def test_addNewWorks(self):
"""Test the method of adding new works.
1. Adding new works from a html document.
2. Adding new work from a html document after getting a work for the working list.
"""
c = Crawler()
# Adding new works from a html document
html1 = ('<!DOCTYPE html>'
'<html>'
'<head>'
' <title>aaa</title>'
'</head>'
'<body>'
' <div>'
' <a href="http://www.testurl.com/link1"></a>'
' <a href="/link2"></a>'
' </div>'
'</body>'
'</html>')
expected = [
'http://www.testurl.com/link1', 'http://www.testurl.com/link2'
]
bs = BeautifulSoup(html1, 'html.parser')
url = 'http://www.testurl.com'
c.addNewWorks(url, bs)
self.assertTrue(c.WLM.workExists())
self.assertEqual(c.WLM.records, deque(expected))
# Adding new work from a html document after getting a work for the working list
c.WLM.getWork()
html2 = ('<!DOCTYPE html>'
'<html>'
'<head>'
' <title>aaa</title>'
'</head>'
'<body>'
' <div>'
' <a href="/link3"></a>'
' </div>'
'</body>'
'</html>')
expected = [
'http://www.testurl.com/link2', 'http://www.testurl.com/link3'
]
bs = BeautifulSoup(html2, 'html.parser')
url = 'http://www.testurl.com'
c.addNewWorks(url, bs)
self.assertTrue(c.WLM.workExists())
self.assertEqual(c.WLM.records, deque(expected))
