ArangoTools
The goal of of this library is to provide a declarative set of
invariants at
the database level.
Influenced by the great DX of MongoDB, and the
declarative model, wherever permissions allow arango-tools will create
resources you've said should exist. Both the ensure
and migrate
functions
are idempotent, and create resources only when they don't exist.
Basically, you should be able to state your intent and know when it couldn't
happen. Think dotenv-safe, but for
database stuff.
Installation
npm install arango-tools
Usage
The ensure
function takes an object describing the desired state of the
database; your invariants, the collections and whatnot that must exist for your
program to run. When invariants hold, you're passed a set of accessor functions
that allow you to interact with the database. When they do not, a descriptive
error is raised.
The collection properties writeConcern
, replicationFactor
, schema
and waitForSync
are mutable, and will be updated if you change the options passed to documentcollection
and edgecollection
objects.
const { ensure } = require('arango-tools')
let { query, truncate, drop, transaction, collections } = await ensure({
variables: { username: 'mike', password: 'test', rootPassword: 'secret' },
schema: {
type: 'database',
name: 'myapp',
url: 'http://localhost:8529',
rootPassword: '{{rootPassword}}',
options: [
{ type: 'user', username: '{{username}}', password: '{{password}}' },
{
type: 'documentcollection',
name: 'people',
options: {
schema: {
rule: {
properties: {
nums: { type: 'array', items: { type: 'number', maximum: 6 } },
},
additionalProperties: { type: 'string' },
required: ['nums'],
},
level: 'moderate',
message:
"The document does not contain an array of numbers in attribute 'nums', or one of the numbers is bigger than 6.",
},
waitForSync: false,
},
},
{
type: 'edgecollection',
name: 'likes',
options: { journalSize: 10485760, waitForSync: true },
},
{
type: 'delimiteranalyzer',
name: 'my-delimiter-analyzer',
delimiter: ';',
},
{
type: 'searchview',
name: 'placeview',
options: {
links: {
places: {
fields: {
name: { analyzers: ['text_en', 'my-delimiter-analyzer'] },
description: { analyzers: ['text_en'] },
},
},
},
},
},
{
type: 'geoindex',
on: 'places',
fields: ['lat', 'lng'],
geoJson: true,
},
],
},
})
const { ArangoTools, dbNameFromFile } = require('arango-tools')
let name = dbNameFromFile(__filename)
let { migrate } = await ArangoTools({ rootPass, url })
let { query, truncate, drop, transaction, collections } = await migrate([
{
type: 'database',
databaseName: name,
users: [{ username: 'mike', passwd: 'sekret' }],
},
{
type: 'documentcollection',
databaseName: name,
name: 'widgets',
options: { journalSize: 10485760, waitForSync: true },
},
{
type: 'delimiteranalyzer',
databaseName: name,
name: 'my-delimiter-analyzer',
delimiter: ';',
},
{
type: 'searchview',
databaseName: name,
name: 'myview',
options: {},
},
{
type: 'geoindex',
databaseName: name,
collection: 'places',
options: {
fields: ['pts'],
geoJson: true,
},
},
])
await collections.widgets.save({ foo: 'bar' })
let cursor = await query`
FOR widget IN widgets
FILTER widget.foo === "bar"
RETURN widget
`
await cursor.all()
await drop()
For those moments when you really just want to connect, there is also connectTo
.
const { query, drop, truncate, transaction } = await connectTo({
url: 'http://localhost:8529',
databaseName: 'mydb',
as: {
username: 'mike',
password: 'secret',
},
})
Issues
Currently arango-tools can create a database, a document/edge collection, a search view, delimiter analyzer and a GeoIndex.
Other types and graphs will be added soon.
Analyzers can't be updated in ArangoDB. The closest equivalent is deleting one and replacing it with a new one of the same name. That's the strategy currently for the delimiteranalyzer
, but it's possible that could cause issues with queries actively using the analyzer at the moment it's replaced.
The replicationFactor
collection property is only available when ArangoDB is running in cluster mode. If arango-tools is talking to a cluster, it will create/update that property. If it's talking to a single database, it will just ignore replicationFactor
. Not being too pedantic means a more consistent experience across environments.