def __init__(self, package_name=None, path=None,
                 extract_nations=False):
        self.package_name = package_name
        self.path = path
        self._dois = []
        self.articles_normalized = []
        self.logger = create_logger(
            "Springer",
            filename=join(CFG_LOGDIR, 'scoap3_harvesting.log')
        )

        self.config = load_config(CFG_CONFIG_PATH, {'SPRINGER': []})

        self.packages_delivery = []
        self.doi_package_name_mapping = []

        if not path and package_name:
            self.logger.info("Got package: %s" % (package_name,))
            self.path = self._extract_packages()
        elif not path and not package_name:
            print "Starting harvest"
            self.run()
        self._crawl_springer_and_find_main_xml()

        self.extract_nations = extract_nations
Example #2
0
    def __init__(self, package_name=None, path=None, extract_nations=False):
        if package_name:
            if not package_name.endswith(".zip"):
                raise Exception('package_name variable requires a ZIP file.')

        self.package_name = package_name
        self.path = path
        self._dois = []
        self.articles_normalized = []
        self.logger = create_logger("Oxford",
                                    filename=join(CFG_LOGDIR,
                                                  'scoap3_harvesting.log'))

        self.config = load_config(CFG_CONFIG_PATH, {'OXFORD': []})

        self.packages_delivery = []
        self.doi_package_name_mapping = []

        if not path and package_name:
            self.logger.info("Got package: %s" % (package_name, ))
            self.path = self._extract_packages()
        elif not path and not package_name:
            print("Starting harvest")
            self.run()
        self._crawl_oxford_and_find_main_xml()
        self.extract_nations = extract_nations
Example #3
0
    def __init__(self, package_name=None, path=None,
                 extract_nations=False):
        if package_name:
            if not package_name.endswith(".zip"):
                raise Exception('package_name variable requires a ZIP file.')

        self.package_name = package_name
        self.path = path
        self._dois = []
        self.articles_normalized = []
        self.logger = create_logger(
            "Oxford",
            filename=join(CFG_LOGDIR, 'scoap3_harvesting.log')
        )

        self.config = load_config(CFG_CONFIG_PATH, {'OXFORD': []})

        self.packages_delivery = []
        self.doi_package_name_mapping = []

        if not path and package_name:
            self.logger.info("Got package: %s" % (package_name,))
            self.path = self._extract_packages()
        elif not path and not package_name:
            print("Starting harvest")
            self.run()
        self._crawl_oxford_and_find_main_xml()
        self.extract_nations = extract_nations
Example #4
0
    def __init__(self, logger):
        self.ftp = None
        self.files_list = []
        self.retrieved_packages = {}
        self.retrieved_packages_unpacked = []
        self.path = None
        self.retrieved_packages_unpacked = []
        self.found_articles = []
        self.found_issues = []
        self.path_r_pkg = []
        self.logger = logger
        self.packages_delivery = []

        self.config = load_config(CFG_CONFIG_PATH, {'ELSEVIER': []})
Example #5
0
    def __init__(self, logger):
        self.ftp = None
        self.files_list = []
        self.retrieved_packages = {}
        self.retrieved_packages_unpacked = []
        self.path = None
        self.retrieved_packages_unpacked = []
        self.found_articles = []
        self.found_issues = []
        self.path_r_pkg = []
        self.logger = logger
        self.packages_delivery = []

        self.config = load_config(CFG_CONFIG_PATH, {'ELSEVIER': []})
Example #6
0
    def __init__(self, package_name=None, path=None, extract_nations=False):
        self.package_name = package_name
        self.path = path
        self._dois = []
        self.articles_normalized = []
        self.logger = create_logger("Springer",
                                    filename=join(CFG_LOGDIR,
                                                  'scoap3_harvesting.log'))

        self.config = load_config(CFG_CONFIG_PATH, {'SPRINGER': []})

        self.packages_delivery = []
        self.doi_package_name_mapping = []

        if not path and package_name:
            self.logger.info("Got package: %s" % (package_name, ))
            self.path = self._extract_packages()
        elif not path and not package_name:
            print "Starting harvest"
            self.run()
        self._crawl_springer_and_find_main_xml()

        self.extract_nations = extract_nations