feat(tests): add HETA 1.2.0 parquet size checks and GeoJSON parity validation

tests/aignostics/application/cli_test.py

@@ -3,6 +3,7 @@
 import contextlib
 import json
 import platform
+import random
 import re
 from collections.abc import Generator
 from datetime import UTC, datetime, timedelta
@@ -11,6 +12,19 @@
 from unittest.mock import MagicMock, patch
 
 import pytest
+from aignx.codegen.exceptions import ForbiddenException
+from aignx.codegen.exceptions import NotFoundException as ApiNotFound
+from aignx.codegen.models import (
+    ItemOutput,
+    ItemResultReadResponse,
+    ItemState,
+    ItemTerminationReason,
+    RunItemStatistics,
+    RunOutput,
+    RunReadResponse,
+    RunState,
+    RunTerminationReason,
+)
 from loguru import logger
 from tenacity import Retrying, retry, stop_after_attempt, wait_exponential
 from typer.testing import CliRunner
@@ -19,7 +33,7 @@
 from aignostics.cli import cli
 from aignostics.platform import LIST_APPLICATION_RUNS_MAX_PAGE_SIZE
 from aignostics.utils import Health, sanitize_path
-from tests.conftest import normalize_output, print_directory_structure
+from tests.conftest import assert_parquet_geojson_parity, normalize_output, print_directory_structure
 from tests.constants_test import (
     HETA_APPLICATION_ID,
     HETA_APPLICATION_VERSION,
@@ -224,7 +238,11 @@ def test_cli_application_dump_schemata(runner: CliRunner, tmp_path: Path, record
         ],
     )
     assert result.exit_code == 0
-    assert "Zipped 11 files" in normalize_output(result.output)
+    output = normalize_output(result.output)
+    zipped_match = re.search(r"Zipped (\d+) files", output)
+    assert zipped_match is not None, f"Expected 'Zipped N files' in output: {output}"
+    zipped_count = int(zipped_match.group(1))
+    assert 12 <= zipped_count <= 16, f"Expected between 12 and 16 schemata files, got {zipped_count}"
     zip_file = sanitize_path(Path(tmp_path / f"{HETA_APPLICATION_ID}_{HETA_APPLICATION_VERSION}_schemata.zip"))
     assert zip_file.exists(), f"Expected zip file {zip_file} not found"
 
@@ -847,8 +865,6 @@ def test_cli_run_list_for_organization(runner: CliRunner) -> None:
 @pytest.mark.unit
 def test_cli_run_list_forbidden_with_organization(runner: CliRunner) -> None:
     """Check ForbiddenException with --for-organization shows org-specific access denied message."""
-    from aignx.codegen.exceptions import ForbiddenException
-
     with patch.object(
         ApplicationService, "application_runs", side_effect=ForbiddenException(status=403, reason="Forbidden")
     ):
@@ -862,8 +878,6 @@ def test_cli_run_list_forbidden_with_organization(runner: CliRunner) -> None:
 @pytest.mark.unit
 def test_cli_run_list_forbidden_without_organization(runner: CliRunner) -> None:
     """Check ForbiddenException without --for-organization shows generic access denied message."""
-    from aignx.codegen.exceptions import ForbiddenException
-
     with patch.object(
         ApplicationService, "application_runs", side_effect=ForbiddenException(status=403, reason="Forbidden")
     ):
@@ -897,18 +911,6 @@ def test_cli_run_describe_not_found(runner: CliRunner, record_property) -> None:
 @pytest.mark.integration
 def test_cli_run_describe_json_includes_items(runner: CliRunner) -> None:
     """Check run describe --format=json includes items in output."""
-    from aignx.codegen.models import (
-        ItemOutput,
-        ItemResultReadResponse,
-        ItemState,
-        ItemTerminationReason,
-        RunItemStatistics,
-        RunOutput,
-        RunReadResponse,
-        RunState,
-        RunTerminationReason,
-    )
-
     mock_run_data = RunReadResponse(
         run_id="test-run-id-123",
         application_id="test-app",
@@ -1111,8 +1113,10 @@ def test_cli_run_execute(runner: CliRunner, tmp_path: Path, record_property) ->
     results_dir = tmp_path / SPOT_1_FILENAME.replace(".tiff", "")
     assert results_dir.is_dir(), f"Expected directory {results_dir} not found"
     files_in_dir = list(results_dir.glob("*"))
-    assert len(files_in_dir) == 9, (
-        f"Expected 9 files in {results_dir}, but found {len(files_in_dir)}: {[f.name for f in files_in_dir]}"
+    expected_count = len(SPOT_1_EXPECTED_RESULT_FILES)
+    assert len(files_in_dir) == expected_count, (
+        f"Expected {expected_count} files in {results_dir}, but found {len(files_in_dir)}: "
+        f"{[f.name for f in files_in_dir]}"
     )
     print(f"Found files in {results_dir}:")
     for filename, expected_size, tolerance_percent in SPOT_1_EXPECTED_RESULT_FILES:
@@ -1133,6 +1137,9 @@ def test_cli_run_execute(runner: CliRunner, tmp_path: Path, record_property) ->
             f"({min_size} to {max_size} bytes, ±{tolerance_percent}% of {expected_size})"
         )
 
+    # Validate parquet <-> GeoJSON parity: area for segmentation, count for cell classification.
+    assert_parquet_geojson_parity(results_dir)
+
     # Validate the execute command exited successfully
     assert result.exit_code == 0
 
@@ -1222,9 +1229,6 @@ def test_cli_run_update_item_metadata_not_dict(runner: CliRunner) -> None:
 @pytest.mark.sequential
 def test_cli_run_dump_and_update_custom_metadata(runner: CliRunner, tmp_path: Path) -> None:
     """Test dumping and updating custom metadata via CLI commands."""
-    import json
-    import random
-
     unique_tag = f"test_metadata_{datetime.now(tz=UTC).timestamp()}"
     with submitted_run(runner, tmp_path, CSV_CONTENT_SPOT0, extra_args=["--tags", unique_tag, "--force"]) as run_id:
         # Step 1: Dump initial custom metadata of run
@@ -1313,11 +1317,8 @@ def test_cli_run_dump_and_update_custom_metadata(runner: CliRunner, tmp_path: Pa
 @pytest.mark.e2e
 @pytest.mark.timeout(timeout=240)
 @pytest.mark.sequential
-def test_cli_run_dump_and_update_item_custom_metadata(runner: CliRunner, tmp_path: Path) -> None:  # noqa: PLR0915
+def test_cli_run_dump_and_update_item_custom_metadata(runner: CliRunner, tmp_path: Path) -> None:
     """Test dumping and updating item custom metadata via CLI commands."""
-    import json
-    import random
-
     unique_tag = f"test_item_metadata_{datetime.now(tz=UTC).timestamp()}"
     # CSV_CONTENT_SPOT0 uses SPOT_0_FILENAME as external_id, which the describe output surfaces
     # as "Item External ID: `...`" — the get_external_id() helper below captures it dynamically.
@@ -1773,8 +1774,6 @@ def test_cli_application_version_document_describe_success(runner: CliRunner, re
 def test_cli_application_version_document_describe_not_found(runner: CliRunner, record_property) -> None:
     """`application version document describe` exits 2 with a clear message on 404."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-03")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_documents = MagicMock()
     fake_documents.details.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
     fake_client = MagicMock()
@@ -1870,8 +1869,6 @@ def test_cli_application_version_document_list_json_empty(runner: CliRunner, rec
 def test_cli_application_version_document_list_resolve_not_found_text(runner: CliRunner, record_property) -> None:
     """`application version document list` exits 2 when the application version cannot be resolved."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-01")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_client = MagicMock()
     fake_client.applications.versions.documents.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
 
@@ -1888,8 +1885,6 @@ def test_cli_application_version_document_list_resolve_not_found_text(runner: Cl
 def test_cli_application_version_document_list_resolve_not_found_json(runner: CliRunner, record_property) -> None:
     """`application version document list --format json` emits structured error on 404."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-01")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_client = MagicMock()
     fake_client.applications.versions.documents.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
 
@@ -1976,8 +1971,6 @@ def test_cli_application_version_document_describe_json_success(runner: CliRunne
 def test_cli_application_version_document_describe_resolve_not_found_text(runner: CliRunner, record_property) -> None:
     """`describe` exits 2 when the application version cannot be resolved (text format)."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-03")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_client = MagicMock()
     fake_client.applications.versions.documents.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
 
@@ -1996,8 +1989,6 @@ def test_cli_application_version_document_describe_resolve_not_found_text(runner
 def test_cli_application_version_document_describe_resolve_not_found_json(runner: CliRunner, record_property) -> None:
     """`describe --format json` emits structured error when version cannot be resolved."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-03")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_client = MagicMock()
     fake_client.applications.versions.documents.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
 
@@ -2026,8 +2017,6 @@ def test_cli_application_version_document_describe_resolve_not_found_json(runner
 def test_cli_application_version_document_describe_not_found_json(runner: CliRunner, record_property) -> None:
     """`describe --format json` emits structured error when the document is missing."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-03")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_documents = MagicMock()
     fake_documents.details.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
     fake_client = MagicMock()
@@ -2111,8 +2100,6 @@ def test_cli_application_version_document_download_resolve_not_found(
 ) -> None:
     """`download` exits 2 when the application version cannot be resolved."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-04")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_client = MagicMock()
     fake_client.applications.versions.documents.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
 
@@ -2142,8 +2129,6 @@ def test_cli_application_version_document_download_not_found(
 ) -> None:
     """`download` exits 2 with a clear message when the document does not exist."""
     record_property("tested-item-id", "TC-APPLICATION-CLI-05-04")
-    from aignx.codegen.exceptions import NotFoundException as ApiNotFound
-
     fake_documents = MagicMock()
     fake_documents.download_to_path.side_effect = ApiNotFound(status=404, reason=API_REASON_NOT_FOUND)
     fake_client = MagicMock()

tests/aignostics/application/gui_test.py

@@ -21,7 +21,7 @@
     _resolve_artifact_url_or_notify,
 )
 from aignostics.cli import cli
-from tests.conftest import assert_notified, normalize_output, print_directory_structure
+from tests.conftest import assert_notified, assert_parquet_geojson_parity, normalize_output, print_directory_structure
 from tests.constants_test import (
     HETA_APPLICATION_ID,
     HETA_APPLICATION_VERSION,
@@ -31,6 +31,7 @@
     SPOT_0_FILESIZE,
     SPOT_0_GS_URL,
     SPOT_1_FILENAME,
+    SPOT_1_FILESIZE,
     SPOT_1_GS_URL,
 )
 
@@ -214,7 +215,7 @@ async def test_gui_download_dataset_via_application_to_run_cancel_to_find_back(
             assert SPOT_1_FILENAME in normalize_output(result.stdout)
             expected_file = Path(tmp_path) / SPOT_1_FILENAME
             assert expected_file.exists(), f"Expected file {expected_file} not found"
-            assert expected_file.stat().st_size == 14681750
+            assert expected_file.stat().st_size == SPOT_1_FILESIZE
 
             # Open the GUI and navigate to Atlas H&E-TME application
             await user.open("/")
@@ -354,7 +355,7 @@ async def test_gui_download_dataset_via_application_to_run_cancel_to_find_back(
 @pytest.mark.flaky(retries=1, delay=5)
 @pytest.mark.timeout(timeout=60 * 10)
 @pytest.mark.sequential  # Helps on Linux with image analysis step otherwise timing out
-async def test_gui_run_download(  # noqa: PLR0915
+async def test_gui_run_download(  # noqa: PLR0914, PLR0915
     user: User, runner: CliRunner, tmp_path: Path, silent_logging: None, record_property
 ) -> None:
     """Test that the user can download a run result via the GUI."""
@@ -440,8 +441,9 @@ async def test_gui_run_download(  # noqa: PLR0915
 
         # Check for files in the results directory
         files_in_results_dir = list(results_dir.glob("*"))
-        assert len(files_in_results_dir) == 9, (
-            f"Expected 9 files in {results_dir}, but found {len(files_in_results_dir)}: "
+        expected_count = len(SPOT_0_EXPECTED_RESULT_FILES)
+        assert len(files_in_results_dir) == expected_count, (
+            f"Expected {expected_count} files in {results_dir}, but found {len(files_in_results_dir)}: "
             f"{[f.name for f in files_in_results_dir]}"
         )
 
@@ -464,6 +466,9 @@ async def test_gui_run_download(  # noqa: PLR0915
                 f"({min_size} to {max_size} bytes, ±{tolerance_percent}% of {expected_size})"
             )
 
+        # Validate parquet <-> GeoJSON parity: area for segmentation, count for cell classification.
+        assert_parquet_geojson_parity(results_dir)
+
 
 @pytest.mark.integration
 @pytest.mark.sequential
@@ -479,12 +484,14 @@ async def test_gui_run_results_pagination_show_more_button_hidden_when_few_resul
     """
     record_property("tested-item-id", "SPEC-APPLICATION-SERVICE, SPEC-GUI-SERVICE")
 
-    # Find a run with fewer items than RESULTS_PAGE_SIZE
+    # Find a run with fewer items than RESULTS_PAGE_SIZE.
+    # Omit has_output so the server-side filter is applied without client-side pagination:
+    # item_count already acts as a proxy (runs with no output show item_count=0 and fail
+    # the 0 < item_count <= RESULTS_PAGE_SIZE check below).
     runs = Service().application_runs(
         application_id=HETA_APPLICATION_ID,
         application_version=HETA_APPLICATION_VERSION,
-        has_output=True,
-        limit=20,
+        limit=5,
     )
 
     # Find a run with few enough items

tests/aignostics/dataset/cli_test.py

@@ -11,7 +11,7 @@
 
 from aignostics.cli import cli
 from tests.conftest import normalize_output
-from tests.constants_test import SPOT_1_FILENAME, SPOT_1_GS_URL
+from tests.constants_test import SPOT_1_FILENAME, SPOT_1_FILESIZE, SPOT_1_GS_URL
 
 SERIES_UID = "1.3.6.1.4.1.5962.99.1.1069745200.1645485340.1637452317744.2.0"
 THUMBNAIL_UID = "1.3.6.1.4.1.5962.99.1.1038911754.1238045814.1637421484298.15.0"
@@ -149,7 +149,7 @@ def test_cli_aignostics_download_sample(runner: CliRunner, tmp_path: Path, recor
 
     expected_file = tmp_path / SPOT_1_FILENAME
     assert expected_file.exists(), f"Expected file {expected_file} not found"
-    assert expected_file.stat().st_size == 14681750
+    assert expected_file.stat().st_size == SPOT_1_FILESIZE
 
 
 @pytest.mark.integration