An koa.js middleware for node-validator.
npm install koa-async-validator --save
- If you want to use checkParams you have to user koa-router or any router that populates ctx.params.
- This middleware is for koa 2.
import util from 'util';
import Koa from 'koa';
import koaValidator from 'koa-async-validator';
import bodyParser from 'koa-bodyparser';
import Router from 'koa-router';
const app = new Koa();
const router = new Router();
app.use(bodyParser());
app.use(koaValidator([options])); // this line must be immediately after bodyParser()!
router.post('/:urlparam', async (ctx, next) => {
// VALIDATION
// checkBody only checks ctx.request.body; none of the other req parameters
// Similarly checkParams only checks in ctx.params (URL params) and
// checkQuery only checks ctx.query (GET params).
ctx.checkBody('postparam', 'Invalid postparam').notEmpty().isInt();
ctx.checkParams('urlparam', 'Invalid urlparam').isAlpha();
ctx.checkQuery('getparam', 'Invalid getparam').isInt();
// OR assert can be used to check on all 3 types of params.
// ctx.assert('postparam', 'Invalid postparam').notEmpty().isInt();
// ctx.assert('urlparam', 'Invalid urlparam').isAlpha();
// ctx.assert('getparam', 'Invalid getparam').isInt();
// SANITIZATION
// as with validation these will only validate the corresponding
// request object
ctx.sanitizeBody('postparam').toBoolean();
ctx.sanitizeParams('urlparam').toBoolean();
ctx.sanitizeQuery('getparam').toBoolean();
// OR find the relevent param in all areas
ctx.sanitize('postparam').toBoolean();
let errors = await ctx.validationErrors();
if (errors) {
ctx.body = `There have been validation errors: ${ util.inspect(errors) }`;
ctx.status = 400;
} else {
ctx.body = {
urlparam: ctx.params.urlparam,
getparam: ctx.params.getparam,
postparam: ctx.params.postparam
}
}
await next();
});
app
.use(router.routes())
.use(router.allowedMethods());
app.listen(8888);
Which will result in:
Needs to be updated
$ curl -d 'postparam=1' http://localhost:8888/test?getparam=1
{"urlparam":"test","getparam":"1","postparam":true}
$ curl -d 'postparam=1' http://localhost:8888/t1est?getparam=1
There have been validation errors: [
{ param: 'urlparam', msg: 'Invalid urlparam', value: 't1est' } ]
$ curl -d 'postparam=1' http://localhost:8888/t1est?getparam=1ab
There have been validation errors: [
{ param: 'getparam', msg: 'Invalid getparam', value: '1ab' },
{ param: 'urlparam', msg: 'Invalid urlparam', value: 't1est' } ]
$ curl http://localhost:8888/test?getparam=1&postparam=1
There have been validation errors: [
{ param: 'postparam', msg: 'Invalid postparam', value: undefined} ]
####skipValidationOnFirstError
{ skipValidationOnFirstError: boolean = false, ... }
The skipValidationOnFirstError
option is set default to false
that works as before, runs all validations and save the errors.
If set to true
will skip next validations when the first sync validation fails. All async validation do not follow this rule. Each validation runs in order that was defiened.
####errorFormatter
{ errorFormatter: (param,msg,value) => string, ... }
The errorFormatter
option can be used to specify a function that can be used to format the objects that populate the error array that is returned in ctx.validationErrors()
. It should return an Object
that has param
, msg
, and value
keys defined.
// In this example, the formParam value is going to get morphed into form body format useful for printing.
app.use(koaValidator({
errorFormatter: (param, msg, value) => {
const namespace = param.split('.');
const root = namespace.shift();
let formParam = root;
while(namespace.length) {
formParam = `${formParam}[${namespace.shift}]`;
}
return {
param : formParam,
msg : msg,
value : value
};
}
}));
####customValidators
{ validatorName: (value, [additional arguments], ctx) => boolean | (value, [additional arguments], ctx) => Promise, ... }
The customValidators
option can be used to add additional validation methods as needed. This option should be an Object
defining the validator names and associated validation functions.
Param value will be sent as is, won't be stringfied.
For customValidators the last argument it is always is koa ctx apart from options.
Define your custom validators:
app.use(koaValidator({
customValidators: {
isArray: (value) => Array.isArray(value),
gte: (param, num) => param >= num,
}
}));
Use them with their validator name:
ctx.checkBody('users', 'Users must be an array').isArray();
ctx.checkQuery('time', 'Time must be an integer great than or equal to 5').isInt().gte(5)
####customSanitizers
{ sanitizerName: (value, [additional arguments]) => any, ... }
The customSanitizers
option can be used to add additional sanitizers methods as needed. This option should be an Object
defining the sanitizer names and associated functions.
Define your custom sanitizers:
app.use(koaValidator({
customSanitizers: {
toSanitizeSomehow: (value) => {
const newValue = value; //some operations
return newValue;
},
}
}));
Use them with their sanitizer name:
ctx.sanitize('address').toSanitizeSomehow();
ctx.check('testparam', 'Error Message').notEmpty().isInt();
ctx.check('testparam.child', 'Error Message').isInt(); // find nested params
ctx.check(['testparam', 'child'], 'Error Message').isInt(); // find nested params
Starts the validation of the specifed parameter, will look for the parameter in req
in the order params
, query
, body
, then validate, you can use 'dot-notation' or an array to access nested values.
If a validator takes in params, you would call it like ctx.assert('reqParam').contains('thisString');
.
Validators are appended and can be chained. See chriso/validator.js for available validators, or add your own.
Alias for ctx.check().
Alias for ctx.check().
Same as ctx.check(), but only looks in ctx.request.body
.
Same as ctx.check(), but only looks in ctx.request.query
.
Same as ctx.check(), but only looks in ctx.request.params
.
Only checks ctx.headers
. This method is not covered by the general ctx.check()
.
If you need to perform asynchronous validation, for example checking a database if a username has been taken already, your custom validator can return a promise or the customValidators should be async functions.
If you are using a promise you have to resolve with a boolean to know if is valid.
If you are using async you need to return a boolean to know if valid or not.
app.use(koaValidator({
customValidators: {
isUsernameAvailable: (username) => new Promise((resolve, reject) => {
User.findOne({ username: username })
.then(user => { resolve(!!user); })
.catch(reject);
});
}
}));
ctx.check('username', 'Username Taken').isUsernameAvailable();
Alternatively you can define all your validations at once using a simple schema. This also enables per-validator error messages. Schema validation will be used if you pass an object to any of the validator methods.
ctx.checkBody({
email: {
notEmpty: true,
isEmail: {
errorMessage: 'Invalid Email'
}
},
password: {
notEmpty: true,
matches: {
options: ['example', 'i'] // pass options to the validator with the options property as an array
// options: [/example/i] // matches also accepts the full expression in the first parameter
},
errorMessage: 'Invalid Password' // Error message for the parameter
},
'name.first': { //
optional: true, // won't validate if field is empty
isLength: {
options: [{ min: 2, max: 10 }],
errorMessage: 'Must be between 2 and 10 chars long' // Error message for the validator, takes precedent over parameter message
},
errorMessage: 'Invalid First Name'
}
});
You can also define a specific location to validate against in the schema by adding in
parameter as shown below:
ctx.check({
email: {
in: 'query',
notEmpty: true,
isEmail: {
errorMessage: 'Invalid Email'
}
}
});
Please remember that the in
attribute will have always highest priority. This mean if you use in: 'query'
then checkQuery() will be called inside even if you do checkParams()
or checkBody()
. For example, all of these calls will check query params for email param:
const schema = {
email: {
in: 'query',
notEmpty: true,
isEmail: {
errorMessage: 'Invalid Email'
}
},
password: {
notEmpty: true,
matches: {
options: ['example', 'i'] // pass options to the validator with the options property as an array
// options: [/example/i] // matches also accepts the full expression in the first parameter
},
errorMessage: 'Invalid Password' // Error message for the parameter
}
};
ctx.check(schema); // will check 'password' no matter where it is but 'email' in query params
ctx.checkQuery(schema); // will check 'password' and 'email' in query params
ctx.checkBody(schema); // will check 'password' in body but 'email' in query params
ctx.checkParams(schema); // will check 'password' in path params but 'email' in query params
Currently supported location are 'body', 'params', 'query'
. If you provide a location parameter that is not supported, the validation process for current parameter will be skipped.
You have two choices on how to get the validation errors:
ctx.assert('email', 'required').notEmpty();
ctx.assert('email', 'valid email required').isEmail();
ctx.assert('password', '6 to 20 characters required').len(6, 20);
let errors = ctx.validationErrors();
let mappedErrors = ctx.validationErrors(true);
errors:
[
{param: "email", msg: "required", value: "<received input>"},
{param: "email", msg: "valid email required", value: "<received input>"},
{param: "password", msg: "6 to 20 characters required", value: "<received input>"}
]
mappedErrors:
{
email: {
param: "email",
msg: "valid email required",
value: "<received input>"
},
password: {
param: "password",
msg: "6 to 20 characters required",
value: "<received input>"
}
}
Note: Using mappedErrors will only provide the last error per param in the chain of validation errors.
You can provide an error message for a single validation with .withMessage()
. This can be chained with the rest of your validation, and if you don't use it for one of the validations then it will fall back to the default.
ctx.assert('email', 'Invalid email')
.notEmpty().withMessage('Email is required')
.isEmail();
let errors = ctx.validationErrors();
errors:
[
{param: 'email', msg: 'Email is required', value: '<received input>'}
{param: 'email', msg: 'Invalid Email', value: '<received input>'}
]
You can use the optional()
method to skip validation. By default, it only skips validation if the key does not exist on the request object. If you want to skip validation based on the property being falsy (null, undefined, etc), you can pass in { checkFalsy: true }
.
ctx.checkBody('email').optional().isEmail();
//if there is no error, ctx.request.body.email is either undefined or a valid mail.
ctx.request.body.comment = 'a <span>comment</span>';
ctx.request.body.username = ' a user ';
ctx.sanitize('comment').escape(); // returns 'a <span>comment</span>'
ctx.sanitize('username').trim(); // returns 'a user'
console.log(ctx.request.body.comment); // 'a <span>comment</span>'
console.log(ctx.request.body.username); // 'a user'
Sanitizes the specified parameter (using 'dot-notation' or array), the parameter will be updated to the sanitized result. Cannot be chained, and will return the result. See chriso/validator.js for available sanitizers, or add your own.
If a sanitizer takes in params, you would call it like ctx.sanitize('reqParam').whitelist(['a', 'b', 'c']);
.
If the parameter is present in multiple places with the same name e.g. ctx.params.comment
& ctx.query.comment
, they will all be sanitized.
Alias for ctx.sanitize().
Same as ctx.sanitize(), but only looks in ctx.body
.
Same as ctx.sanitize(), but only looks in ctx.query
.
Same as ctx.sanitize(), but only looks in ctx.params
.
Only sanitizes ctx.headers
. This method is not covered by the general ctx.sanitize()
.
Koa-router allows you to define regex routes like:
app.get(/\/test(\d+)/, function() {});
You can validate the extracted matches like this:
ctx.assert(0, 'Not a three-digit integer.').len(3, 3).isInt();
See CHANGELOG.md
All this is based on express-validator
Copyright (c) 2016 Luis Carlos Cruz Carballo lcruzc@linkux-it.com, MIT License