def record(): """Return results generator from the Alpha spider.""" spider = phenix_spider.PhenixSpider() response = fake_response_from_file('phenix/test_1.html') selector = Selector(response, type='html') nodes = selector.xpath('//%s' % spider.itertag) return spider.parse_node(response, nodes[0])
def non_thesis(): """Return a heprecord for a Master's thesis (should be None as we don't want them).""" spider = phenix_spider.PhenixSpider() body = """ <ul> <li><b>M.Sc. Author</b>: "This is an Master's thesis, not a PhD", M.Sc. thesis at Master Science University, 2016, <br><br> </ul> """ response = fake_response_from_string(body) node = get_node(spider, '//li', text=body) return spider.parse_node(response, node)