Converting a query result to a Polars DataFrame

[nk9]

I have this macro which converts a struct result from a query!() into a Polars DataFrame:

macro_rules! struct_to_dataframe {
    ($input:expr, [$($field:ident),+]) => {
        {
            let len = $input.len().to_owned();

            // Extract the field values into separate vectors
            $(let mut $field = Vec::with_capacity(len);)*

            for e in $input.into_iter() {
                $($field.push(e.$field);)*
            }
            df! {
                $(stringify!($field) => $field,)*
            }
        }
    };
}

This works great for queries run with query!. However, a query run with sqlx::query returns a set of PgRow objects which return different types for each column. And when I adapt the relevant line:

$($field.push(e.get(stringify!($field)));)*

I get this error:

 1  error[E0282]: type annotations needed
    --> rust/utils.rs:26:32
     |
 26  |             $(let mut $field = Vec::with_capacity(len);)*
     |                                ^^^^^^^^^^^^^^^^^^^^^^^ cannot infer type
     |
    ::: rust/load_all.rs:532:9
     |
 532 |         struct_to_dataframe!(db_eds, [year, state, county, metro_name, ed, district_id]).unwrap();
     |         -------------------------------------------------------------------------------- in this macro invocation

So, can I still use this approach to convert the result of a normal sqlx::query() call to a Polars DataFrame?

[nk9]

I solved this by using query_as() and providing a struct with type information. So:

#[derive(sqlx::FromRow, Debug)]
struct User {
    age: i16,
    name: String,
}

// …

    let users: Vec<User> = sqlx::query_as(
        "SELECT age, name FROM users")
        .fetch_all(&pool)
        .await;

    let users_df: DataFrame = struct_to_dataframe!(users, [age, name]).unwrap();

This way, I am still able to use dot notation to access the attributes, and the type information is available so the compiler is happy.