[PATCH 1/4] uclient-fetch: --header option to pass additional raw HTTP headers

[Sergey Ponomarev]

Andre,

--header 'Content-type: foo/bar' doesn't work as intended because
'Content-type: application/x-www-form-urlencoded' is still added too. So
both headers are sent, which breaks my use case.

Yes, you are right and I described the behaviour in a comment. The
headers overriding doesn't work in this version: it makes code
slightly complicated and slower.
But you can achieve your goal by doing --method=POST
--post-data="OLOLO" --header 'Content-type: foo/bar'

On Thu, 23 Jun 2022 at 09:20, Andre Heider <a.heider at gmail.com> wrote:
>
> On 22/06/2022 12:26, Sergey Ponomarev wrote:
> > Hi Andre and Jo-Philipp,
> >
> > Please add the calloc() error handler yourself because I'm not
> > experienced in C and I haven't time. But for you this may be just 5
> > minutes.
>
> Sure, I had a look, and while adding that wrapper isn't a problem, the
> patchset itself has an issue:
>
> --header 'Content-type: foo/bar' doesn't work as intended because
> 'Content-type: application/x-www-form-urlencoded' is still added too. So
> both headers are sent, which breaks my use case.
>
> I may be looking at that, but I'm time constrained myself too at the moment.
>
> Cheers,
> Andre
>
> > The feature is very important from my point of view. The first version
> > I sent a year(s?) ago but still no progress.
> > If there are any problems please let me know
> >
> > On Wed, 22 Jun 2022 at 08:33, Andre Heider <a.heider at gmail.com> wrote:
> >>
> >> Hi Sergey,
> >>
> >> any update on this series? I'd be interested in the --header option.
> >>
> >> Thanks,
> >> Andre
> >>
> >> On 10/05/2022 11:11, Jo-Philipp Wich wrote:
> >>> Hi Sergey,
> >>>
> >>> a minor nitpick inline below.
> >>>
> >>> On 5/9/22 11:59 PM, Sergey Ponomarev wrote:
> >>>> You can add a custom HTTP header(s) to request:
> >>>>
> >>>>       wget --header='Authorization: Bearer TOKEN' \
> >>>>           --header='If-Modified-Since: Wed, 9 May 2021 12:16:00 GMT' \
> >>>>           https://example.com/
> >>>>
> >>>> Some headers like Authorization or User-Agent may be already set by --password or --user-agent.
> >>>> We may override them but it's a protection from user itself.
> >>>> To keep code concise the logic omitted.
> >>>>
> >>>> Signed-off-by: Sergey Ponomarev <stokito at gmail.com>
> >>>> ---
> >>>> [...]
> >>>> +                    case L_HEADER:
> >>>> +                            if (!raw_headers) {
> >>>> +                                    /* Max possible count of headers is the count of args (argc) - 2
> >>>> +                                     Since the first arg is program and last is a URL.
> >>>> +                                     But user may forget the URL and raw_headers is null terminated so allocate argc */
> >>>> +                                    raw_headers = calloc(argc, sizeof(char *));
> >>>
> >>> Please handle a possible calloc() error here. Maybe consider introducing an
> >>> "xalloc()" or similar helper which wraps calloc() and invokes abort() on NULL
> >>> return.
> >>>
> >>>> +                            }
> >>>> +                            raw_headers[raw_headers_count] = optarg;
> >>>> +                            raw_headers_count++;
> >>>> +                            break;
> >>>>                       case L_POST_DATA:
> >>>>                               post_data = optarg;
> >>>>                               break;
> >>>> [...]
> >>>
> >>>
> >>> _______________________________________________
> >>> openwrt-devel mailing list
> >>> openwrt-devel at lists.openwrt.org
> >>> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
> >>
> >
> >
>


-- 
Sergey Ponomarev,
stokito.com