@@ -1,3 +1,13 @@

+Sun Feb 15 08:50:21 GMT 2004 (njh)

+----------------------------------

+  * libclamav:	added blobClose and blobCmp to reduce likelyhood of scanning the

+			same file twice and reducing the runtime memory usage

+	support video within multiparts

+	better warning of unsupported types that may need to be added later

+	removed some unneeded recursiveness that resulted in some needless

+		scanning twice of the same attachment on messages without

+		any text content

+

 Sun Feb 15 02:03:47 CET 2004 (tk)

 ---------------------------------

   * freshclam: use O_BINARY when saving database under Cygwin

@@ -16,11 +16,14 @@

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

  *

  * $Log: blob.c,v $

+ * Revision 1.7  2004/02/15 08:45:53  nigelhorne

+ * Avoid scanning the same file twice

+ *

  * Revision 1.6  2004/02/10 19:23:54  nigelhorne

  * Change LOG to Log

  *

  */

-static	char	const	rcsid[] = "$Id: blob.c,v 1.6 2004/02/10 19:23:54 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: blob.c,v 1.7 2004/02/15 08:45:53 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -139,6 +142,14 @@ blobAddData(blob *b, const unsigned char *data, size_t len)

 	if(len == 0)

 		return;

+	if(b->isClosed) {

+		/*

+		 * Should be cli_dbgmsg, but I want to see them for now,

+		 * and cli_dbgmsg doesn't support debug levels

+		 */

+		cli_warnmsg("Reopening closed blob\n");

+		b->isClosed = 0;

+	}

 	if(b->data == NULL) {

 		assert(b->len == 0);

 		b->size = len * 4;

@@ -171,3 +182,37 @@ blobGetDataSize(const blob *b)

 	return(b->len);

 }

+

+void

+blobClose(blob *b)

+{

+	b->isClosed = 1;

+

+	if(b->size != b->len) {

+		b->size = b->len;

+		b->data = cli_realloc(b->data, b->size);

+	}

+}

+

+/*

+ * Returns 0 if the blobs are the same

+ */

+int

+blobcmp(const blob *b1, const blob *b2)

+{

+	unsigned long s1, s2;

+

+	assert(b1 != NULL);

+	assert(b2 != NULL);

+

+	if(b1 == b2)

+		return 0;

+

+	s1 = blobGetDataSize(b1);

+	s2 = blobGetDataSize(b2);

+

+	if(s1 != s2)

+		return 1;

+

+	return memcmp(blobGetData(b1), blobGetData(b2), s1);

+}

@@ -23,6 +23,7 @@ typedef struct blob {

 	unsigned	char	*data;	/* the stuff itself */

 	unsigned	long	len;	/* number of bytes of data so far */

 	unsigned	long	size;	/* number of bytes allocated to data so far */

+	int	isClosed;

 #ifdef	CL_DEBUG

 	object_type	magic;	/* verify that this is a blob */

 #endif

@@ -36,3 +37,5 @@ const	char	*blobGetFilename(const blob *b);

 void	blobAddData(blob *b, const unsigned char *data, size_t len);

 unsigned char *blobGetData(const blob *b);

 unsigned	long	blobGetDataSize(const blob *b);

+void	blobClose(blob *b);

+int	blobcmp(const blob *b1, const blob *b2);

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

  *

  * Change History:

  * $Log: mbox.c,v $

+ * Revision 1.44  2004/02/15 08:45:54  nigelhorne

+ * Avoid scanning the same file twice

+ *

  * Revision 1.43  2004/02/14 19:04:05  nigelhorne

  * Handle spaces in boundaries

  *

@@ -120,7 +123,7 @@

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

  *

  */

-static	char	const	rcsid[] = "$Id: mbox.c,v 1.43 2004/02/14 19:04:05 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: mbox.c,v 1.44 2004/02/15 08:45:54 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -548,7 +551,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 	/* Anything left to be parsed? */

 	if(mainMessage && (messageGetBody(mainMessage) != NULL)) {

-		int numberOfAttachments = 0;

+		int numberOfAttachments = 0, numberOfNewAttachments;

 		mime_type mimeType;

 		const char *mimeSubtype;

 		const text *t_line;

@@ -988,6 +991,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						continue;

 					case AUDIO:

 					case IMAGE:

+					case VIDEO:

 						/*

 						 * TODO: it may be nice to

 						 * have an option to throw

@@ -997,7 +1001,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						addAttachment = TRUE;

 						break;

 					default:

-						cli_dbgmsg("Only text and application attachments are supported, type = %d\n",

+						cli_warnmsg("Only text and application attachments are supported, type = %d\n",

 							messageGetMimeType(aMessage));

 						continue;

 					}

@@ -1016,8 +1020,8 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						if(aBlob) {

 							assert(blobGetFilename(aBlob) != NULL);

-							/*if(blobGetDataSize(aBlob) > 0)*/

-								blobList[numberOfAttachments++] = aBlob;

+							blobClose(aBlob);

+							blobList[numberOfAttachments++] = aBlob;

 						}

 					}

 				}

@@ -1027,18 +1031,44 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 					rc = parseEmailBody(NULL, NULL, 0, aText, dir, rfc821Table, subtypeTable);

 					break;

 				}

+

 				/*

 				 * Store any existing attachments at the end of

 				 * the list we've just built up

 				 */

+				numberOfNewAttachments = 0;

 				for(i = 0; i < nBlobs; i++) {

+					int j;

 #ifdef	CL_DEBUG

 					assert(blobs[i]->magic == BLOB);

 #endif

-					blobList[numberOfAttachments++] = blobs[i];

+					for(j = 0; j < numberOfAttachments; j++)

+						if(blobcmp(blobs[i], blobList[j]) == 0)

+							break;

+					if(j >= numberOfAttachments) {

+						assert(numberOfAttachments < MAX_ATTACHMENTS);

+						cli_dbgmsg("Attaching %s to list of blobs\n",

+							blobGetFilename(blobs[i]));

+						blobClose(blobs[i]);

+						blobList[numberOfAttachments++] = blobs[i];

+						numberOfNewAttachments++;

+					} else {

+						cli_warnmsg("Don't scan the same file twice as '%s' and '%s'\n",

+							blobGetFilename(blobs[i]),

+							blobGetFilename(blobList[j]));

+						blobDestroy(blobs[i]);

+					}

 				}

 				/*

+				 * If we've found nothing new save what we have

+				 * and quit - that's this part all done.

+				 */

+				if(numberOfNewAttachments == 0) {

+					rc = parseEmailBody(NULL, blobList, numberOfAttachments, NULL, dir, rfc821Table, subtypeTable);

+					break;

+				}

+				/*

 				 * If there's only one part of the MULTIPART

 				 * we already have the body to decode so

 				 * there's no more work to do.

@@ -1199,6 +1229,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

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

 						if(blobs[i] == NULL)

 							break;

+					blobClose(aBlob);

 					blobs[i] = aBlob;

 					if(i == nBlobs) {

 						nBlobs++;