[PATCH] urlmon: Do not canonicalize Unicode characters.

[Zhiyi Zhang]

This fixes bug 37217

Signed-off-by: Zhiyi Zhang 
---
  dlls/urlmon/tests/uri.c | 35 +++++++++++++++++++++++++++++++++--
  dlls/urlmon/uri.c       | 21 +++++++++++++--------
  2 files changed, 46 insertions(+), 10 deletions(-)

[Jacek Caban]

Hi Zhiyi,

On 02/01/2018 11:56 AM, Zhiyi Zhang wrote:
> This fixes bug 37217
> Signed-off-by: Zhiyi Zhang
>
> dlls/urlmon/tests/uri.c | 35 +++++++++++++++++++++++++++++++++--
>  dlls/urlmon/uri.c       | 21 +++++++++++++--------
>  2 files changed, 46 insertions(+), 10 deletions(-)

The fix might be right, but we'd need more tests to be sure. You're
changing decoding of all URI parts, but add tests only for paths. Please
extend tests to cover all your changes. Also, it would be interesting to
test with other, non-alphanumeric Unicode characters (like LRM, ZWJ or
something invalid). Depending on the outcome of tests, we may want to
change is_unreserved to return true for Unicode characters instead of
changing each caller.

Thanks,
Jacek