View the docs on Docs, and join the
chat at
Working with APIs, more often than not, during development you want to work with a fixed version of the responses provided. This is especially true if the API is still under development, and maybe even still needs input on how to output something. This is what Canned is for!
Canned maps a folder structure to API responses. Given the following directory structure:
/content/index.get.html
/comment/any.get.json
/comment/1/votes/index.get.json
/comment/any/votes/index.get.json
requests like
Accept: application/json
GET /comment/:id
are served from the file /comment/any.get.json as
Content-Type: application/json
{ "content": "I am a comment", "author": "sideshowcoder" }
requests like
Accept: text/html
GET /content/
are served from the file /content/index.get.html as
Content-Type: text/html
<html>
<body>Some html in here</body>
</html>
requests like
Accept: application/json
GET /comment/1/votes
are served from the file /comment/1/index.get.json as
Content-Type: application/json
{ "content": "I am comment 1", "author": "sideshowcoder" }
requests like
Accept: application/json
GET /comment/123456789/votes
are served from the file /comment/any/index.get.json
Content-Type: application/json
{ "content": "I am a wildcard comment for any id", "author": "sideshowcoder" }
The matching works on the filename by treating it as PATH.VERB.CONTENT_TYPE so
index.get.json has the path index the verb is get and the content-type
json. Supported content types are
json => application/json
html => text/html
txt => text/plain
js => application/javascript
So an example is for querying (with canned running on localhost:3000)
$ curl -H "Accept: text/javascript" http://localhost:3000/comment/1
> { "content": "I am a comment", "author": "sideshowcoder" }
Currently Canned supports the basic REST-API mapping, as well as custom method mapping with nested endpoints.
file | resquest
/index.get.json | GET /
/any.get.json | GET /:id
/_search.get.json | GET /search
/comments/index.get.json | GET /comments/
/comments/any.get.json | GET /comments/:id
/comments/_search.get.json | GET /comments/search
/comments/any/index.get.json | GET /comments/:id/
You can even add query parameters to your filenames to return different responses on the same route. If the all query params in a filename match the incoming request, this file will be returned. It will fall back to returning the file with no query params if it exists.
Warning this will be deprecated in the future since canned now supports multiple response based on the request body or GET URL parameters in one file. This is the preferred way since files with ? in the name do not work on Windows
file | resquest
/index?name=Superman.get.json | GET /?name=Superman&NotAllParams=NeedToMatch
/_search?q=hello.get.json | GET /comments/search?q=hello
/_search.get.json | GET /comments/search?iam=soignored
Same support is available for PUT, POST, etc.
/index.post.json | POST serves /... + CORS Headers
/index.put.json | PUT serves /... + CORS Headers
If CORS support is enabled additionally options will be available as a http verb and all requests will serve the CORS Headers as well
/ | OPTIONS serve all the options needed for CORS
/index.get.json | GET serves /... + CORS Headers
If you need some custum return codes, just add them to the file via adding a file header like so
//! statusCode: 201
<html>
<body>Created something successfully! Happy!</body>
</html>
The header will be stripped before sending and the statusCode will be set.
You can also override the default content types by adding a custom content type to the file header:
//! contentType: "application/vnd.custom+xml"
<xml>
<created>1</created>
</xml>
This will be returned with a Content-type: application/vnd.custom+xml header.
Multiple headers need to be written on one single line and comma-separated, like so:
//! statusCode: 201, contentType: "application/vnd.custom+xml"
If you need to send bind custom HTTP headers to the response you can add them as headers to the response file using
the keyword customHeader:
//! customHeader: {"MyCustomHeaderName": "MyCustomHeaderValue"}
In case you need more then one custom header in the response, you can just use the same keyword multiple times:
//! customHeader: {"MyCustomHeaderName": "MyCustomHeaderValue"}
//! customHeader: {"SecondHeaderName": "SecondHeaderValue"}
You can get a different response by specifying request data in variant comments. If the request data matches the comment data the matching response is returned. If there is no match the first response is returned
Note: comments must be on a single line
Custom headers:
//! header: {"authorization": "abc"}
{
"response": "response for abc"
}
//! header: {"authorization": "123"}
{
"response": "response for 123"
}
If you need different responses based on request body then you can specify the request you want matched via body comments:
//! body: {"email": "[email protected]"}
{
"response": "response for [email protected]"
}
//! body: {"email": "[email protected]"}
{
"response": "response for [email protected]"
}
If you need different responses based on request parameters then you can specify them via parameters comments:
//! params: {"foo": "bar"}
{
"response": "response for bar"
}
//! params: {"foo": "baz"}
{
"response": "response for baz"
}
this would match http://my.local.server/my_get_request_path?foo=bar or
http://my.local.server/my_get_request_path?foo=baz respectively.
To use in conjunction with response headers and status codes, just add them on the line above.
//! statusCode: 201
//! header: {"authorization": "abc"}
{
"response": "response for abc"
}
//! statusCode: 201, contentType: "application/my-personal-json"
//! header: {"authorization": "123"}
{
"response": "response for 123"
}
Wildcard responses are also supported, very useful to have 'wildcard' directories, so that if for given a request like:
GET /api/users/1/profile/
you don't have a file in ./canned/api/users/1/profile/index.get.json then
it would look for a file in ./canned/api/users/any/index.get.json or
similar. Wildcards can be specified on the command line via
canned --wildcard myany
This would change the lookup to ./canned/api/users/myany/index.get.json
Most content types support comments natively, like html or javascript. Sadly the probably most used type (JSON) does not :(. So canned actually extends the JSON syntax a little so it can include comments with // or /**/. In case you use the JSON files directly on the backend side as test cases make sure you strip those out as well!
Just install via npm
$ npm install canned
which will install it locally in node_modules, if you want to have it available from anywhere just install globally
$ npm install -g canned
There are 2 ways here, either you embed it somewhere programmatically
var canned = require('canned')
, http = require('http')
, opts = { logger: process.stdout }
can = canned('/path/to/canned/response/folder', opts)
http.createServer(can).listen(3000)
Or just run the provided canned server script
$ canned
Which serves the current folder with canned responses on port 3000
$ canned -p 5000 ./my/responses/
will serve the relative folder via port 5000
If you need canned to respond with some delay, pass delay in ms to response_delay arg
$ canned --response_delay=1000 ./my/reponses/
If you want canned to iterate through all accepted content types in the Accept header, use
$ canned --relaxed_accept=true ./my/reponses/
If for whatever reason you want to turn of CORS support do so via
$ canned --cors=false ./my/responses/
Also if you need additional headers to be served alongside the CORS headers these can be added like this (thanks to runemadsen)
$ canned --headers "Authorization, Another-Header"
To enable CORS programmatically, you can use the following options:
var canned = require('canned')
, http = require('http')
, opts = {
cors: true,
cors_headers: ["Content-Type", "Location"]
}
Optionally, the cors_headers value can be a comma-separated string, as per the CLI option.
Other optional options include:
var opts = {
sanitize: false, // get responses as is without any sanitization
response_delay: 2000, // delay the response for 2 seconds
relaxed_accept: true // iterate through all accepted content types in the `Accept` header
wildcard: 'myany', // specify 'wildcard' directory, e.g. ./canned/api/users/myany/index.get.json
}
For more information checkout the pull request
Already using grunt? Great there is a plugin for that, thanks to jkjustjoshing.
make sure you either install globally or put ./node_modules/.bin in your PATH
make sure /usr/local/share/npm/bin is in your path, this should be true for every install since you won't be able to run any global module bins if not. (like express, and such)
make sure you run a version of node which is 0.10.3 or higher, because it fixes a problem for the encoding handling when reading files
Set the "Content-Type" header to contain "application/json".
- Checkout the repository
- Run the tests and jshint
$ make - Create a topic branch
$ git checkout -b my-new-feature - Code test and make jshint happy!
$ make test$ make hint - Push the branch and create a Pull-Request
I try to review the pull requests as quickly as possible, should it take to long feel free to bug me on twitter
- adding PATCH to default Access-Control-Allow-Method Cors header #113 (@william-mcmillian)
- adding support for delayed responses #114 (@Onatolich)
- adding support to make sanatize optional #115 (@YuliyaMarholina)
- Windows line ending support #102 (@antxxxx)
- cleanup and documentation #95 (@wadtech)
- customHeader handling #110 (@mazoni)
- relaxed handling for accept headers, meaning select the first result that can be
served even if it is not the first accepted content type. option
--relaxed-accept#100 (@CheungJ)
- fix improper handling of carriage return in windows #79 (@git-jiby-me)
- fix handling for urls in request body #90 (@wadtech)
- documentation and test cases for cors headers #91 (@wadtech)
- enable matching raw request body rather than property-based #96 (@ftes)
- fix query string param handling #97 (@wadtech @targoo)
- The regex for matching request, was not considering arrays in the request JSON #82
- For request with a request body, canned was checking content type to exactly match application/json, which is not good as browsers may sent charset as well with the content type. #82
- For matching request and filters with more accuracy, we were converting the values of all keys in request to string before comparing, but this was being done wrong as it was creating string of Objects and arrays as well, which it shouldn’t #82
- support checking the
ACCEPT HEADERfor the response type (thanks git-jiby-me) #81
- support for custom HTTP headers in responses
- fix for matching multiple parameters in response #73 thanks xdemocle
- fix any wildcard in the middle of the path #66 thanks msurdi
- update dependencies and dev-dependencies
- wildcard parameters thanks to msurdi see sideshowcoder#64
- fix support for special characters in the header / params / body matches (@simonprickett, @kevinschumacher, @sideshowcoder)
- support differet statusCodes and content types in multiple response files (@sideshowcoder)
- support for XML headers to support SOAP (@vikalp)
- fix relative path again... (@sideshowcoder)
- fixes for variable responses with JSON body (@bibounde)
- fixes for relative paths on start (@sideshowcoder)
- complex get parameters causing regexp match on file to fail (@sideshowcoder)
- support for multiple responses per file (@hungrydavid)
- support for GET responses without the need for special characters in the filename (@sideshowcoder based on the work by @hungrydavid)
- added support for empty response with 204 for no content (@jkjustjoshing)
- sorry haven't kept a version history, yet. Will now!
- sideshowcoder
- leifg
- runemadsen
- mulderp
- creynders
- jkjustjoshing
- hungrydavid
- bibounde
- vikalp
- simonprickett
- kevinschumacher
- msurdi
- Brendan Rius
- Rocco Russo
- git-jiby-me
- wadtech
- ftes
- targoo
- CheungJ
- antxxxx
- mazoni
- william-mcmillian
- Onatolich
MIT 2013 Philipp Fehre alias @sideshowcoder, or @ischi on twitter