obj-walker

Walk objects like this guy.

Map over an object in a preorder or postoder depth-first manner. Also, provides functions for serializing and deserializng self-referential objects using JSON pointer.

This library is designed to work well with functions that traverse an object in the same way JSON.stringify and JSON.parse do. Namely, preorder and postorder. To mimic that behavior entirely set the jsonCompat option to true.

Custom traversal functions are supported for some functions. This allows you to walk tree-like structures, such as a JSON schema, in a more efficient and logical way. Prefer walkEach in these scenarios.

map, walkEach, walkEachAsync, mapLeaves, compact, and truncate support the option modifyInPlace for in-place modification. Otherwise, the object is deep cloned.

export interface MutationOption {
    /** Set to true to modify the object instead of returning a new object. */
    modifyInPlace?: boolean
}

walker

walker(obj: object, walkFn: WalkFn, options: Options = {}) => void

Generic walking fn that traverses an object in preorder (default) or postorder, calling walkFn for each node. Can be used directly, but probably shouldn't.

export interface Node {
    key: string | undefined
    val: any
    parents: any[]
    path: string[]
    isLeaf: boolean
    isRoot: boolean
}

export interface Options {
    postOrder?: boolean
    jsonCompat?: boolean
    traverse?(x: any): any
}

import { walker } from 'obj-walker'

const obj = {
    a: {
        b: 23,
        c: 24,
    },
    d: {
        e: 'Bob',
        f: [10, 20, 30],
    },
}

const nodes: Node[] = []
const walkFn = (node: Node) => {
    nodes.push(node)
}
walker(obj, walkFn, options)
nodes

Returns an array of nodes. Note this is how walk works, so prefer that fn.

[
  {
    key: undefined,
    parents: [],
    val: { a: { b: 23, c: 24 }, d: { e: 'Bob', f: [10, 20, 30] } },
    path: [],
    isRoot: true,
    isLeaf: false,
  },
  {
    key: 'a',
    val: { b: 23, c: 24 },
    parents: [{ a: { b: 23, c: 24 }, d: { e: 'Bob', f: [10, 20, 30] } }],
    path: ['a'],
    isLeaf: false,
    isRoot: false,
  },
  {
    key: 'b',
    val: 23,
    parents: [
      { b: 23, c: 24 },
      { a: { b: 23, c: 24 }, d: { e: 'Bob', f: [10, 20, 30] } },
    ],
    path: ['a', 'b'],
    isLeaf: true,
    isRoot: false,
  },
  ...
]

walk

walk(obj: object, options: WalkOptions = {}) => Node[]

Walk an object. Returns an array of all nodes in the object in either preorder or postorder.

export interface WalkOptions extends Options {
    leavesOnly?: boolean
}

import { walk } from 'obj-walker'

const obj = {
    a: {
        b: 23,
        c: 24,
    },
    d: {
        e: 'Bob',
        f: [10, 20, 30],
    },
}
walk(obj).map((x) => x.path)

Produces:

[
    [],
    ["a"],
    ["a", "b"],
    ["a", "c"],
    ["d"],
    ["d", "e"],
    ["d", "f"],
    ["d", "f", "0"],
    ["d", "f", "1"],
    ["d", "f", "2"]
]

walkEach

walkEach(obj: object, walkFn: WalkFn, options: options?: WalkOptions & MutationOption) => object

export type WalkFn = (node: Node) => void

Walk over an object calling walkFn for each node. The original object is deep-cloned by default making it possible to simply mutate each node as needed in order to transform the object. The cloned object is returned if options.modifyInPlace is not set to true.

Below I want to walk a MongoDB JSON schema and set additionalProperties to true wherever it exists. I traverse this tree using a custom traverse fn. The original object is not modified.

import { walkEach } from 'obj-walker'

const obj = {
    bsonType: 'object',
    additionalProperties: false,
    required: ['name'],
    properties: {
        _id: {
            bsonType: 'objectId',
        },
        name: { bsonType: 'string' },
        addresses: {
            bsonType: 'array',
            items: {
                bsonType: 'object',
                additionalProperties: false,
                properties: {
                    address: {
                        bsonType: 'object',
                        additionalProperties: false,
                        properties: {
                            zip: { bsonType: 'string' },
                            country: { bsonType: 'string' },
                        },
                    },
                },
            },
        },
    },
}

const traverse = (x: any) => x.properties || (x.items && { items: x.items })
const walkFn = ({ val }: Node) => {
    if ('additionalProperties' in val) {
        val.additionalProperties = true
    }
}
walkEach(obj, walkFn, { traverse })

Produces:

{
  bsonType: 'object',
  additionalProperties: true,
  required: ['name'],
  properties: {
    _id: { bsonType: 'objectId' },
    name: { bsonType: 'string' },
    addresses: {
      bsonType: 'array',
      items: {
        bsonType: 'object',
        additionalProperties: true,
        properties: {
          address: {
            bsonType: 'object',
            additionalProperties: true,
            properties: {
              zip: { bsonType: 'string' },
              country: { bsonType: 'string' },
            },
          },
        },
      },
    },
  },
}

walkEachAsync

Like walkEach but awaits the promise returned by walkFn before proceeding to the next node.

map

Map over an object modifying values with a fn depth-first in a preorder or postorder manner. The output of the mapper fn will be traversed if possible when traversing preorder.

By default, nodes will be excluded by returning undefined. Undefined array values will not be excluded. To customize pass a fn for options.shouldSkip.

map(obj: object, mapper: Mapper, options?: MapOptions & MutationOption) => object

export type Mapper = (node: Node) => any

export type MapOptions = Omit<Options, 'traverse'> & {
    shouldSkip?(val: any, node: Node): boolean
}

import { map } from 'obj-walker'

const obj = {
    a: {
        b: 23,
        c: 24,
    },
    d: {
        e: 'Bob',
        f: [10, null, 30, [31, undefined, 32], 40],
    },
    g: [25, '', { h: [null, 26, 27] }],
    i: 'Frank',
}
map(obj, ({ val }) => (Array.isArray(val) ? _.compact(val) : val))

Produces:

{
  a: { b: 23, c: 24 },
  d: { e: 'Bob', f: [10, 30, [31, 32], 40] },
  g: [25, { h: [26, 27] }],
  i: 'Frank',
}

Postorder

const obj = {
    bob: {
        scores: ['87', 'x97', 95, false],
    },
    joe: {
        scores: [92, 92.5, '73.2', ''],
    },
    frank: {
        scores: ['abc', ''],
    },
}
const result = map(
    obj,
    ({ val, isLeaf }) => {
        if (isLeaf) {
            return parseFloat(val)
        }
        return Array.isArray(val) ? _.compact(val) : val
    },
    { postOrder: true }
)

Produces:

{
  bob: { scores: [87, 95] },
  joe: { scores: [92, 92.5, 73.2] },
  frank: { scores: [] },
}

Custom shouldSkip fn

const obj = {
    bob: {
        scores: ['87', 'x97', 95, false],
    },
    joe: {
        scores: [92, 92.5, '73.2', ''],
    },
    frank: {
        scores: ['abc', ''],
    },
}

const shouldSkip = (val: any, node: Node) =>
    _.isEmpty(val) && !parentIsArray(node)
const result = map(
    obj,
    ({ val, isLeaf }) => {
        if (isLeaf) {
            return parseFloat(val)
        }
        return Array.isArray(val) ? _.compact(val) : val
    },
    { postOrder: true, shouldSkip }
)

Produces:

{
  bob: { scores: [87, 95] },
  joe: { scores: [92, 92.5, 73.2] },
}

mapLeaves

mapLeaves(obj: object, mapper: Mapper, options?: MapOptions) => object

Map over the leaves of an object with a fn. By default, nodes will be excluded by returning undefined. Undefined array values will not be excluded. To customize pass a fn for options.shouldSkip.

import { mapLeaves } from 'obj-walker'

const obj = {
    a: {
        b: 23,
        c: 24,
    },
    d: {
        e: 100,
        f: [10, 20, 30],
    },
}
mapLeaves(obj, ({ val }) => val + 1)

Produces:

{
  a: { b: 24, c: 25 },
  d: { e: 101, f: [11, 21, 31] },
}

findNode

findNode(obj: object, findFn: FindFn, options?: Options) => Node | undefined

Search for a node and short-circuit the tree traversal if it's found.

import { findNode } from 'obj-walker'

const obj = {
    name: 'Joe',
    address: {
        city: 'New York',
        state: 'NY',
        zipCode: '10001',
    },
    likes: ['Stock Market', 'Running'],
}

findNode(obj, (node) => {
    return _.isEqual(node.path, ['address', 'zipCode'])
})

Produces:

{
  key: 'zipCode',
  val: '10001',
  parents: [
    { city: 'New York', state: 'NY', zipCode: '10001' },
    {
      name: 'Joe',
      address: { city: 'New York', state: 'NY', zipCode: '10001' },
      likes: ['Stock Market', 'Running'],
    },
  ],
  path: ['address', 'zipCode'],
  isLeaf: true,
  isRoot: false,
}

flatten

flatten(obj: object, options?: WalkOptions & FlattenOptions) => object

interface FlattenOptions {
    /** Defaults to '.' */
    separator?: string
    /** Flatten objects and not arrays */
    objectsOnly?: boolean
}

Flatten an object's keys. Optionally pass separator to determine what character to join keys with. Defaults to '.'. If an array is passed, an object of path to values is returned unless the objectsOnly option is set.

import { flatten } from 'obj-walker'

const obj = {
    a: {
        b: 23,
        c: 24,
    },
    d: {
        e: 100,
        f: [10, 20, 30],
    },
}
flatten(obj)

Produces:

{
  'a.b': 23,
  'a.c': 24,
  'd.e': 100,
  'd.f.0': 10,
  'd.f.1': 20,
  'd.f.2': 30,
}

unflatten

unflatten(obj: object, options?: UnflattenOptions) => object

interface UnflattenOptions {
    /** Defaults to '.' */
    separator?: string | RegExp
}

Unflatten an object previously flattened. Optionally pass separator to determine what character or RegExp to split keys with. Defaults to '.'.

import { unflatten } from 'obj-walker'

const obj = {
    'a.b': 23,
    'a.c': 24,
    'd.e': 100,
    'd.f.0': 10,
    'd.f.1': 20,
    'd.f.2.g': 30,
    'd.f.2.h.i': 40,
}
unflatten(obj)

Produces:

{
  a: {
    b: 23,
    c: 24,
  },
  d: {
    e: 100,
    f: [10, 20, { g: 30, h: { i: 40 } }],
  },
}

compact

compact(obj: object, options: CompactOptions & MutationOption) => object

interface CompactOptions {
    removeUndefined?: boolean
    removeNull?: boolean
    removeEmptyString?: boolean
    removeFalse?: boolean
    removeNaN?: boolean
    removeEmptyObject?: boolean
    removeEmptyArray?: boolean
    compactArrays?: boolean
    removeFn?: (val: any, node: Node) => boolean
}

Compact an object, removing fields recursively according to the supplied options. All option flags are false by default. If compactArrays is set to true, arrays will be compacted based on the enabled remove option flags.

const obj = {
    a: {
        b: [null, null, 21, '', { b1: null }, { b2: 26 }],
    },
    c: [],
    d: [42, null],
    e: {
        f: {
            g: '',
            h: undefined,
            i: 'null',
        },
    },
}
const result = compact(obj, {
    removeUndefined: true,
    removeEmptyString: true,
    removeNull: true,
    compactArrays: true,
    removeEmptyArray: true,
    removeEmptyObject: true,
    removeFn: (val: any) => val === 'null',
})

Produces:

{
  a: { b: [21, { b2: 26 }] },
  d: [42],
}

truncate

truncate(obj: object, options: TruncateOptions & MutationOption) => object

export interface TruncateOptions {
    /** Max allowed depth of objects/arrays. Default to Infinity */
    maxDepth?: number
    /** What to replace an object/array at the maximum depth with. Defaults to '[Truncated]' */
    replacementAtMaxDepth?: any
    /** Max allowed length of a string. Defaults to Infinity */
    maxStringLength?: number
    /** What to replace the last characters of the truncated string with. Defaults to '...' */
    replacementAtMaxStringLength?: string
    /** Max allowed length of an array. Defaults to Infinity */
    maxArrayLength?: number
    /** Transform instances of Error into plain objects so that truncation can be performed. Defautls to false */
    transformErrors?: boolean
}

Truncate allows you to limit the depth of nested objects/arrays, the length of strings, and the length of arrays. Instances of Error can be converted to plain objects so that the enabled truncation options also apply to the error fields. All truncation methods are opt-in.

Note: For the best performance you should consider setting modifyInPlace to true.

const obj = {
    a: {
        b: 'Frank',
        c: {
            d: 'Joe',
        },
        e: null,
    },
    f: 42,
}
truncate(obj, { depth: 2 })

Produces:

{
  a: {
    b: 'Frank',
    c: '[Truncated]',
    e: null,
  },
  f: 42,
}

size

size(val: any) => number

Estimate the size in bytes.

const obj = {
    a: {
        b: 'hello',
    },
    c: Symbol('hello'),
    d: {
        e: [true, false],
    },
    f: [42, 10n],
}
size(obj)
// 44

Helper fns

These helper fns are exported for your convenience.

export const isObjectOrArray = _.overSome([_.isPlainObject, _.isArray])

export const defShouldSkip = (val: any, node: Node) =>
    val === undefined && !parentIsArray(node)

export const parentIsArray = (node: Node) => {
    const parent = node.parents[0]
    return Array.isArray(parent)
}

export const defTraverse = (x: any) => isObjectOrArray(x) && !_.isEmpty(x) && x

addRefs

addRefs(obj: object, options?: RefOptions): object

Replace duplicate objects refs with pointers to the first object seen.

import { addRefs } from 'obj-walker'

const apiOutput = {
    1: 'foo',
    2: 'bar',
    3: 'baz',
}

const detailsOutput = {
    1: 'bla',
    2: 'bla',
    3: 'bla',
}

const obj = {
    api: {
        input: [1, 2, 3],
        output: apiOutput,
    },
    details: {
        input: apiOutput,
        output: detailsOutput,
    },
    writeToDB: {
        input: detailsOutput,
    },
}
addRefs(obj)

Produces:

{
  api: {
    input: [1, 2, 3],
    output: { '1': 'foo', '2': 'bar', '3': 'baz' },
  },
  details: {
    input: { $ref: '#/api/output' },
    output: { '1': 'bla', '2': 'bla', '3': 'bla' },
  },
  writeToDB: { input: { $ref: '#/details/output' } },
}

deref

deref(obj: object, options?: RefOptions): object

Rehydrate objects by replacing refs with actual objects.

import { deref } from 'obj-walker'

const obj = {
    api: {
        input: [1, 2, 3],
        output: { '1': 'foo', '2': 'bar', '3': 'baz' },
    },
    details: {
        input: { $ref: '#/api/output' },
        output: { '1': 'bla', '2': 'bla', '3': 'bla' },
    },
    writeToDB: { input: { $ref: '#/details/output' } },
}
deref(obj)

Produces:

{
  api: {
    input: [1, 2, 3],
    output: { '1': 'foo', '2': 'bar', '3': 'baz' },
  },
  details: {
    input: { '1': 'foo', '2': 'bar', '3': 'baz' },
    output: { '1': 'bla', '2': 'bla', '3': 'bla' },
  },
  writeToDB: { input: { '1': 'bla', '2': 'bla', '3': 'bla' } },
}

2.4.0

• truncate - Handle top-level Error.

2.3.0

• truncate - Added the ability to pass a function for the replacementAtMaxStringLength option.
• Bumped deps and migrated to Vite.

2.2.0

• Added unflatten to compliment flatten.
• Deprecated walkie. Use walkEach.
• Deprecated walkieAsync. Use walkEachAsync.

2.1.0

• Added handling of top-level scalar values to size.

2.0.0

• BREAKING CHANGE: Changed API for truncate in order to make it more flexible.
• Added size for estimating the size in bytes of an object.

1.10.0

• Added removeFn to compact for arbitrary removal.

1.9.0

• Added objectsOnly option to flatten.

1.8.0

• Added modifyInPlace option for map, walkie, walkieAsync, mapLeaves, compact, and truncate.
• truncate supports Error objects and can truncate strings and arrays based on the stringLength and arrayLength options.
• walker now supports short-circuiting by returning the exported symbol SHORT_CIRCUIT.

1.7.0

• Changed truncate options from maxDepth back to depth since console.dir uses this name. Also, you should probably prefer dtrim.

1.6.0

• Added walkieAsync for walking async.

1.5.0

• Changed truncate options from depth to maxDepth and allow replaceWith to be anything.

1.4.0

• Added truncate for truncating deep objects.

1.3.0

• Make options required for compact.

1.2.0

• Added compact for cleaning up objects/arrays.

1.1.1

• Apply mapping fn to the root node for preorder and postorder map.

1.1.0

• Factor out common code.

1.0.0

• Remove traverse option from all mapping fns.
• findNode fn.
• flatten fn.

0.0.9

• Custom shouldSkip fn for map and mapLeaves.

0.0.8

• Deep clone object for map. Allow postOrder option.

0.0.7

• map iterates over output.
• Added walkie.
• Removed mapKV.

0.0.6

• Generic walker fn.

0.0.5

• Renamed argument.

0.0.4

• Fixed traverse type.

0.0.3

• Fixed traverse logic.

0.0.2

• Fixed bug regarding first-level leaves for addRefs.

0.0.1

• Initial release.