Merge duplicate nodes in macro_to_attributes

- Add duplicate node/key tests to e2e test file - Issue #21 and #22
innolitics · May 7, 2020 · 815793e · 815793e
1 parent ff9a32e
commit 815793e
Show file tree

Hide file tree

Showing 6 changed files with 734 additions and 885 deletions.
diff --git a/dicom_standard/Makefile b/dicom_standard/Makefile
@@ -46,8 +46,8 @@ dist/ciod_to_func_group_macros.json: tmp/raw_ciod_func_group_macro_tables.json
 dist/module_to_attributes.json: tmp/module_to_attributes_no_duplicates.json dist/macros.json dist/ciod_to_func_group_macros.json dist/macro_to_attributes.json
 	$(PYTHONPATH_PREFIX) python3 postprocess_integrate_func_group_macros.py $^ > $@
 
-dist/macro_to_attributes.json: tmp/macros_attributes_partial_references.json dist/references.json
-	$(PYTHONPATH_PREFIX) python3 postprocess_update_reference_links.py $^ > $@
+dist/macro_to_attributes.json: tmp/macros_attributes_updated_references.json
+	$(PYTHONPATH_PREFIX) python3 postprocess_merge_duplicate_nodes.py $< > $@
 
 dist/references.json: tmp/modules_attributes_partial_references.json tmp/raw_section_tables.json
 	$(PYTHONPATH_PREFIX) python3 postprocess_save_references.py $^ > $@
@@ -59,6 +59,9 @@ tmp/module_to_attributes_no_duplicates.json: tmp/modules_attributes_updated_refe
 tmp/modules_attributes_updated_references.json: tmp/modules_attributes_partial_references.json dist/references.json
 	$(PYTHONPATH_PREFIX) python3 postprocess_update_reference_links.py $^ > $@
 
+tmp/macros_attributes_updated_references.json: tmp/macros_attributes_partial_references.json dist/references.json
+	$(PYTHONPATH_PREFIX) python3 postprocess_update_reference_links.py $^ > $@
+
 tmp/modules_attributes_partial_references.json: tmp/modules_attributes_no_references.json
 	$(PYTHONPATH_PREFIX) python3 postprocess_mark_references.py $< > $@
 

diff --git a/dicom_standard/postprocess_merge_duplicate_nodes.py b/dicom_standard/postprocess_merge_duplicate_nodes.py
@@ -18,39 +18,39 @@ def add_conditional_to_description(node):
     node['description'] = formatted_conditional + node['description']
 
 
-def is_duplicate_node(path, module_attr_list):
-    instances = filter(lambda n: n['path'] == path, module_attr_list)
+def is_duplicate_node(path, node_list):
+    instances = filter(lambda n: n['path'] == path, node_list)
     descriptions = map(lambda n: n['description'], instances)
     return len(set(descriptions)) > 1
 
 
-def merge_duplicate_nodes(module_attr_list):
-    path_list = [d['path'] for d in module_attr_list]
+def merge_duplicate_nodes(node_list):
+    path_list = [d['path'] for d in node_list]
     duplicate_paths = [k for k, v in Counter(path_list).items() if v > 1]
     path_to_node = {}
-    for node in module_attr_list:
+    for node in node_list:
         path = node['path']
         if path in path_to_node:
             # Standard workaround: Catch inconsistency in Table C.36.8-1 where "Content Creator's Name" attribute
             # appears twice in same hierarchy without a conditional
             # http://dicom.nema.org/medical/dicom/2019c/output/chtml/part03/sect_C.36.8.html
             if path not in DUPLICATE_PATH_EXCEPTIONS:
                 # Add conditional to description only if the duplicates do not have identical descriptions
-                if is_duplicate_node(path, module_attr_list):
+                if is_duplicate_node(path, node_list):
                     add_conditional_to_description(node)
                     path_to_node[path]['description'] += node['description']
                     path_to_node[path]['externalReferences'].extend(node['externalReferences'])
         else:
             if path in duplicate_paths and path not in DUPLICATE_PATH_EXCEPTIONS:
                 # Add conditional to description only if the duplicates do not have identical descriptions
-                if is_duplicate_node(path, module_attr_list):
+                if is_duplicate_node(path, node_list):
                     add_conditional_to_description(node)
             path_to_node[path] = node
         path_to_node[path].pop('conditional', None)
     return list(path_to_node.values())
 
 
 if __name__ == "__main__":
-    module_attr_list = pl.read_json_data(sys.argv[1])
-    processed_module_attr_list = merge_duplicate_nodes(module_attr_list)
-    pl.write_pretty_json(processed_module_attr_list)
+    node_list = pl.read_json_data(sys.argv[1])
+    processed_node_list = merge_duplicate_nodes(node_list)
+    pl.write_pretty_json(processed_node_list)
diff --git a/dicom_standard/process_macro_attribute_relationship.py b/dicom_standard/process_macro_attribute_relationship.py
@@ -14,7 +14,8 @@ def macro_attr_relationship_table(macro_attr_list):
                 'tag': attribute['tag'],
                 'type': attribute['type'],
                 'linkToStandard': get_standard_link(macro, attribute),
-                'description': attribute['description']
+                'description': attribute['description'],
+                'conditional': attribute.get('conditional'),
             })
     return entries