[POC][Improvement] completion handle aliases

[camilledejoye]

Hi,

That's an attempt to deal with aliases from the import table when completing classes (see phpactor/phpactor#1121).
I tried to not keep it as clean as possible but it still kind of a POC to see how you feel about the idea so don't hesitate: the more critics the better 😄

I decided to put it on the ChainTolerantCompletor because we need the parser to get the import table and I wanted it to work for all class completors without having to deal with them one by one.

The new DoctrineAnnotationCompletor is a bit special so I had to extract the logic into a trait.
As I said in the commit message I did it to because it feels a right use case but if you prefer an helper class with static methods it's also possible of course.

I've mainly used the automated test for now, just one or two manual test in a demo project to make sure it works.
I will update my work environment on Monday so that I can test it in a more realistic code base.

Todos:

• To merge in the same time: [POC][Improvement] completion handle aliases completion-worse-extension#4
• Add/Reuse a cache decorator to avoid doing the same work twice for the same completion request

[dantleech]

Instead of suggesting on En<> => ORM\Entity would it be an idea to instead complete on ORM\\{Entity,Column,Foobar} ? This also makes annotations discoverable ?

[camilledejoye]

Yes, that was the first item in phpactor/phpactor#1121
I just think it might be a bit harder right now because of how the name lookup works but I would definitely like to add this at some point.

[dantleech]

Searching by FQN is trivial to add fortunately ☺️

See PR phpactor/indexer-extension#26

[dantleech]

Given the above is now merged -- would it make sense to do namespace and import name completion instead of this?

• FQN completion:
  • Can be clearly applied (e.g. only when the node is already a partially qualified name ORM\\<>)
  • Can help with discovery
  • Requiring some specification helps (e.g. if I search for Before<> with PHPBench + Behat + etc installed the list is quite extensive)
• Import name completion: e.g. use Doctrine\Annotations; Anno<> => Annotations\\

Not against this - but if this can be accomplished in another way it might be redundant to have?

[camilledejoye]

For me this is two complementary features.
The one in this PR allows to discover aliases that you don't even realize are used and it's more convenient.

With only the search by relative qualified name we still need to type the alias like ORM\.
With Symfonys validator it could be really painful to always have to retype the prefix when you could simply do NotBl<> and have it completed properly given the import table.

While the improvement you've done on the indexer will indeed grant us the possibility to discover available names, which is always useful in new/large code base.
It does not allow to discover existing aliases, often at work I want to import a class but someone already did but used an alias for some reason. When typing the "correct" class name I'm not aware of this alias: this feature will show me a different name letting me know about the alias.

I grant you the last one is really an edge case :)

Bottom line I think we can keep this one, with upgrade if necessary, and create another one to complete relative qualified name.

[dantleech]

you could simply do NotBl<> and have it completed properly given the import table.

Maybe it should be prioritized?

[dantleech]

But that can be a separate thing -- we should add:

$container->register(CompletorClass::class, function () {}, [ CompletionExtension::COMPLETOR => [ 'priority' => 100]])

in the completion extension

[camilledejoye]

I updated with the implementation we discussed earlier.
I will create a PR right away for the completion-worse-extension so that you can test it without figuring out what to register as a service.
Works for me, I will gave it a try at work as much as possible to have a better feeling about it next week-end ;)

PR for completion-worse-extension: phpactor/completion-worse-extension#4