Exemple #1
0
 def setUp(self):
     """
     :return:
     """
     self.wikipedia_abstract_xml = APP_ROOT + "/Data/jawiki-20160901-abstract_part.xml"
     self.wiki_pedia_xml_to_json = WikiPediaXmlToJson(
         self.wikipedia_abstract_xml)
     self.answer_data = APP_ROOT + "/Data/jawiki-20160901-abstract_part0.json"
     self.correct_data = APP_ROOT + "/Data/answer.json"
Exemple #2
0
 def setUp(self):
     """
     :return:
     """
     self.wikipedia_abstract_xml = APP_ROOT + "/../../Data/wiki_image/enwiki-20080103-abstract_part.xml"
     self.wiki_pedia_xml_to_json = WikiPediaXmlToJson(
         self.wikipedia_abstract_xml)
     self.answer_data = APP_ROOT + "/../../Data/wiki_image/enwiki-20080103-abstract_part0.json"
     self.correct_data = APP_ROOT + "/../../Data/wiki_image/answer.json"
                        default='jawiki-20160901-abstract_part.xml',
                        help='set xml file')
    parser.add_argument('--img_flag',
                        '-img',
                        default="False",
                        help='set image Flag')
    parser.add_argument('--set_worker',
                        '-work',
                        default=2,
                        help='set image Flag')
    args = parser.parse_args()
    Image_Flag = ast.literal_eval(args.img_flag)
    if Image_Flag is True:
        wikipedia_abstract_xml = APP_ROOT + "/../Data/wiki_image/" + args.xml_file
    else:
        wikipedia_abstract_xml = APP_ROOT + "/../Data/jawiki-20160901-abstract_dir/" + args.xml_file
    wiki_pedia_xml_to_json = WikiPediaXmlToJson(wikipedia_abstract_xml)
    wiki_pedia_xml_to_json.input(image_Flag=Image_Flag)

    # Multi Process
    with ProcessPoolExecutor() as executor:
        # executor.map(wiki_pedia_xml_to_json.extract_contents, wiki_pedia_xml_to_json.xml_data)
        all_process = []
        for xml_name in wiki_pedia_xml_to_json.xml_data:
            process = executor.submit(wiki_pedia_xml_to_json.extract_contents,
                                      xml_name)
            all_process.append(process)

        for process in as_completed(all_process):
            print(process.result())