Apitest
Apitest is declarative api testing tool with JSON-like DSL.
Read this in other languages: 中文
Installation
Binaries are available in Github Releases. Make sure to put the path to the binary into your PATH.
# linux
curl -L -o apitest https://github.com/sigoden/apitest/releases/latest/download/apitest-linux
chmod +x apitest
sudo mv apitest /usr/local/bin/
# macos
curl -L -o apitest https://github.com/sigoden/apitest/releases/latest/download/apitest-macos
chmod +x apitest
sudo mv apitest /usr/local/bin/
# npm
npm install -g @sigodenjs/apitest
Get Started
Write test file httpbin.jsona
{
test1: {
req: {
url: "https://httpbin.org/post",
method: "post",
headers: {
'content-type':'application/json',
},
body: {
v1: "bar1",
v2: "Bar2",
},
},
res: {
status: 200,
body: { @partial
json: {
v1: "bar1",
v2: "bar2"
}
}
}
}
}
Run test
apitest httpbin.jsona
main
test1 (0.944) ✘
main.test1.res.body.json.v2: bar2 ≠ Bar2
...
The use case test failed. From the error message printed by Apitest, you can see that the actual value of main.test1.res.body.json.v2 is Bar2 instead of bar2.
After we modify bar2 to Bar2, execute Apitest again
apitest httpbin.jsona
main
test1 (0.930) ✔
Features
JSONA DSL
Use JSON-like DSL to write tests. The document is the test.
{
test1: { @describe("user login")
req: {
url: 'http://localhost:3000/login'
method: 'post',
body: {
user: 'jason',
pass: 'a123456,
}
},
res: {
status: 200
body: {
user: 'jason',
token: '', @type
expireIn: 0, @type
}
}
}
}
According to the above use case, I don't need to elaborate, an experienced backend should be able to guess what parameters are passed by this api and what data is returned by the server.
The working principle of Apitest is to construct the request according to the description in the req part and send it to the backend. After receiving the response data from the backend, verify the data according to the description in the res part.
Please don't be scared by DSL. In fact, it is JSON, which loosens some grammatical restrictions (double quotes are not mandatory, comments are supported, etc.), and only one feature is added: comments. In the above example, @describe, @type is Annotation.
Click jsona/spec to view the JSONA specification
By the way, there is a vscode extension supports DSL (jsona) format.
Why use JSONA?
The essence of api testing is to construct and send req data, and receive and verify res data. Data is both the main body and the core, and JSON is the most readable and universal data description format.
Api testing also requires some specific logic. For example, a random number is constructed in the request, and only part of the data given in the response is checked.
JSONA = JSON + Annotation. JSON is responsible for the data part, and annotations are responsible for the logic part. Perfectly fit the interface test requirements.
Data Is Assertion
How to understand? See below.
{
"foo1": 3,
"foo2": ["a", "b"],
"foo3": {
"a": 3,
"b": 4
}
}Assuming that the response data is as above, the test case is as follows:
{
test1: {
req: {
},
res: {
body: {
"foo1": 3,
"foo2": ["a", "b"],
"foo3": {
"a": 3,
"b": 4
}
}
}
}
}
That's right, it's exactly the same. Apitest will compare each part of the data one by one. Any inconsistency will cause the test to fail.
The strategy provided by conventional testing tools is addition. This is very important and I just add an assertion. In Apitest, you can only do subtraction. This data is not concerned. I actively ignore or relax the verification.
For example, the previous test case
{
test1: { @describe("user login")
...
res: {
body: {
user: 'jason',
token: '', @type
expireIn: 0, @type
}
}
}
}
We still checked all the fields. Because the values of token and expireIn are changed, we use @type to tell Apitest to only check the type of the field and ignore the specific value.
Data Is Accessable
Any data of the test case can be testd by subsequent test cases
The following test cases can use all the data of the previous test cases.
{
test1: { @describe("user login")
...
res: {
body: {
token: '', @type
}
}
},
test2: { @describe("create article")
req: {
headers: {
// We access the response data of the previous test case test1.
authorization: `"Bearer " + test1.res.body.token`, @eval
},
}
},
}
Support Mock
With Mock, no longer entangled in fabricating data, Seee @mock
Support Mixin
Use Mixin skillfully, get rid of copy and paste. See @mixin
CI Support
As a command line tool itself, it is very easy to integrate with the back-end ci. And apitest also provides the --ci option to optimize ci.
TDD Support
You can even write only the req part, and after the api has a response, paste the response data directly as the res part. Talk of experience 🐶
In the default mode (not ci), when Apitest encounters a failed test, it will print an error and exit. Apitest has cached test data. You can repeatedly execute wrong test cases, develop and test at the same time, and then enter the follow-up test until you get through.
At the same time, you can also select a test case to execute through the --only option.
User-defiend Functions
You don't need to use this function at all. But I still worry about the need in certain extreme or corner scenes, so I still support it.
Apitest allows users to write custom functions through js to construct request data or verify response data. (Dare to call it a cross-programming language? 🐶), See @jslib
Skip, Delay, Retry & Loop
See #Run
Form, File Upload, GraphQL
See #Http
Annotation
Apitest uses JSONA format to describe test cases.
JSON describes data and annotation describes logic.
@module
Import submodule
scope: entrypoint file
// main.jsona
{
@module("mod1")
}
// mod1.jsona
{
test1: {
req: {
}
}
}
@jslib
Import user-defined functions
scope: entrypoint file
Write functions lib.js
// Make random color e.g. #34FFFF
exports.makeColor = function () {
const letters = '0123456789ABCDEF';
let color = '#';
for (let i = 0; i < 6; i++) {
color += letters[Math.floor(Math.random() * 16)];
}
return color;
}
// Detect date in ISO8601(e.g. 2021-06-02:00:00.000Z) format
exports.isDate = function (date) {
return /^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d{3}Z$/.test(date)
}Use functions
@jslib("lib") // Import js files
{
test1: {
req: {
body: {
// call the `makeColor` function to generate random colors
color:'makeColor()', @eval
}
},
res: {
body: {
// $ indicates the field to be verified, here is `res.body.createdAt`
createdAt:'isDate($)', @eval
// Of course you can use regex directly
updatedAt: `/^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}\.\d{3}Z$/ .test($)`, @eval
}
}
}
}
@mixin
Import mixin file
scope: entrypoint file, group/unit head
First create a file to store the file defined by Mixin
{
createPost: { // Extract routing information to mixin
req: {
url: '/posts',
method: 'post',
},
},
auth1: { // Extract authorization
req: {
headers: {
authorization: `"Bearer " + test1.res.body.token`, @eval
}
}
}
}
@mixin("mixin") // include mixin.jsona
{
createPost1: { @describe("create article 1") @mixin(["createPost", "auth1"])
req: {
body: {
title: "sentence", @mock
}
}
},
createPost2: { @describe("create article 2，with description") @mixin(["createPost", "auth1"])
req: {
body: {
title: "sentence", @mock
description: "paragraph", @mock
}
}
},
}
The more frequently used part, the more suitable it is to be extracted to Mixin.
@client
Setup clients
scope: entrypoint file, group/unit head
Client is responsible for constructing a request according to req, sending it to the server, receiving the response from the server, and constructing res response data.
{
@client({
name: "apiv1",
kind: "http",
options: {
baseURL: "http://localhost:3000/api/v1",
timeout: 30000,
}
})
@client({
name: "apiv2",
kind: "http",
options: {
baseURL: "http://localhost:3000/api/v2",
timeout: 30000,
}
})
test1: { @client("apiv1")
req: {
url: "/posts" // 使用apiv1客户端，所以请求路径是 http://localhost:3000/api/v1/posts
}
},
test2: { @client({name:"apiv2",options:{timeout:30000}})
req: {
url: "/key" // 使用apiv2客户端，所以请求路径是 http://localhost:3000/api/v2/posts
}
},
}
@describe
Give a title
scope: module file, group/unit head
{
@client({name:"default",kind:"echo"})
@describe("This is a module")
group1: { @group @describe("This is a group")
test1: { @describe("A unit in group")
req: {
}
},
group2: { @group @describe("This is a nested group")
test1: { @describe("A unit in nested group")
req: {
}
}
}
}
}
It will be printed as follows
This is a module
This is a group
A unit in group ✔
This is a nested group
A unit in nested group ✔
If the @description is removed, it will be printed as follows
main
group1
test1 ✔
group2
test1 ✔
@group
Mark as case group
scope: group head
The test cases in the group will inherit the group's @client and @mixin. The group also supports Run.
{
group1: { @group @mixin("auth1") @client("apiv1")
test1: {
},
// The mixin of the use case and the mixin of the group will be merged into @mixin(["route1","auth1"])
test2: { @mixin("route1")
},
// The client of the use case will overwrite the client of the group
test3: { @client("echo")
},
group2: { @group // nest group
},
run: {
}
}
}
@eval
Use js expr to generate data (in req) and verify data(in res)
scope: unit block
@eval features:
- can use js builtin functions
- can use jslib functions
- can access environment variables
- can use the data from the previous test
{
test1: { @client("echo")
req: {
v1: "JSON.stringify({a:3,b:4})", @eval // Use JS built-in functions
v2: `
let x = 3;
let y = 4;
x + y
`, @eval // Support code block
v3: "env.FOO", @eval // Access environment variables
v4: 'mod1.test1.res.body.id`, @eval // Access the data of the previous test
}
}
}
@eval in res part with additional features:
$repersent response data in the position- return value true means that the verification passed
- if the return value is not of type bool, the return value and the response data will be checked for congruent matching
{
rest2: {
res: {
v1: "JSON.parse($).a === 3", @eval // $ is `res.v1`
v2: "true", @eval // true force test passed
v4: 'mod1.test1.res.body.id`, @eval // return value congruent matching
}
}
}
@eval accessing use case data with elision
{
test1: {
req: {
v1: 3,
},
res: {
v1: "main.test1.req.v1", @eval
// v1: "test1.req.v1", @eval
// v1: "req.v1", @eval
}
}
}
@mock
Use mock function to generate data
scope: unit req block
Apitest supports nearly 40 mock functions. For a detailed list, see fake-js
{
test1: {
req: {
email: 'email', @mock
username: 'username', @mock
integer: 'integer(-5, 5)', @mock
image: 'image("200x100")', @mock
string: 'string("alpha", 5)', @mock
date: 'date', @mock // iso8601 format // 2021-06-03T07:35:55Z
date1: 'date("yyyy-mm-dd HH:MM:ss")' @mock // 2021-06-03 15:35:55
date2: 'date("unix")', @mock // unix epoch 1622705755
date3: 'date("","3 hours 15 minutes")', @mock
date4: 'date("","2 weeks ago")', @mock
ipv6: 'ipv6', @mock
sentence: 'sentence', @mock
cnsentence: 'cnsentence', @mock
}
}
}
@file
Use file
scope: unit req block
{
test1: {
req: {
headers: {
'content-type': 'multipart/form-data',
},
body: {
field: 'my value',
file: 'bar.jpg', @file // upload file `bar.jpg`
}
},
}
}
@trans
Transform data
scope: unit block
{
test1: { @client("echo")
req: {
v1: { @trans(`JSON.stringify($)`)
v1: 1,
v2: 2,
}
},
res: {
v1: `{"v1":1,"v2":2}`,
}
},
test2: { @client("echo")
req: {
v1: `{"v1":1,"v2":2}`,
},
res: {
v2: { @trans(`JSON.parse($)`)
v1: 1,
v2: 2,
}
}
}
}
@every
A set of assertions are passed before the test passes
scope: unit res block
{
test1: { @client("echo")
req: {
v1: "integer(1, 10)", @mock
},
res: {
v1: [ @every
"$ > -1", @eval
"$ > 0", @eval
]
}
}
}
@some
If one of a set of assertions passes, the test passes
scope: unit res block
{
test1: { @client("echo")
req: {
v1: "integer(1, 10)", @mock
},
res: {
v1: [ @some
"$ > -1", @eval
"$ > 10", @eval
]
}
}
}
@partial
Mark only partial verification instead of congruent verification
scope: unit res block
{
test1: { @client("echo")
req: {
v1: 2,
v2: "a",
},
res: { @partial
v1: 2,
}
},
test2: { @client("echo")
req: {
v1: [
1,
2
]
},
res: {
v1: [ @partial
1
]
}
}
}
@type
Mark only verifies the type of data
scope: unit res block
{
test1: { @client("echo")
req: {
v1: null,
v2: true,
v3: "abc",
v4: 12,
v5: 12.3,
v6: [1, 2],
v7: {a:3,b:4},
},
res: {
v1: null, @type
v2: false, @type
v3: "", @type
v4: 0, @type
v5: 0.0, @type
v6: [], @type
v7: {}, @type
}
},
}
@optional
Marker field is optional
scope: unit res block
{
test1: { @client("echo")
req: {
v1: 3,
// v2: 4, optional field
},
res: {
v1: 3,
v2: 4, @optional
}
}
}
@nullable
Marker field can be null
scope: unit res block
{
test1: { @client("echo")
req: {
v1: null,
// v1: 3,
},
res: {
v1: 3, @nullable
}
}
}
Run
In some scenarios, use cases may not need to be executed, or they may need to be executed repeatedly. It is necessary to add a run option to support this feature.
Skip
{
test1: { @client("echo")
req: {
},
run: {
skip: `mod1.test1.res.status === 200`, @eval
}
}
}
run.skipskip the test when true
Delay
Run the test case after waiting for a period of time
{
test1: { @client("echo")
req: {
},
run: {
delay: 1000,
}
}
}
run.delaydelay in ms
Retry
{
test1: { @client("echo")
req: {
},
run: {
retry: {
stop:'$run.count > 2', @eval
delay: 1000,
}
},
}
}
variables:
$run.countrecords the number of retries.
options:
run.retry.stopwhether to stop retryrun.retry.delayinterval between each retry (ms)
Loop
{
test1: { @client("echo")
req: {
v1:'$run.index', @eval
v2:'$run.item', @eval
},
run: {
loop: {
delay: 1000,
items: [
'a',
'b',
'c',
]
}
},
}
}
variables:
$run.itemcurrent loop data$run.indexcurrent loop data index
options:
run.loop.itemsiter pass to$run.itemrun.loop.delayinterval between each cycle (ms)
Dump
{
test1: { @client("echo")
req: {
},
run: {
dump: true,
}
}
}
run.dumpforce print req/res data when true
Client
The req and res data structure of the test case is defined by the client
The client is responsible for constructing a request according to req, sending it to the server, receiving the response from the server, and constructing res response data.
If the test case does not use the @client annotation to specify a client, the client is the default.
If there is no default client defined in the entry file. Apitest will automatically insert @client({name:"default",kind:"http"}) with http as the default client
Apitest provides two kinds of clients.
Echo
The echo client does not send any request, and directly returns the data in the req part as the res data.
{
test1: {@client("echo")
req: {// Just fill in any data
},
res: {// equal to req
}
}
}
Http
http client handles http/https requests/responses.
{
test1: { @client({options:{timeout: 10000}}) // Custom client parameters
req: {
url: "https://httpbin.org/anything/{id}", // request url
// http methods, `post`, `get`, `delete`, `put`, `patch`
method: "post",
query: { // ?foo=v1&bar=v2
foo: "v1",
bar: "v2",
},
// url path params, `/anything/{id}` => `/anything/33`
params: {
id: 33,
},
headers: {
'x-key': 'v1'
},
body: { // request body
}
},
res: {
status: 200,
headers: {
'x-key': 'v1'
},
body: { // response body
}
}
}
}
Options
{
// `baseURL` will be prepended to `url` unless `url` is absolute.
baseURL: '',
// `timeout` specifies the number of milliseconds before the request times out.
// If the request takes longer than `timeout`, the request will be aborted.
timeout: 0,
// `maxRedirects` defines the maximum number of redirects to follow in node.js.
// If set to 0, no redirects will be followed.
maxRedirects: 0,
// `headers` is default request headers
headers: {
},
// `proxy` configures http(s) proxy, you can also use HTTP_PROXY, HTTPS_PROXY
// environment variables
proxy: "http://user:pass@localhost:8080"
}Cookies
{
test1: {
req: {
url: "https://httpbin.org/cookies/set",
query: {
k1: "v1",
k2: "v2",
},
},
res: {
status: 302,
headers: { @partial
'set-cookie': [], @type
},
body: "", @type
}
},
test2: {
req: {
url: "https://httpbin.org/cookies",
headers: {
Cookie: `test1.res.headers["set-cookie"]`, @eval
}
},
res: {
body: { @partial
cookies: {
k1: "v1",
k2: "v2",
}
}
},
},
}x-www-form-urlencoded
Add the request header "content-type": "application/x-www-form-urlencoded"
{
test2: { @describe('test form')
req: {
url: "https://httpbin.org/post",
method: "post",
headers: {
'content-type':"application/x-www-form-urlencoded"
},
body: {
v1: "bar1",
v2: "Bar2",
}
},
res: {
status: 200,
body: { @partial
form: {
v1: "bar1",
v2: "Bar2",
}
}
}
},
}
multipart/form-data
Add the request header "content-type": "multipart/form-data"
Combined with @file annotation to implement file upload
{
test3: { @describe('test multi-part')
req: {
url: "https://httpbin.org/post",
method: "post",
headers: {
'content-type': "multipart/form-data",
},
body: {
v1: "bar1",
v2: "httpbin.jsona", @file
}
},
res: {
status: 200,
body: { @partial
form: {
v1: "bar1",
v2: "", @type
}
}
}
}
}
graphql
{
vars: { @describe("share variables") @client("echo")
req: {
v1: 10,
}
},
test1: { @describe("test graphql")
req: {
url: "https://api.spacex.land/graphql/",
body: {
query: `\`query {
launchesPast(limit: ${vars.req.v1}) {
mission_name
launch_date_local
launch_site {
site_name_long
}
}
}\`` @eval
}
},
res: {
body: {
data: {
launchesPast: [ @partial
{
"mission_name": "", @type
"launch_date_local": "", @type
"launch_site": {
"site_name_long": "", @type
}
}
]
}
}
}
}
}
Cli
usage: apitest [options] [target]
Options:
-h, --help Show help [boolean]
-V, --version Show version number [boolean]
--ci Whether to run in ci mode [boolean]
--reset Whether to continue with last case [boolean]
--dry-run Check syntax then print all cases [boolean]
--env Specific test enviroment like prod, dev [string]
--only Run specific module/case [string]
--dump Force print req/res data [boolean]
Multiple Test Environments
Apitest supports multiple test environments, which can be specified by the --env option.
// Pre-release environment main.jsona
{
@client({
options: {
url: "http://pre.example.com/api"
}
})
@module("mod1")
}
// Local environment main.local.jsona
{
@client({
options: {
url: "http://localhost:3000/api"
}
})
@module("mod1")
@module("mod2") // Only local test module
}
# By default, tests/main.local.jsona is selected
apitest tests
# Select tests/main.local.jsona
apitest tests --env localApitest allows to specify main.jsona
apitest tests/main.jsona
apitest tests/main.local.jsonaSpecify a specific main.jsona, you can still use the --env option
# Select tests/main.local.jsona
apitest tests/main.jsona --env localNormal Mode
- Start execution from the last failed test case, print error details and exit when encountering a failed test case
- If there is option
--reset, it will start from the beginning instead of where it failed last time - If there is the option
--only mod1.test1, only the selected test case will be executed
CI Mode
- Ignore the cache and execute the test case from scratch
- Continue to execute the failed test case
- After all test cases are executed, errors will be printed uniformly