def test_rebase_url(self):
"""rebase_url"""
from StillWeb.sw_urllib import rebase_url
# Simple rebasing
self.assertEqual("http://a/new/b/c", rebase_url(url="http://a/b/c", src="http://a/", dest="http://a/new/"))
self.assertEqual("http://a/new/b/c", rebase_url("http://a/old/b/c", "http://a/old/", "http://a/new/"))
# When the URL can't be rebased
self.assertEqual("http://xyz/foo", rebase_url("http://xyz/foo", "http://a/old/", "http://a/new/"))
self.assertEqual("http://a/foo", rebase_url("/foo", "http://a/bar/", "http://xyz/new/"))
self.assertRaises(ValueError, rebase_url, "/foo", "http://a/bar/", "http://xyz/new/", must_rebase=True)
def cb(url, criterion):
# Resolve the link URL with respect to the current (fake) URL
link_url = rfc3986_urljoin(fake_current_url, url)
# Convert the fake URL into an absolute real URL
link_url = rebase_url(link_url, fake_base_url, real_base_url)
if not always_absolute:
# Convert the absolute URL into a relative URL
link_url = relative_url(link_url, real_current_url)
return link_url
def __init__(self, framework, orig_target_url):
if not orig_target_url.startswith("/"):
raise AssertionError("orig_target_url must start with /")
self.orig_target_url = orig_target_url
self.target_url = strip_index_from_url(orig_target_url)
self.pathtuple = pathtuple_from_target_url(orig_target_url)
if "output_dir" in framework.plugins['vars'].vars:
self.output_dir = framework.plugins['vars'].vars['output_dir']
self.output_filename = os.path.join(self.output_dir, *self.pathtuple)
if "source_dir" in framework.plugins['vars'].vars:
self.source_dir = framework.plugins['vars'].vars['source_dir']
self.source_filename = os.path.join(self.source_dir, *self.pathtuple)
if "base_url" in framework.plugins['vars'].vars:
self.base_url = framework.plugins['vars'].vars['base_url']
self.current_url = rebase_url(self.target_url, generate_fake_url(), self.base_url)
def rewrite_links(node, match_criteria, target_url, base_url, always_absolute=False):
# We generate a fake URL so links like <a href="/">...</a> will resolve
# to the top-level URL of the *site* rather than of the *server*.
fake_base_url = generate_fake_url()
fake_current_url = rfc3986_urljoin(fake_base_url, target_url)
real_base_url = base_url
real_current_url = rebase_url(fake_current_url, fake_base_url, real_base_url)
def cb(url, criterion):
# Resolve the link URL with respect to the current (fake) URL
link_url = rfc3986_urljoin(fake_current_url, url)
# Convert the fake URL into an absolute real URL
link_url = rebase_url(link_url, fake_base_url, real_base_url)
if not always_absolute:
# Convert the absolute URL into a relative URL
link_url = relative_url(link_url, real_current_url)
return link_url
LinkRewriter(match_criteria).rewrite_links(node, cb)
