Constellation Query Language misclassified as SQL because of cql extension

[eighty4]

Searching for .cql code samples, I found a .cql extension query language called Constellation Query Language. Is language assignment strictly performed on file extension or is there a way to disqualify this other CQL from being marked as SQL language?

[lildude]

If you don’t want the file identified as SQL you can use an override to identify it as another language, including Text, or completely ignore it. You can’t use an override to select a language Linguist doesn’t know about though.

[eighty4]

This solution is to declare it as a different language from the user's repository that's using Constellation Query Language, correct?

Is Linguist lumping both CQLs as SQL only on the file ext? How useful is the SQL classification from Linguist if any file with a .cql extension is classified as SQL?

I've seen discussions and issues about CassandraQL having its own language classification. Would that be possible by parsing the content of the file? Is that done for any other language classifications?

[eighty4]

Looks like the language detection feature is done with TextMate/Sublime.

Does that mean Linguist will attempt a grammar based on a file extension, and if it is not invalidated by the grammar, it resolves to that language? Or is the language resolved by the extension and the grammar only controls highlighting, and in this case, a .cql file will be highlighted with an SQL grammar?

[eighty4]

Here are two TextMate grammars from VSCode plugins:

https://github.com/gnana997/cassandra-lens/blob/main/syntaxes/cql.tmLanguage.json
https://github.com/lawrencekgrant/vscode-cql/blob/master/syntaxes/cql.tmLanguage

[lildude]

This solution is to declare it as a different language from the user's repository that's using Constellation Query Language, correct?

Yes, or add support for that language to Linguist, but you'll need to prove propularity meets our requirements and you'll need a heuristic (ie a regex) to differentiate it from the SQL files.

Is Linguist lumping both CQLs as SQL only on the file ext?

Yes, because SQL is the only language associated with that extension.

How useful is the SQL classification from Linguist if any file with a .cql extension is classified as SQL?

It's only as useful as the languages Linguist knows about. .cql was added to the SQL definition over 9 years ago. I don't know anything about Constellation Query Language but it's possible this pre-dates the existance of Constellation Query Language and no one until you has noticed 😁

Would that be possible by parsing the content of the file? Is that done for any other language classifications?

Yes. It's done using heuristics which are just regular expressions.

Looks like the language detection feature is done with TextMate/Sublime.

It's not. Language detection is done using the various files within Linguist. The grammar files are only used for syntax highlighting once Linguist has identified the languages. These are only provided for the convenience of the syntax highlighting engine. They're not used by Linguist at all.

Does that mean Linguist will attempt a grammar based on a file extension, and if it is not invalidated by the grammar, it resolves to that language? Or is the language resolved by the extension and the grammar only controls highlighting, and in this case, a .cql file will be highlighted with an SQL grammar?

The grammar isn't involved at all. See How Linguist Works.

Here are two TextMate grammars from VSCode plugins:

Linguist relies on community contributions. If you want support for Constellation Query Language you or someone else in that community can follow the instructions in the CONTRIBUTING.md file and submit a PR to add support.

[eighty4]

My interest was in defining cql as a Linguist language for Cassandra Query Language.

I've since noticed that there's a third CQL that gets lumped into SQL because of CassandraQL being tagged as SQL by only checking the extension.

My use case was the GitHub search API and I've gotten around it by explicitly searching orgs and repos for code samples that I need.

I have read How Linguist Works, but still not sure if there's anything I could do to create a CQL that's specific to Cassandra Query Language, and how to prevent the two other CQL data definition and modeling languages from being classified as SQL (or Cassandra Query Language).