def test_get_local_url(self):
self.assertEqual(
uu.get_local_url('/home/bart/doc', 'http://doc.com/foo/bar.txt'),
'/home/bart/doc/foo/bar.txt')
self.assertEqual(
uu.get_local_url('/home/bart/doc/', 'http://doc.com/foo/bar/'),
'/home/bart/doc/foo/bar/bar__root.html')
self.assertEqual(uu.get_local_url('/home/bart/doc', 'http://doc.com/'),
'/home/bart/doc/root__root.html')
def page_error(self, input_url, pages):
local_url = get_local_url(self.output_url,
get_url_without_hash(input_url))
self.logger.info(
'This page could not be downloaded: {0} in {1}'.format(
input_url, local_url))
error_page = DocumentPage(input, None, [])
pages[local_url] = error_page
def page_error(self, input_url, pages):
local_url = get_local_url(self.output_url,
get_url_without_hash(input_url))
self.logger.info(
'This page could not be downloaded: {0} in {1}'.format(
input_url, local_url))
error_page = DocumentPage(input, None, [])
pages[local_url] = error_page
def make_copy(self, url_to_copy, binary=False):
destination_url = get_local_url(self.output_url, url_to_copy)
try:
ensure_path_exists(destination_url)
except:
raise Exception('Could not make copy of {0} in {1}'.format(
url_to_copy, destination_url))
download_file(url_to_copy, get_path_from_url(destination_url),
force=False, binary=binary)
return destination_url
def make_copy(self, url_to_copy, binary=False):
destination_url = get_local_url(self.output_url, url_to_copy)
try:
ensure_path_exists(destination_url)
except:
raise Exception('Could not make copy of {0} in {1}'.format(
url_to_copy, destination_url))
download_file(url_to_copy,
get_path_from_url(destination_url),
force=False,
binary=binary)
return destination_url
def process_page_links(self, tree, local_url, url):
link_tags = self.links(tree)
links = []
for link_tag in link_tags:
attributes = link_tag.attrib
href = ''
if 'href' in attributes:
href = attributes['href']
link_url = get_url_without_hash(urlparse.urljoin(url, href))
local_url_to = get_local_url(self.output_url, link_url)
local_url_to = get_sanitized_url(local_url_to)
link = DocumentLink(link_url, local_url_to)
links.append(link)
else:
continue
return links
def process_page_links(self, tree, local_url, url):
link_tags = self.links(tree)
links = []
for link_tag in link_tags:
attributes = link_tag.attrib
href = ''
if 'href' in attributes:
href = attributes['href']
link_url = get_url_without_hash(urlparse.urljoin(url, href))
local_url_to = get_local_url(self.output_url, link_url)
local_url_to = get_sanitized_url(local_url_to)
link = DocumentLink(link_url, local_url_to)
links.append(link)
else:
continue
return links
def test_get_local_url(self):
self.assertEqual(uu.get_local_url("/home/bart/doc", "http://doc.com/foo/bar.txt"), "/home/bart/doc/foo/bar.txt")
self.assertEqual(
uu.get_local_url("/home/bart/doc/", "http://doc.com/foo/bar/"), "/home/bart/doc/foo/bar/bar__root.html"
)
self.assertEqual(uu.get_local_url("/home/bart/doc", "http://doc.com/"), "/home/bart/doc/root__root.html")
