Fixed issue with table location on storage root (#3094)

closes #3091

Author: FastLee

src/databricks/labs/ucx/hive_metastore/locations.py

@@ -178,7 +178,15 @@ def _external_locations(self) -> Iterable[ExternalLocation]:
             curr = queue.pop()
             num_children = len(curr.children)  # 0 - take parent
             if curr.location and (num_children > 1 or num_children == 0):
-                if curr.parent and num_children == 0 and not curr.is_jdbc():  # one table having the prefix
+                # Checking if the parent location is a valid location for external location.
+                # If the table location is a leaf location (a foldermore than 2 levels from the root)
+                # the parent folder will be considered as external location.
+                # If the table location is a root location (a folder at the root level) it will be considered as
+                # external location.
+
+                if (
+                    curr.parent and curr.parent.is_valid() and num_children == 0 and not curr.is_jdbc()
+                ):  # one table having the prefix
                     curr = curr.parent
                 assert curr.location is not None
                 external_location = ExternalLocation(curr.location, len(list(curr.all_tables())))

tests/unit/hive_metastore/test_locations.py

@@ -25,6 +25,8 @@
         "s3://databricks-e2demofieldengwest/b169/b50",
         "s3a://databricks-datasets-oregon/delta-sharing/share/open-datasets.share",
         "s3n://bucket-name/path-to-file-in-bucket",
+        "s3n://bucket-name/path-to-another-file",
+        "s3n://bucket-name",
         "gcs://test_location2/test2/table2",
         "abfss://cont1@storagetest1.dfs.core.windows.net/test2/table3",
     ],
@@ -164,8 +166,10 @@ def test_external_locations():
     tables_crawler.snapshot.return_value = [
         table_factory(["s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-1/Location/Table", ""]),
         table_factory(["s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-1/Location/Table2", ""]),
+        table_factory(["s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-1/Location2/Table", ""]),
         table_factory(["s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-23/testloc/Table3", ""]),
         table_factory(["s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-23/anotherloc/Table4", ""]),
+        table_factory(["s3://root_location", ""]),
         table_factory(["gcs://test_location2/a/b/table2", ""]),
         table_factory(["dbfs:/mnt/ucx/database1/table1", ""]),
         table_factory(["/dbfs/mnt/ucx/database2/table2", ""]),
@@ -221,7 +225,8 @@ def test_external_locations():
         ExternalLocation('jdbc:mysql://somemysql.us-east-1.rds.amazonaws.com:3306/test_db', 1),
         ExternalLocation('jdbc:providerknown://somedb.us-east-1.rds.amazonaws.com:1234/test_db', 2),
         ExternalLocation('jdbc:providerunknown://somedb.us-east-1.rds.amazonaws.com:1234/test_db', 1),
-        ExternalLocation('s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-1/Location', 2),
+        ExternalLocation('s3://root_location', 1),
+        ExternalLocation('s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-1', 3),
         ExternalLocation('s3://us-east-1-dev-account-staging-uc-ext-loc-bucket-23', 2),
         ExternalLocation('s3://us-east-1-ucx-container', 2),
     ]