Ignore commented code

[jonaskello]

Currently when a string with the gql tag exists but is commented, there will be an exception:

{ GraphQLError: Syntax Error ./xxxx/container-component.tsx (2:1) Cannot parse the unexpected character "/".

[dotansimha]

@jonaskello , you are right, I will fix it soon.
Also, I need to add support for gql(...) (use it as function)

[jonaskello]

IIRC the checks are now done with regexp. I think maybe a better approach would be to parse it into an AST using babel or typescript. That would probably be more accurate than a regexp. But of course it would be more complex to implement and maybe slower too.

[dotansimha]

@jonaskello Yeah sure it makes sense. Do you have a reference or example for checking a code file as AST? I'll give it a try :)
Also, it might be possible to run something like https://www.npmjs.com/package/strip-comments before running the regex...

[dotansimha]

@jonaskello
strip-comments seems to work great (see unit tests)

Please see #192

What do you think?

[dotansimha]

Fixed in 0.8.19 :)

[jonaskello]

IMO, using strip-comments is a simpler solution and therefore better :-). If there are more advanced code-analysis needed later we can look into parsing into an abstract syntax tree at that point.

[danielkcz]

Yea well, this wasn't a good move. See #196 ;) I think it will require an approach that @jonaskello mentioned and go with AST.

Just my two cents. In VSCode I am using https://www.npmjs.com/package/@playlyfe/gql to essentially highlight and intellisense graphql queries. That would mean it can parse it just fine and apparently works cross-language, so it might be worth investigating. Perhaps it's possible it could fulfill a duty of file watcher as well to have generated types refreshed automatically?

[jonaskello]

Regarding parsing to AST I think we could either use babylon as it now has typescript support (see here). Or we could simply use typescript, see here for a simple example of traversing the AST.

[dotansimha]

@jonaskello @FredyC
We recently merged this: #907 , now the codegen will parse the code file as AST and extract the documents from it.