QA engine: add adoption metrics to the QA report

tools/ci_connector_ops/ci_connector_ops/qa_engine/enrichments.py

@@ -5,7 +5,10 @@
 
 import pandas as pd
 
-def get_enriched_catalog(oss_catalog: pd.DataFrame, cloud_catalog: pd.DataFrame) -> pd.DataFrame:
+def get_enriched_catalog(
+    oss_catalog: pd.DataFrame, 
+    cloud_catalog: pd.DataFrame, 
+    adoption_metrics_per_connector_version: pd.DataFrame) -> pd.DataFrame:
     """Merge OSS and Cloud catalog in a single dataframe on their definition id.
     Transformations:
       - Rename columns to snake case.
@@ -15,9 +18,11 @@ def get_enriched_catalog(oss_catalog: pd.DataFrame, cloud_catalog: pd.DataFrame)
     Enrichments:
       - is_on_cloud: determined by the merge operation results.
       - connector_technical_name: built from the docker repository field. airbyte/source-pokeapi -> source-pokeapi.
+      - Adoptions metrics: add the columns from the adoption_metrics_per_connector_version dataframe.
     Args:
         oss_catalog (pd.DataFrame): The open source catalog dataframe.
         cloud_catalog (pd.DataFrame): The cloud catalog dataframe.
+        adoption_metrics_per_connector_version (pd.DataFrame): The crowd sourced adoptions metrics.
 
     Returns:
         pd.DataFrame: The enriched catalog.
@@ -28,15 +33,19 @@ def get_enriched_catalog(oss_catalog: pd.DataFrame, cloud_catalog: pd.DataFrame)
         how="left",
         on="connector_definition_id",
         indicator=True,
-        suffixes=("", "_del"),
+        suffixes=("", "_cloud"),
     )
+
     enriched_catalog.columns = enriched_catalog.columns.str.replace(
         "(?<=[a-z])(?=[A-Z])", "_", regex=True
     ).str.lower() # column names to snake case
+    enriched_catalog = enriched_catalog[[c for c in enriched_catalog.columns if "_cloud" not in c]]
     enriched_catalog["is_on_cloud"] = enriched_catalog["_merge"] == "both"
     enriched_catalog = enriched_catalog.drop(columns="_merge")
     enriched_catalog["connector_name"] = enriched_catalog["name"]
     enriched_catalog["connector_technical_name"] = enriched_catalog["docker_repository"].str.replace("airbyte/", "")
     enriched_catalog["connector_version"] = enriched_catalog["docker_image_tag"]
     enriched_catalog["release_stage"] = enriched_catalog["release_stage"].fillna("unknown")
+    enriched_catalog = enriched_catalog.merge(adoption_metrics_per_connector_version, how="left", on=["connector_definition_id", "connector_version"])
+    enriched_catalog[adoption_metrics_per_connector_version.columns] = enriched_catalog[adoption_metrics_per_connector_version.columns].fillna(0)
     return enriched_catalog

tools/ci_connector_ops/ci_connector_ops/qa_engine/inputs.py

@@ -10,7 +10,6 @@
 import requests
 import pandas as pd
 
-from .constants import CLOUD_CATALOG_URL, OSS_CATALOG_URL
 
 def fetch_remote_catalog(catalog_url: str) -> pd.DataFrame:
     """Fetch a combined remote catalog and return a single DataFrame 
@@ -50,6 +49,3 @@ def fetch_adoption_metrics_per_connector_version() -> pd.DataFrame:
         "total_syncs_count",
         "sync_success_rate",
     ]]
-
-CLOUD_CATALOG = fetch_remote_catalog(CLOUD_CATALOG_URL)
-OSS_CATALOG = fetch_remote_catalog(OSS_CATALOG_URL)

tools/ci_connector_ops/ci_connector_ops/qa_engine/main.py

@@ -3,13 +3,17 @@
 #
 
 
-from .constants import GCS_QA_REPORT_PATH
-from .enrichments import get_enriched_catalog
-from .inputs import CLOUD_CATALOG, OSS_CATALOG
-from .validations import get_qa_report
+from .constants import CLOUD_CATALOG_URL, GCS_QA_REPORT_PATH, OSS_CATALOG_URL
+from . import enrichments, inputs, validations, outputs
 
 
 def main():
-    enriched_catalog = get_enriched_catalog(OSS_CATALOG, CLOUD_CATALOG)
-    qa_report = get_qa_report(enriched_catalog)
-    qa_report.to_json(GCS_QA_REPORT_PATH, orient="records")
+    oss_catalog = inputs.fetch_remote_catalog(OSS_CATALOG_URL)
+    cloud_catalog = inputs.fetch_remote_catalog(CLOUD_CATALOG_URL)
+    adoption_metrics_per_connector_version = inputs.fetch_adoption_metrics_per_connector_version()
+    enriched_catalog = enrichments.get_enriched_catalog(
+        oss_catalog, 
+        cloud_catalog, 
+        adoption_metrics_per_connector_version
+    )
+    validations.get_qa_report(enriched_catalog, len(oss_catalog))

tools/ci_connector_ops/ci_connector_ops/qa_engine/models.py

@@ -7,7 +7,7 @@
 from enum import Enum
 from typing import List
 
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 
 class ConnectorTypeEnum(str, Enum):
     source = "source"
@@ -19,22 +19,28 @@ class ReleaseStageEnum(str, Enum):
     beta = "beta"
     generally_available = "generally_available"
 
+PUBLIC_FIELD = Field(..., is_public=True)
+PRIVATE_FIELD = Field(..., is_public=False)
+
 class ConnectorQAReport(BaseModel):
-    connector_type: ConnectorTypeEnum
-    connector_name: str
-    connector_technical_name: str
-    connector_definition_id: str
-    connector_version: str
-    release_stage: ReleaseStageEnum
-    is_on_cloud: bool
-    is_appropriate_for_cloud_use: bool
-    latest_build_is_successful: bool
-    documentation_is_available: bool
-    number_of_connections: int 
-    number_of_users: int
-    sync_success_rate: float
-    is_eligible_for_promotion_to_cloud: bool
-    report_generation_datetime: datetime
+    connector_type: ConnectorTypeEnum = PUBLIC_FIELD
+    connector_name: str = PUBLIC_FIELD
+    connector_technical_name: str = PUBLIC_FIELD
+    connector_definition_id: str = PUBLIC_FIELD
+    connector_version: str = PUBLIC_FIELD
+    release_stage: ReleaseStageEnum = PUBLIC_FIELD
+    is_on_cloud: bool = PUBLIC_FIELD
+    is_appropriate_for_cloud_use: bool = PUBLIC_FIELD
+    latest_build_is_successful: bool = PUBLIC_FIELD
+    documentation_is_available: bool = PUBLIC_FIELD
+    number_of_connections: int = PRIVATE_FIELD
+    number_of_users: int = PRIVATE_FIELD
+    sync_success_rate: float = PRIVATE_FIELD
+    total_syncs_count: int = PRIVATE_FIELD
+    failed_syncs_count: int = PRIVATE_FIELD
+    succeeded_syncs_count: int = PRIVATE_FIELD
+    is_eligible_for_promotion_to_cloud: bool = PUBLIC_FIELD
+    report_generation_datetime: datetime = PUBLIC_FIELD
 
 class QAReport(BaseModel):
     connectors_qa_report: List[ConnectorQAReport]

tools/ci_connector_ops/ci_connector_ops/qa_engine/outputs.py

@@ -0,0 +1,15 @@
+#
+# Copyright (c) 2023 Airbyte, Inc., all rights reserved.
+#
+
+
+import pandas as pd
+
+from .models import ConnectorQAReport
+
+def persist_qa_report(qa_report: pd.DataFrame, path: str, public_fields_only: bool =True):
+    final_fields = [
+        field.name for field in ConnectorQAReport.__fields__.values() 
+        if field.field_info.extra["is_public"] or not public_fields_only
+    ]
+    qa_report[final_fields].to_json(path, orient="records")

tools/ci_connector_ops/ci_connector_ops/qa_engine/validations.py

@@ -10,7 +10,6 @@
 import requests
 
 from .constants import INAPPROPRIATE_FOR_CLOUD_USE_CONNECTORS
-from .inputs import OSS_CATALOG
 from .models import ConnectorQAReport, QAReport
 
 TRUTHY_COLUMNS_TO_BE_ELIGIBLE = [
@@ -37,7 +36,8 @@ def is_eligible_for_promotion_to_cloud(connector_qa_data: pd.Series) -> bool:
     for col in TRUTHY_COLUMNS_TO_BE_ELIGIBLE
   ])
 
-def get_qa_report(enriched_catalog: pd.DataFrame) -> pd.DataFrame:
+
+def get_qa_report(enriched_catalog: pd.DataFrame, oss_catalog_length: int) -> pd.DataFrame:
     """Perform validation steps on top of the enriched catalog.
     Adds the following columns:
       - documentation_is_available:
@@ -54,6 +54,7 @@ def get_qa_report(enriched_catalog: pd.DataFrame) -> pd.DataFrame:
         Get the sync success rate of the connections with this connector version from our datawarehouse.    
     Args:
         enriched_catalog (pd.DataFrame): The enriched catalog.
+        oss_catalog_length (pd.DataFrame): The length of the OSS catalog, for sanity check.
 
     Returns:
         pd.DataFrame: The final QA report.
@@ -64,9 +65,9 @@ def get_qa_report(enriched_catalog: pd.DataFrame) -> pd.DataFrame:
 
     # TODO YET TO IMPLEMENT VALIDATIONS
     qa_report["latest_build_is_successful"] = False # TODO, tracked in https://github.com/airbytehq/airbyte/issues/21720
-    qa_report["number_of_connections"] = 0 # TODO, tracked in https://github.com/airbytehq/airbyte/issues/21721
-    qa_report["number_of_users"] = 0 # TODO, tracked in https://github.com/airbytehq/airbyte/issues/21721
-    qa_report["sync_success_rate"] = .0 # TODO, tracked in https://github.com/airbytehq/airbyte/issues/21721
+
+    qa_report["is_eligible_for_promotion_to_cloud"] = qa_report.apply(is_eligible_for_promotion_to_cloud, axis="columns")
+    qa_report["report_generation_datetime"] = datetime.utcnow()
 
     qa_report["is_eligible_for_promotion_to_cloud"] = qa_report.apply(is_eligible_for_promotion_to_cloud, axis="columns")
     qa_report["report_generation_datetime"] = datetime.utcnow()
@@ -75,8 +76,8 @@ def get_qa_report(enriched_catalog: pd.DataFrame) -> pd.DataFrame:
     qa_report= qa_report[[field.name for field in ConnectorQAReport.__fields__.values()]]
     # Validate the report structure with pydantic QAReport model.
     QAReport(connectors_qa_report=qa_report.to_dict(orient="records"))
-    if len(qa_report) != len(OSS_CATALOG):
-        raise QAReportGenerationError("The QA report does not contain all the connectors defined in the OSS catalog.")
+    if len(qa_report) != oss_catalog_length:
+      raise QAReportGenerationError("The QA report does not contain all the connectors defined in the OSS catalog.")
     return qa_report
 
 def get_connectors_eligible_for_cloud(qa_report: pd.DataFrame) -> Iterable[ConnectorQAReport]:

tools/ci_connector_ops/setup.py

@@ -23,7 +23,7 @@
 
 
 setup(
-    version="0.1.7",
+    version="0.1.8",
     name="ci_connector_ops",
     description="Packaged maintained by the connector operations team to perform CI for connectors",
     author="Airbyte",

tools/ci_connector_ops/tests/conftest.py

@@ -0,0 +1,57 @@
+#
+# Copyright (c) 2023 Airbyte, Inc., all rights reserved.
+#
+
+
+from datetime import datetime
+import pandas as pd
+import pytest
+
+from ci_connector_ops.qa_engine.constants import OSS_CATALOG_URL, CLOUD_CATALOG_URL
+from ci_connector_ops.qa_engine.inputs import fetch_remote_catalog
+
+@pytest.fixture(scope="module")
+def oss_catalog():
+    return fetch_remote_catalog(OSS_CATALOG_URL)
+
+@pytest.fixture(scope="module")
+def cloud_catalog():
+    return fetch_remote_catalog(CLOUD_CATALOG_URL)
+
+@pytest.fixture(scope="module")
+def adoption_metrics_per_connector_version():
+    return pd.DataFrame([{
+        "connector_definition_id": "dfd88b22-b603-4c3d-aad7-3701784586b1",
+        "connector_version": "2.0.0",
+        "number_of_connections": 0,
+        "number_of_users": 0,
+        "succeeded_syncs_count": 0,
+        "failed_syncs_count": 0,
+        "total_syncs_count": 0,
+        "sync_success_rate": 0.0,
+    }])
+
+@pytest.fixture
+def dummy_qa_report() -> pd.DataFrame:
+    return pd.DataFrame([
+        {
+            "connector_type": "source",
+            "connector_name": "test",
+            "connector_technical_name": "source-test",
+            "connector_definition_id": "foobar",
+            "connector_version": "0.0.0",
+            "release_stage": "alpha",
+            "is_on_cloud": False,
+            "is_appropriate_for_cloud_use": True,
+            "latest_build_is_successful": True,
+            "documentation_is_available": False,
+            "number_of_connections": 0,
+            "number_of_users": 0,
+            "sync_success_rate": .99,
+            "total_syncs_count": 0,
+            "failed_syncs_count": 0,
+            "succeeded_syncs_count": 0,
+            "is_eligible_for_promotion_to_cloud": True,
+            "report_generation_datetime": datetime.utcnow()
+        }
+    ])

tools/ci_connector_ops/tests/test_qa_engine/test_enrichments.py

@@ -8,30 +8,37 @@
 import pandas as pd
 import pytest
 
-from ci_connector_ops.qa_engine import inputs, enrichments
+from ci_connector_ops.qa_engine import enrichments
+
 
 @pytest.fixture
-def enriched_catalog() -> pd.DataFrame:
-    return enrichments.get_enriched_catalog(inputs.OSS_CATALOG, inputs.CLOUD_CATALOG)
+def enriched_catalog(oss_catalog, cloud_catalog, adoption_metrics_per_connector_version) -> pd.DataFrame:
+    return enrichments.get_enriched_catalog(oss_catalog, cloud_catalog, adoption_metrics_per_connector_version)
 
 @pytest.fixture
 def enriched_catalog_columns(enriched_catalog: pd.DataFrame) -> set:
     return set(enriched_catalog.columns)
 
-def test_merge_performed_correctly(enriched_catalog):
-    assert len(enriched_catalog) == len(inputs.OSS_CATALOG)
+def test_merge_performed_correctly(enriched_catalog, oss_catalog):
+    assert len(enriched_catalog) == len(oss_catalog)
 
 def test_new_columns_are_added(enriched_catalog_columns):
     expected_new_columns = {
         "is_on_cloud",
         "connector_name",
         "connector_technical_name",
-        "connector_version"
+        "connector_version",
+        "number_of_connections",
+        "number_of_users",
+        "succeeded_syncs_count",
+        "failed_syncs_count",
+        "total_syncs_count",
+        "sync_success_rate",
     }
     assert expected_new_columns.issubset(enriched_catalog_columns)
 
-def test_no_column_are_removed_and_lowercased(enriched_catalog_columns):
-    for column in inputs.OSS_CATALOG:
+def test_no_column_are_removed_and_lowercased(enriched_catalog_columns, oss_catalog):
+    for column in oss_catalog:
         assert re.sub(r"(?<!^)(?=[A-Z])", "_", column).lower() in enriched_catalog_columns
 
 def test_release_stage_not_null(enriched_catalog):

tools/ci_connector_ops/tests/test_qa_engine/test_inputs.py

@@ -8,9 +8,9 @@
 import pandas as pd
 import pytest
 
-from ci_connector_ops.qa_engine import inputs
+from ci_connector_ops.qa_engine import inputs, constants
 
-@pytest.mark.parametrize("catalog_url", [inputs.OSS_CATALOG_URL, inputs.CLOUD_CATALOG_URL])
+@pytest.mark.parametrize("catalog_url", [constants.OSS_CATALOG_URL, constants.CLOUD_CATALOG_URL])
 def test_fetch_remote_catalog(catalog_url):
     catalog = inputs.fetch_remote_catalog(catalog_url)
     assert isinstance(catalog, pd.DataFrame)

tools/ci_connector_ops/tests/test_qa_engine/test_main.py

@@ -3,34 +3,38 @@
 #
 
 
-import pandas as pd
-import pytest
-
 from ci_connector_ops.qa_engine import main
 
-@pytest.fixture
-def dummy_report() -> pd.DataFrame:
-    return pd.DataFrame([
-        {
-            "connector_type": "source",
-            "connector_name": "test",
-            "docker_image_tag": "0.0.0",
-            "release_stage": "alpha",
-            "is_on_cloud": False,
-            "latest_build_is_successful": False,
-            "documentation_is_available": False,
-            "number_of_connections": 0,
-            "number_of_users": 0,
-            "sync_success_rate": .99
-        }
-    ])
 
-def test_main(tmp_path, mocker, dummy_report):
-    output_path = tmp_path / "output.json"
-    mocker.patch.object(main, "GCS_QA_REPORT_PATH", output_path)
-    mocker.patch.object(main, "get_enriched_catalog")
-    mocker.patch.object(main, "get_qa_report", mocker.Mock(return_value=dummy_report))
+def test_main(mocker, dummy_qa_report):
+    mock_oss_catalog = mocker.Mock(__len__=mocker.Mock(return_value=42))
+    mock_cloud_catalog = mocker.Mock()
+
+    mocker.patch.object(main, "enrichments")
+    mocker.patch.object(main, "outputs")
+    mocker.patch.object(
+        main.inputs, 
+        "fetch_remote_catalog",
+        mocker.Mock(side_effect=[mock_oss_catalog, mock_cloud_catalog]))
+    mocker.patch.object(main.inputs, "fetch_adoption_metrics_per_connector_version")
+    mocker.patch.object(main.validations, "get_qa_report", mocker.Mock(return_value=dummy_qa_report))
+
     main.main()
-    main.get_enriched_catalog.assert_called_with(main.OSS_CATALOG, main.CLOUD_CATALOG)
-    main.get_qa_report.assert_called_with(main.get_enriched_catalog.return_value)
-    assert pd.read_json(output_path).to_dict() == dummy_report.to_dict()
+
+    assert main.inputs.fetch_remote_catalog.call_count == 2
+    main.inputs.fetch_remote_catalog.assert_has_calls(
+        [
+            mocker.call(main.OSS_CATALOG_URL),
+            mocker.call(main.CLOUD_CATALOG_URL)
+        ]
+    )
+    main.enrichments.get_enriched_catalog.assert_called_with(
+        mock_oss_catalog, 
+        mock_cloud_catalog,
+        main.inputs.fetch_adoption_metrics_per_connector_version.return_value
+    )
+    main.validations.get_qa_report.assert_called_with(
+        main.enrichments.get_enriched_catalog.return_value,
+        len(mock_oss_catalog)
+    )
+