+ Added message body conversion: UTF-8 → UCS-2BE

configure.ac

@@ -34,7 +34,7 @@ AC_SUBST([LIBS])
 
 # Checks for header files.
 AC_HEADER_STDC
-AC_CHECK_HEADERS([ctype.h errno.h arpa/inet.h netdb.h netinet/in.h netinet/in_systm.h limits.h sys/poll.h regex.h signal.h stdarg.h stdlib.h stdio.h string.h sys/param.h sys/socket.h sys/time.h unistd.h sys/utsname.h],,[AC_MSG_ERROR([missing required header (see above)])],)
+AC_CHECK_HEADERS([ctype.h errno.h arpa/inet.h netdb.h netinet/in.h netinet/in_systm.h limits.h sys/poll.h regex.h signal.h stdarg.h stdlib.h stdio.h string.h sys/param.h sys/socket.h sys/time.h unistd.h sys/utsname.h iconv.h],,[AC_MSG_ERROR([missing required header (see above)])],)
 AC_CHECK_HEADERS([getopt.h syslog.h])
 AC_HEADER_SYS_WAIT
 AC_HEADER_TIME
@@ -49,7 +49,7 @@ AC_TYPE_SIZE_T
 # Checks for library functions.
 AC_FUNC_MALLOC
 AC_FUNC_SELECT_ARGTYPES
-AC_CHECK_FUNCS([getchar gethostbyname gethostname getopt getpid gettimeofday memset ntohs regcomp select socket strchr strcmp strstr strtol uname],,[AC_MSG_ERROR([missing required function (see above)])])
+AC_CHECK_FUNCS([getchar gethostbyname gethostname getopt getpid gettimeofday memset ntohs regcomp select socket strchr strcmp strstr strtol uname iconv],,[AC_MSG_ERROR([missing required function (see above)])])
 AC_CHECK_FUNCS([calloc getdomainname getopt_long inet_ntop strncasecmp strcasestr syslog])
 
 PKG_PROG_PKG_CONFIG

src/header_f.c

@@ -25,8 +25,10 @@
 
 /* add the given header(s) below the request line */
 void insert_header(char *mes, char *header, int first) {
-	char *ins, *backup;
-
+	char *ins, *backup, *hp, *dhp, *ehp;
+	int max_headers = 100, hcount = 0, c, i;
+	struct hline **hlist = malloc(sizeof(struct hline *) * max_headers);
+
 	if (first) {
 		ins = strchr(mes, '\n');
 		if (ins == NULL) {
@@ -43,6 +45,69 @@ void insert_header(char *mes, char *header, int first) {
 	strncpy(ins, header, strlen(header));
 	strncpy(ins + strlen(header), backup, strlen(backup)+1);
 	free(backup);
+
+	// deduplicate headers
+	hp = strstr(mes, "\r\n");
+	if (hp)
+		hp += 2;
+
+	while (hp && *hp) {
+		if (hcount == max_headers) {
+			max_headers *= 2;
+			hlist = realloc(hlist, sizeof(struct hline *) * max_headers);
+		}
+		c = i = hcount++;
+
+		hlist[i] = malloc(sizeof(struct hline));
+		hlist[i]->begin = hp;
+
+		ehp = strstr(hp, "\r\n");
+		if (ehp) {
+			if (ehp == hp) {
+				hp = NULL;
+				continue;
+			}
+			*ehp = 0;
+			hlist[i]->len = ehp - hp + 2;
+		}
+		else
+			hlist[i]->len = hp - mes + strlen(mes);
+
+		dhp = strchr(hp, ':');
+		if (dhp)
+			hlist[i]->namelen = dhp - hp;
+		else
+			hlist[i]->namelen = 0;
+
+		if (ehp) {
+			*ehp = 0x0D;
+			hp = ehp + 2;
+		}
+		else
+			hp = NULL;
+
+		i = 0;
+		while (i < c) {
+			if ((hlist[i]->namelen == hlist[c]->namelen) && !strncmp(hlist[i]->begin, hlist[c]->begin, hlist[c]->namelen)) {
+				if (hp) {
+					memmove(hlist[c]->begin, hp, hp - mes + strlen(mes) + 1);
+					hp = hlist[c]->begin;
+				}
+				else
+					strcpy(hlist[c]->begin, "\r\n\0");
+
+				--hcount;
+				free(hlist[c]);
+				i = c;
+			}
+			else
+				++i;
+		}
+	}
+
+	for (i = 0; i < hcount; i++)
+		free(hlist[i]);
+	free(hlist);
 }
 
 /* add a Via Header Field in the message. */

src/header_f.h

@@ -20,6 +20,12 @@
 
 #include "shoot.h"
 
+struct hline
+{
+	char *begin;
+	int namelen, len;
+};
+
 void insert_header(char *mes, char *header, int first);
 
 void add_via(char *mes);

src/request.c

@@ -198,12 +198,11 @@ void create_msg(int action, char *req_buff, char *repl_buff, char *username, int
 					FROM_STR, fqdn, lport, c);
 			}
 			req_buff += strlen(req_buff);
-			if (mes_body) {
-				len = strlen(mes_body);
-			}
-			else {
+			if (mes_body)
+				len = mes_body->len;
+			else
 				len = SIPSAK_MES_STR_LEN + strlen(username);
-			}
+
 			sprintf(req_buff, "%s%u\r\n", CON_LEN_STR, len);
 			req_buff += strlen(req_buff);
 			if (con_dis) {
@@ -212,12 +211,7 @@ void create_msg(int action, char *req_buff, char *repl_buff, char *username, int
 			}
 			sprintf(req_buff, "\r\n");
 			req_buff += 2;
-			if (mes_body) {
-				sprintf(req_buff,
-					"%s",
-					mes_body);
-			}
-			else {
+			if (!mes_body) {
 				sprintf(req_buff, "%s%s", SIPSAK_MES_STR, username);
 				req_buff += strlen(req_buff) - 1;
 				*(req_buff) = '.';

src/shoot.c

@@ -1037,8 +1037,11 @@ void shoot(char *buf, int buff_size)
 					/* prevents a strange error */
 					regcomp(&(regexps.authexp), "^SIP/[0-9]\\.[0-9] 40[17] ", REG_EXTENDED|REG_NOSUB|REG_ICASE);
 					insert_auth(request, received);
-					if (verbose > 2)
+					if (verbose > 2) {
 						printf("\nreceived:\n%s\n", received);
+						if (mes_body)
+							printf("<UCS-2BE encoded body>\n");
+					}
 					new_transaction(request, response);
 					continue;
 				} /* if auth...*/

src/sipsak.c

@@ -55,6 +55,7 @@
 #ifdef HAVE_SYSLOG_H
 # include <syslog.h>
 #endif
+#include <iconv.h>
 
 #include "helper.h"
 #include "header_f.h"
@@ -424,14 +425,24 @@ int main(int argc, char *argv[])
 				namebeg=str_to_int(0, optarg);
 				break;
 			case 'B':
-				mes_body=str_alloc(strlen(optarg) + 1);
-				strncpy(mes_body, optarg, strlen(optarg));
+			{
+				iconv_t cd;
+				char *inbuf = optarg, *outbuf;
+				size_t inbytes = strlen(optarg), outbytes = inbytes * 2;
+				mes_body = malloc(sizeof(struct bytearray));
+				mes_body->data = outbuf = str_alloc(outbytes);
+				mes_body->len = outbytes;
+				cd = iconv_open("UCS-2BE", "UTF-8"); // FIX: source charset should be picked from locale
+				iconv(cd, &inbuf, &inbytes, &outbuf, &outbytes);
+				iconv_close(cd);
+				mes_body->len -= outbytes; // some UTF-8 characters 1-byte long, some 2-byte long. All UCS-2 chars 2-byte long
 				break;
+			}
 			case 'c':
 				backup=str_alloc(strlen(optarg)+1);
 				strncpy(backup, optarg, strlen(optarg));
 				parse_uri(backup, &scheme, &user, &host, &port);
-				if (scheme  == NULL) {
+				if (scheme == NULL) {
 					fprintf(stderr, "error: missing scheme in From URI\n");
 					exit_code(2, __PRETTY_FUNCTION__, "missing scheme in From URI");
 				}
@@ -464,9 +475,9 @@ int main(int argc, char *argv[])
 					strncpy(backup, optarg, strlen(optarg));
 					parse_uri(backup, &scheme, &user, &host, &port);
 					if (scheme == NULL) {
-					    fprintf(stderr, "error: REGISTER Contact uri doesn't not contain "
-						   "sip:, sips:, *, or is not empty\n");
-				    	exit_code(2, __PRETTY_FUNCTION__, "unsupported Contact for registration");
+						fprintf(stderr, "error: REGISTER Contact uri doesn't not contain "
+							"sip:, sips:, *, or is not empty\n");
+						exit_code(2, __PRETTY_FUNCTION__, "unsupported Contact for registration");
 					}
 					/*else if (user == NULL) {
 						fprintf(stderr, "error: missing username in Contact uri\n");
@@ -1056,7 +1067,7 @@ int main(int argc, char *argv[])
 	srand(time(0) ^ (getpid() + (getpid() << 15)));
 
 	if (processes > 1) {
-		if (signal(SIGCHLD , sigchld_handler)  == SIG_ERR ) {
+		if (signal(SIGCHLD , sigchld_handler) == SIG_ERR ) {
 			fprintf(stderr, "error: Could not install SIGCHLD handler\n");
 			exit_code(2, __PRETTY_FUNCTION__, "failed to install SIGCHLD handler");
 		}
@@ -1069,7 +1080,7 @@ int main(int argc, char *argv[])
 		}
 
 		if (pid == 0){
-	    	/* child */
+			/* child */
 			upp = (nameend - namebeg + 1) / processes;
 			namebeg = namebeg + upp * i;
 			nameend = namebeg + upp;

src/sipsak.h

@@ -321,6 +321,11 @@ SSL* ssl;
 # endif
 #endif
 
+struct bytearray {
+	char *data;
+	size_t len;
+};
+
 /* lots of global variables. ugly but makes life easier. */
 unsigned long address;
 unsigned int nonce_count, transport;
@@ -331,10 +336,11 @@ int file_b, uri_b, trace, via_ins, usrloc, redirects, rand_rem, replace_b;
 int empty_contact, nagios_warn, fix_crlf, timing, outbound_proxy;
 int timer_t1, timer_t2, timer_final, sysl;
 char *username, *domainname, *password, *replace_str, *hostname, *contact_uri;
-char *mes_body, *con_dis, *auth_username, *from_uri, *headers, *authhash, *local_ip;
+char *con_dis, *auth_username, *from_uri, *headers, *authhash, *local_ip;
 char fqdn[FQDN_SIZE];
 char target_dot[INET_ADDRSTRLEN], source_dot[INET_ADDRSTRLEN];
 char *request, *response, *received, *transport_str;
 regex_t* regex;
+struct bytearray *mes_body;
 
 #endif

src/transport.c

@@ -679,35 +679,50 @@ void create_sockets(struct sipsak_con_data *cd) {
 void send_message(char* mes, struct sipsak_con_data *cd,
 			struct sipsak_counter *sc, struct sipsak_sr_time *srt) {
 	struct timezone tz;
-	int ret = -1;
-
+	int ret = -1, len = strlen(mes), flen = len;
+	char *tmsg;
+
 	if (cd->dontsend == 0) {
-		if (verbose > 2) {
-			printf("\nrequest:\n%s", mes);
+		if (mes_body)
+			flen += mes_body->len;
+
+		tmsg = str_alloc(flen);
+		memcpy(tmsg, mes, len);
+
+		if (verbose > 2)
+			printf("\nrequest [%d bytes]:\n%s", flen, mes);
+
+		if (mes_body) {
+			if (verbose > 2)
+				printf("<UCS-2BE encoded body>\n");
+			memcpy(tmsg + len, mes_body->data, mes_body->len);
 		}
+
 		/* lets fire the request to the server and store when we did */
 		if (cd->csock == -1) {
 			dbg("\nusing un-connected socket for sending\n");
-			ret = sendto(cd->usock, mes, strlen(mes), 0, (struct sockaddr *) &(cd->adr), sizeof(struct sockaddr));
+			ret = sendto(cd->usock, tmsg, flen, 0, (struct sockaddr *) &(cd->adr), sizeof(struct sockaddr));
 		}
 		else {
 			dbg("\nusing connected socket for sending\n");
 #ifdef WITH_TLS_TRANSP
 			if (transport == SIP_TLS_TRANSPORT) {
 # ifdef USE_GNUTLS
-				ret = gnutls_record_send(tls_session, mes, strlen(mes));
+				ret = gnutls_record_send(tls_session, tmsg, flen);
 # else /* USE_GNUTLS */
 #  ifdef USE_OPENSSL
 #  endif /* USE_OPENSSL */
 # endif /* USE_GNUTLS */
 			}
 			else {
 #endif /* TLS_TRANSP */
-				ret = send(cd->csock, mes, strlen(mes), 0);
+				ret = send(cd->csock, tmsg, flen, 0);
 #ifdef WITH_TLS_TRANSP
 			}
 #endif /* TLS_TRANSP */
 		}
+		free(tmsg);
+
 		(void)gettimeofday(&(srt->sendtime), &tz);
 		if (ret==-1) {
 			if (verbose)
@@ -717,8 +732,8 @@ void send_message(char* mes, struct sipsak_con_data *cd,
 		}
 #ifdef HAVE_INET_NTOP
 		if (verbose > 2) {
-			printf("\nsend to: %s:%s:%i\n", transport_str, target_dot, rport);
-    }
+			printf("\nsend to: %s:%s:%i [%d bytes]\n", transport_str, target_dot, rport, ret);
+		}
 #endif
 		sc->send_counter++;
 	}
@@ -936,6 +951,7 @@ int recv_message(char *buf, int size, int inv_trans,
 	int ret = 0;
 	int sock = 0;
 	double tmp_delay;
+	char *ehp;
 #ifdef HAVE_INET_NTOP
 	struct sockaddr_in peer_adr;
 	socklen_t psize = sizeof(peer_adr);
@@ -1042,6 +1058,11 @@ int recv_message(char *buf, int size, int inv_trans,
 #endif // RAW_SUPPORT
 	if (ret > 0) {
 		*(buf+ ret) = '\0';
+
+		// Cut off binary message body
+		if (mes_body && (ehp = strstr(buf, "\r\n\r\n")))
+			ehp[4] = 0;
+
 		if (transport != SIP_UDP_TRANSPORT) {
 			if (verbose > 0)
 				printf("\nchecking message for completeness...\n");
@@ -1083,8 +1104,8 @@ int recv_message(char *buf, int size, int inv_trans,
 		}
 #ifdef HAVE_INET_NTOP
 		if ((verbose > 2) && (getpeername(sock, (struct sockaddr *)&peer_adr, &psize) == 0) && (inet_ntop(peer_adr.sin_family, &peer_adr.sin_addr, &source_dot[0], INET_ADDRSTRLEN) != NULL)) {
-			printf("\nreceived from: %s:%s:%i\n", transport_str, 
-						source_dot, ntohs(peer_adr.sin_port));
+			printf("\nreceived from: %s:%s:%i [%d bytes]\n", transport_str,
+						source_dot, ntohs(peer_adr.sin_port), ret);
 		}
 		else if (verbose > 1 && trace == 0 && usrloc == 0)
 			printf(":\n");