[FFmpeg-devel] [PATCH] avformat/httpauth.c Support RFC7616 [Style fixed]

[Stefano Sabatini]

On date Thursday 2024-04-18 12:49:53 +0200, Stefano Sabatini wrote:
> On date Monday 2024-04-15 19:56:48 +0200, Stefano Sabatini wrote:
> > On date Monday 2024-04-15 02:32:14 +0000, ������ | Eugene wrote:
> > > Update digest authentication in httpauth.c
> > > 
> > > - Refactor make_digest_auth() to support RFC 2617 and RFC 7617
> > > - Add support for SHA-256 and SHA-512/256 algorithms along with MD5
> > > - MD5 and SHA-256 tested, but SHA-512/256 untested due to lack of server
> > > - Replace AVMD5 structure with AVHashContext for hash algorithm flexibility
> > > - Rename update_md5_strings() to update_hash_strings() for consistency
> > > - Address coding style feedback from reviewer:
> > > 
> > > This is a feature update focused on digest authentication enhancements.
> > > Some lint issues in the existing code remain unaddressed in this patch.
> > > 
> > > Signed-off-by: Eugene-bitsensing <eugene at bitsensing.com>
> > > ---
> > 
> > >  libavformat/httpauth.c | 102 +++++++++++++++++++++--------------------
> > >  1 file changed, 53 insertions(+), 49 deletions(-)
> > 
> > add entry to Changelog?
> 
> Updated the patch with a few fixes, please check in attachment.
> 
> [...] 
> > nit++: to avoid semantic overlap I'd rather name this
> > selected_algorithm
> > 
> > > +    if (!*algorithm) {
> > > +        algorithm = "MD5";  // if empty, use MD5 as Default 
> > > +        hashing_algorithm = "MD5";
> > > +    } else if (av_stristr(algorithm, "MD5")) {
> > > +        hashing_algorithm = "MD5";
> > > +    } else if (av_stristr(algorithm, "sha256") || av_stristr(algorithm, "sha-256")) {
> > > +        hashing_algorithm = "SHA256";
> > > +        len_hash = 65; // SHA-2: 64 characters.
> 
> > > +    } else if (av_stristr(algorithm, "sha512-256") || av_stristr(algorithm, "sha-512-256")) {
> > > +        hashing_algorithm = "SHA512_256";
> > > +        len_hash = 65; // SHA-2: 64 characters.
> 
> I'm concerned by this, as it will never be reached because the "sha256"
> branch is always selected instead, that's why this should be made an
> exact match?

Ping to Eugene. I think it would be safe with the use of stricmp.