Implement a new Interface function for datastore_before_update

ckanext/xloader/interfaces.py

@@ -47,3 +47,30 @@ def after_upload(self, context, resource_dict, dataset_dict):
             the resource that was uploaded
         """
         pass
+
+    def datastore_before_update(self, resource_id, existing_info, new_headers):
+        """ Called by the loader just before it is about to modify the
+        DataStore table for a resource (truncate, drop+recreate, or create).
+        It allows plugins to inspect the difference between the current
+        DataStore columns and the ones detected in the incoming file, for
+        example to log an activity when columns are added, removed or
+        renamed.
+
+        :param resource_id: the ID of the resource whose DataStore table is
+            about to be updated.
+        :type resource_id: string
+
+        :param existing_info: a mapping of ``{field_id: info_dict}`` built
+            from the existing DataStore table's Data Dictionary, or ``None``
+            if the DataStore table does not yet exist.
+        :type existing_info: dict or None
+
+        :param new_headers: the list of field dicts that will be written to
+            the DataStore. Each dict has at least an ``id`` and ``type``
+            key, and may include an ``info`` dict for fields that already
+            existed.
+        :type new_headers: list of dicts
+
+        The return value is ignored.
+        """
+        pass

ckanext/xloader/loader.py

@@ -18,10 +18,23 @@
 
 import ckan.plugins as p
 
+from .interfaces import IXloader
 from .job_exceptions import FileCouldNotBeLoadedError, LoaderError
 from .parser import CSV_SAMPLE_LINES, TypeConverter
 from .utils import cleanup_temp_file, datastore_resource_exists, headers_guess, type_guess
 
+
+def _notify_datastore_before_update(resource_id, existing_info, new_headers):
+    """Notify IXloader plugins that the DataStore table for ``resource_id``
+    is about to change. See ``IXloader.datastore_before_update``.
+    """
+    for plugin in p.PluginImplementations(IXloader):
+        plugin.datastore_before_update(
+            resource_id=resource_id,
+            existing_info=existing_info,
+            new_headers=new_headers,
+        )
+
 from ckan.plugins.toolkit import config
 
 import ckanext.datastore.backend.postgres as datastore_db
@@ -356,6 +369,7 @@ def load_csv(csv_filepath, resource_id, mimetype='text/csv', allow_type_guessing
         '''
         fields_match = _fields_match(fields, existing_fields, logger)
         if fields_match == FieldMatch.EXACT_MATCH:
+            _notify_datastore_before_update(resource_id, existing_info, fields)
             logger.info('Clearing records for "%s" from DataStore.', resource_id)
             _clear_datastore_resource(resource_id)
         else:
@@ -366,6 +380,7 @@ def load_csv(csv_filepath, resource_id, mimetype='text/csv', allow_type_guessing
             # then we need to re-guess types
             if allow_type_guessing and fields_match == FieldMatch.MISMATCH:
                 raise LoaderError("File structure has changed, reverting to Tabulator")
+            _notify_datastore_before_update(resource_id, existing_info, fields)
     else:
         fields = [
             {'id': header_name,
@@ -590,6 +605,7 @@ def row_iterator():
         Otherwise 'datastore_create' will append to the existing datastore.
         And if the fields have significantly changed, it may also fail.
         '''
+        _notify_datastore_before_update(resource_id, existing_info, headers_dicts)
         if existing:
             if _fields_match(headers_dicts, existing_fields, logger) == FieldMatch.EXACT_MATCH:
                 logger.info('Clearing records for "%s" from DataStore.', resource_id)