def test_mongin(caplog):
pagename = 'mongin.html'
url = 'https://studies2.hec.fr/jahia/Jahia/cache/offonce/lang/en/mongin/pid/1072'
page = Webpage(url, html=source(pagename))
svars = page.session_variables()
assert 'jsessionid' in svars
testurl = 'https://studies2.hec.fr/jahia/webdav/site/hec/shared/sites/mongin/foo.pdf;jsessionid=123456'
stripped = page.strip_session_variables(testurl)
assert stripped == 'https://studies2.hec.fr/jahia/webdav/site/hec/shared/sites/mongin/foo.pdf;'
def test_broken_html():
pagename = 'healey.html'
url = 'https://blah.org'
page = Webpage(url, html=source(pagename))
targets = set(u for u in page.xpath('//a/@href') if re.search('.pdf$', u, re.I))
assert targets
def test_utf8(caplog):
pagename = 'philpapers-rec.html'
url = 'https://blah.org'
page = Webpage(url, html=source(pagename))
assert 'Analytic' in page.text()