Safetify is a tool for resolving received data from 3rd party applications (eg. API) to make data strongly-typed. Thanks to that we don't need to take care later on about unexpected undefined
values or value type.
To install Safetify simply use npm
:
npm install safetify --save
Received from 3rd party applications data are never strongly-typed so we can get any value in place we expect specifically typed one. Not always something that fits data we made application for. As example we can take bug in API because of which we wouldn't receive specific value or even get empty result. We need to prevent of situations like that one simply by checking for null
s, undefined
values or by mapping data to validated model.
Say we would receive from API data like this:
{
"name": "John Doe",
"age": 30,
"isSingle": false
}
With Safetify we can resolve the data in that way:
import * as Safetify from 'safetify';
let personResolver: Safetify.Object = Safetify.Object<IPerson>({
name: Safetify.String(),
age: Safetify.Number(),
isSingle: Safetify.Boolean()
});
let person: IPerson = personResolver.resolve(data).result;
Or use alternative imports:
import { ObjectResolver, StringResolver, NumberResolver, BooleanResolver } from 'safetify';
let personResolver: ObjectResolver = ObjectResolver<IPerson>({
name: StringResolver(),
age: NumberResolver(),
isSingle: BooleanResolver()
});
let person: IPerson = personResolver.resolve(data).result;
And then data will always be the exact type no matter if we received correct data or not. It prevents of many cases of using methods dedicated to specific types of data. Say we got undefined
value in name
property and now we want to use toUpperCase
method on that data. Basically it would look like that:
let name = person.name; // it's undefined
name.toUpperCase();
This would throw us a TypeError
exception. Safetify in that case will convert failed data to empty string
so in result we will always be operating on string:
let name = person.name; // now it's empty string
name.toUpperCase();
This won't throw any exception of course.
Safetify works both with TypeScript and JavaScript. Look on unit tests or documentation to get more familiar with.
Primitive types have no arguments to pass, ie. we can use resolvers:
StringResolver
NumberResolver
BooleanResolver
just like StringResolver().resolve(<anyString>)
.
In case of advanced types like object
we need to pass structure or values (as we did it in example above).
Additional functionality is a possibility to set nullable and default value (default is available only for primitives and enum).
let personResolver: ObjectResolver = ObjectResolver<IPerson>({
name: StringResolver().defaultsTo('Johnny'),
age: NumberResolver(),
isSingle: BooleanResolver().nullable()
});
After resolving any type of data resolve
method will always return Result
object which has 3 properties:
success
asboolean
,result
as resolved data type,error
if success is false asstring[]
.
If resolving was not succeeded then error
property will always describe what went wrong (and which property in if resolved data was object
or array
)
Safetify provides resolvers for couple of data types:
- AnyResolver - just returns what it gets on input
- NumberResolver - resolves
number
values - StringResolver - resolves
string
values - BooleanResolver - resolves
boolean
values - ArrayResolver - resolves
array
s - ObjectResolver - resolves
object
with specific structure (eg. model) - DictionaryResolver - resolves
object
which is a dictionary (has n key-value pairs of specific type) - PartialResolver - resolvers
object
which is subset of anotherobject
with specific structure (eg. model) - EnumResolver - resolves
enum
(if input is correct value of givenenum
) - DateResolver - resolves
date
s. This includesDate
object, string-date and timestamp - TupleResolver - resolves
tuple
of given structure - OneOfResolver - resolves input which can be one of given type
For examples and more, look on documentation.
Constraints are simple functions to check resolved value under specified condition. Say we want to get only positive numbers. We can make constraint like this:
const positiveResolver: NumberResolver = NumberResolver().constraint((n: number) => n >= 0);
positiveResolver.resolve(5);
Now if value is not positive, Result will have success
as false and error with constraint number.
positiveResolver.resolve(-5);
/*
Returns:
{
success: false,
result: -5,
error: [ 'constraint #0 failed' ]
}
*/
To make it more clear we can set own error in simplest way:
const positiveResolver: NumberResolver = NumberResolver().constraint((n: number) => n >= 0 || 'Value is not positive');
positiveResolver.resolve(-5);
/*
Returns:
{
success: false,
result: -5,
error: [ 'Value is not positive' ]
}
*/
As You saw in examples above, Safetify returns -5
as a result. It's because this is still correct value in terms of type safety. I assume dev would like to get -5
more than NaN
value. In case we would like to change value if constraint is failed, default value can be set:
const positiveResolver: NumberResolver = NumberResolver().constraint((n: number) => n >= 0 || 'Value is not positive', 0);
positiveResolver.resolve(-5);
/*
Returns:
{
success: false,
result: 0,
error: [ 'Value is not positive' ]
}
*/
Much better but sometimes it's still not perfect (especially in this example). Say we want to change all values to positive if there is failed constraint and value is negative. Default value argument can take also default value transform function:
const positiveResolver: NumberResolver = NumberResolver().constraint((n: number) => n >= 0 || 'Value is not positive', (n: number) => Math.abs(n));
positiveResolver.resolve(-5);
/*
Returns:
{
success: false,
result: 5,
error: [ 'Value is not positive' ]
}
*/
And that's all the simple stuff of constraints. Btw. they works only on primitive types resolvers.
Safetify provides bunch of utilities for checking type of values like isDefAndNotNull()
, isObject()
, isDateLike()
etc. You can use them via Safetify.util
property. All functions are described here. Most of the functions are based on Google's Closure Library's Goog functions.
If You want to contribute or fork this library firstly install all dependencies:
npm install
Then You can simply build it with:
npm run build
To run unit tests use:
npm test
To generate documentation use:
npm run jsdoc