Merge 106514 - contentDispositionType misparses the Content-Disposition header in some obscure co...

Merge 106514 - contentDispositionType misparses the Content-Disposition header in some obscure corner cases
https://bugs.webkit.org/show_bug.cgi?id=77577

Reviewed by Eric Seidel.

The contentDispositionType extracts the disposition-type from the
Content-Disposition header.  According to RFC 6266 (and previous RFCs),
the disposition-type must be an RFC 2616 token.  Rather than enforce
this general rule, we had special-cased some examples (including
name=foo and filename=bar).  This patch generalizes our check to
properly validate that the disposition-type is an RFC 2616 token.

In conjunction with some other work in the Chromium network stack, this
causes Chromium to pass the following tests:

  http://greenbytes.de/tech/tc2231/#inlonlyquoted
  http://greenbytes.de/tech/tc2231/#attonlyquoted

Without this patch, these test cases neither trigger a navigation nor a
download in Chromium.  This patch does not appear to cause any visible
change in Safari.  (Safari passes these tests both before and after
this patch.)

* platform/network/HTTPParsers.cpp:
(WebCore::isRFC2616Token):
(WebCore::contentDispositionType):
    - This patch also adds a comment to
      filenameFromHTTPContentDisposition, which explains some of the
      was this function incorrectly implements the requirements in
      RFC 6266.  Resolving these issues is a subject for a future
      patch.
* platform/network/HTTPParsers.h:


TBR=abarth@webkit.org
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=106519