From 795adac96d6e340045e51b7c52690df248ea93d4 Mon Sep 17 00:00:00 2001
From: Bruce Martin <bruce@chanzuckerberg.com>
Date: Fri, 24 Feb 2023 10:08:55 -0800
Subject: [PATCH] Python package: update tiledbsoma dep to 1.0.0rc1 (#225)

* add test for live directory correctness

* upgrade tiledbsoma dependency to 1.0.0rc1

* lint

* enable anon bucket access

* comments
---
 api/python/cell_census/pyproject.toml         |  2 +-
 .../cell_census/tests/test_directory.py       | 22 +++++++++++++++++++
 2 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/api/python/cell_census/pyproject.toml b/api/python/cell_census/pyproject.toml
index e27aaf9dd..fe204a699 100644
--- a/api/python/cell_census/pyproject.toml
+++ b/api/python/cell_census/pyproject.toml
@@ -35,7 +35,7 @@ dependencies= [
     # of TileDB on-disk storage format. Make sure this doesn't fall behind the builder's tiledbsoma version.
     # NOTE: tiledb is also a requirement of the API, but tiledbsoma also has a tiledb dependency, so just use 
     # the same version here
-    "tiledbsoma==1.0rc0",
+    "tiledbsoma==1.0.0rc1",
     "typing_extensions",
     "s3fs",
     "scikit-misc",
diff --git a/api/python/cell_census/tests/test_directory.py b/api/python/cell_census/tests/test_directory.py
index 5d7c1579b..419fc488c 100644
--- a/api/python/cell_census/tests/test_directory.py
+++ b/api/python/cell_census/tests/test_directory.py
@@ -2,6 +2,7 @@
 
 import pytest
 import requests_mock as rm
+import s3fs
 
 import cell_census
 from cell_census._release_directory import CELL_CENSUS_RELEASE_DIRECTORY_URL
@@ -57,3 +58,24 @@ def test_get_census_version_directory(directory_mock: Any) -> None:
 
     for tag in directory:
         assert directory[tag] == cell_census.get_census_version_description(tag)
+
+
+@pytest.mark.live_corpus
+def test_live_directory_contents() -> None:
+    # Sanity check that all directory contents are usable. This uses the
+    # live directory, so it _could_ start failing without a code change.
+    # But given the purpose of this package, that seems like a reasonable
+    # tradeoff, as the data directory should never be "corrupt" or there
+    # is widespread impact on users.
+
+    fs = s3fs.S3FileSystem(anon=True, cache_regions=True)
+
+    directory = cell_census.get_census_version_directory()
+    assert "latest" in directory
+
+    for version, version_description in directory.items():
+        with cell_census.open_soma(census_version=version) as census:
+            assert census is not None
+
+        assert fs.exists(version_description["soma"]["uri"])
+        assert fs.exists(version_description["h5ads"]["uri"])