unstructured-ingest 0.3.8__py3-none-any.whl → 0.3.10__py3-none-any.whl

test/integration/connectors/sql/test_postgres.py

@@ -1,15 +1,18 @@
-import tempfile
-from contextlib import contextmanager
+import json
 from pathlib import Path
 
-import pandas as pd
 import pytest
+from _pytest.fixtures import TopRequest
 from psycopg2 import connect
 
 from test.integration.connectors.utils.constants import DESTINATION_TAG, SOURCE_TAG, env_setup_path
 from test.integration.connectors.utils.docker_compose import docker_compose_context
-from test.integration.connectors.utils.validation import (
-    ValidationConfigs,
+from test.integration.connectors.utils.validation.destination import (
+    StagerValidationConfigs,
+    stager_validation,
+)
+from test.integration.connectors.utils.validation.source import (
+    SourceValidationConfigs,
     source_connector_validation,
 )
 from unstructured_ingest.v2.interfaces import FileData
@@ -25,16 +28,17 @@ from unstructured_ingest.v2.processes.connectors.sql.postgres import (
     PostgresUploadStager,
 )
 
-SEED_DATA_ROWS = 20
+SEED_DATA_ROWS = 10
 
 
-@contextmanager
-def postgres_download_setup() -> None:
+@pytest.fixture
+def source_database_setup() -> str:
+    db_name = "test_db"
     with docker_compose_context(docker_compose_path=env_setup_path / "sql" / "postgres" / "source"):
         connection = connect(
             user="unstructured",
             password="test",
-            dbname="test_db",
+            dbname=db_name,
             host="localhost",
             port=5433,
         )
@@ -43,12 +47,12 @@ def postgres_download_setup() -> None:
                 sql_statment = f"INSERT INTO cars (brand, price) VALUES " f"('brand_{i}', {i})"
                 cursor.execute(sql_statment)
             connection.commit()
-        yield
+        yield db_name
 
 
 @pytest.mark.asyncio
 @pytest.mark.tags(CONNECTOR_TYPE, SOURCE_TAG, "sql")
-async def test_postgres_source():
+async def test_postgres_source(temp_dir: Path, source_database_setup: str):
     connect_params = {
         "host": "localhost",
         "port": 5433,
@@ -56,37 +60,31 @@ async def test_postgres_source():
         "user": "unstructured",
         "password": "test",
     }
-    with postgres_download_setup():
-        with tempfile.TemporaryDirectory() as tmpdir:
-            connection_config = PostgresConnectionConfig(
-                host=connect_params["host"],
-                port=connect_params["port"],
-                database=connect_params["database"],
-                username=connect_params["user"],
-                access_config=PostgresAccessConfig(password=connect_params["password"]),
-            )
-            indexer = PostgresIndexer(
-                connection_config=connection_config,
-                index_config=PostgresIndexerConfig(
-                    table_name="cars", id_column="car_id", batch_size=5
-                ),
-            )
-            downloader = PostgresDownloader(
-                connection_config=connection_config,
-                download_config=PostgresDownloaderConfig(
-                    fields=["car_id", "brand"], download_dir=Path(tmpdir)
-                ),
-            )
-            await source_connector_validation(
-                indexer=indexer,
-                downloader=downloader,
-                configs=ValidationConfigs(
-                    test_id="postgres",
-                    expected_num_files=SEED_DATA_ROWS,
-                    expected_number_indexed_file_data=4,
-                    validate_downloaded_files=True,
-                ),
-            )
+    connection_config = PostgresConnectionConfig(
+        host=connect_params["host"],
+        port=connect_params["port"],
+        database=connect_params["database"],
+        username=connect_params["user"],
+        access_config=PostgresAccessConfig(password=connect_params["password"]),
+    )
+    indexer = PostgresIndexer(
+        connection_config=connection_config,
+        index_config=PostgresIndexerConfig(table_name="cars", id_column="car_id", batch_size=6),
+    )
+    downloader = PostgresDownloader(
+        connection_config=connection_config,
+        download_config=PostgresDownloaderConfig(fields=["car_id", "brand"], download_dir=temp_dir),
+    )
+    await source_connector_validation(
+        indexer=indexer,
+        downloader=downloader,
+        configs=SourceValidationConfigs(
+            test_id="postgres",
+            expected_num_files=SEED_DATA_ROWS,
+            expected_number_indexed_file_data=2,
+            validate_downloaded_files=True,
+        ),
+    )
 
 
 def validate_destination(
@@ -118,63 +116,76 @@ def validate_destination(
 
 @pytest.mark.asyncio
 @pytest.mark.tags(CONNECTOR_TYPE, DESTINATION_TAG, "sql")
-async def test_postgres_destination(upload_file: Path):
+async def test_postgres_destination(upload_file: Path, temp_dir: Path):
     # the postgres destination connector doesn't leverage the file data but is required as an input,
     # mocking it with arbitrary values to meet the base requirements:
     mock_file_data = FileData(identifier="mock file data", connector_type=CONNECTOR_TYPE)
     with docker_compose_context(
         docker_compose_path=env_setup_path / "sql" / "postgres" / "destination"
     ):
-        with tempfile.TemporaryDirectory() as tmpdir:
-            stager = PostgresUploadStager()
-            stager_params = {
-                "elements_filepath": upload_file,
-                "file_data": mock_file_data,
-                "output_dir": Path(tmpdir),
-                "output_filename": "test_db",
-            }
-            if stager.is_async():
-                staged_path = await stager.run_async(**stager_params)
-            else:
-                staged_path = stager.run(**stager_params)
-
-            # The stager should append the `.json` suffix to the output filename passed in.
-            assert staged_path.name == "test_db.json"
-
-            connect_params = {
-                "host": "localhost",
-                "port": 5433,
-                "database": "elements",
-                "user": "unstructured",
-                "password": "test",
-            }
-
-            uploader = PostgresUploader(
-                connection_config=PostgresConnectionConfig(
-                    host=connect_params["host"],
-                    port=connect_params["port"],
-                    database=connect_params["database"],
-                    username=connect_params["user"],
-                    access_config=PostgresAccessConfig(password=connect_params["password"]),
-                )
-            )
+        stager = PostgresUploadStager()
+        staged_path = stager.run(
+            elements_filepath=upload_file,
+            file_data=mock_file_data,
+            output_dir=temp_dir,
+            output_filename=upload_file.name,
+        )
 
-            uploader.run(path=staged_path, file_data=mock_file_data)
+        # The stager should append the `.json` suffix to the output filename passed in.
+        assert staged_path.suffix == upload_file.suffix
 
-            staged_df = pd.read_json(staged_path, orient="records", lines=True)
-            sample_element = staged_df.iloc[0]
-            expected_num_elements = len(staged_df)
-            validate_destination(
-                connect_params=connect_params,
-                expected_num_elements=expected_num_elements,
-                expected_text=sample_element["text"],
-                test_embedding=sample_element["embeddings"],
-            )
+        connect_params = {
+            "host": "localhost",
+            "port": 5433,
+            "database": "elements",
+            "user": "unstructured",
+            "password": "test",
+        }
 
-            uploader.run(path=staged_path, file_data=mock_file_data)
-            validate_destination(
-                connect_params=connect_params,
-                expected_num_elements=expected_num_elements,
-                expected_text=sample_element["text"],
-                test_embedding=sample_element["embeddings"],
+        uploader = PostgresUploader(
+            connection_config=PostgresConnectionConfig(
+                host=connect_params["host"],
+                port=connect_params["port"],
+                database=connect_params["database"],
+                username=connect_params["user"],
+                access_config=PostgresAccessConfig(password=connect_params["password"]),
             )
+        )
+        uploader.precheck()
+        uploader.run(path=staged_path, file_data=mock_file_data)
+
+        with staged_path.open("r") as f:
+            staged_data = json.load(f)
+
+        sample_element = staged_data[0]
+        expected_num_elements = len(staged_data)
+        validate_destination(
+            connect_params=connect_params,
+            expected_num_elements=expected_num_elements,
+            expected_text=sample_element["text"],
+            test_embedding=sample_element["embeddings"],
+        )
+
+        uploader.run(path=staged_path, file_data=mock_file_data)
+        validate_destination(
+            connect_params=connect_params,
+            expected_num_elements=expected_num_elements,
+            expected_text=sample_element["text"],
+            test_embedding=sample_element["embeddings"],
+        )
+
+
+@pytest.mark.parametrize("upload_file_str", ["upload_file_ndjson", "upload_file"])
+def test_postgres_stager(
+    request: TopRequest,
+    upload_file_str: str,
+    tmp_path: Path,
+):
+    upload_file: Path = request.getfixturevalue(upload_file_str)
+    stager = PostgresUploadStager()
+    stager_validation(
+        configs=StagerValidationConfigs(test_id=CONNECTOR_TYPE, expected_count=22),
+        input_file=upload_file,
+        stager=stager,
+        tmp_dir=tmp_path,
+    )

test/integration/connectors/sql/test_singlestore.py

@@ -1,15 +1,18 @@
-import tempfile
-from contextlib import contextmanager
+import json
 from pathlib import Path
 
-import pandas as pd
 import pytest
 import singlestoredb as s2
+from _pytest.fixtures import TopRequest
 
 from test.integration.connectors.utils.constants import DESTINATION_TAG, SOURCE_TAG, env_setup_path
 from test.integration.connectors.utils.docker_compose import docker_compose_context
-from test.integration.connectors.utils.validation import (
-    ValidationConfigs,
+from test.integration.connectors.utils.validation.destination import (
+    StagerValidationConfigs,
+    stager_validation,
+)
+from test.integration.connectors.utils.validation.source import (
+    SourceValidationConfigs,
     source_connector_validation,
 )
 from unstructured_ingest.v2.interfaces import FileData
@@ -26,11 +29,18 @@ from unstructured_ingest.v2.processes.connectors.sql.singlestore import (
     SingleStoreUploadStager,
 )
 
-SEED_DATA_ROWS = 20
+SEED_DATA_ROWS = 10
 
 
-@contextmanager
-def singlestore_download_setup(connect_params: dict) -> None:
+@pytest.fixture
+def source_database_setup() -> dict:
+    connect_params = {
+        "host": "localhost",
+        "port": 3306,
+        "database": "ingest_test",
+        "user": "root",
+        "password": "password",
+    }
     with docker_compose_context(
         docker_compose_path=env_setup_path / "sql" / "singlestore" / "source"
     ):
@@ -40,50 +50,40 @@ def singlestore_download_setup(connect_params: dict) -> None:
                     sql_statment = f"INSERT INTO cars (brand, price) VALUES " f"('brand_{i}', {i})"
                     cursor.execute(sql_statment)
                 connection.commit()
-        yield
+        yield connect_params
 
 
 @pytest.mark.asyncio
 @pytest.mark.tags(CONNECTOR_TYPE, SOURCE_TAG, "sql")
-async def test_singlestore_source():
-    connect_params = {
-        "host": "localhost",
-        "port": 3306,
-        "database": "ingest_test",
-        "user": "root",
-        "password": "password",
-    }
-    with singlestore_download_setup(connect_params=connect_params):
-        with tempfile.TemporaryDirectory() as tmpdir:
-            connection_config = SingleStoreConnectionConfig(
-                host=connect_params["host"],
-                port=connect_params["port"],
-                database=connect_params["database"],
-                user=connect_params["user"],
-                access_config=SingleStoreAccessConfig(password=connect_params["password"]),
-            )
-            indexer = SingleStoreIndexer(
-                connection_config=connection_config,
-                index_config=SingleStoreIndexerConfig(
-                    table_name="cars", id_column="car_id", batch_size=5
-                ),
-            )
-            downloader = SingleStoreDownloader(
-                connection_config=connection_config,
-                download_config=SingleStoreDownloaderConfig(
-                    fields=["car_id", "brand"], download_dir=Path(tmpdir)
-                ),
-            )
-            await source_connector_validation(
-                indexer=indexer,
-                downloader=downloader,
-                configs=ValidationConfigs(
-                    test_id="singlestore",
-                    expected_num_files=SEED_DATA_ROWS,
-                    expected_number_indexed_file_data=4,
-                    validate_downloaded_files=True,
-                ),
-            )
+async def test_singlestore_source(temp_dir: Path, source_database_setup: dict):
+
+    connection_config = SingleStoreConnectionConfig(
+        host=source_database_setup["host"],
+        port=source_database_setup["port"],
+        database=source_database_setup["database"],
+        user=source_database_setup["user"],
+        access_config=SingleStoreAccessConfig(password=source_database_setup["password"]),
+    )
+    indexer = SingleStoreIndexer(
+        connection_config=connection_config,
+        index_config=SingleStoreIndexerConfig(table_name="cars", id_column="car_id", batch_size=6),
+    )
+    downloader = SingleStoreDownloader(
+        connection_config=connection_config,
+        download_config=SingleStoreDownloaderConfig(
+            fields=["car_id", "brand"], download_dir=temp_dir
+        ),
+    )
+    await source_connector_validation(
+        indexer=indexer,
+        downloader=downloader,
+        configs=SourceValidationConfigs(
+            test_id="singlestore",
+            expected_num_files=SEED_DATA_ROWS,
+            expected_number_indexed_file_data=2,
+            validate_downloaded_files=True,
+        ),
+    )
 
 
 def validate_destination(
@@ -102,59 +102,71 @@ def validate_destination(
 
 @pytest.mark.asyncio
 @pytest.mark.tags(CONNECTOR_TYPE, DESTINATION_TAG, "sql")
-async def test_singlestore_destination(upload_file: Path):
+async def test_singlestore_destination(upload_file: Path, temp_dir: Path):
     mock_file_data = FileData(identifier="mock file data", connector_type=CONNECTOR_TYPE)
     with docker_compose_context(
         docker_compose_path=env_setup_path / "sql" / "singlestore" / "destination"
     ):
-        with tempfile.TemporaryDirectory() as tmpdir:
-            stager = SingleStoreUploadStager()
-            stager_params = {
-                "elements_filepath": upload_file,
-                "file_data": mock_file_data,
-                "output_dir": Path(tmpdir),
-                "output_filename": "test_db",
-            }
-            if stager.is_async():
-                staged_path = await stager.run_async(**stager_params)
-            else:
-                staged_path = stager.run(**stager_params)
-
-            # The stager should append the `.json` suffix to the output filename passed in.
-            assert staged_path.name == "test_db.json"
-
-            connect_params = {
-                "host": "localhost",
-                "port": 3306,
-                "database": "ingest_test",
-                "user": "root",
-                "password": "password",
-            }
-
-            uploader = SingleStoreUploader(
-                connection_config=SingleStoreConnectionConfig(
-                    host=connect_params["host"],
-                    port=connect_params["port"],
-                    database=connect_params["database"],
-                    user=connect_params["user"],
-                    access_config=SingleStoreAccessConfig(password=connect_params["password"]),
-                ),
-                upload_config=SingleStoreUploaderConfig(
-                    table_name="elements",
-                ),
-            )
-
-            uploader.run(path=staged_path, file_data=mock_file_data)
-
-            staged_df = pd.read_json(staged_path, orient="records", lines=True)
-            expected_num_elements = len(staged_df)
-            validate_destination(
-                connect_params=connect_params,
-                expected_num_elements=expected_num_elements,
-            )
-
-            uploader.run(path=staged_path, file_data=mock_file_data)
-            validate_destination(
-                connect_params=connect_params,
-                expected_num_elements=expected_num_elements,
-            )
+        stager = SingleStoreUploadStager()
+        staged_path = stager.run(
+            elements_filepath=upload_file,
+            file_data=mock_file_data,
+            output_dir=temp_dir,
+            output_filename=upload_file.name,
+        )
+
+        # The stager should append the `.json` suffix to the output filename passed in.
+        assert staged_path.suffix == upload_file.suffix
+
+        connect_params = {
+            "host": "localhost",
+            "port": 3306,
+            "database": "ingest_test",
+            "user": "root",
+            "password": "password",
+        }
+
+        uploader = SingleStoreUploader(
+            connection_config=SingleStoreConnectionConfig(
+                host=connect_params["host"],
+                port=connect_params["port"],
+                database=connect_params["database"],
+                user=connect_params["user"],
+                access_config=SingleStoreAccessConfig(password=connect_params["password"]),
+            ),
+            upload_config=SingleStoreUploaderConfig(
+                table_name="elements",
+            ),
+        )
+        uploader.precheck()
+        uploader.run(path=staged_path, file_data=mock_file_data)
+
+        with staged_path.open("r") as f:
+            staged_data = json.load(f)
+        expected_num_elements = len(staged_data)
+        validate_destination(
+            connect_params=connect_params,
+            expected_num_elements=expected_num_elements,
+        )
+
+        uploader.run(path=staged_path, file_data=mock_file_data)
+        validate_destination(
+            connect_params=connect_params,
+            expected_num_elements=expected_num_elements,
+        )
+
+
+@pytest.mark.parametrize("upload_file_str", ["upload_file_ndjson", "upload_file"])
+def test_singlestore_stager(
+    request: TopRequest,
+    upload_file_str: str,
+    tmp_path: Path,
+):
+    upload_file: Path = request.getfixturevalue(upload_file_str)
+    stager = SingleStoreUploadStager()
+    stager_validation(
+        configs=StagerValidationConfigs(test_id=CONNECTOR_TYPE, expected_count=22),
+        input_file=upload_file,
+        stager=stager,
+        tmp_dir=tmp_path,
+    )