recce-nightly 1.2.0.20250506__py3-none-any.whl → 1.26.0.20251124__py3-none-any.whl

tests/recce_cloud/test_platform_clients.py

@@ -0,0 +1,333 @@
+"""
+Tests for platform-specific Recce Cloud API clients.
+"""
+
+import os
+from unittest.mock import patch
+from urllib.parse import urlparse
+
+import pytest
+
+from recce_cloud.api.factory import create_platform_client
+from recce_cloud.api.github import GitHubRecceCloudClient
+from recce_cloud.api.gitlab import GitLabRecceCloudClient
+from recce_cloud.ci_providers.base import CIInfo
+
+
+class TestGitHubRecceCloudClient:
+    """Tests for GitHub Actions API client."""
+
+    def test_init(self):
+        """Test client initialization."""
+        client = GitHubRecceCloudClient(token="test_token", repository="owner/repo")
+        assert client.token == "test_token"
+        assert client.repository == "owner/repo"
+        parsed = urlparse(client.api_host)
+        # Accept main domain or subdomains:
+        assert parsed.hostname == "cloud.datarecce.io" or (
+            parsed.hostname and parsed.hostname.endswith(".cloud.datarecce.io")
+        )
+
+    def test_touch_recce_session_pr(self):
+        """Test touch_recce_session for PR context."""
+        client = GitHubRecceCloudClient(token="test_token", repository="owner/repo")
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {
+                "session_id": "test_session_id",
+                "manifest_upload_url": "https://s3.aws.com/manifest",
+                "catalog_upload_url": "https://s3.aws.com/catalog",
+            }
+
+            response = client.touch_recce_session(
+                branch="feature-branch", adapter_type="postgres", cr_number=123, session_type="cr"
+            )
+
+            assert response["session_id"] == "test_session_id"
+            assert response["manifest_upload_url"] == "https://s3.aws.com/manifest"
+            assert response["catalog_upload_url"] == "https://s3.aws.com/catalog"
+
+            # Verify correct API endpoint was called
+            mock_request.assert_called_once()
+            call_args = mock_request.call_args
+            assert call_args[0][0] == "POST"
+            assert "github/owner/repo/touch-recce-session" in call_args[0][1]
+            assert call_args[1]["json"]["branch"] == "feature-branch"
+            assert call_args[1]["json"]["adapter_type"] == "postgres"
+            assert call_args[1]["json"]["pr_number"] == 123
+
+    def test_touch_recce_session_base(self):
+        """Test touch_recce_session for base branch context."""
+        client = GitHubRecceCloudClient(token="test_token", repository="owner/repo")
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {
+                "session_id": "base_session_id",
+                "manifest_upload_url": "https://s3.aws.com/manifest",
+                "catalog_upload_url": "https://s3.aws.com/catalog",
+            }
+
+            client.touch_recce_session(branch="main", adapter_type="snowflake")
+
+            # Verify pr_number is not in the payload when cr_number is None
+            call_args = mock_request.call_args
+            assert "pr_number" not in call_args[1]["json"]
+
+    def test_touch_recce_session_prod_type(self):
+        """Test touch_recce_session with --type prod (should not include pr_number)."""
+        client = GitHubRecceCloudClient(token="test_token", repository="owner/repo")
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {
+                "session_id": "prod_session_id",
+                "manifest_upload_url": "https://s3.aws.com/manifest",
+                "catalog_upload_url": "https://s3.aws.com/catalog",
+            }
+
+            # Even with cr_number detected, session_type="prod" should omit pr_number
+            client.touch_recce_session(branch="main", adapter_type="postgres", cr_number=123, session_type="prod")
+
+            # Verify pr_number is NOT in the payload when session_type is "prod"
+            call_args = mock_request.call_args
+            assert "pr_number" not in call_args[1]["json"]
+            assert call_args[1]["json"]["branch"] == "main"
+            assert call_args[1]["json"]["adapter_type"] == "postgres"
+
+    def test_upload_completed(self):
+        """Test upload_completed notification."""
+        client = GitHubRecceCloudClient(token="test_token", repository="owner/repo")
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {}
+
+            client.upload_completed(session_id="test_session_id")
+
+            mock_request.assert_called_once()
+            call_args = mock_request.call_args
+            assert call_args[0][0] == "POST"
+            assert "github/owner/repo/upload-completed" in call_args[0][1]
+            assert call_args[1]["json"]["session_id"] == "test_session_id"
+
+
+class TestGitLabRecceCloudClient:
+    """Tests for GitLab CI API client."""
+
+    def test_init(self):
+        """Test client initialization."""
+        client = GitLabRecceCloudClient(
+            token="test_token",
+            project_path="group/project",
+            repository_url="https://gitlab.com/group/project",
+        )
+        assert client.token == "test_token"
+        assert client.project_path == "group/project"
+        assert client.repository_url == "https://gitlab.com/group/project"
+
+    def test_touch_recce_session_mr(self):
+        """Test touch_recce_session for MR context."""
+        client = GitLabRecceCloudClient(
+            token="test_token",
+            project_path="group/project",
+            repository_url="https://gitlab.com/group/project",
+        )
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {
+                "session_id": "test_session_id",
+                "manifest_upload_url": "https://s3.aws.com/manifest",
+                "catalog_upload_url": "https://s3.aws.com/catalog",
+            }
+
+            response = client.touch_recce_session(
+                branch="feature-branch",
+                adapter_type="postgres",
+                cr_number=456,
+                commit_sha="abc123def456",
+                session_type="cr",
+            )
+
+            assert response["session_id"] == "test_session_id"
+
+            # Verify correct API endpoint and payload
+            call_args = mock_request.call_args
+            assert call_args[0][0] == "POST"
+            assert "gitlab/group/project/touch-recce-session" in call_args[0][1]
+            payload = call_args[1]["json"]
+            assert payload["branch"] == "feature-branch"
+            assert payload["adapter_type"] == "postgres"
+            assert payload["mr_iid"] == 456
+            assert payload["commit_sha"] == "abc123def456"
+            assert payload["repository_url"] == "https://gitlab.com/group/project"
+
+    def test_touch_recce_session_base(self):
+        """Test touch_recce_session for base branch context."""
+        client = GitLabRecceCloudClient(
+            token="test_token",
+            project_path="group/project",
+            repository_url="https://gitlab.com/group/project",
+        )
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {
+                "session_id": "base_session_id",
+                "manifest_upload_url": "https://s3.aws.com/manifest",
+                "catalog_upload_url": "https://s3.aws.com/catalog",
+            }
+
+            client.touch_recce_session(branch="main", adapter_type="bigquery", commit_sha="base123")
+
+            # Verify mr_iid is not in the payload when cr_number is None
+            call_args = mock_request.call_args
+            assert "mr_iid" not in call_args[1]["json"]
+
+    def test_touch_recce_session_prod_type(self):
+        """Test touch_recce_session with --type prod (should not include mr_iid)."""
+        client = GitLabRecceCloudClient(
+            token="test_token",
+            project_path="group/project",
+            repository_url="https://gitlab.com/group/project",
+        )
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {
+                "session_id": "prod_session_id",
+                "manifest_upload_url": "https://s3.aws.com/manifest",
+                "catalog_upload_url": "https://s3.aws.com/catalog",
+            }
+
+            # Even with cr_number detected, session_type="prod" should omit mr_iid
+            client.touch_recce_session(
+                branch="main",
+                adapter_type="postgres",
+                cr_number=456,
+                commit_sha="abc123def456",
+                session_type="prod",
+            )
+
+            # Verify mr_iid is NOT in the payload when session_type is "prod"
+            call_args = mock_request.call_args
+            assert "mr_iid" not in call_args[1]["json"]
+            assert call_args[1]["json"]["branch"] == "main"
+            assert call_args[1]["json"]["adapter_type"] == "postgres"
+            assert call_args[1]["json"]["commit_sha"] == "abc123def456"
+
+    def test_upload_completed(self):
+        """Test upload_completed notification."""
+        client = GitLabRecceCloudClient(
+            token="test_token",
+            project_path="group/project",
+            repository_url="https://gitlab.com/group/project",
+        )
+
+        with patch.object(client, "_make_request") as mock_request:
+            mock_request.return_value = {}
+
+            client.upload_completed(session_id="test_session_id", commit_sha="commit456")
+
+            mock_request.assert_called_once()
+            call_args = mock_request.call_args
+            assert call_args[0][0] == "POST"
+            assert "gitlab/group/project/upload-completed" in call_args[0][1]
+            payload = call_args[1]["json"]
+            assert payload["session_id"] == "test_session_id"
+            assert payload["commit_sha"] == "commit456"
+
+
+class TestFactoryCreatePlatformClient:
+    """Tests for create_platform_client factory function."""
+
+    def test_create_github_client(self):
+        """Test creating GitHub client."""
+        ci_info = CIInfo(platform="github-actions", repository="owner/repo")
+
+        client = create_platform_client(token="test_token", ci_info=ci_info)
+
+        assert isinstance(client, GitHubRecceCloudClient)
+        assert client.repository == "owner/repo"
+
+    def test_create_github_client_from_env(self):
+        """Test creating GitHub client with environment fallback."""
+        with patch.dict(os.environ, {"GITHUB_REPOSITORY": "owner/repo"}):
+            ci_info = CIInfo(platform="github-actions")
+
+            client = create_platform_client(token="test_token", ci_info=ci_info)
+
+            assert isinstance(client, GitHubRecceCloudClient)
+            assert client.repository == "owner/repo"
+
+    def test_create_github_client_missing_repository(self):
+        """Test error when GitHub repository information is missing."""
+        ci_info = CIInfo(platform="github-actions")
+
+        with patch.dict(os.environ, {}, clear=True):
+            with pytest.raises(ValueError, match="GitHub repository information is required"):
+                create_platform_client(token="test_token", ci_info=ci_info)
+
+    def test_create_gitlab_client(self):
+        """Test creating GitLab client."""
+        with patch.dict(
+            os.environ,
+            {"CI_PROJECT_URL": "https://gitlab.com/group/project"},
+        ):
+            ci_info = CIInfo(platform="gitlab-ci", repository="group/project")
+
+            client = create_platform_client(token="test_token", ci_info=ci_info)
+
+            assert isinstance(client, GitLabRecceCloudClient)
+            assert client.project_path == "group/project"
+            assert client.repository_url == "https://gitlab.com/group/project"
+
+    def test_create_gitlab_client_from_env(self):
+        """Test creating GitLab client with environment fallback."""
+        with patch.dict(
+            os.environ,
+            {
+                "CI_PROJECT_PATH": "group/project",
+                "CI_PROJECT_URL": "https://gitlab.com/group/project",
+            },
+        ):
+            ci_info = CIInfo(platform="gitlab-ci")
+
+            client = create_platform_client(token="test_token", ci_info=ci_info)
+
+            assert isinstance(client, GitLabRecceCloudClient)
+            assert client.project_path == "group/project"
+
+    def test_create_gitlab_client_missing_project_path(self):
+        """Test error when GitLab project path is missing."""
+        with patch.dict(os.environ, {}, clear=True):
+            ci_info = CIInfo(platform="gitlab-ci")
+
+            with pytest.raises(ValueError, match="GitLab project path is required"):
+                create_platform_client(token="test_token", ci_info=ci_info)
+
+    def test_create_gitlab_client_missing_project_url(self):
+        """Test error when GitLab project URL is missing."""
+        with patch.dict(os.environ, {"CI_PROJECT_PATH": "group/project"}, clear=True):
+            ci_info = CIInfo(platform="gitlab-ci", repository="group/project")
+
+            with pytest.raises(ValueError, match="GitLab project URL is required"):
+                create_platform_client(token="test_token", ci_info=ci_info)
+
+    def test_create_client_unsupported_platform(self):
+        """Test error for unsupported platform."""
+        ci_info = CIInfo(platform="unsupported-ci")
+
+        with pytest.raises(ValueError, match="Unsupported platform"):
+            create_platform_client(token="test_token", ci_info=ci_info)
+
+    def test_auto_detect_ci_info(self):
+        """Test automatic CI detection when ci_info is not provided."""
+        with patch.dict(
+            os.environ,
+            {
+                "GITHUB_ACTIONS": "true",
+                "GITHUB_REPOSITORY": "owner/repo",
+                "GITHUB_SHA": "abc123",
+            },
+            clear=True,
+        ):
+            client = create_platform_client(token="test_token")
+
+            assert isinstance(client, GitHubRecceCloudClient)
+            assert client.repository == "owner/repo"

tests/tasks/conftest.py

@@ -1,4 +1,4 @@
 # Share the fixtures for the tasks tests
 
 # noinspection PyUnresolvedReferences
-from tests.adapter.dbt_adapter.conftest import dbt_test_helper
+from tests.adapter.dbt_adapter.conftest import dbt_test_helper  # noqa: F401

tests/tasks/test_histogram.py

@@ -1,6 +1,10 @@
 import pytest
 
-from recce.tasks.histogram import HistogramDiffTask, HistogramDiffCheckValidator, _is_histogram_supported
+from recce.tasks.histogram import (
+    HistogramDiffCheckValidator,
+    HistogramDiffTask,
+    _is_histogram_supported,
+)
 
 
 def test_histogram(dbt_test_helper):
@@ -14,11 +18,7 @@ def test_histogram(dbt_test_helper):
 
     dbt_test_helper.create_model("customers", csv_data, csv_data)
 
-    params = {
-        "model": "customers",
-        "column_name": "age",
-        "column_type": "int"
-    }
+    params = {"model": "customers", "column_name": "age", "column_type": "int"}
 
     task = HistogramDiffTask(params)
     run_result = task.execute()
@@ -30,13 +30,13 @@ def test_histogram(dbt_test_helper):
     #     'bin_edges': [25, 26, ..., 51],
     #     'labels': ['25-26', ..., '51-52']
     # }
-    assert run_result['current']['counts'][0] == 1
-    assert run_result['current']['counts'][-1] == 1
-    assert run_result['current']['total'] == 4
-    assert run_result['min'] == 25
-    assert run_result['max'] == 50
-    assert run_result['bin_edges'][0] == 25
-    assert run_result['bin_edges'][-1] == 51
+    assert run_result["current"]["counts"][0] == 1
+    assert run_result["current"]["counts"][-1] == 1
+    assert run_result["current"]["total"] == 4
+    assert run_result["min"] == 25
+    assert run_result["max"] == 50
+    assert run_result["bin_edges"][0] == 25
+    assert run_result["bin_edges"][-1] == 51
 
 
 def test_histogram_emtpy(dbt_test_helper):
@@ -56,74 +56,66 @@ def test_histogram_emtpy(dbt_test_helper):
     dbt_test_helper.create_model("customers2", csv_data, csv_data_zero)
     dbt_test_helper.create_model("customers3", csv_data_zero, csv_data)
 
-    params = {
-        "model": "customers",
-        "column_name": "age",
-        "column_type": "int"
-    }
+    params = {"model": "customers", "column_name": "age", "column_type": "int"}
 
     task = HistogramDiffTask(params)
     run_result = task.execute()
 
-    assert len(run_result['current']['counts']) == 0
-    assert run_result['current']['total'] == 0
-    assert run_result['min'] is None
-    assert run_result['max'] is None
-    assert len(run_result['bin_edges']) == 0
+    assert len(run_result["current"]["counts"]) == 0
+    assert run_result["current"]["total"] == 0
+    assert run_result["min"] is None
+    assert run_result["max"] is None
+    assert len(run_result["bin_edges"]) == 0
 
-    params = {
-        "model": "customers2",
-        "column_name": "age",
-        "column_type": "int"
-    }
+    params = {"model": "customers2", "column_name": "age", "column_type": "int"}
 
     task = HistogramDiffTask(params)
     run_result = task.execute()
-    assert run_result['base']['counts'][0] == 1
-    assert run_result['base']['counts'][-1] == 1
-    assert run_result['base']['total'] == 4
-    assert run_result['current']['counts'][0] == 0
-    assert run_result['current']['counts'][-1] == 0
-    assert run_result['current']['total'] == 0
-    assert run_result['min'] == 25
-    assert run_result['max'] == 50
-    assert run_result['bin_edges'][0] == 25
-    assert run_result['bin_edges'][-1] == 51
-
-    params = {
-        "model": "customers3",
-        "column_name": "age",
-        "column_type": "int"
-    }
+    assert run_result["base"]["counts"][0] == 1
+    assert run_result["base"]["counts"][-1] == 1
+    assert run_result["base"]["total"] == 4
+    assert run_result["current"]["counts"][0] == 0
+    assert run_result["current"]["counts"][-1] == 0
+    assert run_result["current"]["total"] == 0
+    assert run_result["min"] == 25
+    assert run_result["max"] == 50
+    assert run_result["bin_edges"][0] == 25
+    assert run_result["bin_edges"][-1] == 51
+
+    params = {"model": "customers3", "column_name": "age", "column_type": "int"}
 
     task = HistogramDiffTask(params)
     run_result = task.execute()
-    assert run_result['base']['counts'][0] == 0
-    assert run_result['base']['counts'][-1] == 0
-    assert run_result['base']['total'] == 0
-    assert run_result['current']['counts'][0] == 1
-    assert run_result['current']['counts'][-1] == 1
-    assert run_result['current']['total'] == 4
-    assert run_result['min'] == 25
-    assert run_result['max'] == 50
-    assert run_result['bin_edges'][0] == 25
-    assert run_result['bin_edges'][-1] == 51
+    assert run_result["base"]["counts"][0] == 0
+    assert run_result["base"]["counts"][-1] == 0
+    assert run_result["base"]["total"] == 0
+    assert run_result["current"]["counts"][0] == 1
+    assert run_result["current"]["counts"][-1] == 1
+    assert run_result["current"]["total"] == 4
+    assert run_result["min"] == 25
+    assert run_result["max"] == 50
+    assert run_result["bin_edges"][0] == 25
+    assert run_result["bin_edges"][-1] == 51
 
 
 def test_validator():
     def validate(params: dict = {}, view_options: dict = {}):
-        HistogramDiffCheckValidator().validate({
-            'name': 'test',
-            'type': 'histogram_diff',
-            'params': params,
-            'view_options': view_options,
-        })
-
-    validate({
-        "model": "customers",
-        "column_name": "age",
-        "column_type": "int",
-    })
+        HistogramDiffCheckValidator().validate(
+            {
+                "name": "test",
+                "type": "histogram_diff",
+                "params": params,
+                "view_options": view_options,
+            }
+        )
+
+    validate(
+        {
+            "model": "customers",
+            "column_name": "age",
+            "column_type": "int",
+        }
+    )
 
     with pytest.raises(ValueError):
         validate({})

tests/tasks/test_lineage.py

@@ -3,40 +3,53 @@ import pytest
 
 def test_validator():
     from recce.tasks.lineage import LineageDiffCheckValidator
+
     validator = LineageDiffCheckValidator()
 
     def validate(params: dict):
-        validator.validate({
-            'name': 'test',
-            'type': 'schema_diff',
-            'params': params,
-        })
+        validator.validate(
+            {
+                "name": "test",
+                "type": "schema_diff",
+                "params": params,
+            }
+        )
 
     # Select all models
     validate({})
 
     # Select by selector
-    validate({
-        'select': 'customers',
-        'exclude': 'customers',
-        'packages': ['jaffle_shop'],
-        'view_mode': 'all',
-    })
+    validate(
+        {
+            "select": "customers",
+            "exclude": "customers",
+            "packages": ["jaffle_shop"],
+            "view_mode": "all",
+        }
+    )
 
     # packages should be an array
     with pytest.raises(ValueError):
-        validate({
-            'packages': 'jaffle_shop',
-        })
+        validate(
+            {
+                "packages": "jaffle_shop",
+            }
+        )
 
     # view_mode should be 'all' or 'changed_models'
-    validate({
-        'view_mode': None,
-    })
-    validate({
-        'view_mode': 'all',
-    })
+    validate(
+        {
+            "view_mode": None,
+        }
+    )
+    validate(
+        {
+            "view_mode": "all",
+        }
+    )
     with pytest.raises(ValueError):
-        validate({
-            'view_mode': 'abc',
-        })
+        validate(
+            {
+                "view_mode": "abc",
+            }
+        )