chameleon

chameleon

What is chameleon?

chameleon is a caching reverse proxy.

chameleon supports recording and replaying requests with the ability to customize how responses are stored.

Why is chameleon useful?

• Proxy rate-limited APIs for local development
• Create reliable test environments
• Test services in places you normally couldn't due to firewalling, etc (CI servers being common)
• Improve speed of tests by never leaving your local network
• Inspect recorded APIs responses for exploratory testing
• Stub out unimplemented API endpoints during development

What can't I do with chameleon?

• Have tests that exercise a given service right now as results are cached
• Total control on how things are cached, frequency, rate-limiting, etc (pull requests are welcome, though!)

How to get chameleon?

chameleon has no runtime dependencies. You can download a prebuilt binary for your platform.

If you have Go installed, you may go get github.com/nickpresta/chameleon to download it to your $GOPATH directory.

How to use chameleon

• Check out the example directory for a small app that uses chameleon to create reliable tests with a custom hasher.

To run chameleon, you can:

chameleon -data ./httpbin -url http://httpbin.org -verbose

The directory httpbin must already exist before running.

See chameleon -help for more information.

Specifying custom hash

There may be a reason in your tests to manually create responses - perhaps the backing service doesn't exist yet, or in test mode a service behaves differently than production. When this is the case, you can create custom responses and signal to chameleon the hash you want to use for a given request.

Set the chameleon-request-hash header with a unique hash value (which is a valid filename) and chameleon will look for that hash in the spec.json file and for all subsequent requests.

This allows you to not only have total control over the response for a given request but also makes your test code easier to reason about -- it is clear where the "special case" response is coming from.

Getting the hash for a given request

All responses from chameleon will have a chameleon-request-hash header set which is the hash used for that request. This header is present even if you did not set it on the incoming request.

Preseeding the cache

If you want to configure the cache at runtime without having to depend on an external service, you may preseed the cache via HTTP. This is particularly useful for mocking out services which don't yet exist.

To preseed a request, issue a JSON POST request to chameleon at the _seed endpoint with the following payload:

Request

Response

Repeated, duplicate requests to preseed the cache will be discarded and the cache unaffected.

Successful new preseed requests will return an HTTP 201 CREATED on success or HTTP 500 INTERNAL SERVER ERROR. Duplicate preseed requests will return an HTTP 200 OK on success or HTTP 500 INTERNAL SERVER ERROR on failure.

Here is an example of preseeding the cache with a JSON response for a GET request for /foobar.

import requests

preseed = json.dumps({
    'Request': {
        'Body': '',
        'URL': '/foobar',
        'Method': 'GET',
    },
    'Response': {
        'Body': '{"key": "value"}',
        'Headers': {
            'Content-Type': 'application/json',
            'Other-Header': 'something-else',
        },
        'StatusCode': 200,
    },
})

response = requests.post('http://localhost:6005/_seed', data=preseed)
if response.status_code in (200, 201):
    # Created, or duplicate
else:
    # Error, print it out
    print(response.content)

# Continue tests as normal
# Making requests to `/foobar` will return `{"key": "value"}`
# without hitting the proxied service

Check out the example directory to see preseeding in action.

How chameleon caches responses

chameleon makes a hash for a given request URI, request method and request body and uses that to cache content. What that means:

• a request of GET /foo/ will be cached differently than GET /bar/
• a request of GET /foo/5 will be cached differently than GET /foo/6
• a request of DELETE /foo/5 will be cached differently than DELETE /foo/6
• a request of POST /foo with a body of {"hi":"hello} will be cached differently than a request of POST /foo with a body of {"spam":"eggs"}. To ignore the request body, set a header of chameleon-no-hash-body to any value. This will instruct chameleon to ignore the body as part of the hash.

Writing custom hasher

You can specify a custom hasher, which could be any program in any language, to determine what makes a request unique.

chameleon will communicate with this program via STDIN/STDOUT and feed the hasher a serialized Request (see below). You are then responsible for returning data to chameleon to be used for that given request (which will be hashed).

This feature is especially useful if you have to cache content based on the body of a request (XML payload, specific keys in JSON payload, etc).

See the example hasher for a sample hasher that emulates the default hasher.

Structure of Request

Below is an example Request serialized to JSON.

{
    "BodyBase64":"eyJmb28iOiAiYmFyIn0=",
    "ContentLength":14,
    "Headers":{
        "Accept":[
            "application/json"
        ],
        "Accept-Encoding":[
            "gzip, deflate"
        ],
        "Authorization":[
            "Basic dXNlcjpwYXNzd29yZA=="
        ],
        "Connection":[
            "keep-alive"
        ],
        "Content-Length":[
            "14"
        ],
        "Content-Type":[
            "application/json; charset=utf-8"
        ],
        "User-Agent":[
            "HTTPie/0.7.2"
        ]
    },
    "Method":"POST",
    "URL":{
        "Host":"httpbin.org",
        "Path":"/post",
        "RawQuery":"q=search+term%23home",
        "Scheme":"https"
    }
}

Getting help

Please open an issue for any bugs encountered, features requests, or general troubleshooting.

Authors

Nick Presta (@NickPresta)

Thanks to @mdibernardo for the inspiration.

License

Please see LICENSE