feat: Improve read_csv SQL table reading function defaults (better …

…handle dates) (#16866)
pola-rs · Jun 11, 2024 · d4462f7 · d4462f7
1 parent 806feb5
commit d4462f7
Show file tree

Hide file tree

Showing 2 changed files with 37 additions and 6 deletions.
diff --git a/crates/polars-sql/src/table_functions.rs b/crates/polars-sql/src/table_functions.rs
@@ -79,17 +79,20 @@ impl PolarsTableFunctions {
 
     #[cfg(feature = "csv")]
     fn read_csv(&self, args: &[FunctionArg]) -> PolarsResult<(String, LazyFrame)> {
-        polars_ensure!(!args.is_empty(), SQLSyntax: "`read_csv` expected a path");
+        polars_ensure!(args.len() == 1, SQLSyntax: "`read_csv` expects a single file path; found {:?} arguments", args.len());
 
         use polars_lazy::frame::LazyFileListReader;
         let path = self.get_file_path_from_arg(&args[0])?;
-        let lf = LazyCsvReader::new(&path).finish()?;
+        let lf = LazyCsvReader::new(&path)
+            .with_try_parse_dates(true)
+            .with_missing_is_null(true)
+            .finish()?;
         Ok((path, lf))
     }
 
     #[cfg(feature = "parquet")]
     fn read_parquet(&self, args: &[FunctionArg]) -> PolarsResult<(String, LazyFrame)> {
-        polars_ensure!(!args.is_empty(), SQLSyntax: "`read_parquet` expected a path");
+        polars_ensure!(args.len() == 1, SQLSyntax: "`read_parquet` expects a single file path; found {:?} arguments", args.len());
 
         let path = self.get_file_path_from_arg(&args[0])?;
         let lf = LazyFrame::scan_parquet(&path, Default::default())?;
@@ -98,15 +101,15 @@ impl PolarsTableFunctions {
 
     #[cfg(feature = "ipc")]
     fn read_ipc(&self, args: &[FunctionArg]) -> PolarsResult<(String, LazyFrame)> {
-        polars_ensure!(!args.is_empty(), SQLSyntax: "`read_ipc` expected a path");
+        polars_ensure!(args.len() == 1, SQLSyntax: "`read_ipc` expects a single file path; found {:?} arguments", args.len());
 
         let path = self.get_file_path_from_arg(&args[0])?;
         let lf = LazyFrame::scan_ipc(&path, Default::default())?;
         Ok((path, lf))
     }
     #[cfg(feature = "json")]
     fn read_ndjson(&self, args: &[FunctionArg]) -> PolarsResult<(String, LazyFrame)> {
-        polars_ensure!(!args.is_empty(), SQLSyntax: "`read_json` expected a path");
+        polars_ensure!(args.len() == 1, SQLSyntax: "`read_ndjson` expects a single file path; found {:?} arguments", args.len());
 
         use polars_lazy::frame::LazyFileListReader;
         use polars_lazy::prelude::LazyJsonLineReader;
@@ -125,7 +128,7 @@ impl PolarsTableFunctions {
             ))) => Ok(s.to_string()),
             _ => polars_bail!(
                 SQLSyntax:
-                "only a single quoted string is accepted for the parameter; found: {}", arg,
+                "expected a valid file path as a single-quoted string; found: {}", arg,
             ),
         }
     }

diff --git a/py-polars/tests/unit/sql/test_miscellaneous.py b/py-polars/tests/unit/sql/test_miscellaneous.py
@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+from datetime import date
 from pathlib import Path
 
 import pytest
@@ -241,3 +242,30 @@ def test_values_clause_table_registration() -> None:
         res2 = ctx.execute("SELECT x, y FROM tbl")
         for res in (res1, res2):
             assert res.to_dict(as_series=False) == {"x": [-1], "y": [1]}
+
+
+def test_read_csv(tmp_path: Path) -> None:
+    # check empty string vs null, parsing of dates, etc
+    df = pl.DataFrame(
+        {
+            "label": ["lorem", None, "", "ipsum"],
+            "num": [-1, None, 0, 1],
+            "dt": [
+                date(1969, 7, 5),
+                date(1999, 12, 31),
+                date(2077, 10, 10),
+                None,
+            ],
+        }
+    )
+    csv_target = tmp_path / "test_sql_read.csv"
+    df.write_csv(csv_target)
+
+    res = pl.sql(f"SELECT * FROM read_csv('{csv_target}')").collect()
+    assert_frame_equal(df, res)
+
+    with pytest.raises(
+        SQLSyntaxError,
+        match="`read_csv` expects a single file path; found 3 arguments",
+    ):
+        pl.sql("SELECT * FROM read_csv('a','b','c')")