def parse_schema(big_query_schema, jx_path, nested_path, es_path):
output = OrderedDict()
if any(ApiName(e.name) == REPEATED for e in big_query_schema):
big_query_schema = [e for e in big_query_schema if ApiName(e.name) == REPEATED]
for e in big_query_schema:
json_type = bq_type_to_json_type[e.field_type]
name = unescape_name(ApiName(e.name))
full_name = jx_path + (name,)
full_es_path = es_path + (e.name,)
if e.field_type == "RECORD":
if e.mode == "REPEATED":
output[name] = parse_schema(
e.fields, full_name, (jx_path,) + nested_path, full_es_path
)
else:
output[name] = parse_schema(
e.fields, full_name, nested_path, full_es_path
)
else:
if e.mode == "REPEATED":
output[name] = {NESTED_TYPE: json_type}
else:
output[name] = json_type
return output
def _untype_dict(value):
output = None
for k, v in value.items():
k = unescape_name(ApiName(k))
if k == NESTED_TYPE:
return _untype_list(v)
elif k in typed_to_bq_type:
vv = _untype_value(v)
if vv is not None:
# WE ASSUME A PRIMITIVE VALUE, NOT SOME GREATER STRUCTURE
return vv
else:
new_v = _untype_value(v)
if new_v is not None:
if output is None:
output = {k: new_v}
else:
output[k] = new_v
return output