npm install ifrit
const Ifrit = require("../dist")
const mongodb = require("mongodb")
const schema = require("./schema.json")
/*
* Names and classes of the first two good guys, ordered by age
*/
const query = `
SELECT name, details.biographical.class AS class, details.biographical.age AS age
WHERE NOT(bad_guy) AND details.physical.gender = "male"
ORDER BY details.biographical.age
LIMIT 2
`
mongodb.MongoClient
.connect("mongodb://localhost:27017/ifrit")
.then((db) => {
return db
.collection("mages")
.aggregate(Ifrit.compile.mongodb(schema, query))
.toArray()
})
.then(console.log)
.catch(console.err)
Ifrit is available as a Node.js module as well as a PureScript module. The JavaScript documentation is accessible on github pages. A documentation for PureScript is published on pursuit.
Different scenarios are available on the repository in the examples folder. Here's a
summary of all examples:
{
"name": "string",
"bad_guy": "boolean",
"details": {
"biographical": {
"age": "number",
"class": "string"
},
"physical": {
"gender": "string",
"height": "number"
}
},
"spells": [{
"name": "string",
"power": "number"
}]
}
SELECT name
WHERE bad_guy = true
SELECT name, MIN(details.biographical.age) AS min_age
WHERE details.physical.gender = "female"
GROUP BY NULL
SELECT AVG(spells_power) AS power
FROM (
SELECT AVG(spells.power), details.biographical.class AS class
WHERE details.physical.height < 170
)
GROUP BY class
SELECT name, details.biographical.class AS class, details.biographical.age AS age
WHERE NOT(bad_guy) AND details.physical.gender = "male"
ORDER BY details.biographical.age
LIMIT 2
SELECT name, AVG(details.physical.height)
WHERE details.physical.gender = "female"
GROUP BY NULL
ORDER BY details.physical.height
LIMIT 3
Ifrit is a compiler from SQL to NoSQL for data aggregation. NoSQL databases are great at modelling structured, unstructured and polymorphic data and usually offer powerful map-reduce based API when it comes to data aggregation. However, it is truly challenging to leverage those API in a web service.
In a nutshell, Ifrit:
Ifrit builds an Abstract Syntax Tree (AST) which represents only syntactycally correct requests. Then, it generates a request corresponding to a specific driver. So far, MongoDB is the only target driver available.
_ __
\ /
\ +----------+ +-----------+ +----------+ +----------+ /
SQL = =| tokenize +---------> parse +--------> verify +--------> generate |= = NoSQL
/ +----------+ +-----------+ +----------+ +----------+ \
_/ \__
input
SELECT COUNT(_id) AS nb_txs WHERE amount > 1000 GROUP BY account.currency
output
[
{
"$match": {
"amount": {
"$gt": 1000
}
}
},
{
"$group": {
"_id": "$account.currency",
"nb_txs": {
"$sum": 1
}
}
}
]
Ifrit acts on a single collection at a time and does not support joins. Therefore, a schema is required in order to verify the request (semantically and security wise). Schemas are defined as JSON objects in a declarative syntaxe.
Ifrit supports the following primitive types: number, string, boolean and null. Arrays
and objects can be declared by nesting primitive types in JSON arrays [] or objects {}.
Ifrit can only see what's defined in a schema. The compilation will fail if the request tries to use or select a field not present in the schema. This can be used to control the scope of what elements are accessible via the query.
{
"amount": "number",
"account": {
"country": "string",
"currency": "string"
},
"items": [{
"price": "number",
"description": "string"
}]
}
| type | support |
|---|---|
| projection | SELECT, AS, FROM |
| grouping | GROUP BY |
| filtering | WHERE, LIMIT, OFFSET |
| sorting | ORDER BY, DESC, ASC |
| operators | AND, OR, NOT, =, !=, >, < |
| function | applicable type |
|---|---|
| AVG | number |
| COUNT | any |
| MAX | number |
| MIN | number |
| SUM | number |
⚠ Ifrit relies on a strict order of clauses ⚠
- (1)
SELECT- (2)
FROM- (3)
WHERE- (4)
GROUP BY- (5)
ORDER BY- (6)
(ASC | DESC)- (7)
LIMIT- (8)
OFFSET
Ifrit is case-sensitive, e.g. AS != as, NULL != null, etc.
Ifrit doesn't support the * selector.
ORDER BY can't be use with NULL.
Ifrit can't JOIN from other collections, the FROM can only be used to defined derived
tables, i.e, define a multi-level pipeline of map / reduce operations.
Aggregation functions can be applied to numbers when used with GROUP BY or directly to
array of numbers when apply without. Ifrit also supports nested notation for array of objects
(e.g. SELECT AVG(items.price)).
When no alias is specified, the property of the output schema is named after the selector.
(with MongoDB, . in names are replaced with _).
On a classic i7 quad-core, 16Gb DDR5 RAM / Ubuntu 16.04
SELECT age
> 9,795 ops/sec ±0.40% (91 runs sampled)
SELECT class AS klass, COUNT(bonus)
> 4,791 ops/sec ±0.83% (90 runs sampled)
SELECT AVG(age) GROUP BY class
> 5,754 ops/sec ±0.58% (94 runs sampled)
SELECT is_master WHERE age > 14 AND age < 20
> 4,586 ops/sec ±0.65% (93 runs sampled)
SELECT AVG(power) AS avg_pow FROM (SELECT AVG(spells.power), age) WHERE age > 18 GROUP BY NULL
> 2,378 ops/sec ±0.63% (93 runs sampled)
* joker in selectSupport for the following keyword:
SELECTFROMWHEREGROUP BYORDER BY(ASC | DESC)LIMITOFFSETSupport for the following boolean operators:
ANDORSupport for the following binary operators:
><=!=Support for the following unary operators:
NOTSupport for the following functions (projections & aggregations):
AVGCOUNTMAXMINSUMSupport for aliases inside SELECT
Support for nested objects
Support for derived tables
| Driver | Version |
|---|---|
| MongoDB | ~3.4 |
FAQs
An SQL to NoSQL compiler for data aggregation
We found that ifrit demonstrated a not healthy version release cadence and project activity because the last version was released a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Research
A malicious npm package disguised as a WhatsApp client is exploiting authentication flows with a remote kill switch to exfiltrate data and destroy files.
Security News
PyPI now supports digital attestations, enhancing security and trust by allowing package maintainers to verify the authenticity of Python packages.
Security News
GitHub removed 27 malicious pull requests attempting to inject harmful code across multiple open source repositories, in another round of low-effort attacks.