Expose real string quotes in a tree to don't mislead about tree-sitter-python

[ahlinc]

The exposed token is ", but the underlying source text has the full original token.

Yes, the token has a correct range that allows to access a correct piece in an original input.
But with predicates that is much slower because they work outside of a query engine as post filtering and require access to an input.

If there will be correct terminals then queries like bellow would work without predicates:

; Strings with only triple single quotes
  (string "'''" @quote) @str

; Strings with triple quotes
  (string [ "'''" "\"\"\"" ] @quote) @str

And in a visual tree representation there will be no misunderstanding:

I think it should become a bad practice to hide actual terminals that define syntax structure and replace them to something another. But for now this exist in this grammar for historical reasons.

[tausbn]

I'm not sure I see what the problem is. The exposed token is ", but the underlying source text has the full original token.

For instance, the following query can be used to match only strings delimited with triple (single) quotes:

(
  (string "\"" @quote) @str
  (#match? @quote "'''")
)

[andymass]

Have a similar problem, I'm looking for a query to target only the quote characters, not the entire string. For many languages this works with (_ "\"" "\"") or similar, but not for python. Is there an easy query like this?