diff --git a/gui/tests/artifacts.js b/gui/tests/artifacts.js
index b142bacc40731b5aff2cef5dd0c389ac4d792fbf..2860bc5fc74917192b7b1d82707eb95958f6fca9 100644
--- a/gui/tests/artifacts.js
+++ b/gui/tests/artifacts.js
@@ -71691,6 +71691,17 @@ window.nomadArtifacts = {
"type_data": "bool"
},
"default": false
+ },
+ {
+ "m_def": "nomad.metainfo.metainfo.Quantity",
+ "m_parent_index": 2,
+ "m_parent_sub_section": "quantities",
+ "name": "key_quantity",
+ "type": {
+ "type_kind": "python",
+ "type_data": "str"
+ },
+ "default": null
}
]
},
diff --git a/nomad/metainfo/metainfo.py b/nomad/metainfo/metainfo.py
index 3962b36cb798e1083753f434776fdf310894d44b..a7a61edfca9412f1bd8473bc112c95e35fbd7174 100644
--- a/nomad/metainfo/metainfo.py
+++ b/nomad/metainfo/metainfo.py
@@ -16,6 +16,7 @@
# limitations under the License.
#
from __future__ import annotations
+
import base64
from copy import deepcopy
import importlib
@@ -26,14 +27,15 @@ import re
import sys
from collections.abc import Iterable as IterableABC
from functools import reduce
-from pydantic import parse_obj_as, ValidationError, BaseModel, Field
from typing import (
Any, Callable as TypingCallable, Dict, Iterable, List, Optional, Set, Tuple, Type, TypeVar, Union, cast, ClassVar)
+
import docstring_parser
import jmespath
import numpy as np
import pandas as pd
import pint
+from pydantic import parse_obj_as, ValidationError, BaseModel, Field
from nomad.config import process
from nomad.metainfo.util import (
@@ -915,8 +917,8 @@ class Context:
pass
def create_reference(
- self, section: MSection, quantity_def: Quantity, value: MSection,
- global_reference: bool = False
+ self, section: MSection, quantity_def: Quantity, value: MSection,
+ global_reference: bool = False
) -> str:
'''
Returns a reference for the given target section (value) based on the given context.
@@ -1240,11 +1242,35 @@ class MSection(metaclass=MObjectMeta): # TODO find a way to make this a subclas
return super().__setattr__(name, value)
+ @property
+ def m_key(self):
+ if (subsection := self.m_parent_sub_section) is None:
+ return None
+
+ if subsection.key_quantity is not None:
+ key_quantity = subsection.key_quantity
+ elif (label := subsection.more.get('label_quantity')) is not None:
+ key_quantity = label
+ else:
+ return None
+
+ quantity_def = self.m_def.all_quantities.get(key_quantity)
+ if quantity_def.type != str:
+ raise TypeError(f'Key quantity {key_quantity} must be of type str.')
+
+ if self.m_is_set(quantity_def):
+ return self.m_get(quantity_def)
+
+ return None
+
def __getattr__(self, name):
# The existence of __getattr__ will make mypy and pylint ignore 'missing' dynamic
# attributes and functions and wrong types of those.
# Ideally we have a plugin for both that add the correct type info
+ if name == 'm_key':
+ return self.get(name)
+
if name in self.m_def.all_aliases:
return getattr(self, self.m_def.all_aliases[name].name)
@@ -1842,7 +1868,9 @@ class MSection(metaclass=MObjectMeta): # TODO find a way to make this a subclas
categories: List[Union[Category, Type['MCategory']]] = None,
include: TypingCallable[[Definition, MSection], bool] = None,
exclude: TypingCallable[[Definition, MSection], bool] = None,
- transform: TypingCallable[[Definition, MSection, Any, str], Any] = None) -> Dict[str, Any]:
+ transform: TypingCallable[[Definition, MSection, Any, str], Any] = None,
+ subsection_as_dict: bool = False,
+ ) -> Dict[str, Any]:
'''
Returns the data of this section as a (json serializable) dictionary.
@@ -1889,6 +1917,8 @@ class MSection(metaclass=MObjectMeta): # TODO find a way to make this a subclas
might have to ensure that the result is JSON-serializable. By
default, values are serialized to JSON according to the quantity
type.
+ subsection_as_dict: If true, try to serialize subsections as dictionaries.
+ Only possible when the keys are unique. Otherwise, serialize as list.
'''
if isinstance(self, Definition) and not with_out_meta:
with_meta = True
@@ -1899,7 +1929,9 @@ class MSection(metaclass=MObjectMeta): # TODO find a way to make this a subclas
include_derived=include_derived,
resolve_references=resolve_references,
exclude=exclude,
- transform=transform)
+ transform=transform,
+ subsection_as_dict=subsection_as_dict
+ )
assert not (include is not None and exclude is not None), 'You can only include or exclude, not both.'
@@ -2197,9 +2229,19 @@ class MSection(metaclass=MObjectMeta): # TODO find a way to make this a subclas
if sub_section_def.repeats:
if self.m_sub_section_count(sub_section_def) > 0:
is_set = True
- yield name, [
- None if item is None else item.m_to_dict(**kwargs)
- for item in self.m_get_sub_sections(sub_section_def)]
+ subsections = self.m_get_sub_sections(sub_section_def)
+ subsection_keys: list = [item.m_key for item in subsections if item and item.m_key]
+ has_dup: bool = 0 < len(subsection_keys) != len(set(subsection_keys))
+ if not has_dup and subsection_as_dict:
+ serialised_dict: dict = {}
+ for index, item in enumerate(subsections):
+ if item is None:
+ continue
+ item_key = item.m_key if item.m_key else index
+ serialised_dict[item_key] = item.m_to_dict(**kwargs)
+ yield name, serialised_dict
+ else:
+ yield name, [None if item is None else item.m_to_dict(**kwargs) for item in subsections]
else:
sub_section = self.m_get_sub_section(sub_section_def, -1)
if sub_section is not None:
@@ -2265,9 +2307,12 @@ class MSection(metaclass=MObjectMeta): # TODO find a way to make this a subclas
sub_section_value = dct.get(name)
sub_section_cls = sub_section_def.sub_section.section_cls
if sub_section_def.repeats:
- for sub_section_dct in sub_section_value:
- sub_section = None if sub_section_dct is None else sub_section_cls.m_from_dict(
- sub_section_dct, m_parent=self, m_context=m_context)
+ for sub_section_dct in sub_section_value if isinstance(
+ sub_section_value, list) else sub_section_value.values():
+ sub_section = None
+ if sub_section_dct is not None:
+ sub_section = sub_section_cls.m_from_dict(
+ sub_section_dct, m_parent=self, m_context=m_context)
section.m_add_sub_section(sub_section_def, sub_section)
else:
sub_section = sub_section_cls.m_from_dict(
@@ -3542,6 +3587,7 @@ class SubSection(Property):
sub_section: Quantity = _placeholder_quantity
repeats: Quantity = _placeholder_quantity
+ key_quantity: Quantity = _placeholder_quantity
def __get__(self, obj, type=None):
# the class attribute case
@@ -4339,6 +4385,7 @@ Section.has_variable_names = has_variable_names
Section.path = section_path
SubSection.repeats = Quantity(type=bool, name='repeats', default=False)
+SubSection.key_quantity = Quantity(type=str, name='key_quantity', default=None)
SubSection.sub_section = Quantity(
type=SectionReference, name='sub_section',
diff --git a/nomad/metainfo/util.py b/nomad/metainfo/util.py
index 9be07a7bbd053c078d11062ec1cfc62f9cbc15f6..3952c512b1abde895d6db854326f1790a99d07cb 100644
--- a/nomad/metainfo/util.py
+++ b/nomad/metainfo/util.py
@@ -343,6 +343,15 @@ class MSubSectionList(list):
def __setitem__(self, key, value):
raise NotImplementedError('You can only append subsections.')
+ def __getitem__(self, item):
+ if isinstance(item, str):
+ for sub_section in self:
+ if sub_section.m_key == item:
+ return sub_section
+ raise KeyError(f'No subsection keyed {item} found.')
+
+ return super().__getitem__(item)
+
def __delitem__(self, key):
old_value = self[key]
list.__delitem__(self, key)
diff --git a/tests/metainfo/test_metainfo.py b/tests/metainfo/test_metainfo.py
index 9a610b26bc461224852899e2830c66e3556eba31..bd0035aac4d7582abe7fb6bfbf24cd87dbcf171b 100644
--- a/tests/metainfo/test_metainfo.py
+++ b/tests/metainfo/test_metainfo.py
@@ -859,3 +859,49 @@ class TestEnvironment:
sub_section_system = env.resolve_definition('systems', SubSection)
assert sub_section_system.m_def == SubSection.m_def
assert sub_section_system.name == 'systems'
+
+
+@pytest.mark.parametrize('as_dict', [True, False])
+@pytest.mark.parametrize('add_key', [True, False])
+@pytest.mark.parametrize('str_type', [True, False])
+@pytest.mark.parametrize('dup_key', [True, False])
+def test_serialise_as_dict(as_dict, add_key, str_type, dup_key):
+ class TestSection(MSection):
+ q = Quantity(type=str if str_type else int)
+
+ class TestContainer(MSection):
+ s = SubSection(sub_section=TestSection, repeats=True, key_quantity='q' if add_key else None)
+
+ container = TestContainer()
+
+ def __key(_i):
+ if str_type:
+ return 'abc' if dup_key else f'abc{_i}'
+
+ return 0 if dup_key else _i
+
+ for i in range(3):
+ section = TestSection(q=__key(i))
+ container.m_add_sub_section(TestContainer.s, section)
+
+ kwarg = {'subsection_as_dict': as_dict}
+ if not str_type and add_key:
+ with pytest.raises(TypeError):
+ for i in range(3):
+ _ = container.s[i].m_key
+ else:
+ for i in range(3):
+ assert container.s[i].q == __key(i)
+ if add_key and not dup_key:
+ assert container.s[f'abc{i}'].q == f'abc{i}'
+ else:
+ with pytest.raises(KeyError):
+ _ = container.s[f'abc{i}'].q
+
+ json_dict = container.m_to_dict(**kwarg)
+
+ if not as_dict or add_key and dup_key:
+ assert isinstance(json_dict['s'], list)
+ else:
+ assert isinstance(json_dict['s'], dict)
+ assert json_dict == TestContainer.m_from_dict(json_dict).m_to_dict(**kwarg)