Infer TLD for China based on region in account (#2040)

Co-authored-by: Viswajith Venugopal <[email protected]>

Author: sfc-gh-vvenugopal

DESCRIPTION.md

@@ -10,6 +10,7 @@ Source code is also available at: https://github.com/snowflakedb/snowflake-conne
 
 - v3.12.2 (TBD)
   - Improved implementation of `snowflake.connector.util_text.random_string` to avoid collisions.
+  - If the account specifies a region, and that region is in China, the TLD is now inferred to be snowflakecomputing.cn.
 
 - v3.12.1(August 20,2024)
   - Fixed a bug that logged the session token when renewing a session.

src/snowflake/connector/util_text.py

@@ -235,19 +235,24 @@ def _concatenate_statements(
 
 def construct_hostname(region: str | None, account: str) -> str:
     """Constructs hostname from region and account."""
+
+    def _is_china_region(r: str) -> bool:
+        # This is consistent with the Go driver:
+        # https://github.com/snowflakedb/gosnowflake/blob/f20a46475dce322f3f6b97b4a72f2807571e750b/dsn.go#L535
+        return r.lower().startswith("cn-")
+
     if region == "us-west-2":
         region = ""
     if region:
         if account.find(".") > 0:
             account = account[0 : account.find(".")]
-        top_level_domain = (
-            "com"
-            if not any(substring in region for substring in ["cn-", "CN-"])
-            else "cn"
-        )
+        top_level_domain = "cn" if _is_china_region(region) else "com"
         host = f"{account}.{region}.snowflakecomputing.{top_level_domain}"
     else:
-        host = f"{account}.snowflakecomputing.com"
+        top_level_domain = "com"
+        if account.find(".") > 0 and _is_china_region(account.split(".")[1]):
+            top_level_domain = "cn"
+        host = f"{account}.snowflakecomputing.{top_level_domain}"
     return host
 
 

test/unit/test_construct_hostname.py

@@ -48,6 +48,16 @@ def test_construct_hostname_basic():
         == "account1.cn-central-1.snowflakecomputing.cn"
     )
 
+    assert (
+        construct_hostname(None, "account1.cn-central-1")
+        == "account1.cn-central-1.snowflakecomputing.cn"
+    )
+
+    assert (
+        construct_hostname("", "account1.cn-central-1")
+        == "account1.cn-central-1.snowflakecomputing.cn"
+    )
+
     assert (
         construct_hostname("CN-central-1", "account1")
         == "account1.CN-central-1.snowflakecomputing.cn"