@@ -17,6 +17,9 @@

  *

  * Change History:

  * $Log: mbox.c,v $

+ * Revision 1.79  2004/06/22 04:08:01  nigelhorne

+ * Optimise empty lines

+ *

  * Revision 1.78  2004/06/21 10:21:19  nigelhorne

  * Fix crash when a multipart/mixed message contains many parts that need to be scanned as attachments

  *

@@ -222,7 +225,7 @@

  * Compilable under SCO; removed duplicate code with message.c

  *

  */

-static	char	const	rcsid[] = "$Id: mbox.c,v 1.78 2004/06/21 10:21:19 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: mbox.c,v 1.79 2004/06/22 04:08:01 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -266,6 +269,21 @@ static	char	const	rcsid[] = "$Id: mbox.c,v 1.78 2004/06/21 10:21:19 nigelhorne E

 #include "defaults.h"

 #include "str.h"

+#ifdef	CL_DEBUG

+#if __GLIBC__ == 2 && __GLIBC_MINOR__ >= 1

+#define HAVE_BACKTRACE

+#endif

+

+#ifdef HAVE_BACKTRACE

+#include <execinfo.h>

+#include <signal.h>

+#include <syslog.h>

+#endif

+

+static	void	sigsegv(int sig);

+static	void	print_trace(int use_syslog);

+#endif

+

 #if	defined(NO_STRTOK_R) || !defined(CL_THREAD_SAFE)

 #undef strtok_r

 #undef __strtok_r

@@ -395,6 +413,9 @@ cl_mbox(const char *dir, int desc)

 	message *m, *body;

 	FILE *fd;

 	char buffer[LINE_LENGTH];

+#ifdef	CL_DEBUG

+	void (*segv)(int);

+#endif

 	cli_dbgmsg("in mbox()\n");

@@ -436,6 +457,10 @@ cl_mbox(const char *dir, int desc)

 	pthread_mutex_unlock(&tables_mutex);

 #endif

+#ifdef	CL_DEBUG

+	segv = signal(SIGSEGV, sigsegv);

+#endif

+

 	/*

 	 * is it a UNIX style mbox with more than one

 	 * mail message, or just a single mail message?

@@ -519,6 +544,10 @@ cl_mbox(const char *dir, int desc)

 	cli_dbgmsg("cli_mbox returning %d\n", retcode);

+#ifdef	CL_DEBUG

+	signal(SIGSEGV, segv);

+#endif

+

 	return retcode;

 }

@@ -537,27 +566,33 @@ parseEmailHeaders(const message *m, const table_t *rfc821Table)

 	const text *t;

 	message *ret;

+	cli_dbgmsg("parseEmailHeaders\n");

+

 	if(m == NULL)

 		return NULL;

 	ret = messageCreate();

 	for(t = messageGetBody(m); t; t = t->t_next) {

-		char *buffer = strdup(t->t_text);

+		char *buffer;

 #ifdef CL_THREAD_SAFE

 		char *strptr;

 #endif

-		if(buffer == NULL)

-			break;

-

-		cli_chomp(buffer);

+		if(t->t_text) {

+			buffer = strdup(t->t_text);

+			if(buffer == NULL)

+				break;

+			cli_chomp(buffer);

+		} else

+			buffer = NULL;

 		/*

 		 * Section B.2 of RFC822 says TAB or SPACE means

 		 * a continuation of the previous entry.

 		 */

-		if(inHeader && ((buffer[0] == '\t') || (buffer[0] == ' ')))

+		if(inHeader && buffer &&

+		  ((buffer[0] == '\t') || (buffer[0] == ' ')))

 			inContinuationHeader = TRUE;

 		if(inContinuationHeader) {

@@ -569,22 +604,23 @@ parseEmailHeaders(const message *m, const table_t *rfc821Table)

 			/*

 			 * Add all the arguments on the line

 			 */

-			for(ptr = strtok_r(buffer, ";", &strptr); ptr; ptr = strtok_r(NULL, ":", &strptr))

-				messageAddArgument(ret, ptr);

-			free(buffer);

+			if(buffer) {

+				for(ptr = strtok_r(buffer, ";", &strptr); ptr; ptr = strtok_r(NULL, ":", &strptr))

+					messageAddArgument(ret, ptr);

+				free(buffer);

+			}

 		} else if(inHeader) {

-			cli_dbgmsg("Deal with header %s\n", buffer);

-

 			/*

 			 * A blank line signifies the end of the header and

 			 * the start of the text

 			 */

-			if(strlen(buffer) == 0) {

+			if((buffer == NULL) || (strlen(buffer) == 0)) {

 				cli_dbgmsg("End of header information\n");

 				inContinuationHeader = inHeader = FALSE;

 			} else if(parseEmailHeader(ret, buffer, rfc821Table) == CONTENT_TYPE)

 				inContinuationHeader = continuationMarker(buffer);

-			free(buffer);

+			if(buffer)

+				free(buffer);

 		} else {

 			/*cli_dbgmsg("Add line to body '%s'\n", buffer);*/

 			messageAddLine(ret, buffer, 0);

@@ -787,11 +823,12 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 				 * Ignore blank lines. There shouldn't be ANY

 				 * but some viruses insert them

 				 */

-				while((t_line = t_line->t_next) != NULL) {

-					cli_chomp(t_line->t_text);

-					if(strlen(t_line->t_text) != 0)

-						break;

-				}

+				while((t_line = t_line->t_next) != NULL)

+					if(t_line->t_text) {

+						cli_chomp(t_line->t_text);

+						if(strlen(t_line->t_text) != 0)

+							break;

+					}

 				if(t_line == NULL) {

 					cli_dbgmsg("Empty part\n");

@@ -805,6 +842,10 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						inMimeHead, inhead, boundary, line, t_line->t_next ? t_line->t_next->t_text : "(null)");*/

 					if(inMimeHead) {

+						if(line == NULL) {

+							inhead = inMimeHead = 0;

+							continue;

+						}

 						/*

 						 * Handle continuation lines

 						 * because the previous line

@@ -836,7 +877,8 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						inMimeHead = continuationMarker(line);

 						messageAddArgument(aMessage, line);

 					} else if(inhead) {

-						if(strlen(line) == 0) {

+						if(line == NULL) {

+							/* empty line */

 							inhead = 0;

 							continue;

 						}

@@ -872,7 +914,9 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						 */

 						inMimeHead = continuationMarker(line);

 						if(!inMimeHead)

-							if(t_line->t_next && ((t_line->t_next->t_text[0] == '\t') || (t_line->t_next->t_text[0] == ' ')))

+							if(t_line->t_next &&

+							   t_line->t_next->t_text &&

+							  ((t_line->t_next->t_text[0] == '\t') || (t_line->t_next->t_text[0] == ' ')))

 								inMimeHead = TRUE;

 						parseEmailHeader(aMessage, line, rfc821Table);

@@ -1540,7 +1584,8 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 				 * won't be scanned

 				 */

 				for(t = t_line; t; t = t->t_next)

-					if((strncasecmp(t->t_text, encoding, sizeof(encoding) - 1) == 0) &&

+					if(t->t_text &&

+					   (strncasecmp(t->t_text, encoding, sizeof(encoding) - 1) == 0) &&

 					   (strstr(t->t_text, "7bit") == NULL))

 						break;

 				if(t && ((b = textToBlob(t_line, NULL)) != NULL)) {

@@ -1646,6 +1691,8 @@ boundaryStart(const char *line, const char *boundary)

 	 * notice the extra '-'

 	 */

 	/*cli_dbgmsg("boundaryStart: line = '%s' boundary = '%s'\n", line, boundary);*/

+	if(line == NULL)

+		return 0;	/* empty line */

 	if(strstr(line, boundary) != NULL) {

 		cli_dbgmsg("found %s in %s\n", boundary, line);

 		return 1;

@@ -1667,6 +1714,8 @@ endOfMessage(const char *line, const char *boundary)

 {

 	size_t len;

+	if(line == NULL)

+		return 0;

 	if(*line++ != '-')

 		return 0;

 	if(*line++ != '-')

@@ -1789,6 +1838,7 @@ strstrip(char *s)

 {

 	if(s == (char *)NULL)

 		return(0);

+

 	return(strip(s, strlen(s) + 1));

 }

@@ -1804,7 +1854,8 @@ continuationMarker(const char *line)

 {

 	const char *ptr;

-	assert(line != NULL);

+	if(line == NULL)

+		return FALSE;

 #ifdef	CL_DEBUG

 	cli_dbgmsg("continuationMarker(%s)\n", line);

@@ -2030,3 +2081,43 @@ saveFile(const blob *b, const char *dir)

 	return (close(fd) >= 0);

 }

+

+#ifdef	CL_DEBUG

+static void

+sigsegv(int sig)

+{

+	signal(SIGSEGV, SIG_DFL);

+	print_trace(0);

+	exit(SIGSEGV);

+}

+

+static void

+print_trace(int use_syslog)

+{

+#ifdef HAVE_BACKTRACE

+	void *array[10];

+	size_t size;

+	char **strings;

+	size_t i;

+	pid_t pid = getpid();

+

+	size = backtrace(array, 10);

+	strings = backtrace_symbols(array, size);

+

+	syslog(LOG_ERR, "Backtrace of pid %d:", pid);

+

+	if(use_syslog == 0)

+		cli_dbgmsg("Backtrace of pid %d:\n", pid);

+	else 

+		syslog(LOG_ERR, "Backtrace of pid %d:", pid);

+

+	for(i = 0; i < size; i++)

+		if(use_syslog)

+			syslog(LOG_ERR, "bt[%d]: %s", i, strings[i]);

+		else

+			cli_dbgmsg("%s\n", strings[i]);

+

+	free(strings);

+#endif

+}

+#endif

@@ -17,6 +17,9 @@

  *

  * Change History:

  * $Log: message.c,v $

+ * Revision 1.61  2004/06/22 04:08:02  nigelhorne

+ * Optimise empty lines

+ *

  * Revision 1.60  2004/06/16 08:07:39  nigelhorne

  * Added thread safety

  *

@@ -177,7 +180,7 @@

  * uuencodebegin() no longer static

  *

  */

-static	char	const	rcsid[] = "$Id: message.c,v 1.60 2004/06/16 08:07:39 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: message.c,v 1.61 2004/06/22 04:08:02 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -244,7 +247,7 @@ static	const	struct	encoding_map {

 	{	"7bit",			NOENCODING	},

 	{	"text/plain",		NOENCODING	},

 	{	"quoted-printable",	QUOTEDPRINTABLE	},	/* rfc1522 */

-	{	"base64",		BASE64		},

+	{	"base64",		BASE64		},	/* rfc2045 */

 	{	"8bit",			EIGHTBIT	},

 	{	"8 bit",		EIGHTBIT	},	/* incorrect */

 	{	"x-uuencode",		UUENCODE	},

@@ -792,23 +795,21 @@ messageAddLine(message *m, const char *line, int takeCopy)

 	m->body_last->t_next = NULL;

-	if(takeCopy) {

-		m->body_last->t_text = strdup((line) ? line : "");

-		if(m->body_last->t_text == NULL) {

-			cli_errmsg("messageAddLine: out of memory\n");

-			return -1;

-		}

-		assert(m->body_last->t_text != NULL);

-	} else {

-		assert(line != NULL);

-		m->body_last->t_text = (char *)line;

-	}

-

-	/*

-	 * See if this line marks the start of a non MIME inclusion that

-	 * will need to be scanned

-	 */

 	if(line && *line) {

+		if(takeCopy) {

+			m->body_last->t_text = strdup(line);

+			if(m->body_last->t_text == NULL) {

+				cli_errmsg("messageAddLine: out of memory\n");

+				return -1;

+			}

+			// cli_chomp(m->body_last->t_text);

+		} else

+			m->body_last->t_text = (char *)line;

+

+		/*

+		 * See if this line marks the start of a non MIME inclusion that

+		 * will need to be scanned

+		 */

 		if((m->encoding == NULL) &&

 		   (strncasecmp(line, encoding, sizeof(encoding) - 1) == 0) &&

 		   (strstr(line, "7bit") == NULL))

@@ -826,7 +827,9 @@ messageAddLine(message *m, const char *line, int takeCopy)

 			(isdigit(line[8])) &&

 			(line[9] == ' ')))

 				m->uuencode = m->body_last;

-	}

+	} else

+		m->body_last->t_text = NULL;

+

 	return 1;

 }

@@ -961,7 +964,8 @@ messageToBlob(message *m)

 		 * See RFC1741

 		 */

 		while((t_line = t_line->t_next) != NULL)

-			blobAddData(tmp, (unsigned char *)t_line->t_text, strlen(t_line->t_text));

+			if(t_line->t_text)

+				blobAddData(tmp, (unsigned char *)t_line->t_text, strlen(t_line->t_text));

 		data = blobGetData(tmp);

@@ -1324,13 +1328,12 @@ messageToText(const message *m)

 				last = last->t_next;

 			}

-			if(last)

-				last->t_text = strdup((char *)data);

-

-			if((last == NULL) || (last->t_text == NULL))

+			if(last == NULL)

 				break;

-			if(messageGetEncoding(m) == BASE64)

+			last->t_text = data[0] ? strdup((char *)data) : NULL;

+

+			if(line && messageGetEncoding(m) == BASE64)

 				if(strchr(line, '='))

 					break;

 		}

@@ -1471,7 +1474,6 @@ decodeLine(const message *m, const char *line, unsigned char *buf, size_t buflen

 	char *copy;

 	assert(m != NULL);

-	assert(line != NULL);

 	assert(buf != NULL);

 	switch(messageGetEncoding(m)) {

@@ -1483,11 +1485,20 @@ decodeLine(const message *m, const char *line, unsigned char *buf, size_t buflen

 		case NOENCODING:

 		case EIGHTBIT:

 		default:	/* unknown encoding type - try our best */

+			if(line == NULL) {	/* empty line */

+				*buf++ = '\n';

+				break;

+			}

 			buf = (unsigned char *)strrcpy((char *)buf, line);

 			/* Put the new line back in */

 			return (unsigned char *)strrcpy((char *)buf, "\n");

 		case QUOTEDPRINTABLE:

+			if(line == NULL) {	/* empty line */

+				*buf++ = '\n';

+				break;

+			}

+			

 			softbreak = FALSE;

 			while(*line) {

 				if(*line == '=') {

@@ -1523,6 +1534,8 @@ decodeLine(const message *m, const char *line, unsigned char *buf, size_t buflen

 			break;

 		case BASE64:

+			if(line == NULL)

+				break;

 			/*

 			 * RFC1521 sets the maximum length to 76 bytes

 			 * but many e-mail clients ignore that

@@ -1537,14 +1550,14 @@ decodeLine(const message *m, const char *line, unsigned char *buf, size_t buflen

 			/*

 			 * Klez doesn't always put "=" on the last line

 			 */

-			/*buf = decode(line, buf, base64, p2 == NULL);*/

-			buf = decode(copy, buf, base64, FALSE);

+			buf = decode(copy, buf, base64, (p2 == NULL) && ((strlen(copy) & 3) == 0));

+			/*buf = decode(copy, buf, base64, FALSE);*/

 			free(copy);

 			break;

 		case UUENCODE:

-			if(*line == '\0')	/* empty line */

+			if((line == NULL) || (*line == '\0'))	/* empty line */

 				break;

 			if(strncasecmp(line, "begin ", 6) == 0)

 				break;

@@ -1580,6 +1593,7 @@ decode(const char *in, unsigned char *out, unsigned char (*decoder)(char), bool

 {

 	unsigned char b1, b2, b3, b4;

+	/*cli_dbgmsg("decode %s (len %d ifFast %d)\n", in, strlen(in), isFast);*/

 	if(isFast)

 		/* Fast decoding if not last line */

 		while(*in) {

@@ -1609,23 +1623,17 @@ decode(const char *in, unsigned char *out, unsigned char (*decoder)(char), bool

 				b2 = '\0';

 				nbytes = 1;

 			} else {

-				assert(*in != '\0');

-

 				b2 = (*decoder)(*in++);

 				if(*in == '\0') {

 					b3 = '\0';

 					nbytes = 2;

 				} else {

-					assert(*in != '\0');

-

 					b3 = (*decoder)(*in++);

 					if(*in == '\0') {

 						b4 = '\0';

 						nbytes = 3;

 					} else {

-						assert(*in != '\0');

-

 						b4 = (*decoder)(*in++);

 						nbytes = 4;

 					}

@@ -16,6 +16,9 @@

  *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

  *

  * $Log: text.c,v $

+ * Revision 1.7  2004/06/22 04:08:02  nigelhorne

+ * Optimise empty lines

+ *

  * Revision 1.6  2004/05/05 09:37:52  nigelhorne

  * Removed textClean - not needed in clamAV

  *

@@ -27,7 +30,7 @@

  *

  */

-static	char	const	rcsid[] = "$Id: text.c,v 1.6 2004/05/05 09:37:52 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: text.c,v 1.7 2004/06/22 04:08:02 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -56,7 +59,8 @@ textDestroy(text *t_head)

 {

 	while(t_head) {

 		text *t_next = t_head->t_next;

-		free(t_head->t_text);

+		if(t_head->t_text)

+			free(t_head->t_text);

 		free(t_head);

 		t_head = t_next;

 	}

@@ -147,9 +151,11 @@ textCopy(const text *t_head)

 		assert(last != NULL);

-		last->t_text = strdup(t_head->t_text);

-

-		assert(last->t_text != NULL);

+		if(t_head->t_text) {

+			last->t_text = strdup(t_head->t_text);

+			assert(last->t_text != NULL);

+		} else

+			last->t_text = NULL;

 		t_head = t_head->t_next;

 	}

@@ -183,9 +189,11 @@ textAdd(text *t_head, const text *t)

 		assert(t_head != NULL);

-		t_head->t_text = strdup(t->t_text);

-

-		assert(t_head->t_text != NULL);

+		if(t->t_text) {

+			t_head->t_text = strdup(t->t_text);

+			assert(t_head->t_text != NULL);

+		} else

+			t_head->t_text = NULL;

 		t = t->t_next;

 	}

@@ -237,12 +245,16 @@ textToBlob(const text *t, blob *b)

 	}

 	for(t1 = t; t1; t1 = t1->t_next)

-		s += strlen(t1->t_text) + 1;

+		if(t1->t_text)

+			s += strlen(t1->t_text) + 1;

+		else

+			s++;

 	blobGrow(b, s);

 	do {

-		blobAddData(b, (unsigned char *)t->t_text, strlen(t->t_text));

+		if(t->t_text)

+			blobAddData(b, (unsigned char *)t->t_text, strlen(t->t_text));

 		blobAddData(b, (unsigned char *)"\n", 1);

 	} while((t = t->t_next) != NULL);

@@ -14,10 +14,15 @@

  *  You should have received a copy of the GNU General Public License

  *  along with this program; if not, write to the Free Software

  *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

+ *

+ * $Log: text.h,v $

+ * Revision 1.3  2004/06/22 04:08:02  nigelhorne

+ * Optimise empty lines

+ *

  */

 typedef struct text {

-	char	*t_text;

+	char	*t_text;	/* NULL if the line is empty */

 	struct	text	*t_next;

 } text;