Identical operands rule using syntaxmap

[PaulTaykalo]

Rewrite identical operands rule which using syntax map and tokens access instead of full Regex solution.

This solutions search for the operator first and then tries to find operands to the right and to the left of the operator. This is done by searching tokens which has dots between them.

|   asd    ?? |      self   .    a      ==    self   .    a
| [token]     |    [token]     [token]       [token]     [token]
| identifier  |    keyword    identifier     keyword    identifier

Operands are matching if the next is true:

• they have same tokens count
• they have tokens of the same type
• they have identical tokens strings
• if there's a token to the left of the left token, string between them doesn't have the nil-coalescing operator

[PaulTaykalo]

That's weird. Will fix in evening 😊

[SwiftLintBot]

	1 Warning
⚠️	This PR may need tests.

	12 Messages
📖	Linting Aerial with this PR took 3.09s vs 2.63s on master (17% slower)
📖	Linting Alamofire with this PR took 6.38s vs 4.24s on master (50% slower)
📖	Linting Firefox with this PR took 16.16s vs 13.59s on master (18% slower)
📖	Linting Kickstarter with this PR took 38.37s vs 28.59s on master (34% slower)
📖	Linting Moya with this PR took 2.54s vs 2.37s on master (7% slower)
📖	Linting Nimble with this PR took 2.53s vs 2.69s on master (5% faster)
📖	Linting Quick with this PR took 0.98s vs 1.2s on master (18% faster)
📖	Linting Realm with this PR took 4.02s vs 4.19s on master (4% faster)
📖	Linting SourceKitten with this PR took 1.78s vs 1.78s on master (0% slower)
📖	Linting Sourcery with this PR took 4.61s vs 5.13s on master (10% faster)
📖	Linting Swift with this PR took 21.81s vs 26.25s on master (16% faster)
📖	Linting WordPress with this PR took 32.41s vs 29.6s on master (9% slower)

Generated by 🚫 Danger

[PaulTaykalo]

Well, at least it's faster

[PaulTaykalo]

Message: Linting Aerial with this PR took 2.11s vs 2.12s on master (0% faster)
Message: Linting Alamofire with this PR took 3.76s vs 3.75s on master (0% slower)
Message: Linting Firefox with this PR took 11.88s vs 12.12s on master (1% faster)
Message: Linting Kickstarter with this PR took 27.16s vs 27.91s on master (2% faster)
Message: Linting Moya with this PR took 2.38s vs 2.3s on master (3% slower)
Message: Linting Nimble with this PR took 2.29s vs 2.41s on master (4% faster)
Message: Linting Quick with this PR took 0.95s vs 0.97s on master (2% faster)
Message: Linting Realm with this PR took 3.52s vs 3.67s on master (4% faster)
Message: Linting SourceKitten with this PR took 1.51s vs 1.62s on master (6% faster)
Message: Linting Sourcery with this PR took 4.0s vs 4.14s on master (3% faster)
Message: Linting Swift with this PR took 18.9s vs 19.78s on master (4% faster)
Message: Linting WordPress with this PR took 30.3s vs 31.38s on master (3% faster)

[jpsim]

👍 nice work! Removing regex use is probably one of the more impactful ways to speed up SwiftLint 😄

Please add a changelog entry, consider addressing the nitpicks I highlighted, and merge when ready.

[jpsim]

Oops, looks like this introduces some new false positives according to OSSCheck's comment.

[PaulTaykalo]

@jpsim yeah. also, it seems that this solution also can handle bit complex examples like

a?.b == a?.b

[PaulTaykalo]

Let's call it a day.

[jpsim]

Let's call it a day.

Let's fix the remaining false positives first please. They're missing from the latest OSSCheck comment, but here's one that I copied from before: https://github.com/mozilla-mobile/firefox-ios/blob/48798608a791bbaea7ba3e22d58694318a40da0b/Shared/Extensions/URLExtensions.swift#L363:108

[PaulTaykalo]

Let's call it a day.

Let's fix the remaining false positives first please. They're missing from the latest OSSCheck comment, but here's one that I copied from before: https://github.com/mozilla-mobile/firefox-ios/blob/48798608a791bbaea7ba3e22d58694318a40da0b/Shared/Extensions/URLExtensions.swift#L363:108

it's done in the latest commits.
The issue was with Optional chainings like a?.b == b
Another one with the fact that we wrent' checking that there is only an operator between right and left tokens.
So examples like this were triggering function(something: nil) != nil

[jpsim]

@PaulTaykalo you should now have push access, feel free to merge when you're ready.