From 3b7155dfa8c766f310facc77278936eb7babdabd Mon Sep 17 00:00:00 2001
From: Cathy Ouyang <cathyo@google.com>
Date: Fri, 20 Oct 2023 16:54:30 -0700
Subject: [PATCH] fix: Blob.from_string parse storage uri with regex

---
 google/cloud/storage/blob.py | 13 +++++++------
 tests/unit/test_blob.py      | 14 +++++++++++---
 2 files changed, 18 insertions(+), 9 deletions(-)

diff --git a/google/cloud/storage/blob.py b/google/cloud/storage/blob.py
index aebf24c26..33998f81a 100644
--- a/google/cloud/storage/blob.py
+++ b/google/cloud/storage/blob.py
@@ -134,7 +134,9 @@
     "Blob.download_as_string() is deprecated and will be removed in future. "
     "Use Blob.download_as_bytes() instead."
 )
-
+_GS_URL_REGEX_PATTERN = re.compile(
+    r"(?P<scheme>gs)://(?P<bucket_name>[a-z0-9_.-]+)/(?P<object_name>.+)"
+)
 
 _DEFAULT_CHUNKSIZE = 104857600  # 1024 * 1024 B * 100 = 100 MB
 _MAX_MULTIPART_SIZE = 8388608  # 8 MB
@@ -403,12 +405,11 @@ def from_string(cls, uri, client=None):
         """
         from google.cloud.storage.bucket import Bucket
 
-        scheme, netloc, path, query, frag = urlsplit(uri)
-        if scheme != "gs":
+        match = _GS_URL_REGEX_PATTERN.match(uri)
+        if not match:
             raise ValueError("URI scheme must be gs")
-
-        bucket = Bucket(client, name=netloc)
-        return cls(path[1:], bucket)
+        bucket = Bucket(client, name=match.group("bucket_name"))
+        return cls(match.group("object_name"), bucket)
 
     def generate_signed_url(
         self,
diff --git a/tests/unit/test_blob.py b/tests/unit/test_blob.py
index cb164f6e2..d5058e23c 100644
--- a/tests/unit/test_blob.py
+++ b/tests/unit/test_blob.py
@@ -5819,13 +5819,21 @@ def test_from_string_w_valid_uri(self):
         from google.cloud.storage.blob import Blob
 
         client = self._make_client()
-        uri = "gs://BUCKET_NAME/b"
-        blob = Blob.from_string(uri, client)
+        basic_uri = "gs://bucket_name/b"
+        blob = Blob.from_string(basic_uri, client)
 
         self.assertIsInstance(blob, Blob)
         self.assertIs(blob.client, client)
         self.assertEqual(blob.name, "b")
-        self.assertEqual(blob.bucket.name, "BUCKET_NAME")
+        self.assertEqual(blob.bucket.name, "bucket_name")
+
+        nested_uri = "gs://bucket_name/path1/path2/b#name"
+        blob = Blob.from_string(nested_uri, client)
+
+        self.assertIsInstance(blob, Blob)
+        self.assertIs(blob.client, client)
+        self.assertEqual(blob.name, "path1/path2/b#name")
+        self.assertEqual(blob.bucket.name, "bucket_name")
 
     def test_from_string_w_invalid_uri(self):
         from google.cloud.storage.blob import Blob