How do I select a subset of columns with diesel-rs? - postgresql

Question

I am struggling for several hours now to query a subset of the available columns of a table as well as including a calculation in it. I know that it is not the best way to execute the calculation in the select query, but for now, I am just working on a prototype and it should be feasible for that.
I am using diesel-rs as a ORM for all my database actions in my back-end implementation. The data will be stored in a PostgresSQL server. The full table - as stored in the database - is created using the following query:
CREATE TABLE airports
(
id SERIAL PRIMARY KEY,
icao_code VARCHAR(4) NOT NULL UNIQUE, -- the official ICAO code of the airport
last_update TIMESTAMP NOT NULL, -- when were the information updated the last time?
country VARCHAR(2) NOT NULL, -- two letter country code
longitude REAL NOT NULL, -- with 6 decimal places
latitude REAL NOT NULL, -- with 6 decimal places
name VARCHAR NOT NULL -- just a human readable name of the airport
);
Running diesel migrations run generates the airports table definition and querying the database works without any issue.
Now I am trying to query a list of all airports (their ICAO code) with the corresponding coordinates as well as the distance to a supplied coordinate. Therefore, I created the following diesel-rs table! macro myself
table! {
airport_by_distance (icao_code) {
icao_code -> Varchar,
longitude -> Float8,
latitude -> Float8,
distance -> Float8,
}
}
as well as the struct corresponding to the diesel-rs definition:
#[derive(QueryableByName)]
#[table_name = "airport_by_distance"]
struct AirportByDistance {
icao_code: String,
longitude: f64,
latitude: f64,
distance: f64,
}
The following snipped - as of my understanding - should query the required information:
use diesel::dsl::sql_query;
let latitude = 4.000001;
let longitude = 47.000001;
let query_sql = format!("SELECT icao_code, longitude, latitude, (3959.0 * acos(cos(radians({lat})) * cos(radians(latitude)) * cos(radians(longitude) - radians({long})) + sin(radians({lat})) * sin(radians(latitude)))) AS distance FROM airports ORDER BY distance;", lat=latitude, long=longitude);
let result = match sql_query(query_sql).load::<AirportByDistance>(database_connection) {
Ok(result) => result,
Err(error) => {
error!("{:?}", error);
return Err(());
}
};
Unfortunately, executing the load method, results in a DeserializationError(Custom { kind: UnexpectedEof, error: "failed to fill whole buffer" }) error.
The query which was executed was:
SELECT icao_code,
longitude,
latitude,
(3959.0 * acos(cos(radians(4.000001)) * cos(radians(latitude)) * cos(radians(longitude) - radians(47.000001)) +
sin(radians(4.000001)) * sin(radians(latitude)))) AS distance
FROM airports
ORDER BY distance;
I took it and executed it manually, but it worked flawlessly. I even tried removing the calculation and just selecting a subset of columns, but with no luck either.
Right now I am not sure what I am doing wrong. How can I fix this issue?
EDIT with the fixed code: For the ones who are interested how to code would look like after using the helpful advice of Rasmus:
The table! macro is completely gone and the data definition struct looks like this:
#[derive(Queryable)]
struct AirportByDistance {
icao_code: String,
longitude: f32,
latitude: f32,
distance: f64,
}
The code for the querying the database is as follows:
let result = match airports.select(
(
icao_code,
longitude,
latitude,
sql::<Double>(
&format!("(3959.0 * acos(cos(radians({lat})) * cos(radians(latitude)) * cos(radians(longitude) - radians({long})) + sin(radians({lat})) * sin(radians(latitude)))) AS distance", lat=latitude_reference, long=longitude_reference)
)
)
).load::<AirportByDistance>(database_connection)
{
Ok(result) => result,
Err(error) => {
error!("{:?}", error);
return Err(());
}
};
for airport in result {
info!(
"AIRPORT: {} has {}nm distance",
airport.icao_code, airport.distance
);
}

Answer 1

I think the problem is that the deserializer don't know the raw types of the queried columns.
Try to use typed diesel names/values as much as possible, and explicit sql strings only where needed. And I don't think the "fake" table declaration airports_by_distance helps. Maybe something like this:
use diesel::sql_types::Double;
let result = a::airports
.select((
a::icao_code,
a::longitude,
a::latitude,
sql::<Double>(&format!("(3959.0 * acos(cos(radians({lat})) * cos(radians(latitude)) * cos(radians(longitude) - radians({long})) + sin(radians({lat})) * sin(radians(latitude)))) AS distance", lat=latitude, long=longitue)
))
.load::<AirportByDistance>(&db)?
(Using the table! macro manually is basically just telling diesel that such a table will exist in the actual database when running the program. If that is not true, you will get runtime errors.)