fix: Column selection wasn't applied when reading CSV with no rows (p…

…ola-rs#16739)

crates/polars-io/src/csv/read/read_impl.rs

@@ -64,17 +64,18 @@ pub(crate) fn cast_columns(
     };
 
     if parallel {
-        let cols = df
-            .get_columns()
-            .iter()
-            .map(|s| {
-                if let Some(fld) = to_cast.iter().find(|fld| fld.name().as_str() == s.name()) {
-                    cast_fn(s, fld)
-                } else {
-                    Ok(s.clone())
-                }
-            })
-            .collect::<PolarsResult<Vec<_>>>()?;
+        let cols = POOL.install(|| {
+            df.get_columns()
+                .into_par_iter()
+                .map(|s| {
+                    if let Some(fld) = to_cast.iter().find(|fld| fld.name().as_str() == s.name()) {
+                        cast_fn(s, fld)
+                    } else {
+                        Ok(s.clone())
+                    }
+                })
+                .collect::<PolarsResult<Vec<_>>>()
+        })?;
         *df = unsafe { DataFrame::new_no_checks(cols) }
     } else {
         // cast to the original dtypes in the schema
@@ -473,7 +474,20 @@ impl<'a> CoreReader<'a> {
 
         // An empty file with a schema should return an empty DataFrame with that schema
         if bytes.is_empty() {
-            let mut df = DataFrame::from(self.schema.as_ref());
+            let mut df = if projection.len() == self.schema.len() {
+                DataFrame::from(self.schema.as_ref())
+            } else {
+                DataFrame::from(
+                    &projection
+                        .iter()
+                        .map(|&i| self.schema.get_at_index(i).unwrap())
+                        .map(|(name, dtype)| Field {
+                            name: name.clone(),
+                            dtype: dtype.clone(),
+                        })
+                        .collect::<Schema>(),
+                )
+            };
             if let Some(ref row_index) = self.row_index {
                 df.insert_column(0, Series::new_empty(&row_index.name, &IDX_DTYPE))?;
             }

py-polars/tests/unit/io/test_csv.py

@@ -2160,3 +2160,24 @@ def test_read_csv_dtypes_deprecated() -> None:
         schema={"a": pl.Int8, "b": pl.Int8, "c": pl.Int8},
     )
     assert_frame_equal(df, expected)
+
+
+def test_projection_applied_on_file_with_no_rows_16606(tmp_path: Path) -> None:
+    tmp_path.mkdir(exist_ok=True)
+
+    path = tmp_path / "data.csv"
+
+    data = """\
+a,b,c,d
+"""
+
+    with path.open("w") as f:
+        f.write(data)
+
+    columns = ["a", "b"]
+
+    out = pl.read_csv(path, columns=columns).columns
+    assert out == columns
+
+    out = pl.scan_csv(path).select(columns).collect().columns
+    assert out == columns