Source Hubspot: fix 401 for associations (#15156)

* Revert "Source Hubspot: revert v0.1.78 (#15144)"

This reverts commit cbdb897.

* #379 source hubspot: fix 401 when reading associations

* #379 source hubspot: fix 401 when reading associations

* #379 source hubspot: upd changelog

* auto-bump connector version [ci skip]

Co-authored-by: Octavia Squidington III <[email protected]>

airbyte-config/init/src/main/resources/seed/source_definitions.yaml

@@ -414,7 +414,7 @@
 - name: HubSpot
   sourceDefinitionId: 36c891d9-4bd9-43ac-bad2-10e12756272c
   dockerRepository: airbyte/source-hubspot
-  dockerImageTag: 0.1.79
+  dockerImageTag: 0.1.80
   documentationUrl: https://docs.airbyte.io/integrations/sources/hubspot
   icon: hubspot.svg
   sourceType: api

airbyte-config/init/src/main/resources/seed/source_specs.yaml

@@ -3705,7 +3705,7 @@
     supportsNormalization: false
     supportsDBT: false
     supported_destination_sync_modes: []
-- dockerImage: "airbyte/source-hubspot:0.1.79"
+- dockerImage: "airbyte/source-hubspot:0.1.80"
   spec:
     documentationUrl: "https://docs.airbyte.io/integrations/sources/hubspot"
     connectionSpecification:

airbyte-integrations/connectors/source-hubspot/Dockerfile

@@ -34,5 +34,5 @@ COPY source_hubspot ./source_hubspot
 ENV AIRBYTE_ENTRYPOINT "python /airbyte/integration_code/main.py"
 ENTRYPOINT ["python", "/airbyte/integration_code/main.py"]
 
-LABEL io.airbyte.version=0.1.79
+LABEL io.airbyte.version=0.1.80
 LABEL io.airbyte.name=airbyte/source-hubspot

airbyte-integrations/connectors/source-hubspot/integration_tests/conftest.py

@@ -0,0 +1,19 @@
+#
+# Copyright (c) 2022 Airbyte, Inc., all rights reserved.
+#
+
+import json
+
+import pytest
+
+
+@pytest.fixture(scope="session", name="config")
+def config_fixture():
+    with open("secrets/config.json", "r") as config_file:
+        return json.load(config_file)
+
+
+@pytest.fixture(scope="session", name="oauth_config")
+def oauth_config_fixture():
+    with open("secrets/config_oauth.json", "r") as config_file:
+        return json.load(config_file)

airbyte-integrations/connectors/source-hubspot/integration_tests/test_associations.py

@@ -0,0 +1,57 @@
+#
+# Copyright (c) 2022 Airbyte, Inc., all rights reserved.
+#
+
+import logging
+
+import pytest
+from airbyte_cdk.models import ConfiguredAirbyteCatalog, Type
+from source_hubspot.source import SourceHubspot
+
+
+@pytest.fixture
+def source():
+    return SourceHubspot()
+
+
+@pytest.fixture
+def associations(config, source):
+    streams = source.streams(config)
+    return {stream.name: getattr(stream, "associations", []) for stream in streams}
+
+
+@pytest.fixture
+def configured_catalog(config, source):
+    streams = source.streams(config)
+    return {
+        "streams": [
+            {
+                "stream": stream.as_airbyte_stream(),
+                "sync_mode": "incremental",
+                "cursor_field": [stream.cursor_field],
+                "destination_sync_mode": "append",
+            }
+            for stream in streams
+            if stream.supports_incremental and getattr(stream, "associations", [])
+        ]
+    }
+
+
+@pytest.mark.parametrize("auth", ("api_key", "oauth"))
+def test_incremental_read_fetches_associations(auth, config, oauth_config, configured_catalog, source, associations):
+    configuration = oauth_config if auth == "oauth" else config
+    messages = source.read(logging.getLogger("airbyte"), configuration, ConfiguredAirbyteCatalog.parse_obj(configured_catalog), {})
+
+    association_found = False
+    for message in messages:
+        if message and message.type != Type.RECORD:
+            continue
+        record = message.record
+        stream, data = record.stream, record.data
+        # assume at least one association id is present
+        stream_associations = associations[stream]
+        for association in stream_associations:
+            if data.get(association):
+                association_found = True
+                break
+    assert association_found

airbyte-integrations/connectors/source-hubspot/source_hubspot/source.py

@@ -15,7 +15,7 @@
 from airbyte_cdk.utils.event_timing import create_timer
 from airbyte_cdk.utils.traced_exception import AirbyteTracedException
 from requests import HTTPError
-from source_hubspot.constants import API_KEY_CREDENTIALS, OAUTH_CREDENTIALS, PRIVATE_APP_CREDENTIALS
+from source_hubspot.constants import API_KEY_CREDENTIALS
 from source_hubspot.streams import (
     API,
     Campaigns,
@@ -82,15 +82,10 @@ def get_api(config: Mapping[str, Any]) -> API:
         return API(credentials=credentials)
 
     def get_common_params(self, config) -> Mapping[str, Any]:
-        start_date = config.get("start_date")
+        start_date = config["start_date"]
         credentials = config["credentials"]
         api = self.get_api(config=config)
-        common_params = dict(api=api, start_date=start_date, credentials=credentials)
-
-        credentials_title = credentials.get("credentials_title")
-        if credentials_title == OAUTH_CREDENTIALS or credentials_title == PRIVATE_APP_CREDENTIALS:
-            common_params["authenticator"] = api.get_authenticator()
-        return common_params
+        return dict(api=api, start_date=start_date, credentials=credentials)
 
     def streams(self, config: Mapping[str, Any]) -> List[Stream]:
         credentials = config.get("credentials", {})

airbyte-integrations/connectors/source-hubspot/source_hubspot/streams.py

@@ -245,13 +245,19 @@ def _property_wrapper(self) -> IURLPropertyRepresentation:
             return APIv1Property(properties)
         return APIv3Property(properties)
 
-    def __init__(self, api: API, start_date: str = None, credentials: Mapping[str, Any] = None, **kwargs):
+    def __init__(self, api: API, start_date: Union[str, pendulum.datetime], credentials: Mapping[str, Any] = None, **kwargs):
         super().__init__(**kwargs)
         self._api: API = api
-        self._start_date = pendulum.parse(start_date)
+        self._credentials = credentials
 
-        if credentials["credentials_title"] == API_KEY_CREDENTIALS:
+        self._start_date = start_date
+        if isinstance(self._start_date, str):
+            self._start_date = pendulum.parse(self._start_date)
+        creds_title = self._credentials["credentials_title"]
+        if creds_title == API_KEY_CREDENTIALS:
             self._session.params["hapikey"] = credentials.get("api_key")
+        elif creds_title in (OAUTH_CREDENTIALS, PRIVATE_APP_CREDENTIALS):
+            self._authenticator = api.get_authenticator()
 
     def backoff_time(self, response: requests.Response) -> Optional[float]:
         if response.status_code == codes.too_many_requests:
@@ -642,6 +648,51 @@ def _flat_associations(self, records: Iterable[MutableMapping]) -> Iterable[Muta
             yield record
 
 
+class AssociationsStream(Stream):
+    """
+    Designed to read associations of CRM objects during incremental syncs, since Search API does not support
+    retrieving associations.
+    """
+
+    http_method = "POST"
+    filter_old_records = False
+
+    def __init__(self, parent_stream: Stream, identifiers: Iterable[Union[int, str]], *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self.parent_stream = parent_stream
+        self.identifiers = identifiers
+
+    @property
+    def url(self):
+        """
+        although it is not used, it needs to be implemented because it is an abstract property
+        """
+        return ""
+
+    def path(
+        self,
+        *,
+        stream_state: Mapping[str, Any] = None,
+        stream_slice: Mapping[str, Any] = None,
+        next_page_token: Mapping[str, Any] = None,
+    ) -> str:
+        return f"/crm/v4/associations/{self.parent_stream.entity}/{stream_slice}/batch/read"
+
+    def scopes(self) -> Set[str]:
+        return self.parent_stream.scopes
+
+    def stream_slices(self, sync_mode: SyncMode, cursor_field: List[str] = None, stream_state: Mapping[str, Any] = None) -> Iterable[str]:
+        return self.parent_stream.associations
+
+    def request_body_json(
+        self,
+        stream_state: Mapping[str, Any],
+        stream_slice: Mapping[str, Any] = None,
+        next_page_token: Mapping[str, Any] = None,
+    ) -> Optional[Mapping]:
+        return {"inputs": [{"id": str(id_)} for id_ in self.identifiers]}
+
+
 class IncrementalStream(Stream, ABC):
     """Stream that supports state and incremental read"""
 
@@ -807,6 +858,24 @@ def _process_search(
 
         return list(stream_records.values()), raw_response
 
+    def _read_associations(self, records: Iterable) -> Iterable[Mapping[str, Any]]:
+        records_by_pk = {record[self.primary_key]: record for record in records}
+        identifiers = list(map(lambda x: x[self.primary_key], records))
+        associations_stream = AssociationsStream(
+            api=self._api, start_date=self._start_date, credentials=self._credentials, parent_stream=self, identifiers=identifiers
+        )
+        slices = associations_stream.stream_slices(sync_mode=SyncMode.full_refresh)
+
+        for _slice in slices:
+            logger.info(f"Reading {_slice} associations of {self.entity}")
+            associations = associations_stream.read_records(stream_slice=_slice, sync_mode=SyncMode.full_refresh)
+            for group in associations:
+                current_record = records_by_pk[group["from"]["id"]]
+                associations_list = current_record.get(_slice, [])
+                associations_list.extend(association["toObjectId"] for association in group["to"])
+                current_record[_slice] = associations_list
+        return records_by_pk.values()
+
     def read_records(
         self,
         sync_mode: SyncMode,
@@ -826,15 +895,15 @@ def read_records(
                     stream_state=stream_state,
                     stream_slice=stream_slice,
                 )
-
+                records = self._read_associations(records)
             else:
                 records, raw_response = self._read_stream_records(
                     stream_slice=stream_slice,
                     stream_state=stream_state,
                     next_page_token=next_page_token,
                 )
+                records = self._flat_associations(records)
             records = self._filter_old_records(records)
-            records = self._flat_associations(records)
 
             for record in records:
                 cursor = self._field_to_datetime(record[self.updated_at_field])

airbyte-integrations/connectors/source-hubspot/unit_tests/test_source.py

@@ -50,7 +50,7 @@ def test_check_connection_empty_config(config):
 def test_check_connection_invalid_config(config):
     config.pop("start_date")
 
-    with pytest.raises(TypeError):
+    with pytest.raises(KeyError):
         SourceHubspot().check_connection(logger, config=config)
 
 
@@ -406,6 +406,8 @@ def test_search_based_stream_should_not_attempt_to_get_more_than_10k_records(req
     requests_mock.register_uri("POST", test_stream.url, responses)
     test_stream._sync_mode = None
     requests_mock.register_uri("GET", "/properties/v2/company/properties", properties_response)
+    requests_mock.register_uri("POST", "/crm/v4/associations/company/contacts/batch/read", [{"status_code": 200, "json": {"results": []}}])
+
     records, _ = read_incremental(test_stream, {})
     # The stream should not attempt to get more than 10K records.
     # Instead, it should use the new state to start a new search query.

docs/integrations/sources/hubspot.md

@@ -129,7 +129,8 @@ Now that you have set up the HubSpot source connector, check out the following H
 
 | Version | Date       | Pull Request                                             | Subject                                                                                                                                        |
 |:--------|:-----------|:---------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------|
-| 0.1.79  | 2022-07-28 | [15144](https://github.com/airbytehq/airbyte/pull/15144) | Revert v0.1.78 due to permission issues                                                                                                        |
+| 0.1.80  | 2022-08-01 | [15156](https://github.com/airbytehq/airbyte/pull/15156) | Fix 401 error while retrieving associations using OAuth                                                                                        |
+| 0.1.79  | 2022-07-28 | [15144](https://github.com/airbytehq/airbyte/pull/15144) | Revert v0.1.78 due to permission issues                                                                                                        | 
 | 0.1.78  | 2022-07-28 | [15099](https://github.com/airbytehq/airbyte/pull/15099) | Fix to fetch associations when using incremental mode                                                                                          |
 | 0.1.77  | 2022-07-26 | [15035](https://github.com/airbytehq/airbyte/pull/15035) | Make PropertyHistory stream read historic data not limited to 30 days                                                                          |
 | 0.1.76  | 2022-07-25 | [14999](https://github.com/airbytehq/airbyte/pull/14999) | Partially revert changes made in v0.1.75                                                                                                       |