BUG: Recognize chained fsspec URLs

Author: snitish

pandas/io/common.py

@@ -71,7 +71,7 @@
 
 _VALID_URLS = set(uses_relative + uses_netloc + uses_params)
 _VALID_URLS.discard("")
-_RFC_3986_PATTERN = re.compile(r"^[A-Za-z][A-Za-z0-9+\-+.]*://")
+_RFC_3986_PATTERN = re.compile(r"^[A-Za-z][A-Za-z0-9+\-+.]*(::[A-Za-z0-9+\-+.]+)?://")
 
 BaseBufferT = TypeVar("BaseBufferT", bound=BaseBuffer)
 

pandas/tests/io/json/test_pandas.py

@@ -1753,6 +1753,7 @@ def test_read_timezone_information(self):
         [
             "s3://example-fsspec/",
             "gcs://another-fsspec/file.json",
+            "filecache::s3://yet-another-fsspec/file.json",
             "https://example-site.com/data",
             "some-protocol://data.txt",
         ],

pandas/tests/io/test_common.py

@@ -501,6 +501,15 @@ def test_is_fsspec_url():
     assert icom.is_fsspec_url("RFC-3986+compliant.spec://something")
 
 
+def test_is_fsspec_url_chained():
+    # GH#48978 Support chained fsspec URLs
+    # See https://filesystem-spec.readthedocs.io/en/latest/features.html#url-chaining.
+    assert icom.is_fsspec_url("filecache::s3://pandas/test.csv")
+    assert not icom.is_fsspec_url("filecache:s3://pandas/test.csv")
+    assert not icom.is_fsspec_url("filecache:::s3://pandas/test.csv")
+    assert not icom.is_fsspec_url("filecache::://pandas/test.csv")
+
+
 @pytest.mark.parametrize("encoding", [None, "utf-8"])
 @pytest.mark.parametrize("format", ["csv", "json"])
 def test_codecs_encoding(encoding, format):