Dynamodels is a dead simple typescript overlay to easily manage DynamoDB CRUD operations.
It provides helpers for pagination, filtering, sorting and more !
Install dynamodels package from NPM public registry.
-
Using npm:
npm i dynamodels -
Using yarn:
yarn add dynamodels
The only thing you need to do is extending base model class Model<T> providing a type defintion for your entity, a table name, a hashkey, and optionally, a range key.
Here is an example for table with a composite key:
// Import dynamodels Base Model
import Model from 'dynamodels';
// Type definition for your entity
interface IAlbum {
artist: string;
album: string;
year?: number;
genres?: string[];
}
export class Album extends Model<IAlbum> {
// DynamoDB table name
protected tableName = 'my_albums';
// The keys of the table. Here it is a composite key (artist,album)
protected hashkey = 'artist';
protected rangekey = 'album';
// Optionally override constructor
constructor(item?: IAlbum) {
super(item);
}
}You have the option to either directly provide the table name, as shown in the previous example, or retrieve it from the AWS Systems Manager (SSM) Parameter Store by specifying the ARN of the SSM parameter. The value will be automatically replaced at runtime.
Here is an example of using table name as SSM Parameter ARN
interface IAlbum {
artist: string;
album: string;
year?: number;
genres?: string[];
}
export class Album extends Model<IAlbum> {
protected tableName = 'arn:aws:ssm:<aws-region>:<aws-account-id>:parameter/ParameterName';
protected hashkey = 'artist';
protected rangekey = 'album';
constructor(item?: IAlbum) {
super(item);
}
}Here is another example for a table with a simple hashkey:
import Model from 'dynamodels';
interface IUser {
email: string;
// ..other fields
}
export class User extends Model<IUser> {
protected tableName = 'my_users';
protected hashkey = 'email';
}You can either call the create method or the save method.
The create method will throw if the hashkey or the hashkey/rangekey pair already exists.
The save method will overwrite the existing item.
const classic = new Album({
artist: 'Pink Floyd',
album: 'Dark Side of the Moon',
year: 1973,
genre: ['Rock', 'Psychadelic', 'Progressive'],
});
// Item will be saved
await album.save();
// Will throw, as item already exists
try {
await album.create();
} catch (e) {
if (e.name === 'EALREADYEXISTS') {
// Do something
}
// Do something else
}You can also directly pass in argument the item to save.
const albums = new Album();
await album.create({
artist: 'Bob Marley & The Wailers',
album: "Burnin'",
year: 1973,
genre: ['Reggea'],
});You can use Joi objects to validate the data to save.
If object don't pass Joi validation an error is thrown.
Define your Joi schema in your model.
export class Album extends Model<IAlbum> {
protected tableName = 'my_albums';
protected hashkey = 'artist';
protected rangekey = 'album';
protected schema = Joi.object().keys({
artist: Joi.string().required(),
album: Joi.string().required(),
year: Joi.number().required(),
genres: Joi.array(Joi.string()).optional(),
});
}Model validation is automatically enforced when creating/saving entities:
// Will throw as year must be a number
await album.save({
artist: 'Bob Marley & The Wailers',
album: "Burnin'",
year: '1973',
genre: ['Reggea'],
});Table has a simple hashkey:
const users = new User();
const user = await user.get('some-user@domain.com'); // {email: some-user@domain.com, ...}Table has a composite key:
const albums = new Album();
const nvrmind = await album.get('Nirvana', 'Nevermind'); // {artist: 'Nirvana', album: 'Nevermind', year: 1991...}Note: For table with a composite key, range key is mandatory. This will throw an exception.
await album.get('ACDC'); // Bad RequestYou can also just check if an item exists:
const albums = new Album();
if (await album.exists('The Fugees', 'The Score')) {
return 'Kill me softly';
}Note: It is not advised to use scan operations as it is time-consuming.
To retrieve all the entries of a table use a DynamoDB scan operation.
const albums = new Album();
const result = await albums.scan().exec();This will return the first 1MB of matching result and the last evaluated key.
If you want to retrieve all items beyond this 1MB limit, use execAll;
const result = await albums.scan().execAll();You can use pagination helpers to get paginated result
const albums = new Album();
const result = await albums.scan().paginate({ size: 50 }).exec();This will return the 50 first items and the last evaluated key. To fetch the next page, simply use:
const nextPage = await albums.scan().paginate(result.nextPage)exec();Natively, dynamoDB performs filter operations after having retrieved a page of result.
This leads to inconsistent pages size. Let's say you target a page size of 50. DynamoDB fetch the first page which length is 50. After applying filters on this first page you ends up with 13 results, and maybe 32 on the seconds, 7 on the third and so on.
If you want to force page size to be same despite filtering, you can use PaginationMode.CONSTANT_PAGE_SIZE option.
const albums = new Album();
const result = await albums
.scan()
.filter({
year: 1969, // Summer of love
})
.paginate({
mode: PaginationMode.CONSTANT_PAGE_SIZE,
size: 50,
})
.exec();Under the hood, dynamodels will fetch as many pages as it is necessary to fill the 50 results matching filters.
A filtering helper method is also available.
For instance to retrieve albums released after 1973 (included), use the following query:
const albums = new Album();
const result = await albums
.scan()
.filter({
year: ge(1973),
})
.exec();The filter accept an object where keys are the fields on which you to filter and value can be either:
- just the target value of the field, in this case the equal
EQoperator is used - a call to a filter operator helper method.
Note: if you want to filter on a field which is also a Amazon reserved keyword, dynamodels with automatically escape it ✨
Available filter operators are:
- Equal:
eq(value: string | number | Buffer) - Not Equal:
neq(value: string | number | Buffer) - In:
isIn(values: Array<string | number | Buffer>) - Lesser than:
lt(value: string | number | Buffer) - Lesser or equal than:
le(value: string | number | Buffer) - Greater than:
gt(value: string | number | Buffer) - Greater or equal than:
le(value: string | number | Buffer) - Between boundaries:
between(lower: string | number | Buffer, upper: string | number | Buffer) - Contains substring
contains(value: string) - Do not contains substring:
notContains(value: string) - Begin with:
contains(value: string) - Is null:
isNull() - Is not null:
notNull()
For string, utf-8 alphabetical order is used.
For binary, unsigned byte-wise comparison is used.
Check official DynamoDB documentation for more details.
For complex conditions, i.e. conditions with compositions of AND/OR or NOT clauses, dynamodels provides a fluid synthax to easily write them.
const albums = new Album();
const result = await albums
.scan()
.filter(attr('year').lt(1970)
.or(attr('year').ge(1980))
.and(not(attr('artist').beginsWith('Bob')))
.exec();The library also provides helpers to build dynamoDB queries.
The synthax is simmilar to scan operations.
Key conditions can be added with the keys helpers method.
For instance to retrieve all the album for a given artist.
const albums = new Album();
const result = await albums
.query()
.filter({
artist: 'The Rolling Stones',
})
.exec();You can combine key condition if your table has a composite key.
In this case both condition are applied: it is a AND not a OR.
Available key conditions operators are:
- Equal:
eq(value: string | number | Buffer) - Lesser than:
lt(value: string | number | Buffer) - Lesser or equal than:
le(value: string | number | Buffer) - Greater than:
gt(value: string | number | Buffer) - Greater or equal than:
le(value: string | number | Buffer) - Between boundaries:
between(lower: string | number | Buffer, upper: string | number | Buffer) - Begin with:
contains(value: string)
You can use, if you prefer, the same fluid sythax than for filter conditions.
const albums = new Album();
const result = await albums.query().keys(attr('artist').eq('Bob Dylan')).exec();Just be aware that:
- Key condition on hash key is mandatory.
- Only
eq()operator can be used on hash key. - You can only use
andbetween hash key condition and optional range key condition. - Only the operators listed above can be used on range key condition.
Otherwise dynamoDB will throw a ValidationException. Dynamodels will not check these prerequisites for you.
You can also specify the index which is used.
Let's say you have the following global secondary index called year_index on your albums table: pk=artist, sk=year.
You can retrieve all the albums from Deep Purple release before 1976:
const albums = new Album();
const result = await albums
.query('year_index')
.keys({
artist: 'Deep Purple',
year: lt(1976),
})
.exec();The following synthax using index method is equivalent:
const albums = new Album();
const result = await albums
.query()
.index('year_index')
.filter({
artist: 'Deep Purple',
year: lt(1976),
})
.exec();Pagination work the same way than for scan operations.
const albums = new Album();
const result = await albums.query().keys({ artist: 'Dire Straits' }).paginate({ size: 50 }).exec();This will return the 50 first items and the last evaluated key. To fetch the next page, simply use:
const nextPage = albums.query().keys({ artist: 'Dire Straits' }).paginate(result.nextPage)exec();Filtering process is the same for query and scan operations. See above.
You can use sort helpers to sort the result based on the range key.
const albums = new Album();
// From oldest to newest
const result = await albums
.query('year_index')
.filter({
artist: 'James brown',
})
.sort('asc')
.exec();
// From newest to oldest
const result = await albums
.query('year_index')
.filter({
artist: 'James brown',
})
.sort('desc')
.exec();To perform a batch get operations, simply give the keys in argument:
const albums = new Album();
const result = await albums.batchGet([
{ artist: 'Janis Joplin', album: "I Got Dem Ol' Kozmic Blues Again Mama!" },
{ artist: 'Creedence Clearwater Revival', album: 'Willy and the Poor Boys' },
{ artist: 'The Beatles', album: "Sgt. Pepper's Lonely Hearts Club Band" },
{ artist: 'Queen', album: 'A Night at the Opera' },
{ artist: 'The Clash', album: ' London Calling' },
]);Batch get operations are limited to 100 items.
Under the hood, if you give more than 100 keys or keys pair, dynamodels will split the operation in chunks of 100 items.
For instance a batchGet operation with 642 keys will be automatically split in 7 batches.
A wrapper around putItem operations is provided.
The synthax is the following:
If table has a simple hashkey:
const users = new User();
await user.update('some-user@domain.com', {
password: put(hashSync('n3wP4ssW0rd', 10)),
additional_details: remove(),
});Table has a composite key:
const albums = new Album();
await album.update('Jimi Hendrix', 'Are You Experienced', {
year: add(1967),
genre: put(['Rock', 'Blues', 'Psychadelic']),
});The three helpers method, add, remove, and putprovide convenient synthax to easily build DoumentClient.UpdateAttributes objects.
Table has a simple hashkey:
const users = new User();
await user.delete('some-user@domain.com');Table has a composite key:
const albums = new Album();
await album.delete('Nirvana', 'Nevermind');