def get_theme_suggestions(search_core: SolrCollection, in_context: str) -> list:
"""
Get theme suggestions within a given context and use the number of
occurrences of a theme within the context as weight
:param search_core: The search core to get theme suggestions from
:param in_context: The context
:return: The list of theme suggestions
"""
context_entities = search_core.select_all_documents(
'sys_type:"{0}"'.format(in_context), ['theme'], id_field='sys_id'
)
counts = {}
for context_entity in context_entities:
if 'theme' not in context_entity:
continue
for theme in context_entity['theme']:
if theme not in counts:
counts[theme] = 0
counts[theme] += 1
synonyms_uri_nl = search_core.select_managed_synonyms('uri_nl')
return [{
'theme': synonyms_uri_nl[theme] if theme in synonyms_uri_nl else theme,
'weight': count,
'payload': theme,
'type': 'theme',
'language': ['nl', 'en'],
'in_context_of': in_context
} for theme, count in counts.items()]
def manage_uri_synonyms(collection: SolrCollection) -> None:
logging.info('> managing uri_synonyms resources')
uri_synonyms = {'uri_nl': {}, 'uri_en': {}}
legacy_license_files = ['ckan_license.json', 'overheid_license.json']
for valuelist in os.listdir(os.getenv('VALUELIST_DIR')):
if not valuelist.endswith('.json'):
continue
if valuelist in legacy_license_files:
continue
filepath = os.path.join(os.getenv('VALUELIST_DIR'), valuelist)
for uri, properties in utils.load_json_file(filepath).items():
uri_synonyms['uri_nl'][uri] = properties['labels']['nl-NL']
uri_synonyms['uri_en'][uri] = properties['labels']['en-US']
for lang in uri_synonyms.keys():
logging.info('managed resource: synonyms|%s', lang)
current_uri_synonyms = collection.select_managed_synonyms(lang)
if not current_uri_synonyms:
current_uri_synonyms = {}
logging.info(' current: %s synonyms', len(current_uri_synonyms))
synonyms_to_add = {
key: value
for key, value in uri_synonyms[lang].items()
if key not in current_uri_synonyms.keys()
}
logging.info(' adding: %s synonyms', len(synonyms_to_add))
if len(synonyms_to_add) > 0:
collection.add_managed_synonyms(lang, synonyms_to_add)
def manage_label_synonyms(collection: SolrCollection,
language_code: str) -> None:
logging.info('> managing label resource')
current_label_synonyms = collection.select_managed_synonyms(
'label_{0}'.format(language_code))
logging.info(' current: %s synonyms', len(current_label_synonyms))
label_synonyms = utils.load_resource('labels_{0}'.format(language_code))
synonyms_to_add = {
key: value
for key, value in label_synonyms.items()
if key not in current_label_synonyms.keys()
}
logging.info(' adding: %s synonyms', len(synonyms_to_add))
if len(synonyms_to_add) > 0:
collection.add_managed_synonyms('label_{0}'.format(language_code),
synonyms_to_add)
def manage_hierarchy_theme(collection: SolrCollection) -> None:
logging.info('> managing hierarchy_theme resources')
for hierarchy_item in ['hierarchy_theme', 'hierarchy_theme_query']:
logging.info('managed resource: synonyms|{0}'.format(hierarchy_item))
hierarchy_data = utils.load_resource(hierarchy_item)
current_data = collection.select_managed_synonyms(hierarchy_item)
if not current_data:
current_data = {}
logging.info(' current: %s synonyms', len(current_data))
data_to_add = {
key: value
for key, value in hierarchy_data.items()
if key not in current_data.keys()
}
logging.info(' adding: %s synonyms', len(data_to_add))
if len(data_to_add) > 0:
collection.add_managed_synonyms(hierarchy_item, data_to_add)