fix: [ERP-3351] Handle consent section and question codes with underscores

* Build field name codification into get_schema_field_name for consistent use

Author: ppettitau

rdrf/report/report_builder.py

@@ -20,7 +20,11 @@
 )
 from report.clinical_data_csv_util import ClinicalDataCsvUtil
 from report.models import ReportCdeHeadingFormat
-from report.schema import codify, create_dynamic_schema, get_schema_field_name
+from report.schema import (
+    create_dynamic_schema,
+    get_schema_field_name,
+    to_camel_case,
+)
 from report.utils import (
     get_flattened_json_path,
     get_graphql_result_value,
@@ -89,7 +93,10 @@ def _build_query_from_variants(self, variants, fields):
         # e.g. variants=["itemCode1", "itemCode2"]
         if any(isinstance(item, str) for item in variants):
             return [
-                GqlQuery().fields(fields).query(codify(header)).generate()
+                GqlQuery()
+                .fields(fields)
+                .query(get_schema_field_name(header))
+                .generate()
                 for header in variants
             ]
 
@@ -102,14 +109,16 @@ def _build_query_from_variants(self, variants, fields):
                 sorted_variants, lambda x: x[0]
             ):
                 group_items = list(items)
-                group_name_field = get_schema_field_name(group_name)
+                group_name_field = get_schema_field_name(
+                    to_camel_case(group_name)
+                )
                 next_group_items = [item[1:] for item in group_items]
                 if len(next_group_items[0]) == 1:
                     # We've reached the end of the list so build the lowest level query
                     query_fields = [
                         GqlQuery()
                         .fields(fields)
-                        .query(get_schema_field_name(item))
+                        .query(get_schema_field_name(to_camel_case(item)))
                         .generate()
                         for nested_item in next_group_items
                         for item in nested_item
@@ -255,7 +264,7 @@ def _get_form_data(form):
                     field_section = (
                         GqlQuery()
                         .fields(
-                            map(get_schema_field_name, section["cdes"]),
+                            list(map(get_schema_field_name, section["cdes"])),
                             name=get_schema_field_name(section_code),
                         )
                         .generate()
@@ -400,7 +409,7 @@ def add_fields(subvariant_index=None):
                                     )
                                     item_pointer = "_".join(
                                         [
-                                            get_schema_field_name(codify(field))
+                                            get_schema_field_name(field)
                                             for field in items
                                         ]
                                     )

rdrf/report/schema.py

@@ -86,10 +86,6 @@ def to_camel_case(name):
     return lower_dot_case_name
 
 
-def codify(str):
-    return re.sub(r"[^a-zA-Z0-9]+", "", str)
-
-
 def validate_fields(fields, valid_fields, label):
     invalid_fields = list(set(fields).difference(valid_fields))
     if len(invalid_fields) > 0:
@@ -186,7 +182,7 @@ def resolve_values(parent, _info):
             "resolve_values": resolve_values,
         }
         for wg_type in registry.working_group_types.all():
-            field_name = get_schema_field_name(codify(wg_type.name))
+            field_name = get_schema_field_name(wg_type.name)
             wg_type_fields.update(
                 {
                     field_name: graphene.Field(DataSummaryItemSummaryType),
@@ -422,14 +418,20 @@ def resolve_last_updated(parent, _info):
             return datetime.fromisoformat(form_timestamp)
 
 
-def get_schema_field_name(s):
-    if not _graphql_field_pattern.match(s):
-        new_str = f"field{s}"
-        assert _graphql_field_pattern.match(new_str), (
-            f"Cannot use field '{s}' in graphql schema"
-        )
-        return new_str
-    return s
+def get_schema_field_name(raw):
+    def _codify(name):
+        return re.sub(r"[^a-zA-Z0-9]+", "", name)
+
+    def _safe_field_name(name):
+        if not _graphql_field_pattern.match(name):
+            new_str = f"field{name}"
+            assert _graphql_field_pattern.match(
+                new_str
+            ), f"Cannot use field '{name}' in graphql schema"
+            return new_str
+        return name
+
+    return _safe_field_name(_codify(to_camel_case(raw)))
 
 
 def get_section_fields(_section_key, section_cdes):
@@ -496,9 +498,9 @@ def get_cfg_forms(cfg_key, cfg_model):
 
             def form_resolver(parent, _info, form_model):
                 _patient, _contexts, clinical_data = parent
-                assert len(clinical_data) <= 1, (
-                    "Too many clinical data records to resolve form"
-                )
+                assert (
+                    len(clinical_data) <= 1
+                ), "Too many clinical data records to resolve form"
 
                 if len(clinical_data) == 0:
                     return None
@@ -744,7 +746,7 @@ def working_groups_resolver(parent, _info, working_group_type):
 
     fields = {}
     for working_group_type in registry.working_group_types.all():
-        field_name = get_schema_field_name(codify(working_group_type.name))
+        field_name = get_schema_field_name(working_group_type.name)
         fields[field_name] = graphene.List(WorkingGroupSchemaType)
         fields[f"resolve_{field_name}"] = partial(
             working_groups_resolver, working_group_type=working_group_type