From 04902fd800532fd9afa2024c08dae1abc0c8eca8 Mon Sep 17 00:00:00 2001
From: Ritchie Vink <ritchie46@gmail.com>
Date: Thu, 9 Feb 2023 14:45:27 +0100
Subject: [PATCH 1/2] fix(rust, python): respect skip_rows in glob parsing csv

---
 polars/polars-lazy/src/frame/csv.rs      | 14 --------------
 py-polars/Cargo.lock                     |  5 +++--
 py-polars/tests/unit/io/test_lazy_csv.py | 24 ++++++++++++++++++++++++
 3 files changed, 27 insertions(+), 16 deletions(-)

diff --git a/polars/polars-lazy/src/frame/csv.rs b/polars/polars-lazy/src/frame/csv.rs
index 5788610b41b2..ea9ccf6c2e36 100644
--- a/polars/polars-lazy/src/frame/csv.rs
+++ b/polars/polars-lazy/src/frame/csv.rs
@@ -302,10 +302,6 @@ impl<'a> LazyCsvReader<'a> {
                     let path = r.map_err(|e| PolarsError::ComputeError(format!("{e}").into()))?;
                     let mut builder = self.clone();
                     builder.path = path;
-                    if builder.skip_rows > 0 {
-                        builder.skip_rows = 0;
-                        builder.n_rows = None;
-                    }
                     // do no rechunk yet.
                     builder.rechunk = false;
                     builder.finish_impl()
@@ -314,16 +310,6 @@ impl<'a> LazyCsvReader<'a> {
             // set to false, as the csv parser has full thread utilization
             concat_impl(&lfs, self.rechunk, false, true)
                 .map_err(|_| PolarsError::ComputeError("no matching files found".into()))
-                .map(|lf| {
-                    if self.skip_rows != 0 || self.n_rows.is_some() {
-                        lf.slice(
-                            self.skip_rows as i64,
-                            self.n_rows.unwrap_or(usize::MAX) as IdxSize,
-                        )
-                    } else {
-                        lf
-                    }
-                })
         } else {
             self.finish_impl()
         }
diff --git a/py-polars/Cargo.lock b/py-polars/Cargo.lock
index 687552e27b8e..dab15d6ca642 100644
--- a/py-polars/Cargo.lock
+++ b/py-polars/Cargo.lock
@@ -84,8 +84,9 @@ dependencies = [
 
 [[package]]
 name = "arrow2"
-version = "0.15.0"
-source = "git+https://github.com/ritchie46/arrow2?branch=mmap_slice2#685cf49da02a1c94a501aab65535f6fbbcd7cbd7"
+version = "0.16.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "7a4c5b03335bc1cb0fd9f5297f8fd3bbfd6fb04f3cb0bc7d6c91b7128cb8336a"
 dependencies = [
  "ahash",
  "arrow-format",
diff --git a/py-polars/tests/unit/io/test_lazy_csv.py b/py-polars/tests/unit/io/test_lazy_csv.py
index 65548d1f4266..6022f7590b52 100644
--- a/py-polars/tests/unit/io/test_lazy_csv.py
+++ b/py-polars/tests/unit/io/test_lazy_csv.py
@@ -100,3 +100,27 @@ def test_lazy_n_rows(foods_file_path: Path) -> None:
 def test_scan_slice_streaming(foods_file_path: Path) -> None:
     df = pl.scan_csv(foods_file_path).head(5).collect(streaming=True)
     assert df.shape == (5, 4)
+
+
+def test_glob_skip_rows() -> None:
+
+    with tempfile.TemporaryDirectory() as temp_dir:
+        for i in range(2):
+            file_path = Path(temp_dir) / f"test_{i}.csv"
+            with open(file_path, "w") as f:
+                f.write(
+                    f"""
+metadata goes here
+file number {i}
+foo,bar,baz
+1,2,3
+4,5,6
+7,8,9
+        """
+                )
+        file_path = Path(temp_dir) / "*.csv"
+        assert pl.read_csv(file_path, skip_rows=2).to_dict(False) == {
+            "foo": [1, 4, 7, 1, 4, 7],
+            "bar": [2, 5, 8, 2, 5, 8],
+            "baz": [3, 6, 9, 3, 6, 9],
+        }

From 5fd1c97b272b3eb8261c04ec42398e15e3dd53a4 Mon Sep 17 00:00:00 2001
From: Ritchie Vink <ritchie46@gmail.com>
Date: Thu, 9 Feb 2023 14:58:16 +0100
Subject: [PATCH 2/2] black

---
 py-polars/tests/unit/io/test_lazy_csv.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/py-polars/tests/unit/io/test_lazy_csv.py b/py-polars/tests/unit/io/test_lazy_csv.py
index 6022f7590b52..cd7ecfb3a361 100644
--- a/py-polars/tests/unit/io/test_lazy_csv.py
+++ b/py-polars/tests/unit/io/test_lazy_csv.py
@@ -103,7 +103,6 @@ def test_scan_slice_streaming(foods_file_path: Path) -> None:
 
 
 def test_glob_skip_rows() -> None:
-
     with tempfile.TemporaryDirectory() as temp_dir:
         for i in range(2):
             file_path = Path(temp_dir) / f"test_{i}.csv"