Added support for conditions in repeatable subsections

Author: JonathanNoky

src/nomad_json_parser/parsers/parser.py

@@ -16,6 +16,7 @@
 # limitations under the License.
 #
 
+import ast
 from typing import (
     TYPE_CHECKING,
 )
@@ -50,6 +51,7 @@
     MainMapper,
     MappedJson,
     MapperRule,
+    RepeatKey,
     RepeatPath,
     RuleCondition,
     SubSectionMapper,
@@ -141,6 +143,13 @@ def create_submapping(subsection, key, logger, archive):  # noqa: PLR0912, PLR09
                 repeat.name = path
                 repeat_paths.append(repeat)
             sectionclass.repeat_paths = repeat_paths
+            if 'repeat_keys' in subsection:
+                keys_list = []
+                for s in subsection['repeat_keys']:
+                    repkey = RepeatKey()
+                    repkey.name = str(s)
+                    keys_list.append(repkey)
+                sectionclass.repeat_keys = keys_list
     if 'repeat_paths' in subsection and 'repeats' not in subsection:
         logger.warning('repeat_paths found but not repeats, ignoring repeat_paths.')
     if 'subsections' in subsection:
@@ -296,7 +305,7 @@ def transform_subclass(  # noqa: PLR0913
     return subclass
 
 
-def checkforvalidkey(path, backjson, submap):
+def checkforvalidkey(path, backjson, submap):  # noqa: PLR0911, PLR0912
     backkey = path['name'].split('*')[1].strip('.')
     if backkey:
         for j in range(len(backkey.split('.'))):
@@ -306,13 +315,30 @@ def checkforvalidkey(path, backjson, submap):
                 return False
     if not isinstance(backjson, dict):
         return False
-    for rule in submap['rules']:
-        rulejson = dict(backjson)
-        for rulekey in rule['source'].split('.'):
-            try:
-                rulejson = rulejson[rulekey]
-            except (KeyError, TypeError):
-                return False
+    if 'repeat_keys' in submap:
+        for key in submap['repeat_keys']:
+            rulejson = dict(backjson)
+            if key['name'].startswith('{'):
+                matching = ast.literal_eval(key['name'])
+                for k in matching.keys():
+                    try:
+                        if not rulejson[k] == matching[k]:
+                            return False
+                    except (KeyError, TypeError):
+                        return False
+            else:
+                try:
+                    rulejson = rulejson[key['name']]
+                except (KeyError, TypeError):
+                    return False
+    else:
+        for rule in submap['rules']:
+            rulejson = dict(backjson)
+            for rulekey in rule['source'].split('.'):
+                try:
+                    rulejson = rulejson[rulekey]
+                except (KeyError, TypeError):
+                    return False
     return True
 
 

src/nomad_json_parser/schema_packages/jsonimport.py

@@ -110,6 +110,13 @@ def normalize(self, archive, logger: BoundLogger) -> None:
         super().normalize(archive, logger)
 
 
+class RepeatKey(ArchiveSection):
+    name = Quantity(type=str)
+
+    def normalize(self, archive, logger: BoundLogger) -> None:
+        super().normalize(archive, logger)
+
+
 class SubSectionMapper(MainMapper):
     main_key = Quantity(
         type=str, description='Key of the main class, where the SubSection is linked.'
@@ -123,6 +130,7 @@ class SubSectionMapper(MainMapper):
         description='Marks a repeatable Subsection, attaches to existing list.',
     )
     repeat_paths = SubSection(section_def=RepeatPath, repeats=True)
+    repeat_keys = SubSection(section_def=RepeatKey, repeats=True)
     subsections = SubSection(section_def='SubSectionMapper', repeats=True)
 
     def normalize(self, archive, logger: BoundLogger) -> None:

tests/data/basesection_example_mapper.json

@@ -18,6 +18,7 @@
         "main_key": "steps",
         "repeats": "True",
         "repeat_paths": ["*.Steps","Step_3"],
+        "repeat_keys": ["step_description", {"time_unit": "seconds", "How_long_did_it_take": 300.0}],
         "rules": {
             "step_name": "name",
             "step_description": "comment",
@@ -32,16 +33,17 @@
         "repeat_paths": ["Chem.*"],
         "rules": {
             "name": "name"
+        },
+        "subsections": {
+            "sample_reference_mapping": {
+                "schema": "nomad.datamodel.metainfo.basesections.v1.CompositeSystem",
+                "main_key": "reference",
+                "is_archive": "True",
+                "rules": {
+                    "labid": "lab_id",
+                    "substance_formula": "name"
+                }
+            }
         }
-    },
-    "sample_reference_mapping": {
-        "schema": "nomad.datamodel.metainfo.basesections.v1.CompositeSystem",
-        "main_key": "samples.reference",
-        "is_archive": "True",
-        "rules": {
-            "labid": "lab_id",
-            "substance_formula": "name"
-        }
-
     }
 }

tests/test_json.py

@@ -45,7 +45,7 @@ def test_normalize_mapper(parsed_mapper_archive, caplog):
     normalize_all(parsed_mapper_archive)
 
     assert parsed_mapper_archive.data.mapper_key == 'basesectionexamplemapper'
-    assert len(parsed_mapper_archive.data.subsection_mappings) == 3  # Noqa: PLR2004
+    assert len(parsed_mapper_archive.data.subsection_mappings) == 2  # Noqa: PLR2004
     assert parsed_mapper_archive.data.main_mapping.name == 'main_schema'