Features

• Decorator and class based declarative transformations.
• Automatically converts nullish or boolean-ish values. (e.g. 'null', 'true')
• Works as middleware, following CSV parser you're using (like node-csv)
• Customizable, define your own conversion rules or custom transform methods.
• Tested, with high code coverage.

Installation

reflect-metadata package is required since we're highly rely on type reflection.

npm install type-csv-transformer reflect-metadata
# or
yarn add type-csv-transformer reflect-metadata

And be sure to reflect-metadata is imported before using type-csv-transformer

import "reflect-metadata";

Last, a little tsconfig.json tweak is needed.

{
    "compilerOptions": {
        ...
        "emitDecoratorMetadata": true,
        "experimentalDecorators": true,
        ...
    }
}

Usage

Simple

class Cat {
  @Column()
  name: string;

  @Column({ name: 'name' })
  hasName: boolean;

  @Column()
  age: number;

  @Column({ defaultValue: false })
  hasLongHair: boolean;

  @Column({ name: 'likes' })
  interest: string;

  @Column()
  birthday: Date;

  @Column({ name: 'hometown' })
  @Nullable()
  home: string | null;

  @Column()
  @Bool()
  isOddEyed: boolean;
}

const cat = {
  name: 'Toma',
  age: 10,
  likes: 'chicken',
  birthday: '2011-12-01',
  hometown: 'null',
  isOddEyed: 'false',
};
const catInstance = transform(Cat, cat);

expect(catInstance).toBeInstanceOf(Cat);
expect(catInstance).toEqual({
  name: cat.name,
  hasName: true,
  age: cat.age,
  hasLongHair: false,
  interest: cat.likes,
  birthday: new Date(cat.birthday),
  home: null,
  isOddEyed: false,
});

Advanced

class Cat {
  @Column()
  name: string;

  @Column({ name: 'name' })
  hasName: boolean;

  @Column()
  age: number;

  @Column({ defaultValue: false })
  hasLongHair: boolean;

  @Column({ name: 'likes' })
  interest: string;

  @Column()
  birthday: Date;

  @Column({ name: 'hometown' })
  @Nullable()
  home: string | null;

  @Column()
  @Bool()
  isOddEyed: boolean;
  
  @Column()
  @Transform<number>(value => value / 1000) // convert grams to kilograms
  weight: number;
}

const cat = {
  name: 'Toma',
  age: 10,
  likes: 'chicken',   
  birthday: '2011-12-01',
  hometown: 'NULL',
  isOddEyed: 'no',
  weight: 5600,
};
// this time, we're setting custom symbols
const catInstance = transform(Cat, cat, { nullSymbols: ['NULL'], trueSymbols: ['yes'], falseSymbols: ['no'] });

expect(catInstance).toBeInstanceOf(Cat);
expect(catInstance).toEqual({
  name: cat.name,
  hasName: true,
  age: cat.age,
  hasLongHair: false,
  interest: cat.likes,
  birthday: new Date(cat.birthday),
  home: null,
  isOddEyed: false,
  weight: 5.6,
});

This library is still in early stage. If this behavior does not fit into your use cases, feel free to open an issue!

API Reference

Decorators

@Column(options?)

@Column must be specified on every class properties you want to convert, otherwise transformer will ignore those properties.

ColumnOptions:

| Property | Default | Description | |--------------------------|:-------:|-------------------------------------------------------------------------------------| | name?: string | - | Specifies origin object's key. Used when it's not identical to class property name. | | allowNull?: boolean | false | If set to false or by default, it will throw error when null values are given. | | defaultValue?: unknown | - | Sets default value to assign when evaluation lead to null values. |

@Nullable(options?)

Specifying @Nullable works like set ColumnOptions.allowNull=true. (You cannot use this decorator when ColumnOptions.allowNull=false.)

And it provides extra options to set nullish conversion symbols.

NullableOptions:

| Property | Default | Description | |-------------------------------------|:-------:|-----------------------------------------------------------------------------------------------------------------------------| | symbols?: Array<string \| number> | - | Sets nullish symbols. Overrides default symbols and global transform() behavior. Values cannot be null or empty array. |

@Bool(options?)

@Bool is utility decorator that can transform boolean-ish values

BoolOptions:

| Property | Default | Description | |------------------------------------------|:-------:|------------------------------------------------------------------------------------------------------------------------------------------------------| | trueSymbols?: Array<string \| number> | - | Sets boolean-ish symbols that matches to true. Overrides default symbols and global transform() behavior. Values cannot be null or empty array. | | falseSymbols?: Array<string \| number> | - | Sets boolean-ish symbols that matches to false. Overrides default symbols and global transform() behavior. Values cannot be null or empty array. |

@Transform(transformFunction)

It provides custom transform method. Used when behaviors of @Nullable or @Bool does not work as expected or to handle other complex cases.

• transformFunction: (value: any) => unknown

There are some caveats when mixing other methods with @Transform.

• @Bool will be ignored when used together.
• @Transform will be ignored when transform leads to defaultValue of @Column.

Methods

transform(cls, object, options?)

Transforms object into class instance. Also provides extra options to control its behavior.

TransformerOptions:

| Property | Default | Description | |--------------------------------------------------|:-----------------------:|---------------------------------------------------| | nullSymbols?: Array<string \| number> \| null | ['null', 'None', ''] | Sets nullish symbols. | | trueSymbols?: Array<string \| number> \| null | ['true', 'True', 1] | Sets boolean-ish symbols that matches to true. | | falseSymbols?: Array<string \| number> \| null | ['false', 'False', 0] | Sets boolean-ish symbols that matches to false. |

Inspirations

At first, I was trying to handle CSVs with class-transformer. That attemp actually works but there're many extra codes and some unwanted behaviors. So I created the type-csv-transformer.

Still this library's concepts and implementations are inspired by class-transformer a lot, but there're some differences.

type-csv-transformer

• concentrates on more specific problems.
• provides many utilities to handle CSV easily.
• has much smaller bundle size.

Contributing

1. Fork this repository
2. Create new feature branch
3. Write your codes and commit
4. Make all tests and linter rules pass
5. Push and make pull request