The algorithm to flatten MongoQueries so that they can be represented by a single query key

compound_index_creater.py

# Converts a Query JSON into a flat key.
#{
#  "where": {
#    "activityDistance": {
#      "$gte": 2.5
#    },
#    "accumulatedTotalDistance": {
#      "$lte": 10695
#    },
#    "
#  "order": "-createdAt",
#}
# to a flat key like :
# "{\"$OR\": [{\"_acl\": {\"$exists\": \"\"}}, {\"_acl.*.r\": \"\"}]}"

ARRAY_QUERY_SET = Set(["$in", "$nin", "$box", "$nearSphere", "$all", "$within"])

class SimpleQueryAST:
  """ represent a query ast in a simplified manne """

  def __init__(self, json):

    self._simple_where_keys = [] # simple where clauses
    self._where_in_keys = [] # keys with where in clause
    self._other_keys = [] # ignored for now.
    self._or_keys = [] # or keys
    self.collection_name = ""
    self.app_key = ""
    self.__flatten_json(json)
    self._where_in_keys.sort()
    self._simple_where_keys.sort()
    self._other_keys.sort()
    self._or_keys.sort()


  def __remove_values(self,json_obj):
    result = {}
    for k,v in json_obj.items():
      if type(v) == dict:
        result[k] = self.__remove_values(v)
      elif k == "$or":
        assert type(v) == list
        temp = []
        for item in v:
          temp.append(self.__remove_values(item))
        result["$OR"] = temp
      elif k in ARRAY_QUERY_SET:
        return "$ARRAY_QUERY"
      elif type(v) == str or type(v) == int or type(v) == float or isinstance(v, datetime.datetime) or type(v) == bool or type(v) == unicode or v == None:
        result[k] = ""
      else:
        sys.stderr.write("COULD NOT UNDERSTAND THE OBJECT %s\n".format(str(json_obj)))
        sys.exit(1)
    return result

  def __flatten_json(self, json_obj):
    self.collection_name = json_obj[COLLECTION_NAME_KEY]
    self.answer = self.__remove_values(json_obj[WHERE_KEY])
    if ORDER_KEY in json_obj:
      self.answer['ORDER'] = sorted(json_obj[ORDER_KEY])
    self.answer_str = json.dumps(self.answer, sort_keys = True)
    
    
if __name__ == "__main__":
  query = SimpleQueryAST({"where":{"_id":{"$in":["1","2"]},"_rperm":{"$in":["*",null,"xyzzy","spoon"]}}) # just a sample query