@/jsonpath

Query JavaScript objects with JSONPath expressions. Robust / safe JSONPath engine for Node.js.

Query Example

import { JSONPath } from '@astronautlabs/jsonpath';

let cities = [
  { name: "London", "population": 8615246 },
  { name: "Berlin", "population": 3517424 },
  { name: "Madrid", "population": 3165235 },
  { name: "Rome",   "population": 2870528 }
];

let names = JSONPath.query(cities, '$..name');

// [ "London", "Berlin", "Madrid", "Rome" ]

Install

Install from npm:

$ npm install @astronautlabs/jsonpath

JSONPath Syntax

Here are syntax and examples adapted from Stefan Goessner's original post introducing JSONPath in 2007.

Given this sample data set, see example expressions below:

{
  "store": {
    "book": [ 
      {
        "category": "reference",
        "author": "Nigel Rees",
        "title": "Sayings of the Century",
        "price": 8.95
      }, {
        "category": "fiction",
        "author": "Evelyn Waugh",
        "title": "Sword of Honour",
        "price": 12.99
      }, {
        "category": "fiction",
        "author": "Herman Melville",
        "title": "Moby Dick",
        "isbn": "0-553-21311-3",
        "price": 8.99
      }, {
         "category": "fiction",
        "author": "J. R. R. Tolkien",
        "title": "The Lord of the Rings",
        "isbn": "0-395-19395-8",
        "price": 22.99
      }
    ],
    "bicycle": {
      "color": "red",
      "price": 19.95
    }
  }
}

Example JSONPath expressions:

Methods

JSONPath.query(obj, pathExpression[, count])

Find elements in obj matching pathExpression. Returns an array of elements that satisfy the provided JSONPath expression, or an empty array if none were matched. Returns only first count elements if specified.

let authors = jp.query(data, '$..author');
// [ 'Nigel Rees', 'Evelyn Waugh', 'Herman Melville', 'J. R. R. Tolkien' ]

JSONPath.paths(obj, pathExpression[, count])

Find paths to elements in obj matching pathExpression. Returns an array of element paths that satisfy the provided JSONPath expression. Each path is itself an array of keys representing the location within obj of the matching element. Returns only first count paths if specified.

let paths = jp.paths(data, '$..author');
// [
//   ['$', 'store', 'book', 0, 'author'] },
//   ['$', 'store', 'book', 1, 'author'] },
//   ['$', 'store', 'book', 2, 'author'] },
//   ['$', 'store', 'book', 3, 'author'] }
// ]

JSONPath.nodes(obj, pathExpression[, count])

Find elements and their corresponding paths in obj matching pathExpression. Returns an array of node objects where each node has a path containing an array of keys representing the location within obj, and a value pointing to the matched element. Returns only first count nodes if specified.

let nodes = jp.nodes(data, '$..author');
// [
//   { path: ['$', 'store', 'book', 0, 'author'], value: 'Nigel Rees' },
//   { path: ['$', 'store', 'book', 1, 'author'], value: 'Evelyn Waugh' },
//   { path: ['$', 'store', 'book', 2, 'author'], value: 'Herman Melville' },
//   { path: ['$', 'store', 'book', 3, 'author'], value: 'J. R. R. Tolkien' }
// ]

JSONPath.value(obj, pathExpression[, newValue])

Returns the value of the first element matching pathExpression. If newValue is provided, sets the value of the first matching element and returns the new value.

JSONPath.parent(obj, pathExpression)

Returns the parent of the first matching element.

JSONPath.apply(obj, pathExpression, fn)

Runs the supplied function fn on each matching element, and replaces each matching element with the return value from the function. The function accepts the value of the matching element as its only parameter. Returns matching nodes with their updated values.

let nodes = jp.apply(data, '$..author', function(value) { return value.toUpperCase() });
// [
//   { path: ['$', 'store', 'book', 0, 'author'], value: 'NIGEL REES' },
//   { path: ['$', 'store', 'book', 1, 'author'], value: 'EVELYN WAUGH' },
//   { path: ['$', 'store', 'book', 2, 'author'], value: 'HERMAN MELVILLE' },
//   { path: ['$', 'store', 'book', 3, 'author'], value: 'J. R. R. TOLKIEN' }
// ]

JSONPath.parse(pathExpression)

Parse the provided JSONPath expression into path components and their associated operations.

let path = jp.parse('$..author');
// [
//   { expression: { type: 'root', value: '$' } },
//   { expression: { type: 'identifier', value: 'author' }, operation: 'member', scope: 'descendant' }
// ]

JSONPath.stringify(path)

Returns a path expression in string form, given a path. The supplied path may either be a flat array of keys, as returned by jp.nodes for example, or may alternatively be a fully parsed path expression in the form of an array of path components as returned by jp.parse.

let pathExpression = jp.stringify(['$', 'store', 'book', 0, 'author']);
// "$.store.book[0].author"

Differences from Original Implementation

This implementation aims to be compatible with Stefan Goessner's original implementation with a few notable exceptions described below.

Evaluating Script Expressions

Script expressions (i.e, (...) and ?(...)) are statically evaluated via static-eval rather than using the underlying script engine directly. That means both that the scope is limited to the instance variable (@), and only simple expressions (with no side effects) will be valid. So for example, ?(@.length>10) will be just fine to match arrays with more than ten elements, but ?(process.exit()) will not get evaluated since process would yield a ReferenceError. This method is even safer than vm.runInNewContext, since the script engine itself is more limited and entirely distinct from the one running the application code. See more details in the implementation of the evaluator.

Grammar

This project uses a formal BNF grammar to parse JSONPath expressions, an attempt at reverse-engineering the intent of the original implementation, which parses via a series of creative regular expressions. The original regex approach can sometimes be forgiving for better or for worse (e.g., $['store] => $['store']), and in other cases, can be just plain wrong (e.g. [ => $).

Other Minor Differences

As a result of using a real parser and static evaluation, there are some arguable bugs in the original library that have not been carried through here:

• strings in subscripts may now be double-quoted
• final step arguments in slice operators may now be negative
• script expressions may now contain . and @ characters not referring to instance variables
• subscripts no longer act as character slices on string elements
• non-ascii non-word characters are no-longer valid in member identifier names; use quoted subscript strings instead (e.g., $['$'] instead of $.$)
• unions now yield real unions with no duplicates rather than concatenated results

License

MIT