atlas_lineage | 🎉 Initial commit.

Signed-off-by: mgorsk1 <gorskimariusz13@gmail.com>

common/setup.cfg

@@ -1,6 +1,6 @@
 [flake8]
 format = pylint
-exclude = .svc,CVS,.bzr,.hg,.git,__pycache__,venv
+exclude = .svc,CVS,.bzr,.hg,.git,__pycache__,venv,.venv
 max-complexity = 10
 max-line-length = 120
 

databuilder/setup.cfg

@@ -8,6 +8,7 @@ exclude =
   .git,
   __pycache__,
   venv,
+  .venv,
   build,
   databuilder/sql_parser/usage/presto/antlr_generated
 max-complexity = 10

frontend/amundsen_application/api/utils/metadata_utils.py

@@ -2,7 +2,6 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import logging
-import urllib.parse
 
 from dataclasses import dataclass
 from marshmallow import EXCLUDE
@@ -29,14 +28,16 @@ def __str__(self) -> str:
 
     @classmethod
     def from_uri(cls, uri: str) -> 'TableUri':
-        parsed = urllib.parse.urlparse(uri)
-        cluster, schema = parsed.netloc.rsplit('.', 1)
-        return TableUri(
-            database=parsed.scheme,
-            cluster=cluster,
-            schema=schema,
-            table=parsed.path.lstrip('/')
-        )
+        """
+        TABLE_KEY_FORMAT = '{db}://{cluster}.{schema}/{tbl}'
+        """
+        pattern = re.compile(r'^(?P<database>.*?)://(?P<cluster>.*)\.(?P<schema>.*?)/(?P<table>.*?)$', re.X)
+
+        groups = pattern.match(uri)
+
+        spec = groups.groupdict() if groups else {}
+
+        return TableUri(**spec)
 
 
 def marshall_table_partial(table_dict: Dict) -> Dict:

frontend/setup.cfg

@@ -1,6 +1,6 @@
 [flake8]
 format = pylint
-exclude = .svc,CVS,.bzr,.hg,.git,__pycache__,venv,venv3,node_modules
+exclude = .svc,CVS,.bzr,.hg,.git,__pycache__,venv,venv3,.venv,node_modules
 max-complexity = 10
 max-line-length = 120
 ignore = W503

frontend/tests/unit/utils/test_metadata_utils.py

@@ -1,13 +1,13 @@
 # Copyright Contributors to the Amundsen project.
 # SPDX-License-Identifier: Apache-2.0
 
-from typing import Dict
 import unittest
+from typing import Dict
 
+from amundsen_application import create_app
 from amundsen_application.api.utils.metadata_utils import _convert_prog_descriptions, _sort_prog_descriptions, \
     _parse_editable_rule, is_table_editable, TableUri
 from amundsen_application.config import MatchRuleObject
-from amundsen_application import create_app
 
 local_app = create_app('amundsen_application.config.TestConfig', 'tests/templates')
 
@@ -197,11 +197,21 @@ def test_schema_table_match_rule(self) -> None:
 class TableUriObject(unittest.TestCase):
     def test_simple_constructor(self) -> None:
         uri = TableUri("db", "clstr", "schm", "tbl")
-        self.assertEqual(str(uri), 'db://clstr.schm/tbl')
+        self.assertEqual('db://clstr.schm/tbl', str(uri))
 
     def test_from_uri_factory(self) -> None:
-        uri = TableUri.from_uri("db://clstr.with.dots.schm/tbl/with/slashes")
-        self.assertEqual(uri.database, 'db')
-        self.assertEqual(uri.cluster, 'clstr.with.dots')
-        self.assertEqual(uri.schema, 'schm')
-        self.assertEqual(uri.table, 'tbl/with/slashes')
+        params = [
+            ('db', 'cluster', 'schema', 'table'),
+            ('db_underscore', 'cluster.with.dots', 'schema', 'table/with/slashes'),
+            ('db.dot', 'cluster.dot', 'schema', 'table.dot'),
+            ('hive_table', 'demo', 'test_schema', 'test_table')
+        ]
+
+        for db, cluster, schema, table in params:
+            with self.subTest():
+                _uri = f'{db}://{cluster}.{schema}/{table}'
+                uri = TableUri.from_uri(_uri)
+                self.assertEqual(db, uri.database)
+                self.assertEqual(cluster, uri.cluster)
+                self.assertEqual(schema, uri.schema)
+                self.assertEqual(table, uri.table)