chore: upgrade to marshmallow 3 (#963)

* chore: upgrade to marshmallow 3 and marshmallow3-annotations

Signed-off-by: Dmitriy Kunitskiy <dkunitskiy@lyft.com>

* typos

Signed-off-by: Dmitriy Kunitskiy <dkunitskiy@lyft.com>

Author: Dmitriy Kunitskiy

.gitignore

@@ -16,6 +16,7 @@ build/
 .coverage
 .mypy_cache
 .pytest_cache
+.python-version
 
 
 npm-debug.log

amundsen_application/api/preview/v0.py

@@ -9,6 +9,7 @@
 
 from flask import Response, jsonify, make_response, request, current_app as app
 from flask.blueprints import Blueprint
+from marshmallow import ValidationError
 from werkzeug.utils import import_string
 
 from amundsen_application.models.preview_data import PreviewDataSchema
@@ -49,11 +50,11 @@ def get_table_preview() -> Response:
         preview_data = json.loads(response.data).get('preview_data')
         if status_code == HTTPStatus.OK:
             # validate the returned table preview data
-            data, errors = PreviewDataSchema().load(preview_data)
-            if not errors:
+            try:
+                data = PreviewDataSchema().load(preview_data)
                 payload = jsonify({'previewData': data, 'msg': 'Success'})
-            else:
-                logging.error('Preview data dump returned errors: ' + str(errors))
+            except ValidationError as err:
+                logging.error('Preview data dump returned errors: ' + str(err.messages))
                 raise Exception('The preview client did not return a valid PreviewData object')
         else:
             message = 'Encountered error: Preview client request failed with code ' + str(status_code)

amundsen_application/api/utils/metadata_utils.py

@@ -5,6 +5,7 @@
 import urllib.parse
 
 from dataclasses import dataclass
+from marshmallow import EXCLUDE
 from typing import Any, Dict, List
 
 from amundsen_common.models.dashboard import DashboardSummary, DashboardSummarySchema
@@ -46,10 +47,9 @@ def marshall_table_partial(table_dict: Dict) -> Dict:
 
     TODO - Unify data format returned by search and metadata.
     """
-    schema = PopularTableSchema(strict=True)
-    # TODO: consider migrating to validate() instead of roundtripping
-    table: PopularTable = schema.load(table_dict).data
-    results = schema.dump(table).data
+    schema = PopularTableSchema()
+    table: PopularTable = schema.load(table_dict, unknown=EXCLUDE)
+    results = schema.dump(table)
     # TODO: fix popular tables to provide these? remove if we're not using them?
     # TODO: Add the 'key' or 'id' to the base PopularTableSchema
     results['key'] = f'{table.database}://{table.cluster}.{table.schema}/{table.name}'
@@ -104,10 +104,9 @@ def marshall_table_full(table_dict: Dict) -> Dict:
     :return: Table Dict with sanitized fields
     """
 
-    schema = TableSchema(strict=True)
-    # TODO: consider migrating to validate() instead of roundtripping
-    table: Table = schema.load(table_dict).data
-    results: Dict[str, Any] = schema.dump(table).data
+    schema = TableSchema()
+    table: Table = schema.load(table_dict)
+    results: Dict[str, Any] = schema.dump(table)
 
     is_editable = is_table_editable(results['schema'], results['name'])
     results['is_editable'] = is_editable
@@ -149,9 +148,9 @@ def marshall_dashboard_partial(dashboard_dict: Dict) -> Dict:
     :param dashboard_dict: Dict of partial dashboard metadata
     :return: partial dashboard Dict
     """
-    schema = DashboardSummarySchema(strict=True)
-    dashboard: DashboardSummary = schema.load(dashboard_dict).data
-    results = schema.dump(dashboard).data
+    schema = DashboardSummarySchema(unknown=EXCLUDE)
+    dashboard: DashboardSummary = schema.load(dashboard_dict)
+    results = schema.dump(dashboard)
     results['type'] = 'dashboard'
     # TODO: Bookmark logic relies on key, opting to add this here to avoid messy logic in
     # React app and we have to clean up later.

amundsen_application/base/base_announcement_client.py

@@ -7,6 +7,7 @@
 from http import HTTPStatus
 
 from flask import jsonify, make_response, Response
+from marshmallow import ValidationError
 
 from amundsen_application.models.announcements import Announcements, AnnouncementsSchema
 
@@ -31,20 +32,16 @@ def _create_error_response(message: str) -> Response:
             return make_response(payload, HTTPStatus.INTERNAL_SERVER_ERROR)
 
         try:
-            try:
-                announcements = self.get_posts()
-            except Exception as e:
-                message = 'Encountered exception getting posts: ' + str(e)
-                return _create_error_response(message)
-
-            # validate the returned object
-            data, errors = AnnouncementsSchema().dump(announcements)
-            if not errors:
-                payload = jsonify({'posts': data.get('posts'), 'msg': 'Success'})
-                return make_response(payload, HTTPStatus.OK)
-            else:
-                message = 'Announcement data dump returned errors: ' + str(errors)
-                return _create_error_response(message)
+            announcements = self.get_posts()
         except Exception as e:
-            message = 'Encountered exception: ' + str(e)
+            message = 'Encountered exception getting posts: ' + str(e)
+            return _create_error_response(message)
+
+        try:
+            data = AnnouncementsSchema().dump(announcements)
+            AnnouncementsSchema().load(data)  # validate returned object
+            payload = jsonify({'posts': data.get('posts'), 'msg': 'Success'})
+            return make_response(payload, HTTPStatus.OK)
+        except ValidationError as err:
+            message = 'Announcement data dump returned errors: ' + str(err.messages)
             return _create_error_response(message)

amundsen_application/base/base_bigquery_preview_client.py

@@ -2,6 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 from http import HTTPStatus
+import logging
 from typing import Dict, List
 from amundsen_application.base.base_preview_client import BasePreviewClient
 from amundsen_application.models.preview_data import (
@@ -10,6 +11,7 @@
     PreviewDataSchema,
 )
 from flask import Response, make_response, jsonify
+from marshmallow import ValidationError
 from google.cloud import bigquery
 
 
@@ -60,13 +62,13 @@ def get_preview_data(self, params: Dict, optionalHeaders: Dict = None) -> Respon
             params["schema"],
             params["tableName"],
         )
-        data = PreviewDataSchema().dump(preview_data)[0]
-        errors = PreviewDataSchema().load(data)[1]
-        payload = jsonify({"preview_data": data})
-
-        if not errors:
+        try:
+            data = PreviewDataSchema().dump(preview_data)
+            PreviewDataSchema().load(data)  # for validation only
             payload = jsonify({"preview_data": data})
             return make_response(payload, HTTPStatus.OK)
-        return make_response(
-            jsonify({"preview_data": {}}), HTTPStatus.INTERNAL_SERVER_ERROR
-        )
+        except ValidationError as err:
+            logging.error("PreviewDataSchema serialization error + " + str(err.messages))
+            return make_response(
+                jsonify({"preview_data": {}}), HTTPStatus.INTERNAL_SERVER_ERROR
+            )

amundsen_application/base/base_superset_preview_client.py

@@ -2,9 +2,11 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import abc
+import logging
 
 from flask import Response as FlaskResponse, make_response, jsonify
 from http import HTTPStatus
+from marshmallow import ValidationError
 from requests import Response
 from typing import Dict
 
@@ -45,12 +47,13 @@ def get_preview_data(self, params: Dict, optionalHeaders: Dict = None) -> FlaskR
             response_dict = response.json()
             columns = [ColumnItem(c['name'], c['type']) for c in response_dict['columns']]
             preview_data = PreviewData(columns, response_dict['data'])
-            data = PreviewDataSchema().dump(preview_data)[0]
-            errors = PreviewDataSchema().load(data)[1]
-            if not errors:
+            try:
+                data = PreviewDataSchema().dump(preview_data)
+                PreviewDataSchema().load(data)  # for validation only
                 payload = jsonify({'preview_data': data})
                 return make_response(payload, response.status_code)
-            else:
+            except ValidationError as err:
+                logging.error("PreviewDataSchema serialization error " + str(err.messages))
                 return make_response(jsonify({'preview_data': {}}), HTTPStatus.INTERNAL_SERVER_ERROR)
         except Exception:
             return make_response(jsonify({'preview_data': {}}), HTTPStatus.INTERNAL_SERVER_ERROR)

amundsen_application/base/examples/example_dremio_preview_client.py

@@ -6,6 +6,7 @@
 from typing import Dict  # noqa: F401
 
 from flask import Response, jsonify, make_response, current_app as app
+from marshmallow import ValidationError
 from pyarrow import flight
 
 from amundsen_application.base.base_superset_preview_client import BasePreviewClient
@@ -79,16 +80,15 @@ def get_preview_data(self, params: Dict, optionalHeaders: Dict = None) -> Respon
             column_items = [ColumnItem(n, t) for n, t in zip(names, types)]
 
             preview_data = PreviewData(column_items, rows)
-
-            data = PreviewDataSchema().dump(preview_data)[0]
-            errors = PreviewDataSchema().load(data)[1]
-            if errors:
-                logging.error(f'Error(s) occurred while building preview data: {errors}')
-                payload = jsonify({'preview_data': {}})
-                return make_response(payload, HTTPStatus.INTERNAL_SERVER_ERROR)
-            else:
+            try:
+                data = PreviewDataSchema().dump(preview_data)
+                PreviewDataSchema().load(data)  # for validation only
                 payload = jsonify({'preview_data': data})
                 return make_response(payload, HTTPStatus.OK)
+            except ValidationError as err:
+                logging.error(f'Error(s) occurred while building preview data: {err.messages}')
+                payload = jsonify({'preview_data': {}})
+                return make_response(payload, HTTPStatus.INTERNAL_SERVER_ERROR)
 
         except Exception as e:
             logging.error(f'Encountered exception: {e}')

amundsen_application/models/announcements.py

@@ -4,7 +4,7 @@
 from marshmallow import Schema, fields, post_dump
 from marshmallow.exceptions import ValidationError
 
-from typing import Dict, List
+from typing import Dict, List, Any
 
 
 class Post:
@@ -29,10 +29,11 @@ class AnnouncementsSchema(Schema):
     posts = fields.Nested(PostSchema, many=True)
 
     @post_dump
-    def validate_data(self, data: Dict) -> None:
+    def validate_data(self, data: Dict, **kwargs: Any) -> Dict:
         posts = data.get('posts', [])
         for post in posts:
             if post.get('date') is None:
                 raise ValidationError('All posts must have a date')
             if post.get('title') is None:
                 raise ValidationError('All posts must have a title')
+        return data

amundsen_application/models/preview_data.py

@@ -1,7 +1,7 @@
 # Copyright Contributors to the Amundsen project.
 # SPDX-License-Identifier: Apache-2.0
 
-from marshmallow import Schema, fields
+from marshmallow import Schema, fields, EXCLUDE
 from typing import List
 
 
@@ -24,6 +24,6 @@ def __init__(self, columns: List = [], data: List = [], error_text: str = '') ->
 
 
 class PreviewDataSchema(Schema):
-    columns = fields.Nested(ColumnItemSchema, many=True)
+    columns = fields.Nested(ColumnItemSchema, many=True, unknown=EXCLUDE)
     data = fields.List(fields.Dict, many=True)
     error_text = fields.Str()

amundsen_application/models/user.py

@@ -28,16 +28,11 @@ def load_user(user_data: Dict) -> User:
         # in the user metadata.
         if _str_no_value(user_data.get('profile_url')) and app.config['GET_PROFILE_URL']:
             user_data['profile_url'] = app.config['GET_PROFILE_URL'](user_data['user_id'])
-        data, errors = schema.load(user_data)
-        return data
+        return schema.load(user_data)
     except ValidationError as err:
         return err.messages
 
 
 def dump_user(user: User) -> Dict:
     schema = UserSchema()
-    try:
-        data, errors = schema.dump(user)
-        return data
-    except ValidationError as err:
-        return err.messages
+    return schema.dump(user)

requirements.txt

@@ -56,12 +56,7 @@ requests==2.25.1
 # A lightweight library for converting complex datatypes to and from native Python datatypes.
 # License: MIT
 # Upstream url: https://github.com/marshmallow-code/marshmallow
-marshmallow>=2.15.3,<3.0
-
-# Allows declaring marshmallow schema through type annotations
-# License: MIT
-# Upstream url: https://github.com/justanr/marshmallow-annotations
-marshmallow-annotations>=2.4.0,<3.0
+marshmallow>=3.0,<=3.6
 
 # A utility library for mocking out the requests Python library.
 # License: Apache 2.0
@@ -73,7 +68,7 @@ SQLAlchemy==1.3.23
 
 # A common package that holds the models deifnition and schemas that are used
 # accross different amundsen repositories.
-amundsen-common==0.6.0
+amundsen-common==0.9.0
 
 # Library for rest endpoints with Flask
 # Upstream url: https://github.com/flask-restful/flask-restful

setup.py

@@ -37,11 +37,11 @@ def build_js() -> None:
 with open(requirements_path) as requirements_file:
     requirements = requirements_file.readlines()
 
-__version__ = '3.5.1'
+__version__ = '3.6.0'
 
 oicd = ['flaskoidc==0.1.1']
 pyarrrow = ['pyarrow==3.0.0']
-bigquery_preview = ['google-cloud-bigquery>=2.8.0,<3.0.0', 'flatten-dict==0.3.0']
+bigquery_preview = ['google-cloud-bigquery>=2.13.1,<3.0.0', 'flatten-dict==0.3.0']
 all_deps = requirements + oicd + pyarrrow + bigquery_preview
 
 setup(

tests/unit/base/test_announcement_client.py

@@ -75,7 +75,7 @@ def test_get_posts_success(self) -> None:
                 {
                     'title': 'Test Title',
                     'date': 'December 31, 1999',
-                    'info_list': ['Test announcement'],
+                    'html_content': 'content',
                 }
             ]
             response = MockClient(success_posts)._get_posts()

tests/unit/base/test_superset_preview_client.py

@@ -135,7 +135,7 @@ def test_post_sql_json_incorrect_data_shape(self) -> None:
     def test_post_sql_json_correct_data_shape(self) -> None:
         """
         Test post_sql_json(), which should result in
-        a response with 500 error and empty preview_data payload
+        a response with 200 status and correct preview_data payload
         :return:
         """
         with app.test_request_context():

tests/unit/models/test_user.py

@@ -4,6 +4,8 @@
 import flask
 import unittest
 
+from marshmallow import ValidationError
+
 from amundsen_application.models.user import load_user, dump_user, UserSchema
 
 app = flask.Flask(__name__)
@@ -76,17 +78,17 @@ def test_raise_error_if_no_display_name(self) -> None:
         :return:
         """
         with app.test_request_context():
-            data, errors = UserSchema().load({})
-            self.assertEqual(len(errors['_schema']), 1)
+            with self.assertRaises(ValidationError):
+                UserSchema().load({})
 
     def test_raise_error_if_no_user_id(self) -> None:
         """
         Error is raised if deserialization of Dict will not generate a user_id
         :return:
         """
         with app.test_request_context():
-            data, errors = UserSchema().load({'display_name': 'Test User'})
-            self.assertEqual(len(errors['_schema']), 1)
+            with self.assertRaises(ValidationError):
+                UserSchema().load({'display_name': 'Test User'})
 
     def test_str_no_value(self) -> None:
         """