DocumentFormatter for flattening documents for use in Solr. Does not include array indices.

custom_flattening.py

from mongo_connector.doc_managers.formatters import DocumentFlattener


class DocumentFlattenerUnordered(DocumentFlattener):
    """Formatter that completely flattens documents, but does not unwind arrays.
    Order of elements within arrays is not recorded.

    An example:
      {"a": 2,
       "b": {
         "c": {
           "d": 5
         }
       },
       "e": [6, 7, 8]
      }

    becomes:
      {"a": 2, "b.c.d": 5, "e": [6, 7, 8]}

    """

    def transform_element(self, key, value):
        if isinstance(value, list):
            result = {}
            for lv in value:
                for inner_k, inner_v in self.transform_element(key, lv):
                    inner_l = result.get(inner_k, [])
                    if isinstance(inner_v, list):
                        inner_l.extend(inner_v)
                    else:
                        inner_l.append(inner_v)
                    result[inner_k] = inner_l
            for k in result:
                yield k, result[k]
        else:
            parent = super(DocumentFlattenerUnordered, self)
            for k, v in parent.transform_element(key, value):
                yield k, v