def test_proxy_url_http_proto(self):
url = "http://localhost/?bizz=iss"
proxied = redirect.proxy_url(url)
self.assertEqual("/click/?url=http%3A//localhost/%3Fbizz%3Diss", proxied)
response = self.client.get(proxied, follow=True)
self.assertEqual(len(response.redirect_chain), 1)
self.assertEqual(response.redirect_chain[0][0], "http://localhost/?bizz=iss")
self.assertEqual(response.redirect_chain[0][1], 302)
def test_proxy_url_no_proto_and_port(self):
url = "localhost:8080"
proxied = redirect.proxy_url(url)
self.assertEqual("/click/?url=localhost%3A8080", proxied)
response = self.client.get(proxied, follow=True)
self.assertEqual(len(response.redirect_chain), 1)
self.assertEqual(response.redirect_chain[0][0], "localhost:8080")
self.assertEqual(response.redirect_chain[0][1], 302)
def test_utils(self):
url = "/?bizz=iss"
proxied = redirect.proxy_url(url)
self.assertEqual("/click/?url=/%3Fbizz%3Diss", proxied)
response = self.client.get(proxied, follow=True)
self.assertEqual(len(response.redirect_chain), 1)
self.assertEqual(response.redirect_chain[0][0], "/?bizz=iss")
self.assertEqual(response.redirect_chain[0][1], 302)
def _clean_html_body(request, email, body, charset):
"""Clean up a html part as best we can
Doesn't catch LXML errors
"""
html_tree = lxml_html.fromstring(body)
# if the HTML doc says its a different encoding, use that
for meta_tag in html_tree.xpath("/html/head/meta"):
if meta_tag.get("http-equiv", None) == "Content-Type":
try:
content = meta_tag.attrib["content"]
content = content.split(";", 1)[1]
charset = dict(HEADER_PARAMS.findall(content))["charset"]
break
except (KeyError, IndexError):
pass
elif "charset" in meta_tag.attrib:
charset = meta_tag.attrib["charset"]
break
try:
# check there's a body for premailer
if html_tree.find("body") is not None:
html_tree = InboxenPremailer(html_tree).transform()
except Exception as exc:
# Yeah, a pretty wide catch, but Premailer likes to throw up everything and anything
messages.info(request, _("Part of this message could not be parsed - it may not display correctly"))
_log.warning("Failed to render CSS for %s: %s", email["eid"], exc)
# Mail Pile uses this, give back if you come up with something better
cleaner = Cleaner(
allow_tags=HTML_ALLOW_TAGS,
kill_tags=["style"], # remove style tags, not attrs
remove_unknown_tags=False,
safe_attrs=HTML_SAFE_ATTRS,
safe_attrs_only=True,
style=False, # keep style attrs
)
html_tree = cleaner.clean_html(html_tree)
# filter images if we need to
if not email["display_images"]:
for img in html_tree.xpath("//img"):
try:
# try to delete src first - we don't want to add a src where there wasn't one already
del img.attrib["src"]
# replace image with 1px png
img.attrib["src"] = staticfiles_storage.url("imgs/placeholder.svg")
email["has_images"] = True
except KeyError:
pass
for link in html_tree.xpath("//a"):
try:
# proxy link
url = link.attrib["href"]
link.attrib["href"] = proxy_url(url)
except KeyError:
pass
# open link in tab
link.attrib["target"] = "_blank"
# and prevent window.opener bug (noopener is only supported in newer
# browsers, plus we already set noreferrer in the head)
link.attrib["rel"] = "noreferrer"
# finally, export to unicode
body = unicode_damnit(etree.tostring(html_tree, method="html"), charset)
return safestring.mark_safe(body)
def test_proxy_url_wrong_proto(self):
url = "mailto:[email protected]"
proxied = redirect.proxy_url(url)
# url should have not changed
self.assertEqual("mailto:[email protected]", proxied)
def _clean_html_body(request, email, body, charset):
"""Clean up a html part as best we can
Doesn't catch LXML errors
"""
html_tree = lxml_html.fromstring(body)
# if the HTML doc says its a different encoding, use that
for meta_tag in html_tree.xpath("/html/head/meta"):
if meta_tag.get("http-equiv", None) == "Content-Type":
content = meta_tag.get("content")
try:
content = content.split(";", 1)[1]
charset = dict(HEADER_PARAMS.findall(content))["charset"]
break
except (KeyError, IndexError):
pass
elif meta_tag.get("charset", None):
charset = meta_tag.get("charset")
break
try:
# check there's a body and header for premailer
if html_tree.find("body"):
html_tree = InboxenPremailer(html_tree).transform()
except Exception as exc:
# Yeah, a pretty wide catch, but Premailer likes to throw up everything and anything
messages.info(request, _("Part of this message could not be parsed - it may not display correctly"))
msg = "Failed to render CSS: %s" % exc
_log.exception(msg, extra={"request": request})
# Mail Pile uses this, give back if you come up with something better
cleaner = Cleaner(
allow_tags=HTML_ALLOW_TAGS,
kill_tags = ["style"], # remove style tags, not attrs
remove_unknown_tags=False,
safe_attrs=HTML_SAFE_ATTRS,
safe_attrs_only=True,
style=False, # keep style attrs
)
html_tree = cleaner.clean_html(html_tree)
# filter images if we need to
if not email["display_images"]:
for img in html_tree.xpath("//img"):
try:
# try to delete src first - we don't want to add a src where there wasn't one already
del img.attrib["src"]
# replace image with 1px png
img.attrib["src"] = staticfiles_storage.url("imgs/placeholder.svg")
email["has_images"] = True
except KeyError:
pass
for link in html_tree.xpath("//a"):
try:
# proxy link
url = link.attrib["href"]
link.attrib["href"] = proxy_url(url)
except KeyError:
pass
# finally, export to unicode
body = _unicode_damnit(etree.tostring(html_tree), charset)
return safestring.mark_safe(body)
