Merge pull request #3729 from mvdbeek/cleanup_tool_cache

Remove outdated items from tool cache when reloading the toolbox

lib/galaxy/queue_worker.py

@@ -86,6 +86,7 @@ def reload_tool(app, **kwargs):
 def reload_toolbox(app, **kwargs):
     log.debug("Executing toolbox reload on '%s'", app.config.server_name)
     reload_count = app.toolbox._reload_count
+    app.tool_cache.cleanup()
     app.toolbox = _get_new_toolbox(app)
     app.toolbox._reload_count = reload_count + 1
 

lib/galaxy/tools/toolbox/cache.py

@@ -1,3 +1,6 @@
+import os
+
+from galaxy.util.hash_util import md5_hash_file
 
 
 class ToolCache(object):
@@ -7,9 +10,20 @@ class ToolCache(object):
     """
 
     def __init__(self):
+        self._hash_by_tool_paths = {}
         self._tools_by_path = {}
         self._tool_paths_by_id = {}
 
+    def cleanup(self):
+        """Remove uninstalled tools from tool cache if they are not on disk anymore or if their content has changed."""
+        paths_to_cleanup = {path: tool.all_ids for path, tool in self._tools_by_path.items() if not os.path.exists(path) or md5_hash_file(path) != self._hash_by_tool_paths[path]}
+        for config_filename, tool_ids in paths_to_cleanup.items():
+            del self._hash_by_tool_paths[config_filename]
+            del self._tools_by_path[config_filename]
+            for tool_id in tool_ids:
+                if tool_id in self._tool_paths_by_id:
+                    del self._tool_paths_by_id[tool_id]
+
     def get_tool(self, config_filename):
         """ Get the tool from the cache if the tool is up to date.
         """
@@ -18,10 +32,13 @@ class ToolCache(object):
     def expire_tool(self, tool_id):
         if tool_id in self._tool_paths_by_id:
             config_filename = self._tool_paths_by_id[tool_id]
+            del self._hash_by_tool_paths[config_filename]
             del self._tool_paths_by_id[tool_id]
             del self._tools_by_path[config_filename]
 
     def cache_tool(self, config_filename, tool):
+        tool_hash = md5_hash_file(config_filename)
         tool_id = str( tool.id )
+        self._hash_by_tool_paths[config_filename] = tool_hash
         self._tool_paths_by_id[tool_id] = config_filename
         self._tools_by_path[config_filename] = tool