As it was asked, here's a recursive object comparison function. And a bit more. Assuming that primary use of such function is object inspection, I have something to say. Complete deep comparison is a bad idea when some differences are irrelevant. For example, blind deep comparison in TDD assertions makes tests unnecessary brittle. For that reason, I'd like to introduce a much more valuable partial diff. It is a recursive analogue of a previous contribution to this thread. It ignores keys not present in a
var bdiff = (a, b) =>
_.reduce(a, (res, val, key) =>
res.concat((_.isPlainObject(val) || _.isArray(val)) && b
? bdiff(val, b[key]).map(x => key + '.' + x)
: (!b || val != b[key] ? [key] : [])),
[]);
BDiff allows checking for expected values while tolerating other properties, which is exactly what you want for automatic inspection. This allows building all kinds of advanced assertions. For example:
var diff = bdiff(expected, actual);
// all expected properties match
console.assert(diff.length == 0, "Objects differ", diff, expected, actual);
// controlled inequality
console.assert(diff.length < 3, "Too many differences", diff, expected, actual);
Returning to the complete solution. Building a full traditional diff with bdiff is trivial:
function diff(a, b) {
var u = bdiff(a, b), v = bdiff(b, a);
return u.filter(x=>!v.includes(x)).map(x=>' < ' + x)
.concat(u.filter(x=>v.includes(x)).map(x=>' | ' + x))
.concat(v.filter(x=>!u.includes(x)).map(x=>' > ' + x));
};
Running above function on two complex objects will output something similar to this:
[
" < components.0.components.1.components.1.isNew",
" < components.0.cryptoKey",
" | components.0.components.2.components.2.components.2.FFT.min",
" | components.0.components.2.components.2.components.2.FFT.max",
" > components.0.components.1.components.1.merkleTree",
" > components.0.components.2.components.2.components.2.merkleTree",
" > components.0.components.3.FFTResult"
]
Finally, in order to have a glimpse into how the values differ, we may want to directly eval() the diff output. For that, we need an uglier version of bdiff that outputs syntactically correct paths:
// provides syntactically correct output
var bdiff = (a, b) =>
_.reduce(a, (res, val, key) =>
res.concat((_.isPlainObject(val) || _.isArray(val)) && b
? bdiff(val, b[key]).map(x =>
key + (key.trim ? '':']') + (x.search(/^\d/)? '.':'[') + x)
: (!b || val != b[key] ? [key + (key.trim ? '':']')] : [])),
[]);
// now we can eval output of the diff fuction that we left unchanged
diff(a, b).filter(x=>x[1] == '|').map(x=>[x].concat([a, b].map(y=>((z) =>eval('z.' + x.substr(3))).call(this, y)))));
That will output something similar to this:
[" | components[0].components[2].components[2].components[2].FFT.min", 0, 3]
[" | components[0].components[2].components[2].components[2].FFT.max", 100, 50]
MIT license ;)
_.isEqual(value, other)
Performs a deep comparison between two values to determine if they are equivalent. lodash.com/docs#isEqual – Lukas Liesis