Auto-generate additional tests

Author: jp-dark

apis/python/remote_tests/carrara/README.md

@@ -23,7 +23,7 @@ export CARRARA_TEST_TEAMSPACE="..."  # Teamspace to use for tests
 
 Run the tests with pytest:
 ```bash
-pytest api/python/remotes_tests/carrara/
+pytest api/python/remotes_tests/carrara/ --carrara -v
 ```
 
 Objects will be created at the specified workspace/teamspace: `tiledb://${CARRARA_TEST_WORKSPACE}/${CARRARA_TEST_TEAMSPACE}/...`. The test should remove all created objects when completed; however, assets may remain in the case of a test failure.

apis/python/remote_tests/carrara/_util.py

@@ -0,0 +1,46 @@
+# Copyright (c) TileDB, Inc.
+# Licensed under the MIT License.
+
+from __future__ import annotations
+
+import pathlib
+import urllib.parse
+from typing import Any
+
+
+def parse_tiledb_uri(uri: str) -> tuple[str, str, str | None]:
+    """Given a tiledb:// URI, return (workspace, teamspace, path)."""
+    p_uri = urllib.parse.urlparse(uri)
+    workspace = p_uri.netloc
+
+    p_path = pathlib.Path(p_uri.path)
+    if len(p_path.parts) < 2:
+        raise ValueError(f"Not a tiledb URI - missing teamspace: {uri}")
+    teamspace = p_path.parts[1]
+
+    path = None if len(p_path.parts) == 2 else str(pathlib.PurePath(*p_path.parts[2:]))
+
+    return workspace, teamspace, path
+
+
+def get_asset_info(uri: str) -> dict[str, Any]:
+    import tiledb.client
+
+    _, teamspace, path = parse_tiledb_uri(uri)
+    return tiledb.client.assets.get_asset(path, teamspace=teamspace).to_dict()
+
+
+def s3_from_tiledb_uri(tiledb_uri: str) -> str:
+    """Given a tiledb:// URI, return the object's S3 URL.
+
+    This relies on the teamspace bucket being owned by the test user."""
+    import tiledb.client
+
+    assert tiledb_uri.startswith("tiledb://")
+
+    _, teamspace, path = parse_tiledb_uri(tiledb_uri)
+    asset_info = tiledb.client.assets.get_asset(path, teamspace=teamspace)
+    s3_url = asset_info.uri
+    assert s3_url.startswith("s3://")
+
+    return s3_url

apis/python/remote_tests/carrara/conftest.py

@@ -21,6 +21,25 @@
 TEST_FOLDER = os.getenv("CARRARA_TEST_FOLDER") or "remote_test"
 BASE_URI = f"tiledb://{WORKSPACE_NAME}/{TEAMSPACE_NAME}/{TEST_FOLDER}"
 
+TESTS_INPUT_DIR = os.path.abspath(
+    os.path.join(os.path.dirname(__file__), "../../../../libtiledbvcf/test/inputs")
+)
+
+
+def pytest_addoption(parser):
+    parser.addoption(
+        "--carrara", action="store_true", default=False, help="run Carrara tests"
+    )
+
+
+def pytest_collection_modifyitems(config, items):
+    if config.getoption("--carrara"):
+        return
+    skip_carrara = pytest.mark.skip(reason="need --carrara option to run")
+    for item in items:
+        if "carrara" in item.keywords:
+            item.add_marker(skip_carrara)
+
 
 @pytest.fixture(scope="session")
 def carrara_login() -> None:
@@ -54,3 +73,9 @@ def carrara_group_path() -> Generator[str, None, None]:
         pass
 
 
+@pytest.fixture(scope="session")
+def vcf_sample_uris() -> list[str]:
+    """Returns paths to local VCF/BCF test files used for ingestion tests."""
+    return [os.path.join(TESTS_INPUT_DIR, s) for s in ["small.bcf", "small2.bcf"]]
+
+

apis/python/remote_tests/carrara/test_01_setup.py

@@ -7,8 +7,11 @@
 
 from __future__ import annotations
 
+import pytest
 import tiledbvcf
 
+
+@pytest.mark.carrara
 def test_dataset_create(carrara_login: None, carrara_group_path: str):
     with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
         ds.create_dataset()

apis/python/remote_tests/carrara/test_02_ingest.py

@@ -0,0 +1,72 @@
+# Copyright (c) TileDB, Inc.
+# Licensed under the MIT License.
+"""
+VCF sample ingestion with Carrara URIs.
+"""
+
+from __future__ import annotations
+
+import pytest
+import tiledbvcf
+
+
+@pytest.mark.carrara
+def test_ingest_samples(
+    carrara_login: None, carrara_group_path: str, vcf_sample_uris: list[str]
+) -> None:
+    """Ingest two BCF samples into a remote Carrara dataset and verify total count."""
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.create_dataset()
+        ds.ingest_samples(vcf_sample_uris)
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        assert ds.count() == 14
+
+
+@pytest.mark.carrara
+def test_ingest_and_region_query(
+    carrara_login: None, carrara_group_path: str, vcf_sample_uris: list[str]
+) -> None:
+    """Ingest samples and verify a region-filtered count."""
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.create_dataset()
+        ds.ingest_samples(vcf_sample_uris)
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        assert ds.count(regions=["1:12700-13400"]) == 6
+
+
+@pytest.mark.carrara
+def test_ingest_and_sample_query(
+    carrara_login: None, carrara_group_path: str, vcf_sample_uris: list[str]
+) -> None:
+    """Ingest samples and verify a sample- and region-filtered count."""
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.create_dataset()
+        ds.ingest_samples(vcf_sample_uris)
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        assert ds.count(samples=["HG00280"], regions=["1:12700-13400"]) == 4
+
+
+@pytest.mark.carrara
+def test_ingest_incremental(
+    carrara_login: None, carrara_group_path: str, vcf_sample_uris: list[str]
+) -> None:
+    """Ingest one sample, verify count, then add a second sample and re-verify."""
+    first, second = vcf_sample_uris[0], vcf_sample_uris[1]
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.create_dataset()
+        ds.ingest_samples([first])
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        count_after_first = ds.count()
+        assert count_after_first > 0
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.ingest_samples([second])
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        assert ds.count() == 14
+        assert ds.count() > count_after_first

apis/python/remote_tests/carrara/test_03_delete.py

@@ -0,0 +1,48 @@
+# Copyright (c) TileDB, Inc.
+# Licensed under the MIT License.
+"""
+Dataset and sample deletion with Carrara URIs.
+"""
+
+from __future__ import annotations
+
+import pytest
+import tiledb
+import tiledbvcf
+
+
+@pytest.mark.carrara
+def test_delete_dataset(
+    carrara_login: None, carrara_group_path: str, vcf_sample_uris: list[str]
+) -> None:
+    """Create and ingest a dataset, then delete it and verify it no longer exists."""
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.create_dataset()
+        ds.ingest_samples(vcf_sample_uris)
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        assert ds.count() == 14
+
+    with tiledb.Group(carrara_group_path, mode="m") as G:
+        G.delete(recursive=True)
+
+    assert tiledb.object_type(carrara_group_path) is None
+
+
+@pytest.mark.carrara
+def test_delete_samples(
+    carrara_login: None, carrara_group_path: str, vcf_sample_uris: list[str]
+) -> None:
+    """Ingest two samples, delete one, and verify only the other remains."""
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.create_dataset()
+        ds.ingest_samples(vcf_sample_uris)
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        assert sorted(ds.samples()) == ["HG00280", "HG01762"]
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="w") as ds:
+        ds.delete_samples(["HG00280"])
+
+    with tiledbvcf.Dataset(carrara_group_path, mode="r") as ds:
+        assert ds.samples() == ["HG01762"]