Add fragments to `executable_doc.validate()`

[nrbnlulu]

Description

In order to validate a document we currently need to have all of its fragments in the same doc.
would be useful if we could inject fragments to the validation context just like we pass a schema.

[lrlna]

Hi @nrbnlulu, could I ask you to sketch out an example usage of what you're looking for?

[nrbnlulu]

i.e

// source here might not have the needed fragments
fn parse_doc(my_ctx: &ParsingContext, source: &str, path: &PathBuf) -> anyhow::Result<()>{
    // maybe pass fragments here too?
    let doc = parser
        .parse_executable(&my_ctx.schema, &my_ctx.available_fragments, source, doc_path)?; 

   let doc_orig = doc_orig.validate(&schema, &my_ctx.available_fragments)?;
   ...
}

[lrlna]

Ah, I see what you mean.

It's a bit unusual to have fragments defined separately, since, for example, if the fragment is used in the operation definition and is not defined, the operation definition is invalid.

I think given the current structure of the Parser struct in apollo-compiler, this is going to be a fair bit of work to implement, as the best way to keep a clean API would be to implement this as a builder. And I am not convinced that's necessarily the way for the ExecutableDocument.

A possible alternative I could give you for the moment would be to concatenate your original source (which would have an impl Into<String>, making it easy to concat) with the additional fragment definitions prior to calling parser.parse_executable(). The Schema struct should already be there and doesn't need to be reparsed, which is the most "expensive" part of the whole process.

[nrbnlulu]

A possible alternative I could give you for the moment would be to concatenate your original source (which would have an impl Into, making it easy to concat) with the additional fragment definitions prior to calling parser.parse_executable(). The Schema struct should already be there and doesn't need to be reparsed, which is the most "expensive" part of the whole process.

yep already done that 🙃 ..

BTW the usecase is a codegen that saves info about fragments defined elsewhere..