def testHTMLSuffix(self):
# https://example.com/hello.html
# domain = example.com
# url_path = hello.html
# save_path = ${DOWNLOAD_DIR}/example.com/hello.html
url = "https://example.com/hello.html"
(domain, url_path) = SavePagePipeline._parse_url(url)
self.assertEqual(domain, "example.com")
self.assertEqual(url_path, "hello.html")
def testRoot(self):
# https://example.com/
# domain = example.com
# url_path = index.html
# save_path = ${DOWNLOAD_DIR}/example.com/index.html
url = "https://example.com/"
(domain, url_path) = SavePagePipeline._parse_url(url)
self.assertEqual(domain, "example.com")
self.assertEqual(url_path, "index.html")
def testNested(self):
# https://example.com/foo/bar.php
# domain = example.com
# url_path = foo/bar.php.html
# save_path = ${DOWNLOAD_DIR}/example.com/foo/bar.php.html
url = "https://example.com/foo/bar.php"
(domain, url_path) = SavePagePipeline._parse_url(url)
self.assertEqual(domain, "example.com")
self.assertEqual(url_path, "foo/bar.php.html")