Related
This question already has answers here:
How can I get query string values in JavaScript?
(73 answers)
Closed 1 year ago.
I have a string like this:
abc=foo&def=%5Basf%5D&xyz=5
How can I convert it into a JavaScript object like this?
{
abc: 'foo',
def: '[asf]',
xyz: 5
}
In the year 2021... Please consider this obsolete.
Edit
This edit improves and explains the answer based on the comments.
var search = location.search.substring(1);
JSON.parse('{"' + decodeURI(search).replace(/"/g, '\\"').replace(/&/g, '","').replace(/=/g,'":"') + '"}')
Example
Parse abc=foo&def=%5Basf%5D&xyz=5 in five steps:
decodeURI: abc=foo&def=[asf]&xyz=5
Escape quotes: same, as there are no quotes
Replace &: abc=foo","def=[asf]","xyz=5
Replace =: abc":"foo","def":"[asf]","xyz":"5
Suround with curlies and quotes: {"abc":"foo","def":"[asf]","xyz":"5"}
which is legal JSON.
An improved solution allows for more characters in the search string. It uses a reviver function for URI decoding:
var search = location.search.substring(1);
JSON.parse('{"' + search.replace(/&/g, '","').replace(/=/g,'":"') + '"}', function(key, value) { return key===""?value:decodeURIComponent(value) })
Example
search = "abc=foo&def=%5Basf%5D&xyz=5&foo=b%3Dar";
gives
Object {abc: "foo", def: "[asf]", xyz: "5", foo: "b=ar"}
Original answer
A one-liner:
JSON.parse('{"' + decodeURI("abc=foo&def=%5Basf%5D&xyz=5".replace(/&/g, "\",\"").replace(/=/g,"\":\"")) + '"}')
2022 ES6/7/8 and on approach
Starting ES6 and on, Javascript offers several constructs in order to create a performant solution for this issue.
This includes using URLSearchParams and iterators
let params = new URLSearchParams('abc=foo&def=%5Basf%5D&xyz=5');
params.get("abc"); // "foo"
Should your use case requires you to actually convert it to object, you can implement the following function:
function paramsToObject(entries) {
const result = {}
for(const [key, value] of entries) { // each 'entry' is a [key, value] tupple
result[key] = value;
}
return result;
}
Basic Demo
const urlParams = new URLSearchParams('abc=foo&def=%5Basf%5D&xyz=5');
const entries = urlParams.entries(); //returns an iterator of decoded [key,value] tuples
const params = paramsToObject(entries); //{abc:"foo",def:"[asf]",xyz:"5"}
Using Object.fromEntries and spread
We can use Object.fromEntries, replacing paramsToObject with Object.fromEntries(entries).
The value pairs to iterate over are the list name-value pairs with the
key being the name and the value being the value.
Since URLParams, returns an iterable object, using the spread operator instead of calling .entries will also yield entries per its spec:
const urlParams = new URLSearchParams('abc=foo&def=%5Basf%5D&xyz=5');
const params = Object.fromEntries(urlParams); // {abc: "foo", def: "[asf]", xyz: "5"}
Note: All values are automatically strings as per the URLSearchParams spec
Multiple same keys
As #siipe pointed out, strings containing multiple same-key values will be coerced into the last available value: foo=first_value&foo=second_value will in essence become: {foo: "second_value"}.
As per this answer: https://stackoverflow.com/a/1746566/1194694 there's no spec for deciding what to do with it and each framework can behave differently.
A common use case will be to join the two same values into an array, making the output object into:
{foo: ["first_value", "second_value"]}
This can be achieved with the following code:
const groupParamsByKey = (params) => [...params.entries()].reduce((acc, tuple) => {
// getting the key and value from each tuple
const [key, val] = tuple;
if(acc.hasOwnProperty(key)) {
// if the current key is already an array, we'll add the value to it
if(Array.isArray(acc[key])) {
acc[key] = [...acc[key], val]
} else {
// if it's not an array, but contains a value, we'll convert it into an array
// and add the current value to it
acc[key] = [acc[key], val];
}
} else {
// plain assignment if no special case is present
acc[key] = val;
}
return acc;
}, {});
const params = new URLSearchParams('abc=foo&def=%5Basf%5D&xyz=5&def=dude');
const output = groupParamsByKey(params) // {abc: "foo", def: ["[asf]", "dude"], xyz: 5}
One liner. Clean and simple.
const params = Object.fromEntries(new URLSearchParams(location.search));
For your specific case, it would be:
const str = 'abc=foo&def=%5Basf%5D&xyz=5';
const params = Object.fromEntries(new URLSearchParams(str));
console.log(params);
2023 One-Liner Approach
For the general case where you want to parse query params to an object:
Object.fromEntries(new URLSearchParams(location.search));
For your specific case:
Object.fromEntries(new URLSearchParams('abc=foo&def=%5Basf%5D&xyz=5'));
Split on & to get name/value pairs, then split each pair on =. Here's an example:
var str = "abc=foo&def=%5Basf%5D&xy%5Bz=5"
var obj = str.split("&").reduce(function(prev, curr, i, arr) {
var p = curr.split("=");
prev[decodeURIComponent(p[0])] = decodeURIComponent(p[1]);
return prev;
}, {});
Another approach, using regular expressions:
var obj = {};
str.replace(/([^=&]+)=([^&]*)/g, function(m, key, value) {
obj[decodeURIComponent(key)] = decodeURIComponent(value);
});
This is adapted from John Resig's "Search and Don’t Replace".
The proposed solutions I found so far do not cover more complex scenarios.
I needed to convert a query string like
https://random.url.com?Target=Offer&Method=findAll&filters%5Bhas_goals_enabled%5D%5BTRUE%5D=1&filters%5Bstatus%5D=active&fields%5B%5D=id&fields%5B%5D=name&fields%5B%5D=default_goal_name
into an object like:
{
"Target": "Offer",
"Method": "findAll",
"fields": [
"id",
"name",
"default_goal_name"
],
"filters": {
"has_goals_enabled": {
"TRUE": "1"
},
"status": "active"
}
}
OR:
https://random.url.com?Target=Report&Method=getStats&fields%5B%5D=Offer.name&fields%5B%5D=Advertiser.company&fields%5B%5D=Stat.clicks&fields%5B%5D=Stat.conversions&fields%5B%5D=Stat.cpa&fields%5B%5D=Stat.payout&fields%5B%5D=Stat.date&fields%5B%5D=Stat.offer_id&fields%5B%5D=Affiliate.company&groups%5B%5D=Stat.offer_id&groups%5B%5D=Stat.date&filters%5BStat.affiliate_id%5D%5Bconditional%5D=EQUAL_TO&filters%5BStat.affiliate_id%5D%5Bvalues%5D=1831&limit=9999
INTO:
{
"Target": "Report",
"Method": "getStats",
"fields": [
"Offer.name",
"Advertiser.company",
"Stat.clicks",
"Stat.conversions",
"Stat.cpa",
"Stat.payout",
"Stat.date",
"Stat.offer_id",
"Affiliate.company"
],
"groups": [
"Stat.offer_id",
"Stat.date"
],
"limit": "9999",
"filters": {
"Stat.affiliate_id": {
"conditional": "EQUAL_TO",
"values": "1831"
}
}
}
I compiled and adapted multiple solutions into one that actually works:
CODE:
var getParamsAsObject = function (query) {
query = query.substring(query.indexOf('?') + 1);
var re = /([^&=]+)=?([^&]*)/g;
var decodeRE = /\+/g;
var decode = function (str) {
return decodeURIComponent(str.replace(decodeRE, " "));
};
var params = {}, e;
while (e = re.exec(query)) {
var k = decode(e[1]), v = decode(e[2]);
if (k.substring(k.length - 2) === '[]') {
k = k.substring(0, k.length - 2);
(params[k] || (params[k] = [])).push(v);
}
else params[k] = v;
}
var assign = function (obj, keyPath, value) {
var lastKeyIndex = keyPath.length - 1;
for (var i = 0; i < lastKeyIndex; ++i) {
var key = keyPath[i];
if (!(key in obj))
obj[key] = {}
obj = obj[key];
}
obj[keyPath[lastKeyIndex]] = value;
}
for (var prop in params) {
var structure = prop.split('[');
if (structure.length > 1) {
var levels = [];
structure.forEach(function (item, i) {
var key = item.replace(/[?[\]\\ ]/g, '');
levels.push(key);
});
assign(params, levels, params[prop]);
delete(params[prop]);
}
}
return params;
};
A concise solution:
location.search
.slice(1)
.split('&')
.map(p => p.split('='))
.reduce((obj, pair) => {
const [key, value] = pair.map(decodeURIComponent);
obj[key] = value;
return obj;
}, {});
This is the simple version, obviously you'll want to add some error checking:
var obj = {};
var pairs = queryString.split('&');
for(i in pairs){
var split = pairs[i].split('=');
obj[decodeURIComponent(split[0])] = decodeURIComponent(split[1]);
}
For Node JS, you can use the Node JS API querystring:
const querystring = require('querystring');
querystring.parse('abc=foo&def=%5Basf%5D&xyz=5&foo=b%3Dar');
// returns the object
Documentation: https://nodejs.org/api/querystring.html
I found $.String.deparam the most complete pre built solution (can do nested objects etc.). Check out the documentation.
Another solution based on the latest standard of URLSearchParams (https://developer.mozilla.org/en-US/docs/Web/API/URLSearchParams)
function getQueryParamsObject() {
const searchParams = new URLSearchParams(location.search.slice(1));
return searchParams
? _.fromPairs(Array.from(searchParams.entries()))
: {};
}
Please note that this solution is making use of
Array.from (https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/from)
and _.fromPairs (https://lodash.com/docs#fromPairs) of lodash for the sake of simplicity.
It should be easy to create a more compatible solution since you have access to searchParams.entries() iterator.
I had the same problem, tried the solutions here, but none of them really worked, since I had arrays in the URL parameters, like this:
?param[]=5¶m[]=8&othr_param=abc¶m[]=string
So I ended up writing my own JS function, which makes an array out of the param in URI:
/**
* Creates an object from URL encoded data
*/
var createObjFromURI = function() {
var uri = decodeURI(location.search.substr(1));
var chunks = uri.split('&');
var params = Object();
for (var i=0; i < chunks.length ; i++) {
var chunk = chunks[i].split('=');
if(chunk[0].search("\\[\\]") !== -1) {
if( typeof params[chunk[0]] === 'undefined' ) {
params[chunk[0]] = [chunk[1]];
} else {
params[chunk[0]].push(chunk[1]);
}
} else {
params[chunk[0]] = chunk[1];
}
}
return params;
}
One of the simplest way to do this using URLSearchParam interface.
Below is the working code snippet:
let paramObj={},
querystring=window.location.search,
searchParams = new URLSearchParams(querystring);
//*** :loop to add key and values to the param object.
searchParams.forEach(function(value, key) {
paramObj[key] = value;
});
There is quite simple and incorrect answer with ES6:
console.log(
Object.fromEntries(new URLSearchParams(`abc=foo&def=%5Basf%5D&xyz=5`))
);
But this one line code do not cover multiple same keys, you have to use something more complicated:
function parseParams(params) {
const output = [];
const searchParams = new URLSearchParams(params);
// Set will return only unique keys()
new Set([...searchParams.keys()])
.forEach(key => {
output[key] = searchParams.getAll(key).length > 1 ?
searchParams.getAll(key) : // get multiple values
searchParams.get(key); // get single value
});
return output;
}
console.log(
parseParams('abc=foo&cars=Ford&cars=BMW&cars=Skoda&cars=Mercedes')
)
Code will generate follow structure:
[
abc: "foo"
cars: ["Ford", "BMW", "Skoda", "Mercedes"]
]
Using ES6, URL API and URLSearchParams API.
function objectifyQueryString(url) {
let _url = new URL(url);
let _params = new URLSearchParams(_url.search);
let query = Array.from(_params.keys()).reduce((sum, value)=>{
return Object.assign({[value]: _params.get(value)}, sum);
}, {});
return query;
}
ES6 one liner (if we can call it that way seeing the long line)
[...new URLSearchParams(location.search).entries()].reduce((prev, [key,val]) => {prev[key] = val; return prev}, {})
One simple answer with build in native Node module.(No third party npm modules)
The querystring module provides utilities for parsing and formatting URL query strings. It can be accessed using:
const querystring = require('querystring');
const body = "abc=foo&def=%5Basf%5D&xyz=5"
const parseJSON = querystring.parse(body);
console.log(parseJSON);
Pretty easy using the URLSearchParams JavaScript Web API,
var paramsString = "abc=foo&def=%5Basf%5D&xyz=5";
//returns an iterator object
var searchParams = new URLSearchParams(paramsString);
//Usage
for (let p of searchParams) {
console.log(p);
}
//Get the query strings
console.log(searchParams.toString());
//You can also pass in objects
var paramsObject = {abc:"forum",def:"%5Basf%5D",xyz:"5"}
//returns an iterator object
var searchParams = new URLSearchParams(paramsObject);
//Usage
for (let p of searchParams) {
console.log(p);
}
//Get the query strings
console.log(searchParams.toString());
##Useful Links
URLSearchParams - Web APIs | MDN
Easy URL Manipulation with URLSearchParams | Web
| Google Developers
NOTE: Not Supported in IE
There is no native solution that I'm aware of. Dojo has a built-in unserialization method if you use that framework by chance.
Otherwise you can implement it yourself rather simply:
function unserialize(str) {
str = decodeURIComponent(str);
var chunks = str.split('&'),
obj = {};
for(var c=0; c < chunks.length; c++) {
var split = chunks[c].split('=', 2);
obj[split[0]] = split[1];
}
return obj;
}
edit: added decodeURIComponent()
/**
* Parses and builds Object of URL query string.
* #param {string} query The URL query string.
* #return {!Object<string, string>}
*/
function parseQueryString(query) {
if (!query) {
return {};
}
return (/^[?#]/.test(query) ? query.slice(1) : query)
.split('&')
.reduce((params, param) => {
const item = param.split('=');
const key = decodeURIComponent(item[0] || '');
const value = decodeURIComponent(item[1] || '');
if (key) {
params[key] = value;
}
return params;
}, {});
}
console.log(parseQueryString('?v=MFa9pvnVe0w&ku=user&from=89&aw=1'))
see log
There's a lightweight library called YouAreI.js that's tested and makes this really easy.
YouAreI = require('YouAreI')
uri = new YouAreI('http://user:pass#www.example.com:3000/a/b/c?d=dad&e=1&f=12.3#fragment');
uri.query_get() => { d: 'dad', e: '1', f: '12.3' }
If you are using URI.js, you can use:
https://medialize.github.io/URI.js/docs.html#static-parseQuery
var result = URI.parseQuery("?foo=bar&hello=world&hello=mars&bam=&yup");
result === {
foo: "bar",
hello: ["world", "mars"],
bam: "",
yup: null
};
console.log(decodeURI('abc=foo&def=%5Basf%5D&xyz=5')
.split('&')
.reduce((result, current) => {
const [key, value] = current.split('=');
result[key] = value;
return result
}, {}))
This seems to be the best solution as it takes multiple parameters of the same name into consideration.
function paramsToJSON(str) {
var pairs = str.split('&');
var result = {};
pairs.forEach(function(pair) {
pair = pair.split('=');
var name = pair[0]
var value = pair[1]
if( name.length )
if (result[name] !== undefined) {
if (!result[name].push) {
result[name] = [result[name]];
}
result[name].push(value || '');
} else {
result[name] = value || '';
}
});
return( result );
}
something
paramsToJSON("x=1&x=2&x=3&y=blah");
console yields => {x: Array[3], y: "blah"} where x is an array as is proper JSON
I later decided to convert it to a jQuery plugin too...
$.fn.serializeURLParams = function() {
var result = {};
if( !this.is("a") || this.attr("href").indexOf("?") == -1 )
return( result );
var pairs = this.attr("href").split("?")[1].split('&');
pairs.forEach(function(pair) {
pair = pair.split('=');
var name = decodeURI(pair[0])
var value = decodeURI(pair[1])
if( name.length )
if (result[name] !== undefined) {
if (!result[name].push) {
result[name] = [result[name]];
}
result[name].push(value || '');
} else {
result[name] = value || '';
}
});
return( result )
}
something
$("a").serializeURLParams();
console yields => {x: Array[3], y: "blah"} where x is an array as is proper JSON
Now, the first will accept the parameters only but the jQuery plugin will take the whole url and return the serialized parameters.
Here's one I use:
var params = {};
window.location.search.substring(1).split('&').forEach(function(pair) {
pair = pair.split('=');
if (pair[1] !== undefined) {
var key = decodeURIComponent(pair[0]),
val = decodeURIComponent(pair[1]),
val = val ? val.replace(/\++/g,' ').trim() : '';
if (key.length === 0) {
return;
}
if (params[key] === undefined) {
params[key] = val;
}
else {
if ("function" !== typeof params[key].push) {
params[key] = [params[key]];
}
params[key].push(val);
}
}
});
console.log(params);
Basic usage, eg.
?a=aa&b=bb
Object {a: "aa", b: "bb"}
Duplicate params, eg.
?a=aa&b=bb&c=cc&c=potato
Object {a: "aa", b: "bb", c: ["cc","potato"]}
Missing keys, eg.
?a=aa&b=bb&=cc
Object {a: "aa", b: "bb"}
Missing values, eg.
?a=aa&b=bb&c
Object {a: "aa", b: "bb"}
The above JSON/regex solutions throw a syntax error on this wacky url:
?a=aa&b=bb&c=&=dd&e
Object {a: "aa", b: "bb", c: ""}
Here's my quick and dirty version, basically its splitting up the URL parameters separated by '&' into array elements, and then iterates over that array adding key/value pairs separated by '=' into an object. I'm using decodeURIComponent() to translate the encoded characters to their normal string equivalents (so %20 becomes a space, %26 becomes '&', etc):
function deparam(paramStr) {
let paramArr = paramStr.split('&');
let paramObj = {};
paramArr.forEach(e=>{
let param = e.split('=');
paramObj[param[0]] = decodeURIComponent(param[1]);
});
return paramObj;
}
example:
deparam('abc=foo&def=%5Basf%5D&xyz=5')
returns
{
abc: "foo"
def:"[asf]"
xyz :"5"
}
The only issue is that xyz is a string and not a number (due to using decodeURIComponent()), but beyond that its not a bad starting point.
//under ES6
const getUrlParamAsObject = (url = window.location.href) => {
let searchParams = url.split('?')[1];
const result = {};
//in case the queryString is empty
if (searchParams!==undefined) {
const paramParts = searchParams.split('&');
for(let part of paramParts) {
let paramValuePair = part.split('=');
//exclude the case when the param has no value
if(paramValuePair.length===2) {
result[paramValuePair[0]] = decodeURIComponent(paramValuePair[1]);
}
}
}
return result;
}
If you need recursion, you can use the tiny js-extension-ling library.
npm i js-extension-ling
const jsx = require("js-extension-ling");
console.log(jsx.queryStringToObject("a=1"));
console.log(jsx.queryStringToObject("a=1&a=3"));
console.log(jsx.queryStringToObject("a[]=1"));
console.log(jsx.queryStringToObject("a[]=1&a[]=pomme"));
console.log(jsx.queryStringToObject("a[0]=one&a[1]=five"));
console.log(jsx.queryStringToObject("http://blabla?foo=bar&number=1234"));
console.log(jsx.queryStringToObject("a[fruits][red][]=strawberry"));
console.log(jsx.queryStringToObject("a[fruits][red][]=strawberry&a[1]=five&a[fruits][red][]=cherry&a[fruits][yellow][]=lemon&a[fruits][yellow][688]=banana"));
This will output something like this:
{ a: '1' }
{ a: '3' }
{ a: { '0': '1' } }
{ a: { '0': '1', '1': 'pomme' } }
{ a: { '0': 'one', '1': 'five' } }
{ foo: 'bar', number: '1234' }
{
a: { fruits: { red: { '0': 'strawberry' } } }
}
{
a: {
'1': 'five',
fruits: {
red: { '0': 'strawberry', '1': 'cherry' },
yellow: { '0': 'lemon', '688': 'banana' }
}
}
}
Note: it's based on locutus parse_str function (https://locutus.io/php/strings/parse_str/).
FIRST U NEED TO DEFINE WHAT'S A GET VAR:
function getVar()
{
this.length = 0;
this.keys = [];
this.push = function(key, value)
{
if(key=="") key = this.length++;
this[key] = value;
this.keys.push(key);
return this[key];
}
}
Than just read:
function urlElement()
{
var thisPrototype = window.location;
for(var prototypeI in thisPrototype) this[prototypeI] = thisPrototype[prototypeI];
this.Variables = new getVar();
if(!this.search) return this;
var variables = this.search.replace(/\?/g,'').split('&');
for(var varI=0; varI<variables.length; varI++)
{
var nameval = variables[varI].split('=');
var name = nameval[0].replace(/\]/g,'').split('[');
var pVariable = this.Variables;
for(var nameI=0;nameI<name.length;nameI++)
{
if(name.length-1==nameI) pVariable.push(name[nameI],nameval[1]);
else var pVariable = (typeof pVariable[name[nameI]] != 'object')? pVariable.push(name[nameI],new getVar()) : pVariable[name[nameI]];
}
}
}
and use like:
var mlocation = new urlElement();
mlocation = mlocation.Variables;
for(var key=0;key<mlocation.keys.length;key++)
{
console.log(key);
console.log(mlocation[mlocation.keys[key]];
}
I needed to also deal with + in the query part of the URL (decodeURIComponent doesn't), so I adapted Wolfgang's code to become:
var search = location.search.substring(1);
search = search?JSON.parse('{"' + search.replace(/\+/g, ' ').replace(/&/g, '","').replace(/=/g,'":"') + '"}',
function(key, value) { return key===""?value:decodeURIComponent(value)}):{};
In my case, I'm using jQuery to get URL-ready form parameters, then this trick to build an object out of it and I can then easily update parameters on the object and rebuild the query URL, e.g.:
var objForm = JSON.parse('{"' + $myForm.serialize().replace(/\+/g, ' ').replace(/&/g, '","').replace(/=/g,'":"') + '"}',
function(key, value) { return key===""?value:decodeURIComponent(value)});
objForm.anyParam += stringToAddToTheParam;
var serializedForm = $.param(objForm);
I have a nested object, something similar to this:
var obj = {
"prop1": {
"prop1A": "A",
"prop1B": {
"prop1BA": "BA"
},
"prop1C": "C"
}
};
My final goal is to filter this object to specific pre defined keys, according to another schema object, for example:
var filterSchema = {
"prop1":["prop1A", {"prop1B":["prop1BA"]}]
};
(The filter keys are predefined, I can structure this object differently if you have a better idea...
The output should be an array. In our case:
["A","BA"]
I managed to do this using recursion over the object. I was wondering if there is more elegant way for achieving this (tried using jQuery's map/extend with no luck)
EDIT
I know that this is a "N" level problem which should be solved by recursion. The difference here is that I have the pre-defined filter which already have the "N" levels. So I though maybe I can filter the Objet using the filter Array and than convert it into an array.
EDIT2
Thanks you all for the different answers. This is my own solution for the problem (which I was looking for a more elegant one in the beginning):
My solution
var obj = {
"prop1": {
"prop1A": "A",
//"prop1B": {
// "prop1BA": "BA"
//},
"prop1C": "C",
"prop1D": "D",
"prop1E": {"prop1E1": "444"}
},
"prop2": "12345"
};
var schemaObj = {
"prop1": {
"prop1A": "true",
"prop1B": {
"prop1BA": "true"
},
"prop1C": "true"
},
"prop2": "true"
};
var resultsArray = [];
var keys = Object.keys(schemaObj);
for(var i=0;i<keys.length;i++){
if(obj[keys[i]]){
parser(schemaObj[keys[i]], obj[keys[i]]);
}
}
function parser(v,o){
if( typeof v === "string" ){
resultsArray.push(o);
}
else{
var keys2 = Object.keys(v);
for(var j=0;j<keys2.length;j++){
if(o[keys2[j]]){
parser(v[keys2[j]], o[keys2[j]]);
}
}
}
}
console.log(resultsArray);
Just a reminder about the question - I already had the recursion solution. I am looking for a different solution
Javascript has eval that allows you to create new code at runtime. A possible solution is to using recursion only once to create a string that looks like:
code = "[obj.prop1.prop1A, obj.prop1.prop1B.prop1BA]"
then you can create a data conversion function with
f = eval("function(obj){return " + code + "]}")
and use it with f(x) to get your array.
This is also the most efficient solution if you have to extract the data many times.
For example:
function mkFilter(schema) {
var parts = [];
function xtract(s, prefix) {
if (typeof s === "string") {
parts.push(prefix + s);
} else if (s && s.constructor === Array) {
s.forEach(function(x){ xtract(x, prefix); });
} else {
for (var f in s) {
xtract(s[f], prefix + f + ".");
}
}
}
xtract(schema, "obj.");
var code = "(function(obj){ return [" + parts.join(", ") + "]; })";
return eval(code);
}
passing schemaFilter as argument mkFilter will return a function that given an object returns the array; with your input:
console.log(mkFilter(filterSchema)(obj));
displays ['A', 'BA']. Of course this approach makes sense if you need to reuse the same filter many times with different objects.
If the object may have missing parts and you don't want the filter to fail but just undefined values in the array the code generator needs to be changed slightly:
var code = "(function(obj){ return [";
parts.forEach(function(p){
var chain = p.split(".");
var expr = "";
for (var i=0; i<chain.length-1; i++) {
expr += chain.slice(0, i+1).join(".") + " && ";
}
code += expr + p + ",";
});
code += "]; })";
This will create in your example a filter evaluating
(function(obj){
return [obj && obj.prop1 && obj.prop1.prop1A,
obj && obj.prop1 && obj.prop1.prop1B &&
obj.prop1.prop1B.prop1BA,];
})
Use jquery's map function. you can try the example code snippet in the console, but jquery must be included
a = { aa: '123', ab: 'asdasd'}
$.map(a, function(key,val){
return val;
});
// The map creates an array with the value you return from the code block.
// Output is ["aa", "ab"]
for reference see
http://api.jquery.com/jQuery.map/
The following function seems to do what you want:
function filter(obj, schema, out) {
var i, schemaItems, schemaItem, isItemLevel;
if (!obj || !schema) return;
out = out || {values: []};
isItemLevel = Array.isArray(schema);
schemaItems = isItemLevel ? schema : Object.keys(schema);
for (i = 0; i < schemaItems.length; i++) {
schemaItem = schemaItems[i];
if (isItemLevel && typeof schemaItem === "string") {
out.values.push(obj[schemaItem]);
} else if (typeof schemaItem === "object") {
filter(obj, schemaItem, out);
} else if (typeof schemaItem === "string") {
filter(obj[schemaItem], schema[schemaItem], out);
}
}
return out.values;
}
called as
var obj = {
"prop1": {
"prop1A": "A",
"prop1B": {
"prop1BA": "BA"
},
"prop1C": "C"
}
};
var filterSchema = {
"prop1":["prop1A", {"prop1B":["prop1BA"]}]
};
filter(obj, filterSchema);
returns:
["A", "BA"]
Take it with a grain of salt, it is by far not tested well-enough and I certainly don't claim it's the most elegant way of solving this.
It works like this:
traverse the items of the schema (which is either an array or an object)
for each schemaItem
if we are in an array and the schemaItem is a string, output the respective property value of obj
else if the schemaItem is itself an object, recurse, but stay at the same level in obj
else if the schemaItem is a string, recurse, drilling into both obj and schema
Given I have a circular reference in a large JavaScript object
And I try JSON.stringify(problematicObject)
And the browser throws
"TypeError: Converting circular structure to JSON"
(which is expected)
Then I want to find the cause of this circular reference, preferably using Chrome developer tools? Is this possible? How do you find and fix circular references in a large object?
Pulled from http://blog.vjeux.com/2011/javascript/cyclic-object-detection.html. One line added to detect where the cycle is. Paste this into the Chrome dev tools:
function isCyclic (obj) {
var seenObjects = [];
function detect (obj) {
if (obj && typeof obj === 'object') {
if (seenObjects.indexOf(obj) !== -1) {
return true;
}
seenObjects.push(obj);
for (var key in obj) {
if (obj.hasOwnProperty(key) && detect(obj[key])) {
console.log(obj, 'cycle at ' + key);
return true;
}
}
}
return false;
}
return detect(obj);
}
Here's the test:
> a = {}
> b = {}
> a.b = b; b.a = a;
> isCyclic(a)
Object {a: Object}
"cycle at a"
Object {b: Object}
"cycle at b"
true
#tmack's answer is definitely what I was looking for when I found this question!
Unfortunately it returns many false positives - it returns true if an object is replicated in the JSON, which isn't the same as circularity. Circularity means that an object is its own child, e.g.
obj.key1.key2.[...].keyX === obj
I modified the original answer, and this is working for me:
function isCyclic(obj) {
var keys = [];
var stack = [];
var stackSet = new Set();
var detected = false;
function detect(obj, key) {
if (obj && typeof obj != 'object') { return; }
if (stackSet.has(obj)) { // it's cyclic! Print the object and its locations.
var oldindex = stack.indexOf(obj);
var l1 = keys.join('.') + '.' + key;
var l2 = keys.slice(0, oldindex + 1).join('.');
console.log('CIRCULAR: ' + l1 + ' = ' + l2 + ' = ' + obj);
console.log(obj);
detected = true;
return;
}
keys.push(key);
stack.push(obj);
stackSet.add(obj);
for (var k in obj) { //dive on the object's children
if (Object.prototype.hasOwnProperty.call(obj, k)) { detect(obj[k], k); }
}
keys.pop();
stack.pop();
stackSet.delete(obj);
return;
}
detect(obj, 'obj');
return detected;
}
Here are a few very simple tests:
var root = {}
var leaf = {'isleaf':true};
var cycle2 = {l:leaf};
var cycle1 = {c2: cycle2, l:leaf};
cycle2.c1 = cycle1
root.leaf = leaf
isCyclic(cycle1); // returns true, logs "CIRCULAR: obj.c2.c1 = obj"
isCyclic(cycle2); // returns true, logs "CIRCULAR: obj.c1.c2 = obj"
isCyclic(leaf); // returns false
isCyclic(root); // returns false
Here is MDN's approach to detecting and fixing circular references when using JSON.stringify() on circular objects: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Errors/Cyclic_object_value :
In a circular structure like the following
var circularReference = {otherData: 123};
circularReference.myself = circularReference;
JSON.stringify() will fail:
JSON.stringify(circularReference);
// TypeError: cyclic object value
To serialize circular references you can use a library that supports them (e.g. cycle.js) or implement a solution by yourself, which will require finding and replacing (or removing) the cyclic references by serializable values.
The snippet below illustrates how to find and filter (thus causing data loss) a cyclic reference by using the replacer parameter of JSON.stringify():
const getCircularReplacer = () => {
const seen = new WeakSet();
return (key, value) => {
if (typeof value === "object" && value !== null) {
if (seen.has(value)) {
return;
}
seen.add(value);
}
return value;
};
};
JSON.stringify(circularReference, getCircularReplacer());
// {"otherData":123}
You can also use JSON.stringify with try/catch
function hasCircularDependency(obj)
{
try
{
JSON.stringify(obj);
}
catch(e)
{
return e.includes("Converting circular structure to JSON");
}
return false;
}
Demo
function hasCircularDependency(obj) {
try {
JSON.stringify(obj);
} catch (e) {
return String(e).includes("Converting circular structure to JSON");
}
return false;
}
var a = {b:{c:{d:""}}};
console.log(hasCircularDependency(a));
a.b.c.d = a;
console.log(hasCircularDependency(a));
This is a fix for both #Trey Mack and #Freddie Nfbnm answers on the typeof obj != 'object' condition. Instead it should test if the obj value is not instance of object, so that it can also work when checking values with object familiarity (for example, functions and symbols (symbols aren't instance of object, but still addressed, btw.)).
I'm posting this as an answer since I can't comment in this StackExchange account yet.
PS.: feel free to request me to delete this answer.
function isCyclic(obj) {
var keys = [];
var stack = [];
var stackSet = new Set();
var detected = false;
function detect(obj, key) {
if (!(obj instanceof Object)) { return; } // Now works with other
// kinds of object.
if (stackSet.has(obj)) { // it's cyclic! Print the object and its locations.
var oldindex = stack.indexOf(obj);
var l1 = keys.join('.') + '.' + key;
var l2 = keys.slice(0, oldindex + 1).join('.');
console.log('CIRCULAR: ' + l1 + ' = ' + l2 + ' = ' + obj);
console.log(obj);
detected = true;
return;
}
keys.push(key);
stack.push(obj);
stackSet.add(obj);
for (var k in obj) { //dive on the object's children
if (obj.hasOwnProperty(k)) { detect(obj[k], k); }
}
keys.pop();
stack.pop();
stackSet.delete(obj);
return;
}
detect(obj, 'obj');
return detected;
}
CircularReferenceDetector
Here is my CircularReferenceDetector class which outputs all the property stack information where the circularly referenced value is actually located at and also shows where the culprit references are.
This is especially useful for huge structures where it is not obvious by the key which value is the source of the harm.
It outputs the circularly referenced value stringified but all references to itself replaced by "[Circular object --- fix me]".
Usage:
CircularReferenceDetector.detectCircularReferences(value);
Note:
Remove the Logger.* statements if you do not want to use any logging or do not have a logger available.
Technical Explanation:
The recursive function goes through all properties of the object and tests if JSON.stringify succeeds on them or not.
If it does not succeed (circular reference), then it tests if it succeeds by replacing value itself with some constant string. This would mean that if it succeeds using this replacer, this value is the being circularly referenced value. If it is not, it recursively goes through all properties of that object.
Meanwhile it also tracks the property stack to give you information where the culprit value is located at.
Typescript
import {Logger} from "../Logger";
export class CircularReferenceDetector {
static detectCircularReferences(toBeStringifiedValue: any, serializationKeyStack: string[] = []) {
Object.keys(toBeStringifiedValue).forEach(key => {
var value = toBeStringifiedValue[key];
var serializationKeyStackWithNewKey = serializationKeyStack.slice();
serializationKeyStackWithNewKey.push(key);
try {
JSON.stringify(value);
Logger.debug(`path "${Util.joinStrings(serializationKeyStack)}" is ok`);
} catch (error) {
Logger.debug(`path "${Util.joinStrings(serializationKeyStack)}" JSON.stringify results in error: ${error}`);
var isCircularValue:boolean;
var circularExcludingStringifyResult:string = "";
try {
circularExcludingStringifyResult = JSON.stringify(value, CircularReferenceDetector.replaceRootStringifyReplacer(value), 2);
isCircularValue = true;
} catch (error) {
Logger.debug(`path "${Util.joinStrings(serializationKeyStack)}" is not the circular source`);
CircularReferenceDetector.detectCircularReferences(value, serializationKeyStackWithNewKey);
isCircularValue = false;
}
if (isCircularValue) {
throw new Error(`Circular reference detected:\nCircularly referenced value is value under path "${Util.joinStrings(serializationKeyStackWithNewKey)}" of the given root object\n`+
`Calling stringify on this value but replacing itself with [Circular object --- fix me] ( <-- search for this string) results in:\n${circularExcludingStringifyResult}\n`);
}
}
});
}
private static replaceRootStringifyReplacer(toBeStringifiedValue: any): any {
var serializedObjectCounter = 0;
return function (key: any, value: any) {
if (serializedObjectCounter !== 0 && typeof(toBeStringifiedValue) === 'object' && toBeStringifiedValue === value) {
Logger.error(`object serialization with key ${key} has circular reference to being stringified object`);
return '[Circular object --- fix me]';
}
serializedObjectCounter++;
return value;
}
}
}
export class Util {
static joinStrings(arr: string[], separator: string = ":") {
if (arr.length === 0) return "";
return arr.reduce((v1, v2) => `${v1}${separator}${v2}`);
}
}
Compiled JavaScript from TypeScript
"use strict";
const Logger_1 = require("../Logger");
class CircularReferenceDetector {
static detectCircularReferences(toBeStringifiedValue, serializationKeyStack = []) {
Object.keys(toBeStringifiedValue).forEach(key => {
var value = toBeStringifiedValue[key];
var serializationKeyStackWithNewKey = serializationKeyStack.slice();
serializationKeyStackWithNewKey.push(key);
try {
JSON.stringify(value);
Logger_1.Logger.debug(`path "${Util.joinStrings(serializationKeyStack)}" is ok`);
}
catch (error) {
Logger_1.Logger.debug(`path "${Util.joinStrings(serializationKeyStack)}" JSON.stringify results in error: ${error}`);
var isCircularValue;
var circularExcludingStringifyResult = "";
try {
circularExcludingStringifyResult = JSON.stringify(value, CircularReferenceDetector.replaceRootStringifyReplacer(value), 2);
isCircularValue = true;
}
catch (error) {
Logger_1.Logger.debug(`path "${Util.joinStrings(serializationKeyStack)}" is not the circular source`);
CircularReferenceDetector.detectCircularReferences(value, serializationKeyStackWithNewKey);
isCircularValue = false;
}
if (isCircularValue) {
throw new Error(`Circular reference detected:\nCircularly referenced value is value under path "${Util.joinStrings(serializationKeyStackWithNewKey)}" of the given root object\n` +
`Calling stringify on this value but replacing itself with [Circular object --- fix me] ( <-- search for this string) results in:\n${circularExcludingStringifyResult}\n`);
}
}
});
}
static replaceRootStringifyReplacer(toBeStringifiedValue) {
var serializedObjectCounter = 0;
return function (key, value) {
if (serializedObjectCounter !== 0 && typeof (toBeStringifiedValue) === 'object' && toBeStringifiedValue === value) {
Logger_1.Logger.error(`object serialization with key ${key} has circular reference to being stringified object`);
return '[Circular object --- fix me]';
}
serializedObjectCounter++;
return value;
};
}
}
exports.CircularReferenceDetector = CircularReferenceDetector;
class Util {
static joinStrings(arr, separator = ":") {
if (arr.length === 0)
return "";
return arr.reduce((v1, v2) => `${v1}${separator}${v2}`);
}
}
exports.Util = Util;
Here is a Node ES6 version mixed from the answers from #Aaron V and #user4976005, it fixes the problem with the call to hasOwnProperty:
const isCyclic = (obj => {
const keys = []
const stack = []
const stackSet = new Set()
let detected = false
const detect = ((object, key) => {
if (!(object instanceof Object))
return
if (stackSet.has(object)) { // it's cyclic! Print the object and its locations.
const oldindex = stack.indexOf(object)
const l1 = `${keys.join('.')}.${key}`
const l2 = keys.slice(0, oldindex + 1).join('.')
console.log(`CIRCULAR: ${l1} = ${l2} = ${object}`)
console.log(object)
detected = true
return
}
keys.push(key)
stack.push(object)
stackSet.add(object)
Object.keys(object).forEach(k => { // dive on the object's children
if (k && Object.prototype.hasOwnProperty.call(object, k))
detect(object[k], k)
})
keys.pop()
stack.pop()
stackSet.delete(object)
})
detect(obj, 'obj')
return detected
})
There's a lot of answers here, but I thought I'd add my solution to the mix. It's similar to #Trey Mack's answer, but that solution takes O(n^2). This version uses WeakMap instead of an array, improving the time to O(n).
function isCyclic(object) {
const seenObjects = new WeakMap(); // use to keep track of which objects have been seen.
function detectCycle(obj) {
// If 'obj' is an actual object (i.e., has the form of '{}'), check
// if it's been seen already.
if (Object.prototype.toString.call(obj) == '[object Object]') {
if (seenObjects.has(obj)) {
return true;
}
// If 'obj' hasn't been seen, add it to 'seenObjects'.
// Since 'obj' is used as a key, the value of 'seenObjects[obj]'
// is irrelevent and can be set as literally anything you want. I
// just went with 'undefined'.
seenObjects.set(obj, undefined);
// Recurse through the object, looking for more circular references.
for (var key in obj) {
if (detectCycle(obj[key])) {
return true;
}
}
// If 'obj' is an array, check if any of it's elements are
// an object that has been seen already.
} else if (Array.isArray(obj)) {
for (var i in obj) {
if (detectCycle(obj[i])) {
return true;
}
}
}
return false;
}
return detectCycle(object);
}
And this is what it looks like in action.
> var foo = {grault: {}};
> detectCycle(foo);
false
> foo.grault = foo;
> detectCycle(foo);
true
> var bar = {};
> detectCycle(bar);
false
> bar.plugh = [];
> bar.plugh.push(bar);
> detectCycle(bar);
true
You can also use Symbols - thanks to that approach you won't have to mutate properties of the original object, apart from adding symbol for marking visited node.
It's cleaner and should be faster than gathering node properties and comparing with the object. It also has optional depth limitation if you don't want to serialize big nested values:
// Symbol used to mark already visited nodes - helps with circular dependencies
const visitedMark = Symbol('VISITED_MARK');
const MAX_CLEANUP_DEPTH = 10;
function removeCirculars(obj, depth = 0) {
if (!obj) {
return obj;
}
// Skip condition - either object is falsy, was visited or we go too deep
const shouldSkip = !obj || obj[visitedMark] || depth > MAX_CLEANUP_DEPTH;
// Copy object (we copy properties from it and mark visited nodes)
const originalObj = obj;
let result = {};
Object.keys(originalObj).forEach((entry) => {
const val = originalObj[entry];
if (!shouldSkip) {
if (typeof val === 'object') { // Value is an object - run object sanitizer
originalObj[visitedMark] = true; // Mark current node as "seen" - will stop from going deeper into circulars
const nextDepth = depth + 1;
result[entry] = removeCirculars(val, nextDepth);
} else {
result[entry] = val;
}
} else {
result = 'CIRCULAR';
}
});
return result;
}
This will result in an object that has all the circular dependencies stripped and also does not go deeper than given MAX_CLEANUP_DEPTH.
Using symbols is safe as long as you don't do any meta-programming stuff on the object - they are transparent and they are not enumerable, hence - they will not show in any standard operations on the object.
Also, returning a new, cleaned up object has an advantage of not mutating the original one if you need to perform any additional operations on it.
If you don't want CIRCULAR marking, you can just modify the code a bit, hence skipping object before actually performing operations on it (inside the loop):
originalObj[visitedMark] = true; // Mark current node as "seen" - will stop from going deeper into circulars
const val = originalObj[entry];
// Skip condition - either object is falsy, was visited or we go too deep
const shouldSkip = val[visitedMark] || depth > MAX_SANITIZATION_DEPTH;
if (!shouldSkip) {
if (typeof val === 'object') { // Value is an object - run object sanitizer
const nextDepth = depth + 1;
result[entry] = removeCirculars(val, nextDepth);
} else {
result[entry] = val;
}
}
I just made this. It may be dirty, but works anyway... :P
function dump(orig){
var inspectedObjects = [];
console.log('== DUMP ==');
(function _dump(o,t){
console.log(t+' Type '+(typeof o));
for(var i in o){
if(o[i] === orig){
console.log(t+' '+i+': [recursive]');
continue;
}
var ind = 1+inspectedObjects.indexOf(o[i]);
if(ind>0) console.log(t+' '+i+': [already inspected ('+ind+')]');
else{
console.log(t+' '+i+': ('+inspectedObjects.push(o[i])+')');
_dump(o[i],t+'>>');
}
}
}(orig,'>'));
}
Then
var a = [1,2,3], b = [a,4,5,6], c = {'x':a,'y':b};
a.push(c); dump(c);
Says
== DUMP ==
> Type object
> x: (1)
>>> Type object
>>> 0: (2)
>>>>> Type number
>>> 1: (3)
>>>>> Type number
>>> 2: (4)
>>>>> Type number
>>> 3: [recursive]
> y: (5)
>>> Type object
>>> 0: [already inspected (1)]
>>> 1: (6)
>>>>> Type number
>>> 2: (7)
>>>>> Type number
>>> 3: (8)
>>>>> Type number
This tells that c.x[3] is equal to c, and c.x = c.y[0].
Or, a little edit to this function can tell you what you need...
function findRecursive(orig){
var inspectedObjects = [];
(function _find(o,s){
for(var i in o){
if(o[i] === orig){
console.log('Found: obj.'+s.join('.')+'.'+i);
return;
}
if(inspectedObjects.indexOf(o[i])>=0) continue;
else{
inspectedObjects.push(o[i]);
s.push(i); _find(o[i],s); s.pop(i);
}
}
}(orig,[]));
}
Here is #Thomas's answer adapted for node:
const {logger} = require("../logger")
// Or: const logger = {debug: (...args) => console.log.call(console.log, args) }
const joinStrings = (arr, separator) => {
if (arr.length === 0) return "";
return arr.reduce((v1, v2) => `${v1}${separator}${v2}`);
}
exports.CircularReferenceDetector = class CircularReferenceDetector {
detectCircularReferences(toBeStringifiedValue, serializationKeyStack = []) {
Object.keys(toBeStringifiedValue).forEach(key => {
let value = toBeStringifiedValue[key];
let serializationKeyStackWithNewKey = serializationKeyStack.slice();
serializationKeyStackWithNewKey.push(key);
try {
JSON.stringify(value);
logger.debug(`path "${joinStrings(serializationKeyStack)}" is ok`);
} catch (error) {
logger.debug(`path "${joinStrings(serializationKeyStack)}" JSON.stringify results in error: ${error}`);
let isCircularValue;
let circularExcludingStringifyResult = "";
try {
circularExcludingStringifyResult = JSON.stringify(value, this.replaceRootStringifyReplacer(value), 2);
isCircularValue = true;
} catch (error) {
logger.debug(`path "${joinStrings(serializationKeyStack)}" is not the circular source`);
this.detectCircularReferences(value, serializationKeyStackWithNewKey);
isCircularValue = false;
}
if (isCircularValue) {
throw new Error(`Circular reference detected:\nCircularly referenced value is value under path "${joinStrings(serializationKeyStackWithNewKey)}" of the given root object\n`+
`Calling stringify on this value but replacing itself with [Circular object --- fix me] ( <-- search for this string) results in:\n${circularExcludingStringifyResult}\n`);
}
}
});
}
replaceRootStringifyReplacer(toBeStringifiedValue) {
let serializedObjectCounter = 0;
return function (key, value) {
if (serializedObjectCounter !== 0 && typeof(toBeStringifiedValue) === 'object' && toBeStringifiedValue === value) {
logger.error(`object serialization with key ${key} has circular reference to being stringified object`);
return '[Circular object --- fix me]';
}
serializedObjectCounter++;
return value;
}
}
}
I converted the answer of Freddie Nfbnm to TypeScript:
export class JsonUtil {
static isCyclic(json) {
const keys = [];
const stack = [];
const stackSet = new Set();
let detected = false;
function detect(obj, key) {
if (typeof obj !== 'object') {
return;
}
if (stackSet.has(obj)) { // it's cyclic! Print the object and its locations.
const oldIndex = stack.indexOf(obj);
const l1 = keys.join('.') + '.' + key;
const l2 = keys.slice(0, oldIndex + 1).join('.');
console.log('CIRCULAR: ' + l1 + ' = ' + l2 + ' = ' + obj);
console.log(obj);
detected = true;
return;
}
keys.push(key);
stack.push(obj);
stackSet.add(obj);
for (const k in obj) { // dive on the object's children
if (obj.hasOwnProperty(k)) {
detect(obj[k], k);
}
}
keys.pop();
stack.pop();
stackSet.delete(obj);
return;
}
detect(json, 'obj');
return detected;
}
}
Just to throw my version into the mix... below is a remix of #dkurzaj 's code (which is itself a remix of #Aaron V 's, #user4976005 's, #Trey Mack 's and finally #Freddie Nfbnm 's [removed?] code) plus #darksinge 's WeakMap idea. So... this thread's Megamix, I guess :)
In my version, a report (rather than console.log'ed entries) is optionally returned as an array of objects. If a report is not required, testing stops on the first sighting of a circular reference (a'la #darksinge 's code).
Further, hasOwnProperty has been removed as Object.keys returns only hasOwnProperty properties (see: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Object/keys ).
function isCyclic(x, bReturnReport) {
var a_sKeys = [],
a_oStack = [],
wm_oSeenObjects = new WeakMap(), //# see: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/WeakMap
oReturnVal = {
found: false,
report: []
}
;
//# Setup the recursive logic to locate any circular references while kicking off the initial call
(function doIsCyclic(oTarget, sKey) {
var a_sTargetKeys, sCurrentKey, i;
//# If we've seen this oTarget before, flip our .found to true
if (wm_oSeenObjects.has(oTarget)) {
oReturnVal.found = true;
//# If we are to bReturnReport, add the entries into our .report
if (bReturnReport) {
oReturnVal.report.push({
instance: oTarget,
source: a_sKeys.slice(0, a_oStack.indexOf(oTarget) + 1).join('.'),
duplicate: a_sKeys.join('.') + "." + sKey
});
}
}
//# Else if oTarget is an instanceof Object, determine the a_sTargetKeys and .set our oTarget into the wm_oSeenObjects
else if (oTarget instanceof Object) {
a_sTargetKeys = Object.keys(oTarget);
wm_oSeenObjects.set(oTarget /*, undefined*/);
//# If we are to bReturnReport, .push the current level's/call's items onto our stacks
if (bReturnReport) {
if (sKey) { a_sKeys.push(sKey) };
a_oStack.push(oTarget);
}
//# Traverse the a_sTargetKeys, pulling each into sCurrentKey as we go
//# NOTE: If you want all properties, even non-enumerables, see Object.getOwnPropertyNames() so there is no need to call .hasOwnProperty (per: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Object/keys)
for (i = 0; i < a_sTargetKeys.length; i++) {
sCurrentKey = a_sTargetKeys[i];
//# If we've already .found a circular reference and we're not bReturnReport, fall from the loop
if (oReturnVal.found && !bReturnReport) {
break;
}
//# Else if the sCurrentKey is an instanceof Object, recurse to test
else if (oTarget[sCurrentKey] instanceof Object) {
doIsCyclic(oTarget[sCurrentKey], sCurrentKey);
}
}
//# .delete our oTarget into the wm_oSeenObjects
wm_oSeenObjects.delete(oTarget);
//# If we are to bReturnReport, .pop the current level's/call's items off our stacks
if (bReturnReport) {
if (sKey) { a_sKeys.pop() };
a_oStack.pop();
}
}
}(x, '')); //# doIsCyclic
return (bReturnReport ? oReturnVal.report : oReturnVal.found);
}
Most of the other answers only show how to detect that an object-tree has a circular-reference -- they don't tell you how to fix those circular references (ie. replacing the circular-reference values with, eg. undefined).
The below is the function I use to replace all circular-references with undefined:
export const specialTypeHandlers_default = [
// Set and Map are included by default, since JSON.stringify tries (and fails) to serialize them by default
{type: Set, keys: a=>a.keys(), get: (a, key)=>key, delete: (a, key)=>a.delete(key)},
{type: Map, keys: a=>a.keys(), get: (a, key)=>a.get(key), delete: (a, key)=>a.set(key, undefined)},
];
export function RemoveCircularLinks(node, specialTypeHandlers = specialTypeHandlers_default, nodeStack_set = new Set()) {
nodeStack_set.add(node);
const specialHandler = specialTypeHandlers.find(a=>node instanceof a.type);
for (const key of specialHandler ? specialHandler.keys(node) : Object.keys(node)) {
const value = specialHandler ? specialHandler.get(node, key) : node[key];
// if the value is already part of visited-stack, delete the value (and don't tunnel into it)
if (nodeStack_set.has(value)) {
if (specialHandler) specialHandler.delete(node, key);
else node[key] = undefined;
}
// else, tunnel into it, looking for circular-links at deeper levels
else if (typeof value == "object" && value != null) {
RemoveCircularLinks(value, specialTypeHandlers, nodeStack_set);
}
}
nodeStack_set.delete(node);
}
For use with JSON.stringify specifically, simply call the function above prior to the stringification (note that it does mutate the passed-in object):
const objTree = {normalProp: true};
objTree.selfReference = objTree;
RemoveCircularLinks(objTree); // without this line, the JSON.stringify call errors
console.log(JSON.stringify(objTree));
if you just need to see the content of that circular object, just use console.table(circularObj)
Try using console.log() on the chrome/firefox browser to identify where the issue encountered.
On Firefox using Firebug plugin, you can debug your javascript line by line.
Update:
Refer below example of circular reference issue and which has been handled:-
// JSON.stringify, avoid TypeError: Converting circular structure to JSON
// Demo: Circular reference
var o = {};
o.o = o;
var cache = [];
JSON.stringify(o, function(key, value) {
if (typeof value === 'object' && value !== null) {
if (cache.indexOf(value) !== -1) {
// Circular reference found, discard key
alert("Circular reference found, discard key");
return;
}
alert("value = '" + value + "'");
// Store value in our collection
cache.push(value);
}
return value;
});
cache = null; // Enable garbage collection
var a = {b:1};
var o = {};
o.one = a;
o.two = a;
// one and two point to the same object, but two is discarded:
JSON.stringify(o);
var obj = {
a: "foo",
b: obj
};
var replacement = {"b":undefined};
alert("Result : " + JSON.stringify(obj,replacement));
Refer example LIVE DEMO
For example if I have something like so:
var Constants = {
scope:{
namespaceA: { A_X: "TEST_AX" , A_Y: "TEST_AY" },
namespaceN: { N_X: "TEST_NX" , N_Y: "TEST_NY" }
}
_mapping: [],
getMapping: function(){...}
}
var flattenList = flatten(Constants.scope); //returns ["TEST_AX","TEST_AY","TEST_NX","TEST_NY"]
var anotherWayFlattened = flatten(Constants.scope.namespaceA,Constants.scope.namespaceB); //returns same result as above
EDIT: one way would be to iterate over the scope via for-each loop but I was looking for something more elegent?
DOUBLE EDIT: ok I just whipped something up like so:
var flattenedList = (function(list){
var flatList = []
$.each(list,function(i,items){
for(var p in items) flatList.push(items[p]);
})
return flatList;
})([Constants.scope.namespaceA,Constants.scope.namespaceB]);
but was wondering if we can avoid passing in the particular property and just pass in Constants and search for the list of namespaces
[Constants.scope.namespaceA,Constants.scope.namespaceB]
I'm wondering why you pass the sub-objects explicitly in an array. Why not just pass the whole Constants.scope object?
var flattenedList = (function(obj){
var flatList = []
for (var prop in obj) {
var items = obj[prop];
for (var p in items)
flatList.push(items[p]);
}
return flatList;
})(Constants.scope);
From your comment it looks like you wanted this:
var flattenedList = (function(obj, test){
var flatList = []
for (var prop in obj) {
if (!test(prop))
continue;
var items = obj[prop];
for (var p in items)
flatList.push(items[p]);
}
return flatList;
})(Constants, function(name) {
return name.substr(0, 9) == "namespace";
// or maybe
return /^namespace[A-Z]$/.test(name);
});
if you wanted to recurse to any (non cyclical!) depth, you could do this :
function flattenList(list, accumulator){
accumulator = accumulator || [];
for(var p in list){
if(list.hasOwnProperty(p)) {
if(typeof list[p] === "string") {
accumulator.push(list[p]);
} else if(typeof list[p] === "object") { // this is not a reliable test!
flattenList(list[p], accumulator);
}
}
}
return accumulator;
}
This code makes a number of assumptions - we only have strings at the end of our objects etc. Alternatively, if you know the depth in advance, your current solution can be optimized by using concat :
var flattenedList = (function(list){
return Array.prototype.concat.apply([], list);
})([Constants.scope.namespaceA,Constants.scope.namespaceB]);
Here's an approach that allows for deeper nesting. I know that wasn't part of the goals, but I found it a more interesting problem. :-)
var flatten = (function() {
var toString = Object.prototype.toString, slice = Array.prototype.slice;
var flatten = function(input, output) {
var value;
output = (toString.call(output) == "[object Array]") ? output : [];
for (name in input) {if (input.hasOwnProperty(name)) {
value = input[name];
if (toString.call(value) == "[object Object]") {
flatten(value, output);
} else {
output.push(value);
}
}};
return output;
};
var merge = function(first, second) {
return first.concat(second);
}
return function() {
return slice.call(arguments).map(flatten).reduce(merge);
};
}());
This allows either approach:
flatten(Constants.scope);
flatten(Constants.scope.namespaceA, Constants.scope.namespaceN);
You can pass in as many separate arguments as you like, or one argument. They'll all be searched to arbitrary depths.
For some environments, you might have to shim Array.prototype functions map and reduce.
Assume I have an object:
var obj = {
foo:"bar",
fizz:"buzz"
};
I need to access a property of that object dynamically like so:
var objSetter = function(prop,val){
obj[prop] = val;
}
No problems there, except for that prop needs to be case insensitive in case the property name is passed into the function as, say, Foo instead of foo.
So how can I point to an object's property by name without regard to case? I would like to avoid iterating the entire object if possible.
Try this:
var myObject = { "mIxeDCaSEKeY": "value" };
var searchKey = 'mixedCaseKey';
var asLowercase = searchKey.toLowerCase();
myObject[Object.keys(myObject).find(key => key.toLowerCase() === asLowercase)];
You can alternatively already provide the searchKey in lowercase.
If you want it as a function:
/**
* #param {Object} object
* #param {string} key
* #return {any} value
*/
function getParameterCaseInsensitive(object, key) {
const asLowercase = key.toLowerCase();
return object[Object.keys(object)
.find(k => k.toLowerCase() === asLowercase)
];
}
If the key can't be found, then it'll return undefined, just like normal.
If you need to support older browsers, then you can use filter instead:
function getParameterCaseInsensitive(object, key) {
const asLowercase = key.toLowercase();
return object[Object.keys(object).filter(function(k) {
return k.toLowerCase() === asLowercase;
})[0]];
}
I suggest using the polyfills for Object.keys() and Array.filter() if you need even older support.
Note: If you want to also check non-enumerable keys, use Object.getOwnPropertyNames() instead of Object.keys().
Nerdy Note: This assumes your Object doesn't have a key undefined (eg: const foo = {[undefined]: 'bar'};). That's just weird.
Compare all the properties of obj with prop.
var objSetter = function(prop,val){
prop = (prop + "").toLowerCase();
for(var p in obj){
if(obj.hasOwnProperty(p) && prop == (p+ "").toLowerCase()){
obj[p] = val;
break;
}
}
}
For this, I prefer using the prototype over a standalone function just for ease of use and expressiveness. I just don't like funneling objects into functions if I don't have to.
Also, while the accepted answer works, I wanted a more comprehensive solution for both getting and setting that would behave as much like the native dot notation or bracket notation as possible.
With that in mind, I created a couple prototype functions for setting/getting an object property without regard to case. You have to remember to be VERY responsible when adding to the Object prototype. Especially when using JQuery and other libraries. Object.defineProperty() with enumerable set to false was used specifically to avoid conflict with JQuery. I also didn't bother naming the functions anything that indicates they are case-insensitive, but you certainly could. I like shorter names.
Here's the getter:
Object.defineProperty(Object.prototype, "getProp", {
value: function (prop) {
var key,self = this;
for (key in self) {
if (key.toLowerCase() == prop.toLowerCase()) {
return self[key];
}
}
},
//this keeps jquery happy
enumerable: false
});
Here's the setter:
Object.defineProperty(Object.prototype, "setProp", {
value: function (prop, val) {
var key,self = this;
var found = false;
if (Object.keys(self).length > 0) {
for (key in self) {
if (key.toLowerCase() == prop.toLowerCase()) {
//set existing property
found = true;
self[key] = val;
break;
}
}
}
if (!found) {
//if the property was not found, create it
self[prop] = val;
}
return val;
},
//this keeps jquery happy
enumerable: false
});
Now that we've created those functions, our code is super clean and concise and just works.
Case-insensitive getting:
var obj = {foo: 'bar', camelCase: 'humpy'}
obj.getProp("FOO"); //returns 'bar'
obj.getProp("fOO"); //returns 'bar'
obj.getProp("CAMELCASE"); //returns 'humpy'
obj.getProp("CamelCase"); //returns 'humpy'
Case-insensitive setting:
var obj = {foo: 'bar', camelCase: 'humpy'}
obj.setProp('CAmelCasE', 'super humpy'); //sets prop 'camelCase' to 'super humpy'
obj.setProp('newProp', 'newval'); //creates prop 'newProp' and sets val to 'newval'
obj.setProp('NewProp', 'anotherval'); //sets prop 'newProp' to 'anotherval'
Yet another variation on those already presented which pushes the iteration down into the Underscore/Lodash findKey function:
var _ = require('underscore');
var getProp = function (obj, name) {
var realName = _.findKey(obj, function (value, key) {
return key.toLowerCase() === name.toLowerCase();
});
return obj[realName];
};
For example:
var obj = { aa: 1, bB: 2, Cc: 3, DD: 4 };
getProp(obj, 'aa'); // 1
getProp(obj, 'AA'); // 1
getProp(obj, 'bb'); // 2
getProp(obj, 'BB'); // 2
getProp(obj, 'cc'); // 3
getProp(obj, 'CC'); // 3
getProp(obj, 'dd'); // 4
getProp(obj, 'DD'); // 4
getProp(obj, 'EE'); // undefined
This answer requires ES6.
const x = { 'aB': 1, 'X-Total-Count': 10, y3: 2 }
console.log(x[Object.keys(x).find(key=>{return key.match(/^ab$/i)})])
console.log(x[Object.keys(x).find(key=>{return key.match(/^x-total-count$/i)})])
console.log(x[Object.keys(x).find(key=>{return key.match(/^y3$/i)})])
It seems to me like a good candidate for Proxy with traps to convert string keys to either upper case or lower case and behaving like a regular object.
This works with either notation: dots or braquets
Here is the code:
'use strict';
function noCasePropObj(obj)
{
var handler =
{
get: function(target, key)
{
//console.log("key: " + key.toString());
if (typeof key == "string")
{
var uKey = key.toUpperCase();
if ((key != uKey) && (key in target))
return target[key];
return target[uKey];
}
return target[key];
},
set: function(target, key, value)
{
if (typeof key == "string")
{
var uKey = key.toUpperCase();
if ((key != uKey) && (key in target))
target[key] = value;
target[uKey] = value;
}
else
target[key] = value;
},
deleteProperty: function(target, key)
{
if (typeof key == "string")
{
var uKey = key.toUpperCase();
if ((key != uKey) && (key in target))
delete target[key];
if (uKey in target)
delete target[uKey];
}
else
delete target[key];
},
};
function checkAtomic(value)
{
if (typeof value == "object")
return new noCasePropObj(value); // recursive call only for Objects
return value;
}
var newObj;
if (typeof obj == "object")
{
newObj = new Proxy({}, handler);
// traverse the Original object converting string keys to upper case
for (var key in obj)
{
if (typeof key == "string")
{
var objKey = key.toUpperCase();
if (!(key in newObj))
newObj[objKey] = checkAtomic(obj[key]);
}
}
}
else if (Array.isArray(obj))
{
// in an array of objects convert to upper case string keys within each row
newObj = new Array();
for (var i = 0; i < obj.length; i++)
newObj[i] = checkAtomic(obj[i]);
}
return newObj; // object with upper cased keys
}
// Use Sample:
var b = {Name: "Enrique", last: "Alamo", AdDrEsS: {Street: "1233 Main Street", CITY: "Somewhere", zip: 33333}};
console.log("Original: " + JSON.stringify(b)); // Original: {"Name":"Enrique","last":"Alamo","AdDrEsS":{"Street":"1233 Main Street","CITY":"Somewhere","zip":33333}}
var t = noCasePropObj(b);
console.log(JSON.stringify(t)); // {"NAME":"Enrique","LAST":"Alamo","ADDRESS":{"STREET":"1233 Main Street","CITY":"Somewhere","ZIP":33333}}
console.log('.NaMe:' + t.NaMe); // .NaMe:Enrique
console.log('["naME"]:' + t["naME"]); // ["naME"]:Enrique
console.log('.ADDreSS["CitY"]:' + t.ADDreSS["CitY"]); // .ADDreSS["CitY"]:Somewhere
console.log('check:' + JSON.stringify(Object.getOwnPropertyNames(t))); // check:["NAME","LAST","ADDRESS"]
console.log('check2:' + JSON.stringify(Object.getOwnPropertyNames(t['AddresS']))); // check2:["STREET","CITY","ZIP"]
You could do this in order to "normalize" prop
var normalizedProp = prop.toLowerCase();
obj[normalizedProp] = val;
const getPropertyNoCase = (obj, prop) => obj[Object.keys(obj).find(key => key.toLowerCase() === prop.toLowerCase() )];
or
const getPropertyNoCase = (obj, prop) => {
const lowerProp = prop.toLowerCase(obj[Object.keys(obj).find(key => key.toLowerCase() === prop.toLowerCase() )];
}
The ES6 example posted by #nilloc is incorrect and will break in use.
Here is a working example:
const x = {'first':5,'X-Total-Count':10,'third':20};
console.log(x[Object.keys(x).reduce((result,key)=>{
if (!result) {
return key.match(/x-total-count/i)
} else {
return result;
}
},null)]);
or better yet, it should return undefined if the key doesn't exist:
const x = {'first':5,'X-Total-Count':10,'third':20};
console.log(x[Object.keys(x).reduce((result,key)=>{
if (!result) {
return key.match(/x-total-count/i) || undefined
} else {
return result;
}
},undefined)]);
One consideration is that the above example will return the last matching key in the object if there are multiple keys that match.
Here is an example with the code made into a function:
/**
* #param {Object} object
* #param {string} key
* #return {string||undefined} value || undefined
*/
function getKeyCase(obj,key) {
const re = new RegExp(key,"i");
return Object.keys(obj).reduce((result,key)=>{
if (!result) {
return key.match(re) || undefined
} else {
return result;
}
},undefined);
const x = {'first':5,'X-Total-Count':10,'third':20};
console.log(x[getKeyCase(x,"x-total-count")]);
Its really sad that the iteration can't be skipped as it seems. For me what is acceptable but may not be for everyone is to shape the object one time via iteration and then use it in regular hashmap fashion.
const hashmap = {
'FOO': 'foo as in function programming',
'bar': 'bar is in baz',
};
const shapedmap = Object.entries(hashmap).reduce(
(acc, [key, val]) => (acc[key.toUpperCase()] = val, acc), {}
);
for (const term of ['foo', 'bar', 'baz']) {
const match = shapedmap[term.toUpperCase()]
match && console.log('awesome, we got the term.', match);
};
Even if it just one time lookup has to be performed, it shouldn't less performant as any other iteration solution since after 1 pass, the lookup speed is constant. (I guess).
This is an old question, but it was the first one I found.
As #ZachSmith says, you can use a Proxy.
Here's some example code:
function lowercase(oldKey) {
// Check that it's a string.
return typeof oldKey === 'string' ? oldKey.toLowerCase() : oldKey;
}
const propertiesMap = new Map(
Object.keys(obj).map(propKey => [lowercase(propKey), obj[propKey]])
);
const caseInsensitiveGetHandler = {
get: function(target, property, receiver) {
return propertiesMap.get(lowercase(property));
}
};
obj = new Proxy(obj, caseInsensitiveGetHandler);
For my use case, I only needed to proxy the object's getter, but you may need to implement more of the Proxy methods.
There is no need for any iteration. Since prop might not be a string, it should be coerced to a string first where appropriate since that's what objects do natively. A simple getter function is:
function objGetter(prop) {
return obj[String(prop).toLowerCase()];
}
If there is a requirement is to restring access to own properties:
function objGetter(prop) {
prop = String(prop).toLowerCase();
if (obj.hasOwnProperty(prop)) {
return obj.prop;
}
}
and a setter:
function objSetter(prop, val) {
obj[String(prop).toLowerCase()] = val;
}
Heres a very simple code to do this
Assuming that data is the array of objects like
data=[{"A":"bc","B":"nn"}]
var data=data.reduce(function(prev, curr) {
var cc = curr; // current value
var K = Object.keys(cc); // get all keys
var n = {};
for (var i = 0; i < K.length; i++) {
var key = K[i];//get hte key
n[key.toLowerCase()] = cc[key] // convert to lowercase and assign
}
prev.push(n) // push to array
return prev;
}, [])
Output will be
data=[{"a":"bc","b":"nn"}]
You might only need to do case-insensitive matching (usually expensive because of object iteration) IF a case-sensitive match (cheap and quick) fails.
Say you have:
var your_object = { "Chicago" : 'hi' , "deTroiT" : 'word' , "atlanta" : 'get r dun' } ;
And you have, for whatever reason, the_value, Detroit:
if( your_object.hasOwnProperty( the_value ) )
{
// do what you need to do here
}
else
{ // since the case-sensitive match did not succeed,
// ... Now try a the more-expensive case-insensitive matching
for( let lvs_prop in your_object )
{ if( the_value.toLowerCase() == lvs_prop.toLowerCase() )
{
// do what you need to do here
break ;
} ;
}
} ;
why would we do it that complicated when we simply can make it all lower case:
var your_object = {
"chickago" : 'hi' ,
"detroit" : 'word',
"atlanta" : 'get r dun',
GetName: function (status) {
return this[status].name;
} };
to call it: your_object.GetName(your_var.toLowerCase());
Another simple way:
function getVal(obj, prop){
var val;
prop = (prop + "").toLowerCase();
for(var p in obj){
if(obj.hasOwnProperty(p) && prop == (p+ "").toLowerCase()){
val = obj[p]
break;
}
}
return val;
}
Use it like this:
var obj = {
foo:"bar",
fizz:"buzz"
};
getVal(obj,"FoO") -> returns "bar"
Here is a nice recursive function that allows you to traverse a javascript object in a case-insensitive way:
let testObject = {'a': {'B': {'cC': [1,2,3]}}}
let testSeq = ['a','b','cc']
function keySequence(o, kseq) {
if(kseq.length==0){ return o; }
let validKeys = Object.keys(o).filter(k=>k.toLowerCase()==kseq[0].toLowerCase());
if(validKeys.length==0) { return `Incorrect Key: ${kseq[0]}` }
return keySequence(o[validKeys[0]], kseq.slice(1))
}
keySequence(testObject, testSeq); //returns [1,2,3]
This will convert everything to lowercase, but in a bind this could help if you are not concerned with retaining case.
var somedata = {
"MixEdCase": 1234
}
var temp = JSON.parse(JSON.stringify(somedata).toLowerCase());
console.log(temp.mixedcase);
// or
console.log(temp["mixedcase"]);
So, you will need to get the object key that matches the case of the existing object, then use this to do your object update.
const obj = {
foo:"bar",
fizz:"buzz"
};
// to get obj.foo or obj.FOO or obj.foO returning "bar"
// create regex expression of case insensitive version of the key string
const regex=passedKey=> new RegExp(`^${passedKey}$`,'gi');
// find the key that matches the string you are passing
const formattedKey=passedKey=>Object.keys(obj).find(key=>regex(passedKey).test(key));
formattedKey('Foo'); // returns foo
formattedKey('FoO'); // returns foo
// consequently you can can use it like wise
obj[formattedKey('Foo')] // returns bar
obj[formattedKey('FoO')] // returns bar
obj[formattedKey('foo')] // returns bar