Ensure tool_data_path exists; assert non-DM skip via persisted XML

install_tool_data_tables used to create tool_data_path as a side effect
of os.makedirs(target_dir); skipping it for non-DM repos means
copy_sample_files now blows up with FileNotFoundError. Explicitly
mkdir(exist_ok=True) before copying.

The non-DM integration test was asserting against the in-memory
tool_data_tables dict, which is populated as tools are loaded
regardless of registration; switch to verify_no_installed_repository_data_table_entries,
which inspects shed_tool_data_table_conf.xml.

Author: natefoo

lib/galaxy/tool_shed/galaxy_install/install_manager.py

@@ -188,6 +188,8 @@ def __handle_repository_contents(
             sample_files = irmm_metadata_dict.get("sample_files", [])
             tool_index_sample_files = stdtm.get_tool_index_sample_files(sample_files)
             tool_data_path = self.app.config.tool_data_path
+            # Used to be created as a side effect of install_tool_data_tables; that no longer runs for non-DM repos.
+            os.makedirs(tool_data_path, exist_ok=True)
             tool_util.copy_sample_files(tool_data_path, tool_index_sample_files, tool_path=tool_path)
             sample_files_copied = [str(s) for s in tool_index_sample_files]
             repository_tools_tups = irmm.get_repository_tools_tups()

test/integration/test_repository_operations.py

@@ -1,4 +1,5 @@
 import os
+import xml.etree.ElementTree as ET
 from collections import namedtuple
 
 from sqlalchemy import select
@@ -62,13 +63,14 @@ def test_tool_with_package_dependency_uninstall(self):
         self.uninstall_repository(*repo)
 
     def test_non_data_manager_install_skips_data_table_registration(self):
-        """Non-Data-Manager repos must not register data tables on install."""
+        """Non-Data-Manager repos must not persist data table entries in shed_tool_data_table_conf.xml."""
         non_dm_repo = ("devteam", "bwa", "051eba708f43")
         non_dm_table_names = {"bwa_indexes", "bwa_mem_indexes"}
         self.install_repository(*non_dm_repo)
-        registered = set(self._app.tool_data_tables.data_tables.keys())
+        shed_conf = self._app.config.shed_tool_data_table_config
+        registered = {t.get("name") for t in ET.parse(shed_conf).getroot().findall("table")}
         leaked = non_dm_table_names & registered
-        assert not leaked, f"Unexpected data tables registered by non-DM repo: {sorted(leaked)}"
+        assert not leaked, f"Unexpected data tables in {shed_conf}: {sorted(leaked)}"
 
     def test_repository_update(self):
         response = self._install_repository(revision=REVISION_4, version="0.0.3", allow_upgraded=True)[0]