[Runtime Environment] Parse special characters in private Git URIs (#…

…28250)

python/ray/_private/runtime_env/packaging.py

@@ -156,8 +156,8 @@ def parse_uri(pkg_uri: str) -> Tuple[Protocol, str]:
                 netloc='_ray_pkg_029f88d5ecc55e1e4d64fc6e388fd103.zip'
             )
             -> ("gcs", "_ray_pkg_029f88d5ecc55e1e4d64fc6e388fd103.zip")
-    For HTTPS URIs, the netloc will have '.' replaced with '_', and
-    the path will have '/' replaced with '_'. The package name will be the
+    For HTTPS URIs, the netloc will have '.', ':', and '@' swapped with '_',
+    and the path will have '/' replaced with '_'. The package name will be the
     adjusted path with 'https_' prepended.
         urlparse(
             "https://github.com/shrekris-anyscale/test_module/archive/HEAD.zip"
@@ -209,9 +209,10 @@ def parse_uri(pkg_uri: str) -> Tuple[Protocol, str]:
     if protocol == Protocol.S3 or protocol == Protocol.GS:
         return (protocol, f"{protocol.value}_{uri.netloc}{uri.path.replace('/', '_')}")
     elif protocol == Protocol.HTTPS:
+        parsed_netloc = uri.netloc.replace(".", "_").replace(":", "_").replace("@", "_")
         return (
             protocol,
-            f"https_{uri.netloc.replace('.', '_')}{uri.path.replace('/', '_')}",
+            f"https_{parsed_netloc}{uri.path.replace('/', '_')}",
         )
     elif protocol == Protocol.FILE:
         return (

python/ray/tests/test_runtime_env_packaging.py

@@ -373,6 +373,49 @@ def test_unzip_with_matching_subdirectory_names(
         )
 
 
+class TestParseUri:
+    @pytest.mark.parametrize(
+        "parsing_tuple",
+        [
+            ("gcs://file.zip", Protocol.GCS, "file.zip"),
+            ("s3://bucket/file.zip", Protocol.S3, "s3_bucket_file.zip"),
+            ("https://test.com/file.zip", Protocol.HTTPS, "https_test_com_file.zip"),
+            ("gs://bucket/file.zip", Protocol.GS, "gs_bucket_file.zip"),
+        ],
+    )
+    def test_parsing_basic(self, parsing_tuple):
+        uri, protocol, package_name = parsing_tuple
+        parsed_protocol, parsed_package_name = parse_uri(uri)
+
+        assert protocol == parsed_protocol
+        assert package_name == parsed_package_name
+
+    @pytest.mark.parametrize(
+        "parsing_tuple",
+        [
+            (
+                "https://username:[email protected]/repo/archive/commit_hash.zip",
+                "https_username_PAT_github_com_repo_archive_commit_hash.zip",
+            ),
+            (
+                (
+                    "https://un:[email protected]/user/repo/-/"
+                    "archive/commit_hash/repo-commit_hash.zip"
+                ),
+                (
+                    "https_un_pwd_gitlab_com_user_repo_-_"
+                    "archive_commit_hash_repo-commit_hash.zip"
+                ),
+            ),
+        ],
+    )
+    def test_parse_private_git_https_uris(self, parsing_tuple):
+        raw_uri, parsed_uri = parsing_tuple
+        parsed_protocol, parsed_package_name = parse_uri(raw_uri)
+        assert parsed_protocol == Protocol.HTTPS
+        assert parsed_package_name == parsed_uri
+
+
 @pytest.mark.skipif(sys.platform == "win32", reason="Fails on windows")
 def test_travel(tmp_path):
     dir_paths = set()
@@ -433,23 +476,6 @@ def handler(path):
     assert dir_paths == visited_dir_paths
 
 
-@pytest.mark.parametrize(
-    "parsing_tuple",
-    [
-        ("gcs://file.zip", Protocol.GCS, "file.zip"),
-        ("s3://bucket/file.zip", Protocol.S3, "s3_bucket_file.zip"),
-        ("https://test.com/file.zip", Protocol.HTTPS, "https_test_com_file.zip"),
-        ("gs://bucket/file.zip", Protocol.GS, "gs_bucket_file.zip"),
-    ],
-)
-def test_parsing(parsing_tuple):
-    uri, protocol, package_name = parsing_tuple
-    parsed_protocol, parsed_package_name = parse_uri(uri)
-
-    assert protocol == parsed_protocol
-    assert package_name == parsed_package_name
-
-
 def test_is_whl_uri():
     assert is_whl_uri("gcs://my-package.whl")
     assert not is_whl_uri("gcs://asdf.zip")