feat: assure it respects .gz endings

Author: Skn0tt

pandas/io/common.py

@@ -875,9 +875,25 @@ def __init__(
         super().__init__(name=name, mode=mode, fileobj=fileobj, **kwargs)
 
     @classmethod
-    def open(cls, mode="r", **kwargs):
+    def open(cls, name=None, mode="r", **kwargs):
         mode = mode.replace("b", "")
-        return super().open(mode=mode, **kwargs)
+        return super().open(name=name, mode=cls.extend_mode(name, mode), **kwargs)
+
+    @classmethod
+    def extend_mode(
+        cls, name: FilePath | ReadBuffer[bytes] | WriteBuffer[bytes], mode: str
+    ) -> str:
+        if mode != "w":
+            return mode
+        if isinstance(name, (os.PathLike, str)):
+            filename = Path(name)
+            if filename.suffix == ".gz":
+                return mode + ":gz"
+            elif filename.suffix == ".xz":
+                return mode + ":xz"
+            elif filename.suffix == ".bz2":
+                return mode + ":bz2"
+        return mode
 
     def infer_filename(self):
         """

pandas/tests/io/parser/test_compression.py

@@ -5,6 +5,7 @@
 
 import os
 from pathlib import Path
+import tarfile
 import zipfile
 
 import pytest
@@ -180,3 +181,26 @@ def test_ignore_compression_extension(all_parsers):
             Path(path_zip).write_text(Path(path_csv).read_text())
 
             tm.assert_frame_equal(parser.read_csv(path_zip, compression=None), df)
+
+
+@skip_pyarrow
+def test_writes_tar_gz(all_parsers):
+    parser = all_parsers
+    data = DataFrame(
+        {
+            "Country": ["Venezuela", "Venezuela"],
+            "Twitter": ["Hugo Chávez Frías", "Henrique Capriles R."],
+        }
+    )
+    with tm.ensure_clean("test.tar.gz") as tar_path:
+        data.to_csv(tar_path, index=False)
+
+        # test that read_csv infers .tar.gz to gzip:
+        tm.assert_frame_equal(parser.read_csv(tar_path), data)
+
+        # test that file is indeed gzipped:
+        with tarfile.open(tar_path, "r:gz") as tar:
+            result = parser.read_csv(
+                tar.extractfile(tar.getnames()[0]), compression="infer"
+            )
+            tm.assert_frame_equal(result, data)