def read_dataset()

in scripts/adapet/ADAPET/src/data/COPAReader.py [0:0]

• 46 lines of code
• 7 McCabe index (conditional complexity)

    def read_dataset(self, split=None, is_eval=False):
        '''
        Read the original dataset

        :param split: partition of the dataset
        :param is_eval:
        '''

        file = self._get_file(split)
        data = []

        with open(file, 'r') as f_in:
            for line in f_in.readlines():
                json_string = json.loads(line)

                premise = json_string["premise"]
                choice1 = json_string["choice1"]
                choice2 = json_string["choice2"]
                question = json_string["question"]
                idx = json_string["idx"]

                if "label" in json_string:
                    lbl = json_string["label"]
                else:
                    lbl = -1

                dict_input = {"premise": premise, "choice1": choice1,
                              "idx": idx, "choice2": choice2, "question": question}
                dict_output = {"lbl": lbl}

                dict_input_output = {"input": dict_input, "output": dict_output}
                data.append(dict_input_output)

        if split == 'train' or split == 'unlabeled':
            mirror_data = []
            for dict_input_output in data:
                dict_input, dict_output = dict_input_output["input"], \
                                          dict_input_output["output"]
                mirror_dict_input = {
                                        "premise": dict_input["premise"],
                                        "choice1": dict_input["choice2"],
                                        "choice2": dict_input["choice1"],
                                        "idx": dict_input["idx"],
                                        "question": dict_input["question"]
                }
                mirror_dict_output = {"lbl": 1 if dict_output["lbl"] == 0 else 0}
                mirror_dict_input_output = {
                        "input": mirror_dict_input,
                        "output": mirror_dict_output
                }
                mirror_data.append(mirror_dict_input_output)

            data.extend(mirror_data)

        data = np.asarray(data)

        return data