def test_cannot_resolve_with_wrong_params_dict(self):
# Ommitted params
p = base.resolve("http://www.mediagaon.or.kr/jsp/sch/mnews/newsView.jsp")
self.assertNotIsInstance(p, KindsArticlePage)
with self.assertRaises(base.NotResolvedError):
# Params not declared in pagetype
base.resolve(url="http://www.mediagaon.or.kr/jsp/sch/mnews/newsView.jsp", params={"foo": "bar"})
# Wrong params does not raise NotResolvedError
base.resolve(url="http://www.mediagaon.or.kr/jsp/sch/mnews/newsView.jsp", params={"newsId": "bar"})
def test_raise_exception_on_implicit_fetching(self):
page = base.resolve("http://example.com")
with self.assertRaises(base.NotFetchedYetError):
page.get_links()
with self.assertRaises(base.NotFetchedYetError):
page.get_links(fetch=False)
def test_select_and_parse_properties(self):
p = base.resolve("https://googleblog.blogspot.kr/2015/11/google-gobble-thanksgiving-trends-on.html")
p.fetch()
title = p.get_properties()["title"]
body = p.get_properties()["body"]
self.assertEqual(title, "Google gobble: Thanksgiving trends on Search")
self.assertTrue(body.startswith("In just a few hours"))
def test_parse_example_webpages(self):
page = base.resolve("http://example.com")
self.assertIsInstance(page, base.Hypertext)
page.fetch()
links = page["links"]
self.assertEqual(len(links), 1)
self.assertEqual("http://www.iana.org/domains/example", links[0])
def test_resolve_with_params_and_parse_properties(self):
p = base.resolve(
"http://www.mediagaon.or.kr/jsp/sch/mnews/newsView.jsp", params={"newsId": "01100101.20151102100000159"}
)
self.assertIsInstance(p, KindsArticlePage)
p.fetch()
title = p.get_properties()["title"]
self.assertEqual(title, "[한·중·일 정상회의] 3국 정상, 회의 전엔 ‘미소 촬영’ 회견 땐 웃음기 ‘싹’…비빔밥으로 만찬")
def test_resolve_with_multiple_params(self):
p = base.resolve(
url="http://www.mediagaon.or.kr/jsp/sch/mnews/search.jsp",
params={"startDate": "2015.01.01", "endDate": "2015.10.05"},
)
self.assertIsInstance(p, KindsSearchPage)
p.fetch()
print(p["links"])
def test_parse_cern(self):
page = base.resolve("http://info.cern.ch")
self.assertIsInstance(page, base.Hypertext)
page.fetch()
links = page["links"]
self.assertEqual(len(links), 4)
self.assertEqual(
[ item for item in links ],
[
"http://info.cern.ch/hypertext/WWW/TheProject.html",
"http://line-mode.cern.ch/www/hypertext/WWW/TheProject.html",
"http://home.web.cern.ch/topics/birth-web",
"http://home.web.cern.ch/about"
]
)
def test_resolve_url_with_pattern_vars(self):
p = base.resolve("https://googleblog.blogspot.kr/2015/11/google-gobble-thanksgiving-trends-on.html")
self.assertIsInstance(p, GoogleBlogPage)
def test_resolve_pagetype_with_url(self):
p = base.resolve("http://info.cern.ch/hypertext/WWW/TheProject.html")
self.assertIsInstance(p, W3Page)
def test_base_hypertext_does_not_parse_any_content(self):
page = base.resolve("http://info.cern.ch")
properties = page.get_properties(fetch=True)
self.assertEqual(properties.keys(), {"links"})
def test_cannot_parse_example_dot_com_with_post_method(self):
with self.assertRaises(base.NotResolvedError):
page = base.resolve("http://example.com", method="POST")
