the request/response, in the actual code, yes?

[bethesque]

If this was a get request, how would you specify which event would actually be queried for?

[bethesque]

Ah, I think I see what you mean about not specifying the full body, you're talking about the request, not the response. We allow extra values in the response, but not in the request, because we don't want data to leak out. There is a way to turn off this request validation in the ruby pact library, but because I think it's a bad thing (and Martin Fowler agrees with me!), I've not advertised it :P

Do you really think it's a good feature to have?

[uglyog]

yeah, sorry. This is in the consumer test specifying the request matching.

[kenbot]

We absolutely need to keep the Ruby and JVM pacts in sync, regarding
matching capabilities. It looks like, from the discussion linked above,
that the Ruby issue hasn't been fixed either, right? If so, we have a
blank slate to come up with a scheme that can do whatever matching we need
maybe with JSON path-navigation syntax or regexes or whatever.

The implementation complexity could easily get out of hand here, so we
would need to choose wisely.

On 2 May 2014 12:17, Ronald Holshausen notifications@github.com wrote:

As an example, I currently need to match a body of the following form:

{
    "event_id": "26c7343d-a19c-4bb7-88a2-107079037b17",
    "timestamp": "2014-04-14T07:02:59.154Z",
    "host": "rholshausen.engr.acx",
    "ip_address": "0:0:0:0:0:0:0:1%0",
    "event_type": "DOCUMENT:EVENT_VIEW_VIEWED",
    "user": {
        "id": 1879049107,
        "username": "poleary",
        "organisation": 1879048492
    }, "document": {
        "id": 1879099218
    }
}

But I'm only interested in the values for event_type, user.id and
document.id. And I don't care about the structure, which can change based
on some context.

I would like to say something like (in ruby like pseudocode):

.uponReceiving("a PUT request with an Event in JSON format", "PUT",
"/events/[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}", headers,
aBodyThatMatches(
'/eventType' => "DOCUMENT:EVENT_VIEW_VIEWED",
'/user/id' => 1879049107,
'/document/id' => 1879099218
)
)

This code can match both a JSON and XML response.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/33#issuecomment-41977594
.

[bethesque]

I believe there are actually three features we need to discuss here

1. How to add regular expression matching in a way that is language independent (the ruby version has the capability, however, the serialisation of the matcher is ruby specific, so can't be used for a java verification)
2. How to match both request and response bodies in a flexible way that allows /flexible/style/paths to be used.
3. Whether or not there should be the ability to allow extra fields in a request body (ie. allow data to "leak")

[kenbot]

Incidentally, is there any thought to support any other kind of body than
JSON? It would seem inappropriately restrictive to only allow
application/json bodies. This impacts the problem at hand I think.

On 2 May 2014 12:48, bethesque notifications@github.com wrote:

I believe there are actually three features we need to discuss here

1. How to add regular expression matching in a way that is language
   independent (the ruby version has the capability, however, the
   serialisation of the matcher is ruby specific, so can't be used for a java
   verification)
2. How to match in a flexible way that allows /flexible/style/paths to be
   matched.
3. Whether or not there should be the ability to allow extra fields in a
   request body (ie. allow data to "leak")

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/33#issuecomment-41979709
.

[bethesque]

The ruby implementation doesn't just support JSON, it's just that the best matching/diffing functionality is for JSON. You can specify any content type you like, but it will just do a plain text match on it.

There has always been the idea that we'd support XML as well, but nobody involved has been on a project that used XML since we started using it, and nobody has had enough motivation to do it in their free time!

[bethesque]

@uglyog , if you do this, your request body will match if there are extra values in your request body, however, as you probably remember, what gets written to the pact file is the expected request, no the actual request, so you'd be missing the values you didn't specify, which may upset the provider.

  {
    :method => :post,
    :path => '/contract-pricing',
    :headers => {
      'Content-Type' => 'application/json;charset=utf-8',
      'Accept' => 'application/json, text/plain, */*'
    },
    :options => {:allow_unexpected_keys_in_body => true}
  }

As you can see, all the dodgy hacks in pact were made to support Condor :P

[uglyog]

Hmm, it looks like the pact-jvm does not allow unexpected keys in the body (it is set in a constant to false).

I've done some hackery to get it to work for me by changing the constant to a var, but it should have options as per the ruby one.

I'm wondering if the actual request should be recorded for the pact verification. For my case there are values that I will not know before hand, but they will be required for the validation with the provider to succeed.

[thetrav]

The request doesn't allow extra keys, the response does, I thought that was in line with the ruby version.

There's a PactServerConfig case class which could / should be used for settings like this

[bethesque]

You were right in your assumption Trav, it's not supposed to be a feature, it was a hack I made to support Condor.

Ron, we'd need to move the pact serialization to the mock server if we were to use the actual request. I'm not convinced this is a good thing though. Wouldn't type based matching be a better option for the values that are unknown beforehand?

[uglyog]

Type based matching would work, but then we need a regex format that works across both implementations ;-)