Add unique item property keyword.

[bayleedev]

@epoberezkin my second test is correctly failing, and it should fail, but when I change valid to false, like I believe it should be, it fails still and I'm not sure why.

Can you point me in the right direction?

[coveralls]

Coverage remained the same at 100.0% when pulling 0388bfa on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[epoberezkin]

@blainesch can you change it to false so I can see it fail when it shouldn't? How can I say why it fails otherwise?

btw, the implementation of uniqueItems in Ajv can be helpful.

[bayleedev]

btw, the implementation of uniqueItems in Ajv can be helpful.

So a compile instead of validate ? I can try that :D

[bayleedev]

I pushed the change you were asking for.

[coveralls]

Coverage decreased (-0.2%) to 99.82% when pulling 9ea05a0 on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[epoberezkin]

https://runkit.com/esp/591f1ba7015d6e0012243737 - works here

maybe it's because you have two tests with the same name? not sure. You need to debug.

[epoberezkin]

So a compile instead of validate ? I can try that :D

Yes, implementation can be more more efficient :) (three loops?), but what I meant was that you may find some ideas there and figure out why it's not working here :). Compile vs. validate won't make much difference as you don't need to manipulate the schema - just to iterate the keys...

A more important thing is that it probably should check deep equality rather than by reference, in the same way uniqueItems is specified. If it were to be added to the spec it would definitely be deep equality. Ajv has "equal" function inside (for performance reasons - it's faster than all the libs I saw, probably should be extracted), not sure what is the better way to do here.

[bayleedev]

Looks like I needed to define it in spec/schema-tests.spec.js. It's a wip right now, I'll add some more tests and look into deepEqual as a test case.

[coveralls]

Coverage remained the same at 100.0% when pulling c0b5e9b on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[epoberezkin]

right, it simply wasn't added to the instance

[coveralls]

Coverage remained the same at 100.0% when pulling 6801b0e on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[epoberezkin]

So if you use compile, it will be passed schema (keys), parentSchema and schema compilation context (it). There is it.util. I suggest exposing equal that Ajv uses there (needs change in Ajv) so you can use it.util.equal consistently in the implementations (it is used by enum, constant and uniqueItems):

definition = {
  compile: function(keys, parentSchema, it) {
    var equal = it.util.equal;
    return function(data) {
      // validation here can use equal
    };
  },
  // ...
}

[coveralls]

Coverage remained the same at 100.0% when pulling b23dc1a on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[bayleedev]

Sweet, how's that? :D

[bayleedev]

@epoberezkin any update?

[epoberezkin]

@blainesch I didn't realise it's finished. Can you add a section in readme please?

[bayleedev]

Yeah give me a few minutes [:

[coveralls]

Coverage remained the same at 100.0% when pulling f9a4dfb on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[coveralls]

Coverage remained the same at 100.0% when pulling edc78d5 on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[coveralls]

Coverage remained the same at 100.0% when pulling edc78d5 on blainesch:feature/unique-item-properties into ff2c769 on epoberezkin:master.

[bayleedev]

Thanks!