On Sep 15, 2013, at 5:54 AM, Dani Tweig <dani.tw...@gmail.com> wrote:
> thanks for your response.
> I did as you advices and i am having problem destring one of the server
> headers.
> I still see with wireshark that it is not destroyed.
>
> here is the relevant piece of code:
>
> switch (event) {
>
> case TS_EVENT_HTTP_SEND_RESPONSE_HDR:
> {
> TSMBuffer bufp;
> TSMLoc hdr_loc;
> TSMLoc field_loc;
>
> TSHttpTxnServerRespGet(txnp, &bufp, &hdr_loc);
> field_loc = TSMimeHdrFieldFind(bufp, hdr_loc, "Content-Disposition",
> -1);
> if (field_loc == TS_NULL_MLOC) {
> fprintf(stderr,"[add_header] Error while getting field !!!\n");
> return 0;
> }else{
> fprintf(stderr,"[add_header] removing Content-Disposition !!!\n");
> }
> TSMimeHdrFieldDestroy(bufp, hdr_loc, field_loc);
> TSHandleMLocRelease(bufp, TS_NULL_MLOC, hdr_loc);
> }
Looking at other plugins that use TSMimeHdrFieldDestroy, they typically loop
over duplicate headers using TSMimeHdrFieldNextDup. Maybe you need that for
your case too.
>
> On Fri, Sep 13, 2013 at 12:40 AM, Uri Shachar <ushac...@hotmail.com> wrote:
>
>> On Wed, 11 Sep 2013 21:49:56 Dani wrote:
>>> Subject: can traffic server transform delay the headers response ?
>> ...
>>> i would like to edit the http headers returned to the browser based on
>> the
>>> analysis i do in the transform, but i found that the headers are already
>>> returned to the browser BEFORE i get a chance to analyse the files
>> content
>>> with the transform mechanism.
>>
>> All you need to do is add a hookpoint on TS_HTTP_SEND_RESPONSE_HDR_HOOK,
>> and only reenable when you have finished your analysis + modified the
>> headers.
>>
>> --Uri
>
>
>
>
> --
> Dani Tweig
