def to_utf8(self):
cnt = 0
for site_url in self.get_url_list():
site_info = self.web_sites.get_web_site(site_url)
if site_info.redirect_to is not None and TUrlUtf8Encode.is_idna_string(
site_info.redirect_to):
site_info.redirect_to = TUrlUtf8Encode.convert_url_from_idna(
site_info.redirect_to)
if site_info.redirect_to == site_url and site_info.reach_status == TWebSiteReachStatus.abandoned:
site_info.redirect_to = None
site_info.reach_status = TWebSiteReachStatus.normal
cnt += 1
if TUrlUtf8Encode.is_idna_string(site_url):
site_info.url = TUrlUtf8Encode.convert_url_from_idna(site_url)
cnt += 1
self.logger.info("{} conversions made".format(cnt))
def add_web_site(self, site_url: str):
# russian domain must be in utf8
assert not TUrlUtf8Encode.is_idna_string(site_url)
assert site_url.startswith("http")
for x in self.office_web_sites:
assert x.url != site_url
s = TDeclarationWebSite(parent_office=self)
s.url = site_url
self.office_web_sites.append(s)