You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@trafficserver.apache.org by Tin Zaw <tz...@yahoo.com> on 2010/01/08 04:34:00 UTC
Is INKMimeHdrFieldInsert() working as intended?
OK, it's me, sorry to bug you again.
I am following replace-header plugin example to implement a plugin that sets a cookie.
Here is my call back.
case INK_EVENT_HTTP_SEND_RESPONSE_HDR:
replace_header(txnp, contp);
And below is replace_header(). But when I run it, everything seems fine according to logs, but I don't find an extra header (I intercept the TS's response using another HTTP, burpSuite, proxy running on desktop with the browser) setting cookie in the response.
Logs show:
[Jan 8 03:25:32.019] Server {3086841568} DIAG: (add-cookie) ENTER replace_header.
[Jan 8 03:25:32.019] Server {3086841568} DIAG: (add-cookie) EXIT replace_header.
So, am I doing something wrong in code, or forgetting to set something in config? Or could it be something deeper?
Many thanks in advance!
static void
replace_header(INKHttpTxn txnp, INKCont contp)
{
INKMBuffer resp_bufp;
INKMLoc resp_loc;
INKMLoc field_loc;
const char *cookie_content = "terms=; path=/; expires=Thu, 01-Jan-1970 00:00:00 GMT";
INKDebug("add-cookie", "ENTER replace_header.");
if (!INKHttpTxnServerRespGet(txnp, &resp_bufp, &resp_loc)) {
INKDebug("add-cookie", "INKHttpTxnServerRespGet failed.");
goto done;
}
/* create a new field in the header */
field_loc = INKMimeHdrFieldCreate(resp_bufp, resp_loc);
if (!field_loc) {
INKDebug("add-cookie", "INKMimeHdrFieldCreate returns null.");
INKHandleMLocRelease(resp_bufp, INK_NULL_MLOC, resp_loc);
goto done;
}
/* set its name */
if (INK_SUCCESS != INKMimeHdrFieldNameSet(resp_bufp, resp_loc, field_loc, INK_MIME_FIELD_SET_COOKIE, INK_MIME_LEN_SET_COOKIE)) {
INKDebug("add-cookie", "INKMimeHdrFieldNameSet failed.");
}
/* set its value */
//INKMimeHdrFieldValueInsert(resp_bufp, resp_loc, field_loc, "none", 4, -1);
if (INK_SUCCESS != INKMimeHdrFieldValueStringInsert(resp_bufp, resp_loc, field_loc, -1, cookie_content, -1)) {
INKDebug("add-cookie", "INKMimeHdrFieldValueStringInsert failed.");
}
/* insert it into the header */
if (INK_SUCCESS != INKMimeHdrFieldInsert(resp_bufp, resp_loc, field_loc, -1)) {
INKDebug("add-cookie", "INKMimeHdrFieldInsert failed.");
}
INKHandleMLocRelease(resp_bufp, resp_loc, field_loc);
INKHandleMLocRelease(resp_bufp, INK_NULL_MLOC, resp_loc);
done:
INKHttpTxnReenable(txnp, INK_EVENT_HTTP_CONTINUE);
INKDebug("add-cookie", "EXIT replace_header.");
}
Re: Is INKMimeHdrFieldInsert() working as intended?
Posted by Steve Jiang <sj...@apache.org>.
INKHttpTxnServerRespGet retrieves the response from the origin server
(from READ_RESPONSE_HDR). TS builds a response to send to the client
from the server response between READ_RESPONSE_HDR and
SEND_RESPONSE_HDR, so changing the server response at the latter hook
will be too late.
If you want to change what is sent to the client you can either:
a) set your hook at READ_RESPONSE_HDR_HOOK and change the response from
INKHttpTxnServerRespGet
b) keep it at SEND_RESPONSE_HDR_HOOK and modify the response from
INKHttpTxnClientRespGet
I hope that was a clear enough explanation.
--Steve
Tin Zaw wrote, on 1/7/10 7:34 PM:
> OK, it's me, sorry to bug you again.
>
> I am following replace-header plugin example to implement a plugin that sets a cookie.
>
> Here is my call back.
>
> case INK_EVENT_HTTP_SEND_RESPONSE_HDR:
> replace_header(txnp, contp);
>
> And below is replace_header(). But when I run it, everything seems fine according to logs, but I don't find an extra header (I intercept the TS's response using another HTTP, burpSuite, proxy running on desktop with the browser) setting cookie in the response.
>
> Logs show:
> [Jan 8 03:25:32.019] Server {3086841568} DIAG: (add-cookie) ENTER replace_header.
> [Jan 8 03:25:32.019] Server {3086841568} DIAG: (add-cookie) EXIT replace_header.
>
> So, am I doing something wrong in code, or forgetting to set something in config? Or could it be something deeper?
>
> Many thanks in advance!
>
>
>
> static void
> replace_header(INKHttpTxn txnp, INKCont contp)
> {
> INKMBuffer resp_bufp;
> INKMLoc resp_loc;
> INKMLoc field_loc;
>
> const char *cookie_content = "terms=; path=/; expires=Thu, 01-Jan-1970 00:00:00 GMT";
>
> INKDebug("add-cookie", "ENTER replace_header.");
>
> if (!INKHttpTxnServerRespGet(txnp, &resp_bufp, &resp_loc)) {
> INKDebug("add-cookie", "INKHttpTxnServerRespGet failed.");
> goto done;
> }
>
> /* create a new field in the header */
> field_loc = INKMimeHdrFieldCreate(resp_bufp, resp_loc);
> if (!field_loc) {
> INKDebug("add-cookie", "INKMimeHdrFieldCreate returns null.");
> INKHandleMLocRelease(resp_bufp, INK_NULL_MLOC, resp_loc);
> goto done;
> }
>
> /* set its name */
> if (INK_SUCCESS != INKMimeHdrFieldNameSet(resp_bufp, resp_loc, field_loc, INK_MIME_FIELD_SET_COOKIE, INK_MIME_LEN_SET_COOKIE)) {
> INKDebug("add-cookie", "INKMimeHdrFieldNameSet failed.");
> }
>
> /* set its value */
> //INKMimeHdrFieldValueInsert(resp_bufp, resp_loc, field_loc, "none", 4, -1);
> if (INK_SUCCESS != INKMimeHdrFieldValueStringInsert(resp_bufp, resp_loc, field_loc, -1, cookie_content, -1)) {
> INKDebug("add-cookie", "INKMimeHdrFieldValueStringInsert failed.");
> }
>
> /* insert it into the header */
> if (INK_SUCCESS != INKMimeHdrFieldInsert(resp_bufp, resp_loc, field_loc, -1)) {
> INKDebug("add-cookie", "INKMimeHdrFieldInsert failed.");
> }
> INKHandleMLocRelease(resp_bufp, resp_loc, field_loc);
> INKHandleMLocRelease(resp_bufp, INK_NULL_MLOC, resp_loc);
>
> done:
> INKHttpTxnReenable(txnp, INK_EVENT_HTTP_CONTINUE);
> INKDebug("add-cookie", "EXIT replace_header.");
> }
Re: Is INKMimeHdrFieldInsert() working as intended?
Posted by Steve Jiang <sj...@apache.org>.
INKHttpTxnServerRespGet retrieves the response from the origin server
(from READ_RESPONSE_HDR). TS builds a response to send to the client
from the server response between READ_RESPONSE_HDR and
SEND_RESPONSE_HDR, so changing the server response at the latter hook
will be too late.
If you want to change what is sent to the client you can either:
a) set your hook at READ_RESPONSE_HDR_HOOK and change the response from
INKHttpTxnServerRespGet
b) keep it at SEND_RESPONSE_HDR_HOOK and modify the response from
INKHttpTxnClientRespGet
I hope that was a clear enough explanation.
--Steve
Tin Zaw wrote, on 1/7/10 7:34 PM:
> OK, it's me, sorry to bug you again.
>
> I am following replace-header plugin example to implement a plugin that sets a cookie.
>
> Here is my call back.
>
> case INK_EVENT_HTTP_SEND_RESPONSE_HDR:
> replace_header(txnp, contp);
>
> And below is replace_header(). But when I run it, everything seems fine according to logs, but I don't find an extra header (I intercept the TS's response using another HTTP, burpSuite, proxy running on desktop with the browser) setting cookie in the response.
>
> Logs show:
> [Jan 8 03:25:32.019] Server {3086841568} DIAG: (add-cookie) ENTER replace_header.
> [Jan 8 03:25:32.019] Server {3086841568} DIAG: (add-cookie) EXIT replace_header.
>
> So, am I doing something wrong in code, or forgetting to set something in config? Or could it be something deeper?
>
> Many thanks in advance!
>
>
>
> static void
> replace_header(INKHttpTxn txnp, INKCont contp)
> {
> INKMBuffer resp_bufp;
> INKMLoc resp_loc;
> INKMLoc field_loc;
>
> const char *cookie_content = "terms=; path=/; expires=Thu, 01-Jan-1970 00:00:00 GMT";
>
> INKDebug("add-cookie", "ENTER replace_header.");
>
> if (!INKHttpTxnServerRespGet(txnp, &resp_bufp, &resp_loc)) {
> INKDebug("add-cookie", "INKHttpTxnServerRespGet failed.");
> goto done;
> }
>
> /* create a new field in the header */
> field_loc = INKMimeHdrFieldCreate(resp_bufp, resp_loc);
> if (!field_loc) {
> INKDebug("add-cookie", "INKMimeHdrFieldCreate returns null.");
> INKHandleMLocRelease(resp_bufp, INK_NULL_MLOC, resp_loc);
> goto done;
> }
>
> /* set its name */
> if (INK_SUCCESS != INKMimeHdrFieldNameSet(resp_bufp, resp_loc, field_loc, INK_MIME_FIELD_SET_COOKIE, INK_MIME_LEN_SET_COOKIE)) {
> INKDebug("add-cookie", "INKMimeHdrFieldNameSet failed.");
> }
>
> /* set its value */
> //INKMimeHdrFieldValueInsert(resp_bufp, resp_loc, field_loc, "none", 4, -1);
> if (INK_SUCCESS != INKMimeHdrFieldValueStringInsert(resp_bufp, resp_loc, field_loc, -1, cookie_content, -1)) {
> INKDebug("add-cookie", "INKMimeHdrFieldValueStringInsert failed.");
> }
>
> /* insert it into the header */
> if (INK_SUCCESS != INKMimeHdrFieldInsert(resp_bufp, resp_loc, field_loc, -1)) {
> INKDebug("add-cookie", "INKMimeHdrFieldInsert failed.");
> }
> INKHandleMLocRelease(resp_bufp, resp_loc, field_loc);
> INKHandleMLocRelease(resp_bufp, INK_NULL_MLOC, resp_loc);
>
> done:
> INKHttpTxnReenable(txnp, INK_EVENT_HTTP_CONTINUE);
> INKDebug("add-cookie", "EXIT replace_header.");
> }