use crate::duckdb_source::_helpers::catalog::{catalog_tables, described_columns};
use crate::duckdb_source::_helpers::connection::text_rows;
use crate::duckdb_source::_helpers::query::subquery;
use crate::duckdb_source::_helpers::type_map::source_column;
use crate::duckdb_source::constants::CATALOG_QUERIES;
use crate::duckdb_source::errors::DuckDbSourceError;
use crate::duckdb_source::models::DuckDbReader;
use crate::postgres_source::models::SourceTable;
use crate::project::models::SourceConfig;
pub async fn read_catalog(
reader: &DuckDbReader,
source: &SourceConfig,
) -> Result<Vec<SourceTable>, DuckDbSourceError> {
let named: Vec<(&String, &str)> = source
.tables
.iter()
.filter_map(|(name, table)| Some((name, table.sql.as_deref()?)))
.collect();
let catalog = match source.include.is_empty() {
true => 0,
false => CATALOG_QUERIES.len(),
};
let queries: Vec<String> = CATALOG_QUERIES[..catalog]
.iter()
.map(|query| (*query).to_owned())
.chain(
named
.iter()
.map(|(_, sql)| format!("DESCRIBE SELECT * FROM {}", subquery(sql))),
)
.collect();
let answers = text_rows(reader, queries).await.map_err(|(index, error)| {
match index
.checked_sub(catalog)
.and_then(|query| named.get(query))
{
Some((name, _)) => DuckDbSourceError::Invalid(format!("query table {name}: {error}")),
None => error,
}
})?;
let (listed, described) = answers.split_at(catalog);
let mut tables: Vec<SourceTable> = match catalog {
0 => Vec::new(),
_ => catalog_tables(listed, (&source.include, &source.exclude))?,
};
for ((name, _), rows) in named.iter().zip(described) {
tables.push(SourceTable {
split_pages: None,
schema: String::new(),
name: (*name).clone(),
columns: described_columns(rows)
.iter()
.map(|(column, data_type, nullable)| source_column(column, data_type, *nullable))
.collect(),
primary_key: Vec::new(),
indexed_columns: Vec::new(),
estimated_rows: -1,
estimated_bytes: -1,
comment: None,
listed_keys: Vec::new(),
});
}
Ok(tables)
}