git@github.com:stalniy/ucast.git
npm install @ucast/mongo
This package is a part of [ucast] ecosystem. It provides a parser that can parse MongoDB query into conditions AST.
[ucast]: https://github.com/stalniy/ucast
``sh`
npm i @ucast/mongoor
yarn add @ucast/mongoor
pnpm add @ucast/mongo
To parse MongoDB query into conditions AST, you need to create a MongoQueryParser instance:
`js
import { MongoQueryParser, allParsingInstructions } from '@ucast/mongo';
const parser = new MongoQueryParser(allParsingInstructions);
const ast = parser.parse({
id: 1,
active: true
});
`
To create a parser you need to pass in it parsing instruction. Parsing instruction is an object that defines how a particular operator should be parsed. There are 3 types of ParsingInstruction, one for each AST node type:
* field represents an instruction for an operator which operates in a field context only. For example, operators $eq, $lt, $not, $regex
* represents an instruction for an operator that aggregates nested queries. For example, operators $and, $or, $nor
* represents an instruction for an operator which operates in a document context only. For example, $where or $jsonSchema
It's important to understand that it's not required that parsing instruction with type field should be parsed into FieldCondition. It can be parsed into CompoundCondition as it's done for $not operator.
A parsing instruction is an object of 3 fields:
`ts`
const parsingInstruction = {
type: 'field' | 'document' | 'compound',
validate?(instruction, value) { // optional
// throw exception if something is wrong
},
parse?(instruction, schema, context) { // optional
/*
* custom logic to parse operator,
* returns FieldCondition | DocumentCondition | CompoundCondition
*/
}
}
Some operators like $and and $or optimize their parsing logic, so if one of that operators contain a single condition it will be resolved to that condition without additional wrapping. They also recursively collapse conditions from nested operators with the same name. Let's see an example to understand what this means:
`js`
const ast = parser.parse({
a: 1
$and: [
{ b: 2 },
{ c: 3 }
]
});
console.dir(ast, { depth: null })
/*
CompoundCondition {
operator: "and",
value: [
FieldCondition { operator: "eq", field: "a", value: 1 },
FieldCondition { operator: "eq", field: "b", value: 2 },
FieldCondition { operator: "eq", field: "c", value: 3 },
]
}
*/
This optimization logic helps to speed up interpreter's execution time, instead of going deeply over tree-like structure we have a plain structure of all conditions under a single compound condition.
Pay attention: parser removes $ prefix from operator names
In order for an operator to be parsed, it needs to define a parsing instruction. Let's implement a custom instruction which checks that object corresponds to a particular json schema.
First of all, we need to understand on which level this operator operates (field or document). In this case, $jsonSchema clearly operates on document level. It doesn't contain nested MongoDB queries, so it's not a compound instruction. So, we are left only with document one.
To test that document corresponds to provided json schema, we use ajv but it's also possible to use a library of your preference.
`js
// operators/jsonSchema.js
import { DocumentInstruction, DocumentCondition } from '@ucast/core';
import Ajv from 'ajv';
export const $jsonSchema: DocumentInstruction = {
type: 'document',
validate(instruction, value) {
if (!value || typeof value !== 'object') {
throw new Error("${instruction.name}" expects to receive an object)`
}
},
parse(instruction, schema) {
const ajv = new Ajv();
return new DocumentCondition(instruction.name, ajv.compile(schema));
}
};
In order to use this operator, we need to pass this instruction into MongoQueryParser constructor:
`js
import { MongoQueryParser, allParsingInstructions } from '@ucast/core';
import { $jsonSchema } from './operators/jsonSchema';
const parser = new MongoQueryParser({
...allParsingInstructions,
$jsonSchema
});
const ast = parser.parse({
$jsonSchema: {
type: 'object',
properties: {
firstName: { type: 'string' },
lastName: { type: 'string' },
},
additionalProperties: false,
}
});
console.dir(ast, { depth: null });
/*
DocumentCondition { operator: "jsonSchema", value: [Function: validate] }
*/
`
The only thing which is left is to implement a corresponding JavaScript interpreter:
`js``
function jsonSchema(condition, object) { // interpreter
return condition.value(object);
}
Want to file a bug, contribute some code, or improve documentation? Excellent! Read up on guidelines for [contributing]
[contributing]: https://github.com/stalniy/ucast/blob/master/CONTRIBUTING.md