@@ -1,3 +1,9 @@

+Sat Aug 21 12:59:43 BST 2004 (njh)

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

+  * libclamav:	Changed the handling of miltipart messages, that is scanning

+  			emails with attachments. Reports on impact on memory

+			usage and speed welcome to clamav-devel@clamav.net

+

 Fri Aug 20 21:05:04 CEST 2004 (tk)

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

   * libclamav/Makefile.am: add line.[ch]

@@ -16,12 +16,15 @@

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

  *

  * $Log: line.c,v $

+ * Revision 1.2  2004/08/21 11:57:57  nigelhorne

+ * Use line.[ch]

+ *

  * Revision 1.1  2004/08/20 11:58:20  nigelhorne

  * First draft

  *

  */

-static	char	const	rcsid[] = "$Id: line.c,v 1.1 2004/08/20 11:58:20 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: line.c,v 1.2 2004/08/21 11:57:57 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -33,16 +36,17 @@ static	char	const	rcsid[] = "$Id: line.c,v 1.1 2004/08/20 11:58:20 nigelhorne Ex

 #include "line.h"

 #include "others.h"

+#ifdef	OLD

 line_t *

 lineCreate(const char *data)

 {

-	line_t *ret = cli_malloc(sizeof(struct line));

+	line_t *ret = (line_t *)li_malloc(sizeof(struct line));

 	if(ret == NULL)

 		return NULL;

-	ret->l_data = strdup(data);

-	if(ret->l_data == NULL) {

+	ret->l_str = strdup(data);

+	if(ret->l_str == NULL) {

 		free(ret);

 		return NULL;

 	}

@@ -61,8 +65,53 @@ lineLink(line_t *line)

 line_t *

 lineUnlink(line_t *line)

 {

+	/*printf("%d:\n\t'%s'\n", line->l_refs, line->l_str);*/

+

 	if(--line->l_refs == 0) {

-		free(line->l_data);

+		free(line->l_str);

+		free(line);

+		return NULL;

+	}

+	return line;

+}

+

+const char *

+lineGetData(const line_t *line)

+{

+	return line ? line->l_str : NULL;

+}

+#else

+line_t *

+lineCreate(const char *data)

+{

+	line_t *ret = (line_t *)cli_malloc(strlen(data) + 2);

+

+	if(ret == NULL)

+		return NULL;

+

+	ret[0] = (char)1;

+	strcpy(&ret[1], data);

+

+	return ret;

+}

+

+line_t *

+lineLink(line_t *line)

+{

+	if(line[0] == 127) {

+		cli_warnmsg("lineLink: linkcount too large\n");

+		return NULL;

+	}

+	line[0]++;

+	return line;

+}

+

+line_t *

+lineUnlink(line_t *line)

+{

+	/*printf("%d:\n\t'%s'\n", (int)line[0], &line[1]);*/

+

+	if(--line[0] == 0) {

 		free(line);

 		return NULL;

 	}

@@ -72,5 +121,6 @@ lineUnlink(line_t *line)

 const char *

 lineGetData(const line_t *line)

 {

-	return line ? line->l_data : NULL;

+	return line ? &line[1] : NULL;

 }

+#endif

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

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

  *

  * $Log: line.h,v $

+ * Revision 1.3  2004/08/21 11:57:57  nigelhorne

+ * Use line.[ch]

+ *

  * Revision 1.2  2004/08/20 19:06:45  kojm

  * add line.[ch]

  *

@@ -27,10 +30,16 @@

 #ifndef __LINE_H

 #define __LINE_H

+#ifdef	OLD

+/* easier to read, but slower */

+

 typedef struct line {

-	char	*l_data;	/* the line's contents */

+	char	*l_str;	/* the line's contents */

 	unsigned int	l_refs;	/* the number of references to the data */

 } line_t;

+#else

+typedef	char	line_t;	/* first byte is the ref count */

+#endif

 line_t	*lineCreate(const char *data);

 line_t	*lineLink(line_t *line);

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

  *

  * Change History:

  * $Log: mbox.c,v $

+ * Revision 1.108  2004/08/21 11:57:57  nigelhorne

+ * Use line.[ch]

+ *

  * Revision 1.107  2004/08/20 04:55:07  nigelhorne

  * FOLLOWURL

  *

@@ -309,7 +312,7 @@

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

  *

  */

-static	char	const	rcsid[] = "$Id: mbox.c,v 1.107 2004/08/20 04:55:07 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: mbox.c,v 1.108 2004/08/21 11:57:57 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -347,6 +350,7 @@ static	char	const	rcsid[] = "$Id: mbox.c,v 1.107 2004/08/20 04:55:07 nigelhorne

 #include "table.h"

 #include "mbox.h"

 #include "blob.h"

+#include "line.h"

 #include "text.h"

 #include "message.h"

 #include "others.h"

@@ -409,9 +413,9 @@ typedef enum	{ FALSE = 0, TRUE = 1 } bool;

 #undef	WITH_CURL

 #endif

-static	message	*parseEmailHeaders(message *m, const table_t *rfc821Table, bool destroy);

+static	message	*parseEmailHeaders(const message *m, const table_t *rfc821Table);

 static	int	parseEmailHeader(message *m, const char *line, const table_t *rfc821Table);

-static	int	parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, const char *dir, table_t *rfc821Table, table_t *subtypeTable, unsigned int options);

+static	int	parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, const char *dir, const table_t *rfc821Table, const table_t *subtypeTable, unsigned int options);

 static	int	boundaryStart(const char *line, const char *boundary);

 static	int	endOfMessage(const char *line, const char *boundary);

 static	int	initialiseTables(table_t **rfc821Table, table_t **subtypeTable);

@@ -615,7 +619,7 @@ cli_mbox(const char *dir, int desc, unsigned int options)

 				/*

 				 * End of a message in the mail box

 				 */

-				body = parseEmailHeaders(m, rfc821, TRUE);

+				body = parseEmailHeaders(m, rfc821);

 				if(body == NULL) {

 					messageReset(m);

 					continue;

@@ -637,7 +641,7 @@ cli_mbox(const char *dir, int desc, unsigned int options)

 				cli_dbgmsg("Finished processing message\n");

 			} else

 				lastLineWasEmpty = (bool)(buffer[0] == '\0');

-			if(messageAddLine(m, buffer, 1) < 0)

+			if(messageAddStr(m, buffer) < 0)

 				break;

 		} while(fgets(buffer, sizeof(buffer), fd) != NULL);

@@ -655,27 +659,25 @@ cli_mbox(const char *dir, int desc, unsigned int options)

 		 * FIXME: files full of new lines and nothing else are

 		 * handled ungracefully...

 		 */

-		do

+		do {

 			/*

-			 * No need to preprocess such as cli_chomp() since

-			 * that'll be done by parseEmailHeaders()

-			 *

 			 * TODO: this needlessly creates a message object,

 			 * it'd be better if parseEmailHeaders could also

 			 * read in from a file. I do not want to lump the

 			 * parseEmailHeaders code here, that'd be a duplication

 			 * of code I want to avoid

 			 */

-			if(messageAddLine(m, buffer, 1) < 0)

+			(void)cli_chomp(buffer);

+			if(messageAddStr(m, buffer) < 0)

 				break;

-		while(fgets(buffer, sizeof(buffer), fd) != NULL);

+		} while(fgets(buffer, sizeof(buffer), fd) != NULL);

 	}

 	fclose(fd);

 	retcode = 0;

-	body = parseEmailHeaders(m, rfc821, TRUE);

+	body = parseEmailHeaders(m, rfc821);

 	messageDestroy(m);

 	if(body) {

 		/*

@@ -711,14 +713,12 @@ cli_mbox(const char *dir, int desc, unsigned int options)

  * of the message in memory, the upside is that it makes for easier parsing

  * of encapsulated messages, and in the long run uses less memory in those

  * scenarios

- * BUT: if 'destroy' is set, the caller has given us a hint than 'm' will

- * not be used again before it is destroyed, so we can trash it

  */

 static message *

-parseEmailHeaders(message *m, const table_t *rfc821, bool destroy)

+parseEmailHeaders(const message *m, const table_t *rfc821)

 {

 	bool inHeader = TRUE;

-	text *t;

+	const text *t;

 	message *ret;

 	bool anyHeadersFound = FALSE;

@@ -729,26 +729,15 @@ parseEmailHeaders(message *m, const table_t *rfc821, bool destroy)

 	ret = messageCreate();

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

-		char *buffer;

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

+		const char *buffer;

 #ifdef CL_THREAD_SAFE

 		char *strptr;

 #endif

-		if(t->t_text) {

-			if(destroy) {

-				buffer = t->t_text;

-				t->t_text = NULL;

-			} else {

-				buffer = strdup(t->t_text);

-				if(buffer == NULL)

-					break;

-			}

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

-				free(buffer);

-				buffer = NULL;

-			}

-		} else

+		if(t->t_line)

+			buffer = lineGetData(t->t_line);

+		else

 			buffer = NULL;

 		/*

@@ -760,13 +749,12 @@ parseEmailHeaders(message *m, const table_t *rfc821, bool destroy)

 			/*

 			 * Add all the arguments on the line

 			 */

-			if(buffer) {

-				const char *ptr;

+			const char *ptr;

+			char *copy = strdup(buffer);

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

-					messageAddArgument(ret, ptr);

-				free(buffer);

-			}

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

+				messageAddArgument(ret, ptr);

+			free(copy);

 		} else if(inHeader) {

 			/*

 			 * A blank line signifies the end of the header and

@@ -779,11 +767,10 @@ parseEmailHeaders(message *m, const table_t *rfc821, bool destroy)

 				if((parseEmailHeader(ret, buffer, rfc821) >= 0) ||

 				   (strncasecmp(buffer, "From ", 5) == 0))

 					anyHeadersFound = TRUE;

-				free(buffer);

 			}

 		} else {

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

-			if(messageAddLine(ret, buffer, 0) < 0)

+			if(messageAddLine(ret, t->t_line) < 0)

 				break;

 		}

 	}

@@ -861,7 +848,7 @@ parseEmailHeader(message *m, const char *line, const table_t *rfc821)

  *	2 for success, attachments not saved

  */

 static int	/* success or fail */

-parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, const char *dir, table_t *rfc821Table, table_t *subtypeTable, unsigned int options)

+parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, const char *dir, const table_t *rfc821Table, const table_t *subtypeTable, unsigned int options)

 {

 	message **messages;	/* parts of a multipart message */

 	int inhead, inMimeHead, i, rc = 1, htmltextPart, multiparts = 0;

@@ -965,7 +952,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 			}

 			do

-				if(boundaryStart(t_line->t_text, boundary))

+				if(boundaryStart(lineGetData(t_line->t_line), boundary))

 					break;

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

@@ -1017,8 +1004,9 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 				 * but some viruses insert them

 				 */

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

-					if(t_line->t_text &&

-					   (cli_chomp(t_line->t_text) > 0))

+					if(t_line->t_line &&

+					   /*(cli_chomp(t_line->t_text) > 0))*/

+					   (strlen(lineGetData(t_line->t_line)) > 0))

 						break;

 				if(t_line == NULL) {

@@ -1038,7 +1026,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 				}

 				do {

-					const char *line = t_line->t_text;

+					const char *line = lineGetData(t_line->t_line);

 					/*cli_dbgmsg("inMimeHead %d inhead %d boundary %s line '%s' next '%s'\n",

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

@@ -1115,11 +1103,16 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						 * put white space after the ;

 						 */

 						inMimeHead = continuationMarker(line);

-						if(!inMimeHead)

-							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;

+						if(!inMimeHead) {

+							const text *next = t_line->t_next;

+

+							if(next && next->t_line) {

+								const char *data = lineGetData(next->t_line);

+

+								if((data[0] == '\t') || (data[0] == ' '))

+									inMimeHead = TRUE;

+							}

+						}

 						parseEmailHeader(aMessage, line, rfc821Table);

 					} else if(boundaryStart(line, boundary)) {

@@ -1137,7 +1130,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						/* t_line = NULL;*/

 						break;

 					} else {

-						if(messageAddLine(aMessage, line, 1) < 0)

+						if(messageAddLine(aMessage, t_line->t_line) < 0)

 							break;

 						lines++;

 					}

@@ -1342,7 +1335,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 							/*

 							 * No plain text version

 							 */

-							messageAddLine(aMessage, "No plain text alternative", 1);

+							messageAddStr(aMessage, "No plain text alternative");

 						assert(messageGetBody(aMessage) != NULL);

 						break;

 					case TEXT:

@@ -1404,7 +1397,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 							messages[i] = NULL;

 							continue;

 						}

-						messageAddLineAtTop(aMessage,

+						messageAddStrAtTop(aMessage,

 							"Received: by clamd");

 #ifdef	SAVE_TO_DISC

 						/*

@@ -1447,27 +1440,6 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 						 * be an attachment

 						 */

 						cli_dbgmsg("Found multipart inside multipart\n");

-#if	0

-						if(aMessage) {

-							body = parseEmailHeaders(aMessage, rfc821Table, TRUE);

-							if(body) {

-								assert(aMessage == messages[i]);

-								messageDestroy(messages[i]);

-								messages[i] = NULL;

-

-								if(mainMessage && (mainMessage != messageIn))

-									messageDestroy(mainMessage);

-

-								/*t = messageToText(body);

-								rc = parseEmailBody(body, blobs, nBlobs, t, dir, rfc821Table, subtypeTable, options);*/

-								rc = parseEmailBody(body, blobs, nBlobs, aText, dir, rfc821Table, subtypeTable, options);

-								/*textDestroy(t);*/

-

-								cli_dbgmsg("Finished recursion\n");

-

-								mainMessage = body;

-							}

-#else

 						if(aMessage) {

 							/*

 							 * The headers were parsed when reading in the

@@ -1478,7 +1450,6 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 							assert(aMessage == messages[i]);

 							messageDestroy(messages[i]);

 							messages[i] = NULL;

-#endif

 						} else {

 							rc = parseEmailBody(NULL, blobs, nBlobs, NULL, dir, rfc821Table, subtypeTable, options);

 							if(mainMessage && (mainMessage != messageIn))

@@ -1663,7 +1634,7 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 			}

 			if((strcasecmp(mimeSubtype, "rfc822") == 0) ||

 			   (strcasecmp(mimeSubtype, "delivery-status") == 0)) {

-				message *m = parseEmailHeaders(mainMessage, rfc821Table, FALSE);

+				message *m = parseEmailHeaders(mainMessage, rfc821Table);

 				if(m) {

 					cli_dbgmsg("Decode rfc822");

@@ -1810,12 +1781,15 @@ parseEmailBody(message *messageIn, blob **blobsIn, int nBlobs, text *textIn, con

 				 * must remain otherwise non bounce messages

 				 * won't be scanned

 				 */

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

-					if(t->t_text &&

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

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

-					   (strstr(t->t_text, "8bit") == NULL))

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

+					const char *txt = lineGetData(t->t_line);

+

+					if(txt &&

+					   (strncasecmp(txt, encoding, sizeof(encoding) - 1) == 0) &&

+					   (strstr(txt, "7bit") == NULL) &&

+					   (strstr(txt, "8bit") == NULL))

 						break;

+				}

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

 					cli_dbgmsg("Found a bounce message\n");

@@ -2399,14 +2373,14 @@ checkURLs(message *m, const char *dir)

 				continue;

 			if(*p2 == '\0')

 				continue;

-			if(n == MAX_URLS) {

-				cli_warnmsg("Not all URLs will be scanned\n");

-				break;

-			}

 			if(tableFind(t, p2) == 1) {

 				cli_dbgmsg("URL %s already downloaded\n", p2);

 				continue;

 			}

+			if(n == MAX_URLS) {

+				cli_warnmsg("Not all URLs will be scanned\n");

+				break;

+			}

 			(void)tableInsert(t, p2, 1);

 			cli_dbgmsg("Downloading URL %s to be scanned\n", p2);

 			strncpy(name, p2, sizeof(name));

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

  *

  * Change History:

  * $Log: message.c,v $

+ * Revision 1.72  2004/08/21 11:57:57  nigelhorne

+ * Use line.[ch]

+ *

  * Revision 1.71  2004/08/13 09:28:16  nigelhorne

  * Remove incorrect comment style

  *

@@ -210,7 +213,7 @@

  * uuencodebegin() no longer static

  *

  */

-static	char	const	rcsid[] = "$Id: message.c,v 1.71 2004/08/13 09:28:16 nigelhorne Exp $";

+static	char	const	rcsid[] = "$Id: message.c,v 1.72 2004/08/21 11:57:57 nigelhorne Exp $";

 #if HAVE_CONFIG_H

 #include "clamav-config.h"

@@ -240,6 +243,7 @@ static	char	const	rcsid[] = "$Id: message.c,v 1.71 2004/08/13 09:28:16 nigelhorn

 #include <pthread.h>

 #endif

+#include "line.h"

 #include "mbox.h"

 #include "table.h"

 #include "blob.h"

@@ -259,6 +263,7 @@ static	char	const	rcsid[] = "$Id: message.c,v 1.71 2004/08/13 09:28:16 nigelhorn

 typedef enum { FALSE = 0, TRUE = 1 } bool;

+static	void	messageIsEncoding(message *m);

 static	unsigned char	*decodeLine(message *m, const char *line, unsigned char *buf, size_t buflen);

 static unsigned char *decode(message *m, const char *in, unsigned char *out, unsigned char (*decoder)(char), bool isFast);

 static	void	squeeze(char *s);

@@ -814,16 +819,41 @@ messageGetEncoding(const message *m)

 	return(m->encodingType);

 }

+int

+messageAddLine(message *m, line_t *line)

+{

+	assert(m != NULL);

+

+	if(m->body_first == NULL)

+		m->body_last = m->body_first = (text *)cli_malloc(sizeof(text));

+	else {

+		m->body_last->t_next = (text *)cli_malloc(sizeof(text));

+		m->body_last = m->body_last->t_next;

+	}

+

+	if(m->body_last == NULL)

+		return -1;

+

+	m->body_last->t_next = NULL;

+

+	if(line && lineGetData(line)) {

+		m->body_last->t_line = lineLink(line);

+

+		messageIsEncoding(m);

+	} else

+		m->body_last->t_line = NULL;

+

+	return 1;

+}

+

 /*

  * Add the given line to the end of the given message

  * If needed a copy of the given line is taken which the caller must free

  * Line must not be terminated by a \n

  */

 int

-messageAddLine(message *m, const char *line, int takeCopy)

+messageAddStr(message *m, const char *data)

 {

-	static const char encoding[] = "Content-Transfer-Encoding";

-	static const char binhex[] = "(This file must be converted with BinHex 4.0)";

 	assert(m != NULL);

 	if(m->body_first == NULL)

@@ -838,40 +868,18 @@ messageAddLine(message *m, const char *line, int takeCopy)

 	m->body_last->t_next = NULL;

-	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;

+	if(data && *data) {

+		m->body_last->t_line = lineCreate(data);

-		/*

-		 * 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))

-			m->encoding = m->body_last;

-		else if(/*(m->bounce == NULL) &&*/

-			(cli_filetype(line, strlen(line)) == CL_MAILFILE))

-				m->bounce = m->body_last;

-		else if((m->binhex == NULL) &&

-			(strncasecmp(line, binhex, sizeof(binhex) - 1) == 0))

-				m->binhex = m->body_last;

-		else if((m->uuencode == NULL) &&

-			((strncasecmp(line, "begin ", 6) == 0) &&

-			(isdigit(line[6])) &&

-			(isdigit(line[7])) &&

-			(isdigit(line[8])) &&

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

-				m->uuencode = m->body_last;

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

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

+			return -1;

+		}

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

+

+		messageIsEncoding(m);

 	} else

-		m->body_last->t_text = NULL;

+		m->body_last->t_line = NULL;

 	return 1;

 }

@@ -882,14 +890,14 @@ messageAddLine(message *m, const char *line, int takeCopy)

  * Line must not be terminated by a \n

  */

 int

-messageAddLineAtTop(message *m, const char *line)

+messageAddStrAtTop(message *m, const char *data)

 {

 	text *oldfirst;

 	assert(m != NULL);

 	if(m->body_first == NULL)

-		return messageAddLine(m, line, 1);

+		return messageAddLine(m, lineCreate(data));

 	oldfirst = m->body_first;

 	m->body_first = (text *)cli_malloc(sizeof(text));

@@ -899,16 +907,46 @@ messageAddLineAtTop(message *m, const char *line)

 	}

 	m->body_first->t_next = oldfirst;

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

+	m->body_first->t_line = lineCreate((data) ? data : "");

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

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

+	if(m->body_first->t_line == NULL) {

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

 		return -1;

 	}

 	return 1;

 }

 /*

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

+ * will need to be scanned

+ */

+static void

+messageIsEncoding(message *m)

+{

+	static const char encoding[] = "Content-Transfer-Encoding";

+	static const char binhex[] = "(This file must be converted with BinHex 4.0)";

+	const char *line = lineGetData(m->body_last->t_line);

+

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

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

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

+		m->encoding = m->body_last;

+	else if(/*(m->bounce == NULL) &&*/

+		(cli_filetype(line, strlen(line)) == CL_MAILFILE))

+			m->bounce = m->body_last;

+	else if((m->binhex == NULL) &&

+		(strncasecmp(line, binhex, sizeof(binhex) - 1) == 0))

+			m->binhex = m->body_last;

+	else if((m->uuencode == NULL) &&

+		((strncasecmp(line, "begin ", 6) == 0) &&

+		(isdigit(line[6])) &&

+		(isdigit(line[7])) &&

+		(isdigit(line[8])) &&

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

+			m->uuencode = m->body_last;

+}

+

+/*

  * Returns a pointer to the body of the message. Note that it does NOT return

  * a copy of the data

  */

@@ -961,7 +999,7 @@ messageToBlob(message *m)

 			return NULL;

 		}

-		filename = cli_strtok(t_line->t_text, 2, " ");

+		filename = cli_strtok(lineGetData(t_line->t_line), 2, " ");

 		if(filename == NULL) {

 			cli_dbgmsg("UUencoded attachment sent with no filename\n");

@@ -1013,8 +1051,10 @@ messageToBlob(message *m)

 		 * See RFC1741

 		 */

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

-			if(t_line->t_text)

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

+			if(t_line->t_line) {

+				const char *d = lineGetData(t_line->t_line);

+				blobAddData(tmp, (unsigned char *)d, strlen(d));

+			}

 		data = blobGetData(tmp);

@@ -1239,9 +1279,6 @@ messageToBlob(message *m)

 		return b;

 	} else {

-		/*

-		 * Discard attachments with no filename

-		 */

 		filename = (char *)messageFindArgument(m, "filename");

 		if(filename == NULL) {

 			filename = (char *)messageFindArgument(m, "name");

@@ -1282,7 +1319,7 @@ messageToBlob(message *m)

 	do {

 		unsigned char data[1024];

 		unsigned char *uptr;

-		const char *line = t_line->t_text;

+		const char *line = lineGetData(t_line->t_line);

 		if(messageGetEncoding(m) == UUENCODE) {

 			/*

@@ -1353,14 +1390,12 @@ messageToText(message *m)

 				last = last->t_next;

 			}

-			if((last == NULL) ||

-			   ((last->t_text = strdup(t_line->t_text)) == NULL)) {

-				if(last)

-					free(last);

+			if(last == NULL) {

 				if(first)

 					textDestroy(first);

 				return NULL;

 			}

+			last->t_line = lineLink(t_line->t_line);

 		}

 	else {

 		if(messageGetEncoding(m) == UUENCODE) {

@@ -1380,7 +1415,7 @@ messageToText(message *m)

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

 			unsigned char data[1024];

 			unsigned char *uptr;

-			const char *line = t_line->t_text;

+			const char *line = lineGetData(t_line->t_line);

 			if(messageGetEncoding(m) == BASE64) {

 				/*

@@ -1410,7 +1445,7 @@ messageToText(message *m)

 			if(last == NULL)

 				break;

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

+			last->t_line = ((data[0] != '\n') && data[0]) ? lineCreate((char *)data) : NULL;

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

 				if(strchr(line, '='))

@@ -1418,11 +1453,9 @@ messageToText(message *m)

 		}

 		if(m->base64chars) {

 			unsigned char data[4];

-			unsigned char *ptr;

 			memset(data, '\0', sizeof(data));

-			ptr = decode(m, NULL, data, base64, FALSE);

-			if(ptr) {

+			if(decode(m, NULL, data, base64, FALSE)) {

 				if(first == NULL)

 					first = last = cli_malloc(sizeof(text));

 				else {

@@ -1431,7 +1464,7 @@ messageToText(message *m)

 				}

 				if(last != NULL)

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

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

 			}

 			m->base64chars = 0;

 		}

@@ -1642,10 +1675,11 @@ decodeLine(message *m, const char *line, unsigned char *buf, size_t buflen)

 			if(copy == NULL)

 				break;

+			squeeze(copy);

+

 			p2 = strchr(copy, '=');

 			if(p2)

 				*p2 = '\0';

-			squeeze(copy);

 			/*

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

@@ -1714,7 +1748,6 @@ squeeze(char *s)

  * decoded. After the last line is found, decode will be called with in = NULL

  * to flush these out

  */

-#if	1

 static unsigned char *

 decode(message *m, const char *in, unsigned char *out, unsigned char (*decoder)(char), bool isFast)

 {

@@ -1866,84 +1899,6 @@ decode(message *m, const char *in, unsigned char *out, unsigned char (*decoder)(

 	}

 	return out;

 }

-#else

-static unsigned char *

-decode(message *m, const char *in, unsigned char *out, unsigned char (*decoder)(char), bool isFast)

-{

-	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) {

-			b1 = (*decoder)(*in++);

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

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

-			/*

-			 * Put this line here to help on some compilers which

-			 * can make use of some architecure's ability to

-			 * multiprocess when different variables can be

-			 * updated at the same time - here b3 is used in

-			 * one line, b1/b2 in the next and b4 in the next after

-			 * that, b3 and b4 rely on in but b1/b2 don't

-			 */

-			*out++ = (b1 << 2) | ((b2 >> 4) & 0x3);

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

-			*out++ = (b2 << 4) | ((b3 >> 2) & 0xF);

-			*out++ = (b3 << 6) | (b4 & 0x3F);

-		}

-	else

-		/* Slower decoding for last line */

-		while(*in) {

-			int nbytes;

-

-			b1 = (*decoder)(*in++);

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

-				b2 = '\0';

-				nbytes = 1;

-			} else {

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

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

-					b3 = '\0';

-					nbytes = 2;

-				} else {

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

-

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

-						b4 = '\0';

-						nbytes = 3;

-					} else {

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

-						nbytes = 4;

-					}

-				}

-			}

-

-			switch(nbytes) {

-				case 3:

-					b4 = '\0';

-					/* fall through */

-				case 4:

-					*out++ = (b1 << 2) | ((b2 >> 4) & 0x3);

-					*out++ = (b2 << 4) | ((b3 >> 2) & 0xF);

-					*out++ = (b3 << 6) | (b4 & 0x3F);

-					break;

-				case 2:

-					*out++ = (b1 << 2) | ((b2 >> 4) & 0x3);

-					*out++ = b2 << 4;

-					break;

-				case 1:

-					*out++ = b1 << 2;

-					break;

-				default:

-					assert(0);

-			}

-			if(nbytes != 4)

-				break;

-		}

-	return out;

-}

-#endif

 static unsigned char

 hex(char c)

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

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

  *

  * $Log: message.h,v $

+ * Revision 1.13  2004/08/21 11:57:57  nigelhorne

+ * Use line.[ch]

+ *

  * Revision 1.12  2004/07/20 14:35:29  nigelhorne

  * Some MYDOOM.I were getting through

  *

@@ -83,8 +86,9 @@ void	messageAddArguments(message *m, const char *arg);