From 2feb807e1eee2271141c96ed37f184296dead659 Mon Sep 17 00:00:00 2001
From: IlyaFaer <gurovilya9@gmail.com>
Date: Wed, 26 Jun 2024 11:24:45 +0400
Subject: [PATCH] feat(filesystem): use only netloc and scheme for fingerprint

---
 dlt/common/storages/configuration.py            | 15 +++++++++++++--
 tests/load/filesystem/test_filesystem_client.py | 16 +++++++++++-----
 2 files changed, 24 insertions(+), 7 deletions(-)

diff --git a/dlt/common/storages/configuration.py b/dlt/common/storages/configuration.py
index 09beb0015e..b2bdb3a7b6 100644
--- a/dlt/common/storages/configuration.py
+++ b/dlt/common/storages/configuration.py
@@ -111,8 +111,19 @@ def resolve_credentials_type(self) -> Type[CredentialsConfiguration]:
         return self.PROTOCOL_CREDENTIALS.get(self.protocol) or Optional[CredentialsConfiguration]  # type: ignore[return-value]
 
     def fingerprint(self) -> str:
-        """Returns a fingerprint of bucket_url"""
-        return digest128(self.bucket_url) if self.bucket_url else ""
+        """Returns a fingerprint of bucket schema and netloc.
+
+        Returns:
+            str: Fingerprint.
+        """
+        if not self.bucket_url:
+            return ""
+
+        if self.is_local_path(self.bucket_url):
+            return digest128("")
+
+        uri = urlparse(self.bucket_url)
+        return digest128(self.bucket_url.replace(uri.path, ""))
 
     def __str__(self) -> str:
         """Return displayable destination location"""
diff --git a/tests/load/filesystem/test_filesystem_client.py b/tests/load/filesystem/test_filesystem_client.py
index fbfd08271b..53e54c2f06 100644
--- a/tests/load/filesystem/test_filesystem_client.py
+++ b/tests/load/filesystem/test_filesystem_client.py
@@ -39,11 +39,17 @@ def logger_autouse() -> None:
 ]
 
 
-def test_filesystem_destination_configuration() -> None:
-    assert FilesystemDestinationClientConfiguration().fingerprint() == ""
-    assert FilesystemDestinationClientConfiguration(
-        bucket_url="s3://cool"
-    ).fingerprint() == digest128("s3://cool")
+@pytest.mark.parametrize(
+    "url, exp",
+    (
+        (None, ""),
+        ("/path/path2", digest128("")),
+        ("s3://cool", digest128("s3://cool")),
+        ("s3://cool.domain/path/path2", digest128("s3://cool.domain")),
+    ),
+)
+def test_filesystem_destination_configuration(url, exp) -> None:
+    assert FilesystemDestinationClientConfiguration(bucket_url=url).fingerprint() == exp
 
 
 @pytest.mark.parametrize("write_disposition", ("replace", "append", "merge"))