CONNECT adaptor: url

This is the URL connection adaptor. A connection adaptor makes and manages a connection to a remote server, and passes queries along the connection. The queries may or may not return replies; delayed queries may set up the expectation of a later reply to a specific listener. Handling a delayed query is not done by the connection adaptor, except that the outgoing query itself is done using a connection. Incoming answers are events which are handled by a LISTEN adaptor.

The canonical use of a CONNECT adaptor is to provide a way for wftk (or other adaptor-using apps, such as the popup UI framework to post information to a remote server via HTTP. The results may be useful; if so, however, the query adaptor will be the place where interpretation is done, and the connection adaptor simply tosses the reply back to the caller. Ideally the HTTP connection adaptor will preparse HTML into an appropriate XML structure, but that may have to wait a while.

The nice thing about this adaptor is that it's a wrapper around the libcurl library, so it automatically supports (cut and paste from the libcurl site) "FTP, FTPS, HTTP, HTTPS, GOPHER, TELNET, DICT, FILE, LDAP, HTTPS certificates, HTTP POST, HTTP PUT, FTP uploading, kerberos, HTTP form based upload, proxies, cookies, user+password authentication, file transfer resume, http proxy tunneling and more!" In other words, I thought for about 15 minutes about writing a simple connection adaptor and realized this would be stupid. For once, I didn't reinvent the wheel.
#include <stdio.h>
#include <stdarg.h>
#include <string.h>
#include <curl.h>
#include <easy.h>
#include <types.h>
#include "../wftk_session.h"
#include "../wftk_internals.h"
The adaptor_info structure is used to pass adaptor info (duh) back to the config module when it's building an adaptor instance. Here's what it contains:
static char *names[] = 

XML * CONNECT_url_init (WFTK_ADAPTOR * ad, va_list args);
XML * CONNECT_url_free (WFTK_ADAPTOR * ad, va_list args);
XML * CONNECT_url_info (WFTK_ADAPTOR * ad, va_list args);
XML * CONNECT_url_send (WFTK_ADAPTOR * ad, va_list args);

static WFTK_API_FUNC vtab[] = 

static struct wftk_adaptor_info _CONNECT_url_info =
Cool. So here's the incredibly complex function which returns a pointer to that:
struct wftk_adaptor_info * CONNECT_url_get_info ()
   return & _CONNECT_url_info;
Thus concludes the communication with the config module. Now on with the actual implementation of functionality. First, the initialization of an adaptor instance. This is pretty straightforward; it creates a cURL instance which persists as a handle. If two subsequent calls are to the same server, cURL will automatically attempt to use persistent connections, which is pretty neat if you think about it.
XML * CONNECT_url_init (WFTK_ADAPTOR * ad, va_list args) {
   CURL * curl;
   const char * parms;

   curl_global_init (CURL_GLOBAL_ALL);

   curl = curl_easy_init();
   if (curl) {
      ad->bindata = (void *) curl;
   } else {
      xml_set (ad->parms, "error", "Unable to initialize libcurl.");
   return (XML *) 0;
And the corresponding call to free up the cURL instance:
XML * CONNECT_url_free (WFTK_ADAPTOR * ad, va_list args)
   CURL * curl = (CURL *) (ad->bindata);

   if (curl) curl_easy_cleanup (curl);

   curl_global_cleanup ();

Next up is the info call, which builds and returns a little XML telling the caller about the adaptor. If the adaptor itself is NULL, then it just returns info about the installed adaptor handler; otherwise it's free to elaborate on the adaptor instance.
XML * CONNECT_url_info (WFTK_ADAPTOR * ad, va_list args) {
   XML * info;

   info = xml_create ("info");
   xml_set (info, "type", "connect");
   xml_set (info, "name", "url");
   xml_set (info, "ver", "1.0.0");
   xml_set (info, "compiled", __TIME__ " " __DATE__);
   xml_set (info, "author", "Michael Roberts");
   xml_set (info, "contact", "");
   xml_set (info, "libcurl", curl_version());
   xml_set (info, "extra_functions", "0");

   return (info);

So. Down to business: we use a connection to send a query and receive its immediate result. This initial stab (Sep. 2, 2001) is just going to worry about sending HTTP POST and GET requests; obviously, later we're going to get fancy with stuff like LDAP and FTP and God only knows what!

To build a query, we need to know several things: the URL to send the request to, whether to use a GET or a POST, and the individual data fields we'll use. If we're working with a GET, then we have to check whether the URL already contains the '?'; if so, then some of the fields are already contained in the URL and we'll just tack our new ones on the end. If not, we'll build the URL complete. If we're sending a POST, then libcurl will already do the right thing for us when we build our data package.

There will also be some additional optional information that the query can contain: user/password authentication, extra headers, and so forth; I'll document them as I add them, but right now I'm just going to postpone the whole thing.

A callback function is used to write the returning chunks of page; this callback just tacks the chunks onto a "return" attribute in the stash XML. After the conversation is completed, then you could technically do something sophisticated with it, but I'm not going to.
int CONNECT_url_writefunction (void * data, size_t size, size_t num, void * xml) {
   XML * stash = (XML *) xml;

   xml_attrncat (stash, "return", data, size * num);
   return size * num;

XML * CONNECT_url_send  (WFTK_ADAPTOR * ad, va_list args)
   CURL * curl = (CURL *) (ad->bindata);

   XML * query = (XML *) 0;
   XML * data  = (XML *) 0;
   XML * stash;
   XML * field;
   char * mark;
   char * mark2;
   char * qmark;
   int post = 0;
   char * built;
   XML * ret;

   if (args) query = va_arg (args, XML *);
   if (!query) {
      xml_set (ad->parms, "error", "No query specified.");
      return (XML *) 0;
   if (args) data = va_arg (args, XML *);
   if (!data) data = query; /* NULL data means to look at the query structure itself. */

   stash = xml_create ("s");
   xml_set (stash, "url", xml_attrval (query, "url"));

   if (!strcmp (xml_attrval (query, "method"), "POST")) post = 1;

   if (post) {
      curl_easy_setopt (curl, CURLOPT_POST, 1L);
      built = "body";
      xml_set (stash, built, "");
      qmark = NULL;
   } else {
      curl_easy_setopt (curl, CURLOPT_HTTPGET, 1L);
      built = "url";
      qmark = strchr (xml_attrval (stash, built), '?');

   field = xml_firstelem (data);
   while (field) {
      if (xml_is (field, "field")) {
         if (qmark) xml_attrcat (stash, built, "&");
         else {
            if (!post) xml_attrcat (stash, built, "?");
            qmark = "";

         mark = curl_escape ((char *)xml_attrval (field, "name"), 0);
         xml_attrcat (stash, built, mark);
         free (mark);
         xml_attrcat (stash, built, "=");
         if (*xml_attrval (field, "value")) {
            mark = curl_escape ((char *) xml_attrval (field, "value"), 0);
            xml_attrcat (stash, built, mark);
            free (mark);
         } else {
            mark = xml_stringcontent (field);
            mark2 = curl_escape (mark, 0);
            xml_attrcat (stash, built, mark2);
            free (mark); free (mark2);

      field = xml_nextelem (field);

   curl_easy_setopt (curl, CURLOPT_URL, xml_attrval (stash, "url"));
   if (post) curl_easy_setopt (curl, CURLOPT_POSTFIELDS, xml_attrval (stash, "body"));

   xml_set (stash, "return", "");

   curl_easy_setopt (curl, CURLOPT_WRITEFUNCTION, CONNECT_url_writefunction);
   curl_easy_setopt (curl, CURLOPT_FILE, stash);

   curl_easy_perform (curl);

   ret = xml_createtext (xml_attrval (stash, "return")); /* Later we'll get fancy. */
   xml_free (stash);

   return ret;

This code and documentation are released under the terms of the GNU license. They are additionally copyright (c) 2000, Vivtek. All rights reserved except those explicitly granted under the terms of the GNU license. This presentation was prepared with LPML. Try literate programming. You'll like it.