✨ NEW: Add verdi database summary

aiida/cmdline/commands/cmd_database.py

@@ -194,3 +194,48 @@ def detect_invalid_nodes():
         echo.echo_success('no integrity violations detected')
     else:
         echo.echo_critical('one or more integrity violations detected')
+
+
+@verdi_database.command('summary')
+@options.VERBOSE()
+def database_summary(verbose):
+    """Summarise the entities in the database."""
+    from aiida.orm import QueryBuilder, Node, Group, Computer, Comment, Log, User
+    data = {}
+
+    # User
+    query_user = QueryBuilder().append(User, project=['email'])
+    data['Users'] = {'count': query_user.count()}
+    if verbose:
+        data['Users']['emails'] = query_user.distinct().all(flat=True)
+
+    # Computer
+    query_comp = QueryBuilder().append(Computer, project=['name'])
+    data['Computers'] = {'count': query_comp.count()}
+    if verbose:
+        data['Computers']['names'] = query_comp.distinct().all(flat=True)
+
+    # Node
+    count = QueryBuilder().append(Node).count()
+    data['Nodes'] = {'count': count}
+    if verbose:
+        node_types = QueryBuilder().append(Node, project=['node_type']).distinct().all(flat=True)
+        data['Nodes']['node_types'] = node_types
+        process_types = QueryBuilder().append(Node, project=['process_type']).distinct().all(flat=True)
+        data['Nodes']['process_types'] = [p for p in process_types if p]
+
+    # Group
+    query_group = QueryBuilder().append(Group, project=['type_string'])
+    data['Groups'] = {'count': query_group.count()}
+    if verbose:
+        data['Groups']['type_strings'] =  query_group.distinct().all(flat=True)
+
+    # Comment
+    count = QueryBuilder().append(Comment).count()
+    data['Comments'] = {'count': count}
+
+    # Log
+    count = QueryBuilder().append(Log).count()
+    data['Logs'] = {'count': count}
+
+    echo.echo_dictionary(data, sort_keys=False, fmt='yaml')

aiida/cmdline/utils/echo.py

@@ -187,7 +187,7 @@ def echo_formatted_list(collection, attributes, sort=None, highlight=None, hide=
             click.secho(template.format(symbol=' ', *values))
 
 
-def _format_dictionary_json_date(dictionary):
+def _format_dictionary_json_date(dictionary, sort_keys=True):
     """Return a dictionary formatted as a string using the json format and converting dates to strings."""
     from aiida.common import json
 
@@ -201,27 +201,39 @@ def default_jsondump(data):
 
         raise TypeError(f'{repr(data)} is not JSON serializable')
 
-    return json.dumps(dictionary, indent=4, sort_keys=True, default=default_jsondump)
+    return json.dumps(dictionary, indent=4, sort_keys=sort_keys, default=default_jsondump)
 
 
-VALID_DICT_FORMATS_MAPPING = OrderedDict((('json+date', _format_dictionary_json_date), ('yaml', yaml.dump),
-                                          ('yaml_expanded', lambda d: yaml.dump(d, default_flow_style=False))))
+def _format_yaml(dictionary, sort_keys=True):
+    """Return a dictionary formatted as a string using the YAML format."""
+    return yaml.dump(dictionary, sort_keys=sort_keys)
 
 
-def echo_dictionary(dictionary, fmt='json+date'):
+def _format_yaml_expanded(dictionary, sort_keys=True):
+    """Return a dictionary formatted as a string using the expanded YAML format."""
+    return yaml.dump(dictionary, sort_keys=sort_keys, default_flow_style=False)
+
+
+VALID_DICT_FORMATS_MAPPING = OrderedDict(
+    (('json+date', _format_dictionary_json_date), ('yaml', _format_yaml), ('yaml_expanded', _format_yaml_expanded))
+)
+
+
+def echo_dictionary(dictionary, fmt='json+date', sort_keys=True):
     """
     Print the given dictionary to stdout in the given format
 
     :param dictionary: the dictionary
     :param fmt: the format to use for printing
+    :param sort_keys: Whether to automatically sort keys
     """
     try:
         format_function = VALID_DICT_FORMATS_MAPPING[fmt]
     except KeyError:
         formats = ', '.join(VALID_DICT_FORMATS_MAPPING.keys())
         raise ValueError(f'Unrecognised printing format. Valid formats are: {formats}')
 
-    echo(format_function(dictionary))
+    echo(format_function(dictionary, sort_keys=sort_keys))
 
 
 def is_stdout_redirected():

docs/source/reference/command_line.rst

@@ -220,6 +220,7 @@ Below is a list with all available subcommands.
     Commands:
       integrity  Check the integrity of the database and fix potential issues.
       migrate    Migrate the database to the latest schema version.
+      summary    Summarise the entities in the database.
       version    Show the version of the database.
 
 

tests/cmdline/commands/test_database.py

@@ -180,3 +180,13 @@ def tests_database_version(run_cli_command, manager):
     result = run_cli_command(cmd_database.database_version)
     assert result.output_lines[0].endswith(backend_manager.get_schema_generation_database())
     assert result.output_lines[1].endswith(backend_manager.get_schema_version_database())
+
+
+@pytest.mark.usefixtures('clear_database_before_test')
+def tests_database_summary(aiida_localhost, run_cli_command):
+    """Test the ``verdi database summary -v`` command."""
+    from aiida import orm
+    node = orm.Dict().store()
+    result = run_cli_command(cmd_database.database_summary, ['--verbose'])
+    assert aiida_localhost.label in result.output
+    assert node.node_type in result.output