def _check_redirect_to_other_host(self, url):
redirect_url = self._redirect_url()
if not redirect_url:
return
if url_utils.hostname_from_url(url) == url_utils.hostname_from_url(redirect_url):
return
msg = "%s redirects to other host %s" % (url, redirect_url)
msg += "\nPlease provide non-redirecting URL"
error_utils.exit_with_message(msg)
def _analyze(self, link):
base_url_host = url_utils.hostname_from_url(self.base_url)
if self.base_href:
link = url_utils.make_absolute_url(self.base_href, link)
link_host = url_utils.hostname_from_url(link)
if not link_host or link_host == base_url_host:
link = url_utils.make_absolute_url(self.base_url, link)
link = url_utils.relative_url(link)
return "internal", link
else:
return "external", link
def _analyze(self, link):
link = url_utils.prepend_missing_scheme(link, self.base_url)
if self.base_href_tag:
link = url_utils.make_absolute_url(self.base_href_tag, link)
link_host = url_utils.hostname_from_url(link)
base_url_host = url_utils.hostname_from_url(self.base_url)
if not link_host or link_host == base_url_host:
link = url_utils.make_absolute_url(self.base_url, link)
link = url_utils.relative_url(link)
return "internal", link
else:
return "external", link
def __init__(self, page_start, sqlite_file, pages_list_file,
config_to_save):
self.page_start = page_start
self.page_host = url_utils.hostname_from_url(page_start)
self.sqlite_file = sqlite_file
self.pages_list = self._parse_pages_list_file(
pages_list_file) if pages_list_file else False
self.config_to_save = config_to_save
self.conn = None
def _get_resource_id(self, url, is_truncated):
page_host = url_utils.hostname_from_url(self.url)
relative_url = url_utils.internal_relative_url(url, page_host)
if relative_url is False:
is_external = 1
else:
url = relative_url
is_external = 0
c = self.conn.cursor()
c.execute(
"SELECT id FROM devtools_resource WHERE url = ? AND is_truncated = ? AND is_external = ?",
(url, is_truncated, is_external))
result = c.fetchone()
if result:
return result[0]
c.execute(
"INSERT INTO devtools_resource (url, is_truncated, is_external) VALUES (?, ?, ?)",
(url, is_truncated, is_external))
return c.lastrowid
def __init__(self):
self.page_host = url_utils.hostname_from_url(config.start_url)
self.pages_list = self._parse_pages_list_file()
self.conn = None