def __init__(self,
channel,
save='fulltext.seed.txt',
db='zhuanli',
dburl='mongodb://localhost/zhuanli'):
CWPParser.__init__(self, channel, channel, db, dburl)
self.seed_saver = LinkSaver(save)
def __init__(self):
CWPParser.__init__(self, 'abs_list', 'abs_list', 'zhuanli')
self.store = PatentStore('abstract')
self.failed_link = LinkSaver('abstract.parser.failed.txt')
self.url_format = 'http://epub.sipo.gov.cn/dxbdl.action?strSources=fmmost&strWhere=%s&recordCursor=0&strLicenseCode=&action=dxbdln'
self.save_count = 0
def __init__(self):
CWPParser.__init__(self, 'fs_court2', 'fs')
self.store = FSCourtStore('fs_court')
def init(self):
print 'job start at', datetime.datetime.now()
return CWPParser.init(self)
def __init__(self):
CWPParser.__init__(self, 'shanghai_court', 'court')
self.an_saver = LinkSaver('ah.%s.txt' % self.name)
def __init__(self):
CWPParser.__init__(self, 'shanghai_court', 'shanghai_court')
self.pagestore = ShanghaiCourtStore('sh_court_2')
self.link_saver = LinkSaver('wrong.id.txt')
def __init__(self, channel, dist_file, name, parser):
CWPParser.__init__(self, channel, name)
self.bin_writer = BinSaver(dist_file)
self.parser = parser
def __init__(self):
CWPParser.__init__(self, 'fs_court', 'fs')
self.saver = LinkSaver('seed.txt')