fix: Fix regression error when scanning AWS presigned URL (pola-rs#24530)

Author: nameexhaustion

crates/polars-io/src/path_utils/mod.rs

@@ -422,7 +422,8 @@ pub fn expand_paths_hive(
                         // Search after offset to prevent matching `.s3.amazonaws.com` (legacy global endpoint URL without region).
                         let region_end = offset + after_scheme[offset..].find(".amazonaws.com/")?;
 
-                        if after_scheme[..region_end].contains('/') {
+                        // Do not convert if '?' (this can be query parameters for AWS presigned URLs).
+                        if after_scheme[..region_end].contains('/') || after_scheme.contains('?') {
                             return None;
                         }
 

py-polars/polars/io/cloud/_utils.py

@@ -54,13 +54,22 @@ def _get_path_scheme(path: str | Path) -> str | None:
 
 
 def _is_aws_cloud(*, scheme: str, first_scan_path: str) -> bool:
-    return any(scheme == x for x in ["s3", "s3a"]) or (
-        (scheme == "http" or scheme == "https")
-        and 0
-        < first_scan_path.find(".s3.")
-        < first_scan_path.find(".amazonaws.com/")
-        < first_scan_path[len(scheme) + 3 :].find("/")
-    )
+    if any(scheme == x for x in ["s3", "s3a"]):
+        return True
+
+    if scheme == "http" or scheme == "https":
+        bucket_end = first_scan_path.find(".s3.")
+        region_end = first_scan_path.find(".amazonaws.com/", bucket_end + 4)
+
+        if (
+            first_scan_path.find("/", len(scheme) + 3, region_end) > 0
+            or "?" in first_scan_path
+        ):
+            return False
+
+        return 0 < bucket_end < region_end
+
+    return False
 
 
 def _is_azure_cloud(scheme: str) -> bool:

py-polars/tests/unit/io/cloud/test_cloud.py

@@ -6,6 +6,7 @@
 import pytest
 
 import polars as pl
+from polars.io.cloud._utils import _is_aws_cloud
 
 
 @pytest.mark.slow
@@ -54,3 +55,32 @@ def f() -> None:
     # Note: We get called 2 times per attempt
     if call_count != 4:
         raise AssertionError(call_count)
+
+
+def test_is_aws_cloud() -> None:
+    assert _is_aws_cloud(
+        scheme="https",
+        first_scan_path="https://bucket.s3.eu-west-1.amazonaws.com/key",
+    )
+
+    # Slash in front of amazonaws.com
+    assert not _is_aws_cloud(
+        scheme="https",
+        first_scan_path="https://bucket/.s3.eu-west-1.amazonaws.com/key",
+    )
+
+    assert not _is_aws_cloud(
+        scheme="https",
+        first_scan_path="https://bucket?.s3.eu-west-1.amazonaws.com/key",
+    )
+
+    # Legacy global endpoint
+    assert not _is_aws_cloud(
+        scheme="https", first_scan_path="https://bucket.s3.amazonaws.com/key"
+    )
+
+    # Has query parameters (e.g. presigned URL).
+    assert not _is_aws_cloud(
+        scheme="https",
+        first_scan_path="https://bucket.s3.eu-west-1.amazonaws.com/key?",
+    )