hitit is a minimalistic tool for testing an HTTP(S) API. It is a stopgap until I publish the next version of kaboot.
The current version of hitit, v1.2.8, is considered to be mostly stable and mostly complete. Suggestions and patches are welcome. Future changes planned are:
- Support for concurrent testing (a.k.a stress testing).
The dependencies of hitit are three:
To install, type npm i hitit.
To use hitit, you need node.js v0.8.0 or newer.
To do a single request, use h.one. This function takes three arguments: state, options and callback.
options must be an object. These are the options for the request; any of them can be undefined.
tag: an optional string that will be printed to the console when the request is started.host: optional string.port: optional integer.method: optional string. If defined, must be a valid HTTP method.path: optional string.headers: optional object.body: can be of any type. See below for more details.code: any valid HTTP status code; defaults to 200. If the response has a different matching code, the request will be considered as a failure. If you want the request to succeed in any case, you can use'*'as the status code.apres: a function that is executed after the request finishes. See below for more details.delay: optional integer that determines how many milliseconds should be waited until the next request.timeout: optional integer that will abort the request afterbody.timeoutseconds elapse of socket inactivity. Defaults to60.https: optional boolean. If true,httpswill be used instead ofhttp.rejectUnauthorized: optional boolean. Iffalse, insecurehttpswill be accepted by default (this is useful when you're testing with a self-signed certificate).raw: optional boolean. If true, the response's body will be returned as a raw buffer.
state must also be an object. It serves two purposes: #1 keep state between requests; and #2 have default values for some request parameters. Regarding keeping state between requests, you can assign any key in this object for your own purposes, as long as is none of the keys that options can have. If you assign a key that is one of the options keys (for example, host), if options.host is undefined, state.host will be considered as the host. This is what enables #2.
Many times it is useful to make a request depending on state. For that reason, any of the keys of options can also be a function; if so, it will be evaluated passing state as its only argument. For example, if state.id is 3 and you define options.path to be function (state) {return 'download/' + state.id}, this is equivalent to setting options.path to 'download/3'. Note that also the matching keys of state are evaluated in this way if they are functions.
The body can be of any type. If it's null or undefined, it will be considered an empty string. If it's neither an array or an object, it will be coerced a string. If it is either an array or an object with body.multipart being undefined, it will be considered a JSON. In this case, it will be automatically stringified and the content-header will be set to application/json (unless you override this default). Finally, if body is an object and body.multipart is defined, hitit will do a multipart/form-data request. body.multipart can be either an object or an array of objects. Each of these objects can represent either a field or a file. In the case of a field, the object will have three keys: type: 'field', name: STRING and value: STRING. In the case of a file, the object will have these fields: type: 'file', name: STRING, either value or path (the first to provide the literal value of the file, the second a path to where the file is) and an optional contentType - in its absence, if path is provided, a mime lookup of the file will be performed.
// Example
body.multipart = {
{type: 'field', name: 'field1', value: 'somedata'},
{type: 'file', name: 'file1', path: 'test/image.jpg', contentType: 'application/octet-stream'},
}The apres is an optional function that is executed after the request, but only if the response's status code matches the expected code. It receives four arguments: state, options, rdata and callback. The only one that needs explanation is rdata: it consists of an object with five keys: code, which is the status code of the response; headers, which contain the headers of the response; body, which contains the body of the response (parsed to a string or to an array/object, in case the content-type header of the response is application/json); time, an array with the time when the requested started and the time when the request ended. And finally, request, which is equal to options.
The apres function can halt or suspend execution depending on its return value. if it returns false, this is considered to be an error and callback will be called with an error. If it returns true, execution will continue. If it returns undefined, execution will be suspended. This is useful for asynchronous operations; if you wish to resume execution, you can call callback with a falsy first argument, indicating the absence of an error.
If you're calling h.one directly, the concept of sequence is irrelevant. However, h.seq invokes h.one, so your tests can exert control flow from inside the apres.
callback is the callback function that is called at the end of the request. It receives two arguments, error and rdata. If there was an error, error will have an error code (-2 if the arguments are invalid, -1 if there was an error during the start of the request, and 0 if the request started but the response server became unresponsive, or if there was a timeout). error.request will contain the request parameters. Also, if the code didn't match the response's status code, rdata will be passed as error. In the absence of error, rdata is received as a second argument.
This function accepts a sequence of requests and executes them in turn. It takes four arguments:
state, an object (the samestatethat will be passed toh.one).sequence, an array with requests.callback, a callback function.map, an optional function that transforms each of the elements inrequests.
callback will always receive error as its first argument and an array of rdatas as its second. If the sequence was completed successfully, error will be undefined.
In the simplest case, sequence can be an array with a number of objects, each of them a valid options object that can be passed to h.one. However, you can have nested arrays with options inside; h.seq will unnest the array for you. This is useful when you have functions returning tests.
If any of the components of sequence has a falsy value or is an empty array, it will be ignored. This is useful for creating conditional tests. However, these conditions cannot depend on the dynamic state of the test; rather, the conditional must depend on a condition that is defined when h.seq is invoked.
Finally, you can use a function for converting each of the elements within sequence to a valid h.one options object. For example, the default h.stdmap function converts the following array:
['a tag', 'a method', 'a path', 'headers', 'a body', 'a code', 'an apres', 'a delay']into
{tag: 'a tag', method: 'a method', path: 'a path', headers: 'headers', body: 'a body', code: 'a code', apres: 'an apres', delay: 'a delay'}
In this case, notice that host and port are not defined and must hence be passed through state.
The complete source code is contained in hitit.js. It is about 240 lines long.
hitit is written by Federico Pereiro (fpereiro@gmail.com) and released into the public domain.