From 78249af38a6227f373469880749251bc705c4daa Mon Sep 17 00:00:00 2001
From: Bo Wang <wang.b.frank@gmail.com>
Date: Mon, 11 Nov 2024 19:22:34 +0100
Subject: [PATCH] Parse the query parameter of url without using
 infer_storage_options (#912)

---
 s3fs/core.py            |  7 ++++---
 s3fs/tests/test_s3fs.py | 37 +++++++++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 3 deletions(-)

diff --git a/s3fs/core.py b/s3fs/core.py
index 9b9a3f57..a4a3691f 100644
--- a/s3fs/core.py
+++ b/s3fs/core.py
@@ -14,7 +14,7 @@
 
 import fsspec  # noqa: F401
 from fsspec.spec import AbstractBufferedFile
-from fsspec.utils import infer_storage_options, tokenize, setup_logging as setup_logger
+from fsspec.utils import tokenize, setup_logging as setup_logger
 from fsspec.asyn import (
     AsyncFileSystem,
     AbstractAsyncStreamedFile,
@@ -391,8 +391,9 @@ def _get_kwargs_from_urls(urlpath):
         Assume that we want to use version_aware mode for
         the filesystem.
         """
-        url_storage_opts = infer_storage_options(urlpath)
-        url_query = url_storage_opts.get("url_query")
+        from urllib.parse import urlsplit
+
+        url_query = urlsplit(urlpath).query
         out = {}
         if url_query is not None:
             from urllib.parse import parse_qs
diff --git a/s3fs/tests/test_s3fs.py b/s3fs/tests/test_s3fs.py
index b64907d1..612db73f 100644
--- a/s3fs/tests/test_s3fs.py
+++ b/s3fs/tests/test_s3fs.py
@@ -79,6 +79,7 @@ def s3_base():
         os.environ["AWS_SECRET_ACCESS_KEY"] = "foo"
     if "AWS_ACCESS_KEY_ID" not in os.environ:
         os.environ["AWS_ACCESS_KEY_ID"] = "foo"
+    os.environ.pop("AWS_PROFILE", None)
 
     print("server up")
     yield
@@ -2162,6 +2163,42 @@ def test_via_fsspec(s3):
         assert f.read() == b"hello"
 
 
+@pytest.mark.parametrize(
+    ["raw_url", "expected_url", "expected_version_aware"],
+    [
+        (
+            "s3://arn:aws:s3:us-west-2:123456789012:accesspoint/abc/123.jpg",
+            "arn:aws:s3:us-west-2:123456789012:accesspoint/abc/123.jpg",
+            False,
+        ),
+        (
+            "s3://arn:aws:s3:us-west-2:123456789012:accesspoint/abc/123.jpg?versionId=some_version_id",
+            "arn:aws:s3:us-west-2:123456789012:accesspoint/abc/123.jpg?versionId=some_version_id",
+            True,
+        ),
+        (
+            "s3://xyz/abc/123.jpg",
+            "xyz/abc/123.jpg",
+            False,
+        ),
+        (
+            "s3://xyz/abc/123.jpg?versionId=some_version_id",
+            "xyz/abc/123.jpg?versionId=some_version_id",
+            True,
+        ),
+    ],
+)
+def test_fsspec_url_to_fs_compatability(
+    s3, raw_url, expected_url, expected_version_aware
+):
+    import fsspec
+
+    fs, url = fsspec.url_to_fs(raw_url)
+    assert isinstance(fs, type(s3))
+    assert fs.version_aware is expected_version_aware
+    assert url == expected_url
+
+
 def test_repeat_exists(s3):
     fn = "s3://" + test_bucket_name + "/file1"
     s3.touch(fn)