httputil: Only strip tabs and spaces from header values

The RFC specifies that only tabs and spaces should be stripped. Removing additonal whitespace characters can lead to framing errors with certain proxies.

Author: bdarnell

tornado/httputil.py

@@ -62,6 +62,9 @@
  from asyncio import Future # noqa: F401
  import unittest # noqa: F401
 
+# To be used with str.strip() and related methods.
+HTTP_WHITESPACE = " \t"
+
 
 @lru_cache(1000)
 def _normalize_header(name: str) -> str:
@@ -171,15 +174,15 @@ def parse_line(self, line: str) -> None:
  # continuation of a multi-line header
  if self._last_key is None:
  raise HTTPInputError("first header line cannot start with whitespace")
- new_part = " " + line.lstrip()
+ new_part = " " + line.lstrip(HTTP_WHITESPACE)
  self._as_list[self._last_key][-1] += new_part
  self._dict[self._last_key] += new_part
  else:
  try:
  name, value = line.split(":", 1)
  except ValueError:
  raise HTTPInputError("no colon in header line")
- self.add(name, value.strip())
+ self.add(name, value.strip(HTTP_WHITESPACE))
 
  @classmethod
  def parse(cls, headers: str) -> "HTTPHeaders":

tornado/test/httputil_test.py

@@ -334,6 +334,25 @@ def test_unicode_newlines(self):
  gen_log.warning("failed while trying %r in %s", newline, encoding)
  raise
 
+ def test_unicode_whitespace(self):
+ # Only tabs and spaces are to be stripped according to the HTTP standard.
+ # Other unicode whitespace is to be left as-is. In the context of headers,
+ # this specifically means the whitespace characters falling within the
+ # latin1 charset.
+ whitespace = [
+ (" ", True), # SPACE
+ ("\t", True), # TAB
+ ("\u00a0", False), # NON-BREAKING SPACE
+ ("\u0085", False), # NEXT LINE
+ ]
+ for c, stripped in whitespace:
+ headers = HTTPHeaders.parse("Transfer-Encoding: %schunked" % c)
+ if stripped:
+ expected = [("Transfer-Encoding", "chunked")]
+ else:
+ expected = [("Transfer-Encoding", "%schunked" % c)]
+ self.assertEqual(expected, list(headers.get_all()))
+
  def test_optional_cr(self):
  # Both CRLF and LF should be accepted as separators. CR should not be
  # part of the data when followed by LF, but it is a normal char