corpusbuilder.py 文件源码

def build_initialize(self):
        """
        Initialize the corpus build.
        """

        self.start_time = time.time()
        self.logger.info("--- Starting ---")
        self.logger.info("Building corpus %s" % self.name)
        self.logger.info("Command line arguments: %s" % " ".join(sys.argv[1:]))
        if not self._widget:
            print("\n%s\n" % textwrap.TextWrapper(width=79).fill(" ".join(self.get_description())))

        # Corpus installers may require additional modules. For example,
        # Gabra is currently distributed as MongoDB files, which are read by
        # using the pymongo library.
        # Unless the user wishes to install only the corpus module, try to
        # import these additional modules, and raise an exception if they are
        # unavailable:
        if not self.arguments.only_module:
            for module, package, url in self.get_modules():
                try:
                    exec("import {}".format(module))
                except ImportError:
                    raise DependencyError(package, url)

        if self.DB.db_type == SQL_MYSQL:
            self.DB.connection.execute("SET NAMES 'utf8'")
            self.DB.connection.execute("SET CHARACTER SET 'utf8mb4'")
            self.DB.connection.execute("SET unique_checks=0")
            self.DB.connection.execute("SET foreign_key_checks=0")