Python Client.listSets方法代码示例

# 需要导入模块: from oaipmh.client import Client [as 别名]
# 或者: from oaipmh.client.Client import listSets [as 别名]
    def scrape(self):
        raise Exception("not finished")
        registry = MetadataRegistry()
        registry.registerReader('oai_dc', oai_dc_reader)
        url = self.setting('pmh-endpoint')
        client = Client(url, registry)

        print "  OAI Repository", url
        print "  Available sets:"
        for s in client.listSets():
            print "   ", s

        oai_set = self.setting('set')
        oai_from = self.setting('from')
        oai_until = self.setting('until')

        kwargs = {}

        if oai_set:
            kwargs['set'] = oai_set

        if oai_from is not None:
            date_args = [int(arg) for arg in oai_from.split("-")]
            kwargs['from_'] = datetime.datetime(*date_args)

        if oai_until is not None:
            date_args = [int(arg) for arg in oai_until.split("-")]
            kwargs['until'] = datetime.datetime(*date_args)

        records = [r for r in client.listRecords(metadataPrefix='oai_dc', **kwargs)]

        data_filepath = os.path.join(self.work_dir(), self.setting('data-file'))
        with open(data_filepath, 'wb') as f:
            print "  picking", len(records), "records"
            pickle.dump(records, f)

# 需要导入模块: from oaipmh.client import Client [as 别名]
# 或者: from oaipmh.client.Client import listSets [as 别名]
def list_sets(target):
    if target is not None:
        client = Client(target['url'], registry)
        setspecs = client.listSets()
        results = []
        if setspecs is not None:
            for setspec in setspecs:
                results.append(convert_setspec(setspec))
        return results

# 需要导入模块: from oaipmh.client import Client [as 别名]
# 或者: from oaipmh.client.Client import listSets [as 别名]
    def list_oai_community_sets(self, repository):

        try:
            registry = MetadataRegistry()
            registry.registerReader('oai_dc', oai_dc_reader)
            client = Client(repository.base_url, registry)
            sets = client.listSets()
        except:
            return

        """ Filter records to build list of community sets """
        self.communities = []
        for i in sets:
            set_id = i[0]
            set_name = i[1]
            """ Build collection tuples (id, human readable name) """
            if set_id[:3] == 'com':
                set_data = []
                set_data.append(set_id)
                set_data.append(set_name)
                self.communities.append(set_data)

        self.communities = sorted(
            self.communities, key=lambda community: community[1])

# 需要导入模块: from oaipmh.client import Client [as 别名]
# 或者: from oaipmh.client.Client import listSets [as 别名]
#### OAI-PMH Client processing 

oai = Client('http://snape.mzk.cz/OAI-script', registry)

id = oai.identify()
print id.repositoryName()
print id.adminEmails()
print id.baseURL()

formats = oai.listMetadataFormats()
pprint formats

# 'marc21'

sets = oai.listSets()
for s in sets:
	print s

# 'MZK03'

recids = oai.listIdentifiers(metadataPrefix='marc21', set='MZK03') # from_='2003-01-01T00:00:00Z', until=''

# for example: 'MZK03-907223' is in the list of maps
# or 356050 *not a map

# 238208 problematic
r = oai.getRecord(identifier='MZK03-1479', metadataPrefix='marc21')

# from lxml import etree
# print etree.tostring(r[1],pretty_print=True)