b151ef55 |
/* |
e82a5185 |
* Copyright (C) 2002-2006 Nigel Horne <njh@bandsman.co.uk> |
b151ef55 |
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* 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.
*/ |
e82a5185 |
static char const rcsid[] = "$Id: mbox.c,v 1.279 2006/02/06 02:36:39 nigelhorne Exp $"; |
8b242bb9 |
#if HAVE_CONFIG_H
#include "clamav-config.h"
#endif |
b151ef55 |
#ifndef CL_DEBUG |
897fd9c7 |
#define NDEBUG /* map CLAMAV debug onto standard */ |
b151ef55 |
#endif
#ifdef CL_THREAD_SAFE |
f5e9abc8 |
#ifndef _REENTRANT |
b151ef55 |
#define _REENTRANT /* for Solaris 2.8 */
#endif |
f5e9abc8 |
#endif |
b151ef55 |
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <assert.h>
#include <string.h>
#include <strings.h>
#include <ctype.h>
#include <time.h>
#include <fcntl.h> |
0bcad2b1 |
#include <sys/param.h> |
b151ef55 |
#include <clamav.h> |
9a7398ee |
#include <dirent.h> |
67a25177 |
#include <limits.h> |
b151ef55 |
|
a77dc192 |
#if defined(HAVE_READDIR_R_3) || defined(HAVE_READDIR_R_2)
#include <stddef.h>
#endif
|
8a88fb93 |
#ifdef CL_THREAD_SAFE
#include <pthread.h>
#endif
|
b151ef55 |
#include "table.h"
#include "mbox.h"
#include "blob.h" |
de617e3e |
#include "line.h" |
b151ef55 |
#include "text.h"
#include "message.h"
#include "others.h"
#include "defaults.h" |
7fca6080 |
#include "str.h" |
22f3b19b |
#include "filetypes.h" |
e82a5185 |
#include "uuencode.h" |
b151ef55 |
|
98685ac1 |
#ifdef CL_DEBUG
#if __GLIBC__ == 2 && __GLIBC_MINOR__ >= 1
#define HAVE_BACKTRACE
#endif |
89e9a596 |
#endif |
98685ac1 |
#ifdef HAVE_BACKTRACE
#include <execinfo.h>
#include <signal.h>
#include <syslog.h>
static void sigsegv(int sig);
static void print_trace(int use_syslog);
#endif
|
07cbf822 |
#if defined(NO_STRTOK_R) || !defined(CL_THREAD_SAFE) |
b151ef55 |
#undef strtok_r
#undef __strtok_r
#define strtok_r(a,b,c) strtok(a,b)
#endif
/* required for AIX and Tru64 */
#ifdef TRUE
#undef TRUE
#endif
#ifdef FALSE
#undef FALSE
#endif
|
6b93ea0c |
typedef enum { FALSE = 0, TRUE = 1 } bool;
|
e82a5185 |
#ifndef isblank
#define isblank(c) (((c) == ' ') || ((c) == '\t'))
#endif
|
3fa72383 |
#define SAVE_TO_DISC /* multipart/message are saved in a temporary file */ |
49674596 |
|
e9bdeb72 |
/*
* Code does exist to run FOLLORURLS on systems without libcurl, however that
* is not recommended so it is not compiled by default |
cbe29191 |
* |
dd35cb90 |
* On Solaris, when using the GNU C compiler, the clamAV build system uses the
* Sun supplied ld instead of the GNU ld causing an error. Therefore you cannot
* use WITH_CURL on Solaris with gcc, you must configure with
* "--without-libcurl". I don't know if it works with Sun's own compiler
* |
cbe29191 |
* Fails to link on Solaris 10 with this error:
* Undefined first referenced
* symbol in file
* __floatdidf /opt/sfw/lib/libcurl.s |
e9bdeb72 |
*/ |
dd35cb90 |
#if C_SOLARIS && __GNUC__ |
cbe29191 |
#undef WITH_CURL
#endif
|
e9bdeb72 |
#ifdef WITH_CURL |
f52d7358 |
#define FOLLOWURLS 5 /*
* Maximum number of URLs scanned in a message |
e82a5185 |
* part. Helps to find Dialer.gen-45. If |
f52d7358 |
* not defined, don't check any URLs
*/ |
e9bdeb72 |
#endif |
3fa72383 |
|
3eb12bae |
#ifdef FOLLOWURLS |
da812a6a |
|
6da40aa1 |
#include "htmlnorm.h"
|
da812a6a |
#ifdef WITH_CURL /* Set in configure */
/*
* To build with WITH_CURL:
* LDFLAGS=`curl-config --libs` ./configure ...
*/ |
88771ffa |
#include <curl/curl.h> |
6736d46f |
/* |
e82a5185 |
* Needs curl >= 7.11 (I've heard that 7.9 can cause crashes and I have seen
* 7.10 segfault, later versions can be flakey as well) |
6736d46f |
* untested)
*/ |
e82a5185 |
#if (LIBCURL_VERSION_NUM < 0x070B00) |
6736d46f |
#undef WITH_CURL /* also undef FOLLOWURLS? */
#endif
#endif /*WITH_CURL*/
|
3eb12bae |
#else /*!FOLLOWURLS*/
#undef WITH_CURL |
6736d46f |
#endif /*FOLLOWURLS*/ |
88771ffa |
|
9a7398ee |
/* |
c29ebe66 |
* Define this to handle messages covered by section 7.3.2 of RFC1341. |
9a7398ee |
* This is experimental code so it is up to YOU to (1) ensure it's secure |
291ac47f |
* (2) periodically trim the directory of old files
*
* If you use the load balancing feature of clamav-milter to run clamd on |
d85c1fad |
* more than one machine you must make sure that .../partial is on a shared |
291ac47f |
* network filesystem |
9a7398ee |
*/ |
d85c1fad |
#define PARTIAL_DIR |
9a7398ee |
|
273cd2bb |
/*#define NEW_WORLD*/
|
00615ec9 |
static int cli_parse_mbox(const char *dir, int desc, unsigned int options); |
b3a5cdd8 |
static message *parseEmailFile(FILE *fin, const table_t *rfc821Table, const char *firstLine, const char *dir); |
de617e3e |
static message *parseEmailHeaders(const message *m, const table_t *rfc821Table); |
8c0250d5 |
static int parseEmailHeader(message *m, const char *line, const table_t *rfc821Table); |
565c449d |
static int parseEmailBody(message *messageIn, text *textIn, const char *dir, const table_t *rfc821Table, const table_t *subtypeTable, unsigned int options); |
b151ef55 |
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);
static int getTextPart(message *const messages[], size_t size);
static size_t strip(char *buf, int len);
static bool continuationMarker(const char *line);
static int parseMimeHeader(message *m, const char *cmd, const table_t *rfc821Table, const char *arg); |
5a01973c |
static void saveTextPart(message *m, const char *dir); |
0674e2af |
static char *rfc2047(const char *in); |
273cd2bb |
static char *rfc822comments(const char *in, char *out); |
9a7398ee |
#ifdef PARTIAL_DIR
static int rfc1341(message *m, const char *dir);
#endif |
82933497 |
static bool usefulHeader(int commandNumber, const char *cmd); |
c840fa41 |
static char *getline_from_mbox(char *buffer, size_t len, FILE *fin); |
3fa72383 |
|
c5ed8336 |
static void checkURLs(message *m, const char *dir); |
da812a6a |
#ifdef WITH_CURL |
314ff77b |
struct arg { |
a95c894a |
const char *url;
const char *dir; |
314ff77b |
char *filename;
};
#ifdef CL_THREAD_SAFE
static void *getURL(void *a);
#else
static void *getURL(struct arg *arg);
#endif |
3fa72383 |
#endif
|
b151ef55 |
/* Maximum line length according to RFC821 */ |
e82a5185 |
#define RFC2821LENGTH 1000 |
b151ef55 |
/* Hashcodes for our hash tables */
#define CONTENT_TYPE 1
#define CONTENT_TRANSFER_ENCODING 2
#define CONTENT_DISPOSITION 3
/* Mime sub types */
#define PLAIN 1
#define ENRICHED 2
#define HTML 3
#define RICHTEXT 4
#define MIXED 5
#define ALTERNATIVE 6
#define DIGEST 7
#define SIGNED 8
#define PARALLEL 9
#define RELATED 10 /* RFC2387 */
#define REPORT 11 /* RFC1892 */ |
fdc8a467 |
#define APPLEDOUBLE 12 /* Handling of this in only noddy for now */ |
49674596 |
#define FAX MIXED /*
* RFC3458
* Drafts stated to treat is as mixed if it is
* not known. This disappeared in the final
* version (except when talking about
* voice-message), but it is good enough for us
* since we do no validation of coversheet
* presence etc. (which also has disappeared
* in the final version)
*/ |
b62a19da |
#define ENCRYPTED 13 /*
* e.g. RFC2015
* Content-Type: multipart/encrypted;
* boundary="nextPart1383049.XCRrrar2yq";
* protocol="application/pgp-encrypted"
*/ |
db09f781 |
#define X_BFILE RELATED /*
* BeOS, expert two parts: the file and it's
* attributes. The attributes part comes as
* Content-Type: application/x-be_attribute
* name="foo"
* I can't find where it is defined, any
* pointers would be appreciated. For now
* we treat it as multipart/related
*/ |
cbc2eaa9 |
#define KNOWBOT 14 /* Unknown and undocumented format? */ |
b151ef55 |
static const struct tableinit {
const char *key;
int value;
} rfc821headers[] = { |
68badbc1 |
/* TODO: make these regular expressions */ |
b759d5eb |
{ "Content-Type", CONTENT_TYPE }, |
39ff42ee |
{ "Content-Transfer-Encoding", CONTENT_TRANSFER_ENCODING },
{ "Content-Disposition", CONTENT_DISPOSITION }, |
b151ef55 |
{ NULL, 0 } |
4fc38d69 |
}, mimeSubtypes[] = { /* see RFC2045 */ |
b151ef55 |
/* subtypes of Text */
{ "plain", PLAIN },
{ "enriched", ENRICHED },
{ "html", HTML },
{ "richtext", RICHTEXT },
/* subtypes of Multipart */
{ "mixed", MIXED },
{ "alternative", ALTERNATIVE },
{ "digest", DIGEST },
{ "signed", SIGNED },
{ "parallel", PARALLEL },
{ "related", RELATED },
{ "report", REPORT }, |
fdc8a467 |
{ "appledouble", APPLEDOUBLE }, |
49674596 |
{ "fax-message", FAX }, |
b62a19da |
{ "encrypted", ENCRYPTED }, |
db09f781 |
{ "x-bfile", X_BFILE }, /* BeOS */ |
cbc2eaa9 |
{ "knowbot", KNOWBOT }, /* ??? */
{ "knowbot-metadata", KNOWBOT }, /* ??? */
{ "knowbot-code", KNOWBOT }, /* ??? */
{ "knowbot-state", KNOWBOT }, /* ??? */ |
b151ef55 |
{ NULL, 0 }
}; |
8a88fb93 |
#ifdef CL_THREAD_SAFE
static pthread_mutex_t tables_mutex = PTHREAD_MUTEX_INITIALIZER;
#endif |
b151ef55 |
|
0dbec6b9 |
#ifndef O_BINARY
#define O_BINARY 0
#endif
|
273cd2bb |
#ifdef NEW_WORLD |
00615ec9 |
|
e82a5185 |
#undef PARTIAL_DIR
|
00615ec9 |
#if HAVE_MMAP
#if HAVE_SYS_MMAN_H
#include <sys/mman.h>
#else /* HAVE_SYS_MMAN_H */
#undef HAVE_MMAP
#endif |
e82a5185 |
#else /*HAVE_MMAP*/
#undef NEW_WORLD
#endif |
00615ec9 |
#endif
|
e82a5185 |
#ifdef NEW_WORLD
/*
* Files larger than this are scanned with the old method, should be
* StreamMaxLength, I guess
* If NW_MAX_FILE_SIZE is not defined, all files go through the
* new method. This definition is for machines very tight on RAM, or
* with large StreamMaxLength values
*/
#define MAX_ALLOCATION 134217728 /* see libclamav/others.c */
#define NW_MAX_FILE_SIZE MAX_ALLOCATION
|
273cd2bb |
struct scanlist { |
e82a5185 |
const char *start;
size_t size;
encoding_type decoder; /* only BASE64 and QUOTEDPRINTABLE for now */
struct scanlist *next; |
273cd2bb |
};
|
e82a5185 |
static struct map {
const char *offset; /* sorted */
const char *word;
struct map *next;
} *map, *tail;
static void create_map(const char *begin, const char *end);
static void add_to_map(const char *offset, const char *word);
static const char *find_in_map(const char *offset, const char *word);
static void free_map(void);
|
00615ec9 |
/*
* This could be the future. Instead of parsing and decoding it just decodes. |
273cd2bb |
* |
00615ec9 |
* USE IT AT YOUR PERIL, a large number of viruses are not detected with this
* method, possibly because the decoded files must be exact and not have
* extra data at the start or end, which this code will produce. |
53ee0b60 |
* |
273cd2bb |
* Currently only supports base64 and quoted-printable
*
* You may also see a lot of warnings. For the moment it falls back to old
* world mode if it doesn't know what to do - that'll be removed.
* The code is untidy...
*
* FIXME: Some mailbox scans are slower with this method. I suspect that it's
* because the scan can proceed to the end of the file rather than the end
* of the attachment which can mean than later emails are scanned many times |
9c8806fb |
* |
e82a5185 |
* FIXME: quoted printable doesn't know when to stop, so size related virus
* matching breaks
*
* TODO: Fall through to cli_parse_mbox() too often
*
* TODO: Add support for systems without mmap()
*
* TODO: partial_dir fall through |
00615ec9 |
*/
int
cli_mbox(const char *dir, int desc, unsigned int options)
{ |
e82a5185 |
char *start, *ptr, *line;
const char *last, *p, *q;
size_t size; |
00615ec9 |
struct stat statb;
message *m;
fileblob *fb; |
53ee0b60 |
int ret = 0; |
8c2e0f32 |
int wasAlloced; |
273cd2bb |
struct scanlist *scanlist, *scanelem; |
00615ec9 |
|
f60a8d41 |
if(dir == NULL) {
cli_warnmsg("cli_mbox called with NULL dir\n");
return CL_ENULLARG;
} |
00615ec9 |
if(fstat(desc, &statb) < 0)
return CL_EOPEN;
size = statb.st_size;
if(size == 0)
return CL_CLEAN;
|
e82a5185 |
#ifdef NW_MAX_FILE_SIZE
if(size > NW_MAX_FILE_SIZE)
return cli_parse_mbox(dir, desc, options);
#endif |
00615ec9 |
|
e82a5185 |
/*cli_warnmsg("NEW_WORLD is new code - use at your own risk.\n");*/
#ifdef PARTIAL_DIR
cli_warnmsg("PARTIAL_DIR doesn't work in the NEW_WORLD yet\n");
#endif |
0d9e07a9 |
|
273cd2bb |
start = mmap(NULL, size, PROT_READ, MAP_PRIVATE, desc, 0);
if(start == MAP_FAILED) |
00615ec9 |
return CL_EMEM;
cli_dbgmsg("mmap'ed mbox\n");
|
273cd2bb |
ptr = cli_malloc(size);
if(ptr) {
memcpy(ptr, start, size); |
8c2e0f32 |
munmap(start, size); |
273cd2bb |
start = ptr; |
e82a5185 |
wasAlloced = 1; |
8c2e0f32 |
} else
wasAlloced = 0;
|
e82a5185 |
/* last points to the last *valid* address in the array */
last = &start[size - 1];
create_map(start, last);
|
273cd2bb |
scanelem = scanlist = NULL;
q = start; |
e82a5185 |
/*
* FIXME: mismatch of const char * and char * here and in later calls
* to find_in_map()
*/
while((p = find_in_map(q, "base64")) != NULL) { |
273cd2bb |
cli_dbgmsg("Found base64\n");
if(scanelem) {
scanelem->next = cli_malloc(sizeof(struct scanlist));
scanelem = scanelem->next;
} else
scanlist = scanelem = cli_malloc(sizeof(struct scanlist));
scanelem->next = NULL;
scanelem->decoder = BASE64;
q = scanelem->start = &p[6]; |
e82a5185 |
if(((p = find_in_map(q, "\nFrom ")) != NULL) ||
((p = find_in_map(q, "base64")) != NULL) ||
((p = find_in_map(q, "quoted-printable")) != NULL)) { |
64ff0d49 |
scanelem->size = (size_t)(p - q); |
273cd2bb |
q = p; |
9c8806fb |
} else {
scanelem->size = (size_t)(last - scanelem->start) + 1;
break;
} |
e82a5185 |
cli_dbgmsg("base64: last %u q %u\n", (unsigned int)last, (unsigned int)q); |
273cd2bb |
assert(scanelem->size <= size); |
00615ec9 |
} |
e82a5185 |
|
273cd2bb |
q = start; |
e82a5185 |
while((p = find_in_map(q, "quoted-printable")) != NULL) { |
9c8806fb |
if(p != q)
switch(p[-1]) {
case ' ':
case ':':
case '=': /* wrong but allow it */
break;
default:
q = &p[16];
cli_dbgmsg("Ignore quoted-printable false positive\n");
continue; /* false positive */
}
|
273cd2bb |
cli_dbgmsg("Found quoted-printable\n"); |
e82a5185 |
#ifdef notdef
/*
* The problem with quoted printable is recognising when to stop
* parsing
*/ |
273cd2bb |
if(scanelem) {
scanelem->next = cli_malloc(sizeof(struct scanlist));
scanelem = scanelem->next;
} else
scanlist = scanelem = cli_malloc(sizeof(struct scanlist));
scanelem->next = NULL;
scanelem->decoder = QUOTEDPRINTABLE;
q = scanelem->start = &p[16]; |
e82a5185 |
cli_dbgmsg("qp: last %u q %u\n", (unsigned int)last, (unsigned int)q);
if(((p = find_in_map(q, "\nFrom ")) != NULL) ||
((p = find_in_map(q, "quoted-printable")) != NULL) ||
((p = find_in_map(q, "base64")) != NULL)) { |
64ff0d49 |
scanelem->size = (size_t)(p - q); |
273cd2bb |
q = p; |
9c8806fb |
cli_dbgmsg("qp: scanelem->size = %u\n", scanelem->size);
} else {
scanelem->size = (size_t)(last - scanelem->start) + 1;
break;
} |
273cd2bb |
assert(scanelem->size <= size); |
e82a5185 |
#else
if(wasAlloced)
free(start);
else
munmap(start, size);
free_map();
return cli_parse_mbox(dir, desc, options);
#endif |
00615ec9 |
}
|
273cd2bb |
if(scanlist == NULL) {
const struct tableinit *tableinit;
bool anyHeadersFound = FALSE; |
9c8806fb |
bool hasuuencode = FALSE; |
e82a5185 |
cli_file_t type; |
273cd2bb |
/* FIXME: message: There could of course be no decoder needed... */
for(tableinit = rfc821headers; tableinit->key; tableinit++) |
e82a5185 |
if(find_in_map(start, tableinit->key)) { |
273cd2bb |
anyHeadersFound = TRUE;
break;
}
|
e82a5185 |
if((!anyHeadersFound) && find_in_map(start, "\nbegin ")) |
9c8806fb |
/* uuencoded part */
hasuuencode = TRUE;
|
e82a5185 |
free_map();
type = cli_filetype(start, size);
if((type == CL_TYPE_UNKNOWN_TEXT) &&
(strncmp(start, "Microsoft Mail Internet Headers", 31) == 0))
type = CL_TYPE_MAIL;
|
8c2e0f32 |
if(wasAlloced)
free(start);
else
munmap(start, size); |
00615ec9 |
|
9c8806fb |
if(anyHeadersFound || hasuuencode) {
/* TODO: reduce the number of falls through here */ |
e82a5185 |
if(hasuuencode)
cli_dbgmsg("New world - fall back to old uudecoder, type %d\n", type);
else
cli_dbgmsg("cli_mbox: unknown encoder, type %d\n", type);
if(type == CL_TYPE_MAIL)
return cli_parse_mbox(dir, desc, options);
cli_dbgmsg("Unknown filetype %d, return CLEAN\n", type);
return CL_CLEAN; |
53ee0b60 |
} |
9c8806fb |
|
e82a5185 |
/* The message could be a plain text phish */
if((type == CL_TYPE_MAIL) && (!(options&CL_DB_NOPHISHING)))
return cli_parse_mbox(dir, desc, options);
cli_dbgmsg("cli_mbox: I believe it's plain text which must be clean\n"); |
273cd2bb |
return CL_CLEAN;
} |
e82a5185 |
free_map();
#if 0
if(wasAlloced) {
const char *max = NULL;
for(scanelem = scanlist; scanelem; scanelem = scanelem->next) {
const char *end = &scanelem->start[scanelem->size];
if(end > max)
max = end;
}
if(max < last)
printf("could free %d bytes\n", (int)(last - max));
}
#endif |
273cd2bb |
for(scanelem = scanlist; scanelem; scanelem = scanelem->next) {
if(scanelem->decoder == BASE64) { |
e82a5185 |
const char *b64start = scanelem->start;
size_t b64size = scanelem->size; |
273cd2bb |
cli_dbgmsg("b64size = %lu\n", b64size); |
e82a5185 |
while((*b64start != '\n') && (*b64start != '\r')) { |
53ee0b60 |
b64start++;
b64size--; |
273cd2bb |
}
/*
* Look for the end of the headers
*/
while(b64start < last) {
if(*b64start == ';') { |
53ee0b60 |
b64start++;
b64size--; |
e82a5185 |
} else if((memcmp(b64start, "\n\n", 2) == 0) ||
(memcmp(b64start, "\r\r", 2) == 0)) {
b64start += 2;
b64size -= 2;
break;
} else if(memcmp(b64start, "\r\n\r\n", 4) == 0) {
b64start += 4;
b64size -= 4;
break;
} else if(memcmp(b64start, "\n \n", 3) == 0) {
/*
* Some viruses are broken and have
* one space character at the end of
* the headers
*/
b64start += 3;
b64size -= 3;
break;
} else if(memcmp(b64start, "\r\n \r\n", 5) == 0) {
/*
* Some viruses are broken and have
* one space character at the end of
* the headers
*/
b64start += 5;
b64size -= 5;
break; |
53ee0b60 |
} |
6c4485f9 |
b64start++; |
273cd2bb |
b64size--; |
6c4485f9 |
} |
53ee0b60 |
|
273cd2bb |
if(b64size > 0L) |
9c8806fb |
while((!isalnum(*b64start)) && (*b64start != '/')) { |
273cd2bb |
if(b64size-- == 0L)
break;
b64start++;
}
if(b64size > 0L) { |
e82a5185 |
int lastline; |
273cd2bb |
cli_dbgmsg("cli_mbox: decoding %ld base64 bytes\n", b64size); |
53ee0b60 |
|
273cd2bb |
line = NULL; |
53ee0b60 |
|
273cd2bb |
m = messageCreate(); |
e82a5185 |
if(m == NULL) {
if(wasAlloced)
free(start);
else
munmap(start, size);
|
273cd2bb |
return CL_EMEM; |
e82a5185 |
} |
273cd2bb |
messageSetEncoding(m, "base64"); |
53ee0b60 |
|
e82a5185 |
lastline = 0;
|
9c8806fb |
do { |
273cd2bb |
int length = 0; |
e82a5185 |
char *newline, *equal; |
53ee0b60 |
|
273cd2bb |
/*printf("%ld: ", b64size); fflush(stdout);*/ |
53ee0b60 |
|
273cd2bb |
for(ptr = b64start; b64size && (*ptr != '\n') && (*ptr != '\r'); ptr++) {
length++;
--b64size;
} |
53ee0b60 |
|
273cd2bb |
/*printf("%d: ", length); fflush(stdout);*/ |
53ee0b60 |
|
e82a5185 |
newline = cli_realloc(line, length + 1);
if(newline == NULL)
break;
line = newline; |
53ee0b60 |
|
273cd2bb |
memcpy(line, b64start, length);
line[length] = '\0'; |
53ee0b60 |
|
e82a5185 |
equal = strchr(line, '=');
if(equal) {
lastline++;
*equal = '\0';
} |
273cd2bb |
/*puts(line);*/ |
53ee0b60 |
|
273cd2bb |
if(messageAddStr(m, line) < 0)
break; |
53ee0b60 |
|
273cd2bb |
if((b64size > 0) && (*ptr == '\r')) { |
e82a5185 |
b64start = ++ptr;
--b64size;
}
if((b64size > 0) && (*ptr == '\n')) {
b64start = ++ptr; |
273cd2bb |
--b64size;
} |
e82a5185 |
if(lastline) |
273cd2bb |
break; |
9c8806fb |
} while(b64size > 0L);
|
273cd2bb |
free(line);
fb = messageToFileblob(m, dir);
messageDestroy(m);
if(fb)
fileblobDestroy(fb);
else
ret = -1; |
00615ec9 |
} |
273cd2bb |
} else if(scanelem->decoder == QUOTEDPRINTABLE) { |
e82a5185 |
const char *quotedstart = scanelem->start;
size_t quotedsize = scanelem->size; |
53ee0b60 |
|
273cd2bb |
cli_dbgmsg("quotedsize = %lu\n", quotedsize);
while(*quotedstart != '\n') { |
53ee0b60 |
quotedstart++;
quotedsize--; |
273cd2bb |
}
/*
* Look for the end of the headers
*/
while(quotedstart < last) {
if(*quotedstart == ';') { |
53ee0b60 |
quotedstart++;
quotedsize--; |
e82a5185 |
} else if((*quotedstart == '\n') || (*quotedstart == '\r')) { |
273cd2bb |
quotedstart++;
quotedsize--;
if((*quotedstart == '\n') || (*quotedstart == '\r')) {
quotedstart++;
quotedsize--;
break;
} |
53ee0b60 |
} |
273cd2bb |
quotedstart++;
quotedsize--; |
53ee0b60 |
} |
00615ec9 |
|
273cd2bb |
while(!isalnum(*quotedstart)) {
quotedstart++;
quotedsize--;
} |
00615ec9 |
|
273cd2bb |
if(quotedsize > 0L) {
cli_dbgmsg("cli_mbox: decoding %ld quoted-printable bytes\n", quotedsize); |
00615ec9 |
|
273cd2bb |
m = messageCreate(); |
e82a5185 |
if(m == NULL) {
if(wasAlloced)
free(start);
else
munmap(start, size);
|
273cd2bb |
return CL_EMEM; |
e82a5185 |
} |
273cd2bb |
messageSetEncoding(m, "quoted-printable"); |
00615ec9 |
|
273cd2bb |
line = NULL; |
00615ec9 |
|
9c8806fb |
do { |
273cd2bb |
int length = 0; |
e82a5185 |
char *newline; |
00615ec9 |
|
273cd2bb |
/*printf("%ld: ", quotedsize); fflush(stdout);*/ |
00615ec9 |
|
273cd2bb |
for(ptr = quotedstart; quotedsize && (*ptr != '\n') && (*ptr != '\r'); ptr++) {
length++;
--quotedsize;
} |
00615ec9 |
|
273cd2bb |
/*printf("%d: ", length); fflush(stdout);*/ |
00615ec9 |
|
e82a5185 |
newline = cli_realloc(line, length + 1);
if(newline == NULL)
break;
line = newline; |
00615ec9 |
|
273cd2bb |
memcpy(line, quotedstart, length);
line[length] = '\0'; |
00615ec9 |
|
273cd2bb |
/*puts(line);*/ |
00615ec9 |
|
273cd2bb |
if(messageAddStr(m, line) < 0)
break; |
53ee0b60 |
|
273cd2bb |
if((quotedsize > 0) && (*ptr == '\r')) { |
e82a5185 |
quotedstart = ++ptr;
--quotedsize;
}
if((quotedsize > 0) && (*ptr == '\n')) {
quotedstart = ++ptr; |
273cd2bb |
--quotedsize;
} |
9c8806fb |
} while(quotedsize > 0L);
|
273cd2bb |
free(line);
fb = messageToFileblob(m, dir);
messageDestroy(m); |
53ee0b60 |
|
273cd2bb |
if(fb)
fileblobDestroy(fb);
else
ret = -1;
} |
00615ec9 |
}
} |
273cd2bb |
scanelem = scanlist;
while(scanelem) {
struct scanlist *n = scanelem->next;
free(scanelem);
scanelem = n;
} |
00615ec9 |
|
8c2e0f32 |
if(wasAlloced)
free(start);
else
munmap(start, size); |
00615ec9 |
|
8c2e0f32 |
/*
* FIXME: Need to run cl_scandir() here and return that value
*/ |
53ee0b60 |
if(ret == 0) |
00615ec9 |
return CL_CLEAN; /* a lie - but it gets things going */ |
53ee0b60 |
|
e82a5185 |
cli_dbgmsg("New world - don't know what to do - fall back to old world\n"); |
273cd2bb |
/* Fall back for now */ |
9c8806fb |
lseek(desc, 0L, SEEK_SET); |
00615ec9 |
return cli_parse_mbox(dir, desc, options);
} |
e82a5185 |
static void
create_map(const char *begin, const char *end)
{
const struct wordlist {
const char *word;
int len;
} wordlist[] = {
{ "base64", 6 },
{ "quoted-printable", 16 },
{ "\nbegin ", 7 },
{ "\nFrom ", 7 },
{ NULL, 0 }
};
if(map) {
cli_warnmsg("create_map called without free_map\n");
free_map();
}
while(begin < end) {
const struct wordlist *word;
for(word = wordlist; word->word; word++) {
if((end - begin) < word->len)
continue;
if(strncasecmp(begin, word->word, word->len) == 0) {
add_to_map(begin, word->word);
break;
}
}
begin++;
}
}
/* To sort map, assume 'offset' is presented in sorted order */
static void
add_to_map(const char *offset, const char *word)
{
if(map) {
tail->next = cli_malloc(sizeof(struct map)); /* FIXME: verify */
tail = tail->next;
} else
map = tail = cli_malloc(sizeof(struct map)); /* FIXME: verify */
tail->offset = offset;
tail->word = word;
tail->next = NULL;
}
static const char *
find_in_map(const char *offset, const char *word)
{
const struct map *item;
for(item = map; item; item = item->next)
if(item->offset >= offset)
if(strcasecmp(word, item->word) == 0)
return item->offset;
return NULL;
}
static void
free_map(void)
{
while(map) {
struct map *next = map->next;
free(map);
map = next;
}
map = NULL;
}
#else /*!NEW_WORLD*/ |
00615ec9 |
int
cli_mbox(const char *dir, int desc, unsigned int options)
{ |
f60a8d41 |
if(dir == NULL) {
cli_warnmsg("cli_mbox called with NULL dir\n");
return CL_ENULLARG;
} |
00615ec9 |
return cli_parse_mbox(dir, desc, options);
}
#endif
|
b151ef55 |
/*
* TODO: when signal handling is added, need to remove temp files when a |
7d3d11d0 |
* signal is received |
b151ef55 |
* TODO: add option to scan in memory not via temp files, perhaps with a |
74b5c349 |
* named pipe or memory mapped file, though this won't work on big e-mails
* containing many levels of encapsulated messages - it'd just take too much
* RAM |
c6259ac5 |
* TODO: parse .msg format files |
fdc8a467 |
* TODO: fully handle AppleDouble format, see |
7d3d11d0 |
* http://www.lazerware.com/formats/Specs/AppleSingle_AppleDouble.pdf |
f54a8635 |
* TODO: ensure parseEmailHeaders is always called before parseEmailBody
* TODO: create parseEmail which calls parseEmailHeaders then parseEmailBody |
f1c33aa0 |
* TODO: Look into TNEF. Is there anything that needs to be done here? |
e82a5185 |
* TODO: Handle unepected NUL bytes in header lines which stop strcmp()s:
* e.g. \0Content-Type: application/binary; |
b151ef55 |
*/ |
00615ec9 |
static int
cli_parse_mbox(const char *dir, int desc, unsigned int options) |
b151ef55 |
{ |
c6259ac5 |
int retcode, i; |
82933497 |
message *body; |
b151ef55 |
FILE *fd; |
e82a5185 |
char buffer[RFC2821LENGTH + 1]; |
0e3b08fc |
#ifdef HAVE_BACKTRACE |
98685ac1 |
void (*segv)(int);
#endif |
49674596 |
static table_t *rfc821, *subtype; |
11466e82 |
#ifdef CL_DEBUG
char tmpfilename[16];
int tmpfd;
#endif |
b151ef55 |
|
e82a5185 |
#ifdef NEW_WORLD
cli_dbgmsg("fall back to old world\n");
#else |
b151ef55 |
cli_dbgmsg("in mbox()\n"); |
e82a5185 |
#endif |
b151ef55 |
|
c6259ac5 |
i = dup(desc);
if((fd = fdopen(i, "rb")) == NULL) {
cli_errmsg("Can't open descriptor %d\n", desc);
close(i); |
7d3d11d0 |
return CL_EOPEN; |
c6259ac5 |
} |
11466e82 |
#ifdef CL_DEBUG
/*
* Copy the incoming mail for debugging, so that if it falls over
* we have a copy of the offending email. This is debugging code
* that you shouldn't of course install in a live environment. I am
* not interested in hearing about security issues with this section
* of the parser.
*/
strcpy(tmpfilename, "/tmp/mboxXXXXXX");
tmpfd = mkstemp(tmpfilename);
if(tmpfd < 0) {
perror(tmpfilename);
cli_errmsg("Can't make debugging file\n");
} else {
FILE *tmpfp = fdopen(tmpfd, "w");
if(tmpfp) {
while(fgets(buffer, sizeof(buffer) - 1, fd) != NULL)
fputs(buffer, tmpfp);
fclose(tmpfp);
rewind(fd);
} else
cli_errmsg("Can't fdopen debugging file\n");
}
#endif |
802c37fc |
if(fgets(buffer, sizeof(buffer) - 1, fd) == NULL) { |
c6259ac5 |
/* empty message */
fclose(fd); |
11466e82 |
#ifdef CL_DEBUG
unlink(tmpfilename);
#endif |
7d3d11d0 |
return CL_CLEAN; |
c6259ac5 |
} |
8a88fb93 |
#ifdef CL_THREAD_SAFE
pthread_mutex_lock(&tables_mutex);
#endif |
49674596 |
if(rfc821 == NULL) {
assert(subtype == NULL); |
7b8fb055 |
|
49674596 |
if(initialiseTables(&rfc821, &subtype) < 0) {
rfc821 = NULL;
subtype = NULL; |
8a88fb93 |
#ifdef CL_THREAD_SAFE
pthread_mutex_unlock(&tables_mutex);
#endif |
7b8fb055 |
fclose(fd); |
11466e82 |
#ifdef CL_DEBUG
unlink(tmpfilename);
#endif |
7d3d11d0 |
return CL_EMEM; |
7b8fb055 |
} |
b151ef55 |
} |
8a88fb93 |
#ifdef CL_THREAD_SAFE
pthread_mutex_unlock(&tables_mutex);
#endif |
b151ef55 |
|
89e9a596 |
#ifdef HAVE_BACKTRACE |
98685ac1 |
segv = signal(SIGSEGV, sigsegv);
#endif
|
f54a8635 |
/* |
9c8806fb |
* Is it a UNIX style mbox with more than one |
f54a8635 |
* mail message, or just a single mail message? |
9c8806fb |
*
* TODO: It would be better if we called cli_scandir here rather than
* in cli_scanmail. Then we could improve the way mailboxes with more
* than one message is handled, e.g. stopping parsing when an infected
* message is stopped, and giving a better indication of which message
* within the mailbox is infected |
f54a8635 |
*/ |
e82a5185 |
if((strncmp(buffer, "From ", 5) == 0) && isalnum(buffer[5])) { |
b151ef55 |
/* |
c6259ac5 |
* Have been asked to check a UNIX style mbox file, which
* may contain more than one e-mail message to decode |
53ee0b60 |
*
* It would be far better for scanners.c to do this splitting
* and do this
* FOR EACH mail in the mailbox
* DO
* pass this mail to cli_mbox --
* scan this file
* IF this file has a virus quit
* THEN
* return CL_VIRUS
* FI
* END
* This would remove a problem with this code that it can
* fill up the tmp directory before it starts scanning |
b151ef55 |
*/ |
82933497 |
bool lastLineWasEmpty;
int messagenumber;
message *m = messageCreate();
if(m == NULL) {
fclose(fd);
#ifdef HAVE_BACKTRACE
signal(SIGSEGV, segv);
#endif |
11466e82 |
#ifdef CL_DEBUG
unlink(tmpfilename);
#endif |
82933497 |
return CL_EMEM;
}
lastLineWasEmpty = FALSE;
messagenumber = 1; |
b151ef55 |
|
c6259ac5 |
do { |
f54a8635 |
cli_chomp(buffer); |
e82a5185 |
if(lastLineWasEmpty && (strncmp(buffer, "From ", 5) == 0) && isalnum(buffer[5])) { |
bf497d0a |
cli_dbgmsg("Deal with email number %d\n", messagenumber++); |
b151ef55 |
/* |
f54a8635 |
* End of a message in the mail box |
b151ef55 |
*/ |
de617e3e |
body = parseEmailHeaders(m, rfc821); |
f2b068fb |
if(body == NULL) {
messageReset(m);
continue;
} |
f54a8635 |
messageDestroy(m);
if(messageGetBody(body)) |
565c449d |
if(!parseEmailBody(body, NULL, dir, rfc821, subtype, options)) { |
09ccd6e0 |
messageReset(body);
m = body;
continue;
} |
b151ef55 |
/* |
f54a8635 |
* Starting a new message, throw away all the |
00615ec9 |
* information about the old one. It would
* be best to be able to scan this message
* now, but cli_scanfile needs arguments
* that haven't been passed here so it can't be
* called |
b151ef55 |
*/ |
f54a8635 |
m = body;
messageReset(body); |
b151ef55 |
|
c6259ac5 |
cli_dbgmsg("Finished processing message\n"); |
f54a8635 |
} else |
a66ca28a |
lastLineWasEmpty = (bool)(buffer[0] == '\0'); |
d1a6ea81 |
|
1ae303c2 |
if(isuuencodebegin(buffer)) { |
64ff0d49 |
/* |
d1a6ea81 |
* Fast track visa to uudecode.
* TODO: binhex, yenc
*/ |
e82a5185 |
if(uudecodeFile(m, buffer, dir, fd) < 0) |
1ae303c2 |
if(messageAddStr(m, buffer) < 0) |
e82a5185 |
break; |
1ae303c2 |
} else |
d1a6ea81 |
if(messageAddStr(m, buffer) < 0)
break; |
802c37fc |
} while(fgets(buffer, sizeof(buffer) - 1, fd) != NULL); |
bf497d0a |
|
82933497 |
fclose(fd);
|
00615ec9 |
cli_dbgmsg("Extract attachments from email %d\n", messagenumber); |
82933497 |
body = parseEmailHeaders(m, rfc821);
messageDestroy(m); |
f2b068fb |
} else { |
7fca6080 |
/*
* It's a single message, parse the headers then the body |
f2b068fb |
*/ |
f1c33aa0 |
if(strncmp(buffer, "P I ", 4) == 0)
/*
* CommuniGate Pro format: ignore headers until
* blank line
*/ |
802c37fc |
while((fgets(buffer, sizeof(buffer) - 1, fd) != NULL) && |
f1c33aa0 |
(strchr("\r\n", buffer[0]) == NULL))
;
/*
* Ignore any blank lines at the top of the message
*/ |
f2b068fb |
while(strchr("\r\n", buffer[0]) && |
c840fa41 |
(getline_from_mbox(buffer, sizeof(buffer) - 1, fd) != NULL)) |
6b93ea0c |
;
|
a78256af |
buffer[sizeof(buffer) - 1] = '\0'; |
802c37fc |
|
b3a5cdd8 |
body = parseEmailFile(fd, rfc821, buffer, dir); |
82933497 |
fclose(fd); |
f2b068fb |
} |
7fca6080 |
|
7d3d11d0 |
/*
* This is not necessarily true, but since the only options are
* CL_CLEAN and CL_VIRUS this is the better choice. It would be
* nice to have CL_CONTINUESCANNING or something like that
*/
retcode = CL_CLEAN; |
c6259ac5 |
|
f2b068fb |
if(body) {
/*
* Write out the last entry in the mailbox
*/
if(messageGetBody(body)) |
565c449d |
if(!parseEmailBody(body, NULL, dir, rfc821, subtype, options)) |
e745ac7e |
retcode = CL_EFORMAT; |
b151ef55 |
|
f2b068fb |
/*
* Tidy up and quit
*/
messageDestroy(body);
} |
b151ef55 |
cli_dbgmsg("cli_mbox returning %d\n", retcode);
|
89e9a596 |
#ifdef HAVE_BACKTRACE |
98685ac1 |
signal(SIGSEGV, segv);
#endif
|
11466e82 |
#ifdef CL_DEBUG
unlink(tmpfilename);
#endif |
b151ef55 |
return retcode;
}
/* |
82933497 |
* Read in an email message from fin, parse it, and return the message |
7fca6080 |
* |
82933497 |
* FIXME: files full of new lines and nothing else are
* handled ungracefully...
*/
static message * |
b3a5cdd8 |
parseEmailFile(FILE *fin, const table_t *rfc821, const char *firstLine, const char *dir) |
82933497 |
{
bool inHeader = TRUE;
bool contMarker = FALSE; |
7e0afd07 |
bool lastWasBlank = FALSE; |
82933497 |
message *ret;
bool anyHeadersFound = FALSE;
int commandNumber = -1; |
7e0afd07 |
char *fullline = NULL, *boundary = NULL; |
82933497 |
size_t fulllinelength = 0; |
e82a5185 |
char buffer[RFC2821LENGTH + 1]; |
82933497 |
cli_dbgmsg("parseEmailFile\n");
ret = messageCreate();
if(ret == NULL)
return NULL;
strcpy(buffer, firstLine);
do { |
9c8806fb |
char *line; |
82933497 |
(void)cli_chomp(buffer);
|
9c8806fb |
line = buffer;
if(line[0] == '\0')
line = NULL; |
82933497 |
/*
* Don't blank lines which are only spaces from headers,
* otherwise they'll be treated as the end of header marker
*/ |
7e0afd07 |
if(lastWasBlank) {
lastWasBlank = FALSE;
if(boundaryStart(buffer, boundary)) {
cli_dbgmsg("Found a header line with space that should be blank\n");
inHeader = FALSE;
}
}
if(boundary) {
free(boundary);
boundary = NULL;
} |
82933497 |
if(inHeader) { |
e82a5185 |
cli_dbgmsg("parseEmailFile: check '%s' contMarker %d fullline %p\n", |
7e0afd07 |
buffer ? buffer : "", (int)contMarker, fullline); |
9c8806fb |
if(line && isspace(line[0])) { |
7e0afd07 |
char copy[sizeof(buffer)];
strcpy(copy, buffer);
strstrip(copy);
if(copy[0] == '\0') {
/* |
097d9527 |
* The header line contains only white
* space. This is not the end of the
* headers according to RFC2822, but
* some MUAs will handle it as though
* it were, and virus writers exploit
* this bug. We can't just break from
* the loop here since that would allow
* other exploits such as inserting a
* white space line before the
* content-type line. So we just have
* to make a best guess. Sigh. |
7e0afd07 |
*/
if(fullline) {
if(parseEmailHeader(ret, fullline, rfc821) < 0)
continue;
free(fullline);
fullline = NULL;
} |
097d9527 |
if((boundary = (char *)messageFindArgument(ret, "boundary")) != NULL) { |
7e0afd07 |
lastWasBlank = TRUE;
continue;
}
}
}
lastWasBlank = FALSE; |
9c8806fb |
if((line == NULL) && (fullline == NULL)) { /* empty line */ |
82933497 |
if(!contMarker) {
/*
* A blank line signifies the end of
* the header and the start of the text
*/ |
0c9750fa |
if(!anyHeadersFound)
/* Ignore the junk at the top */
continue;
|
82933497 |
cli_dbgmsg("End of header information\n");
inHeader = FALSE;
} else
contMarker = FALSE;
} else {
char *ptr;
const char *qptr; |
cdfc0f05 |
int lookahead; |
82933497 |
if(fullline == NULL) { |
e82a5185 |
char cmd[RFC2821LENGTH + 1], out[RFC2821LENGTH + 1]; |
82933497 |
/*
* Continuation of line we're ignoring?
*/ |
e82a5185 |
if(isblank(line[0])) { |
9c8806fb |
contMarker = continuationMarker(line); |
82933497 |
continue;
}
/*
* Is this a header we're interested in?
*/ |
9c8806fb |
if((strchr(line, ':') == NULL) ||
(cli_strtokbuf(line, 0, ":", cmd) == NULL)) {
if(strncmp(line, "From ", 5) == 0) |
82933497 |
anyHeadersFound = TRUE;
continue;
}
|
273cd2bb |
ptr = rfc822comments(cmd, out); |
82933497 |
commandNumber = tableFind(rfc821, ptr ? ptr : cmd);
switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE:
anyHeadersFound = TRUE;
break;
default:
if(!anyHeadersFound)
anyHeadersFound = usefulHeader(commandNumber, cmd);
continue;
} |
9c8806fb |
fullline = strdup(line);
fulllinelength = strlen(line) + 1;
} else if(line != NULL) {
fulllinelength += strlen(line); |
e82a5185 |
ptr = cli_realloc(fullline, fulllinelength);
if(ptr == NULL)
continue;
fullline = ptr; |
9c8806fb |
strcat(fullline, line); |
82933497 |
}
|
9c8806fb |
if(line) {
contMarker = continuationMarker(line); |
82933497 |
|
cdfc0f05 |
if(contMarker)
continue;
} else
contMarker = FALSE; |
82933497 |
assert(fullline != NULL);
lookahead = getc(fin);
if(lookahead != EOF) {
ungetc(lookahead, fin);
/*
* Section B.2 of RFC822 says TAB or
* SPACE means a continuation of the
* previous entry.
*
* Add all the arguments on the line
*/ |
e82a5185 |
if(isblank(lookahead)) |
82933497 |
continue;
}
|
9c8806fb |
if(line) { |
cdfc0f05 |
int quotes = 0;
for(qptr = fullline; *qptr; qptr++)
if(*qptr == '\"')
quotes++; |
82933497 |
|
cdfc0f05 |
if(quotes & 1)
continue;
} |
82933497 |
|
273cd2bb |
ptr = rfc822comments(fullline, NULL); |
82933497 |
if(ptr) {
free(fullline);
fullline = ptr;
}
if(parseEmailHeader(ret, fullline, rfc821) < 0)
continue;
free(fullline);
fullline = NULL;
} |
1ae303c2 |
} else if(line && isuuencodebegin(line)) { |
b3a5cdd8 |
/*
* Fast track visa to uudecode.
* TODO: binhex, yenc
*/ |
e82a5185 |
if(uudecodeFile(ret, line, dir, fin) < 0) |
1ae303c2 |
if(messageAddStr(ret, line) < 0)
break;
} else |
9c8806fb |
if(messageAddStr(ret, line) < 0) |
82933497 |
break; |
c840fa41 |
} while(getline_from_mbox(buffer, sizeof(buffer) - 1, fin) != NULL); |
82933497 |
if(fullline) {
if(*fullline) switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE: |
17bd4b65 |
cli_dbgmsg("parseEmailHeaders: Fullline unparsed '%s'\n", fullline); |
82933497 |
}
free(fullline);
}
if(!anyHeadersFound) {
/*
* False positive in believing we have an e-mail when we don't
*/
messageDestroy(ret);
cli_dbgmsg("parseEmailFile: no headers found, assuming it isn't an email\n");
return NULL;
}
messageClean(ret);
cli_dbgmsg("parseEmailFile: return\n");
return ret;
}
/*
* The given message contains a raw e-mail. |
68be129f |
*
* Returns the message's body with the correct arguments set |
735377bc |
*
* The downside of this approach is that for a short time we have two copies
* 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 |
82933497 |
*
* TODO: remove the duplication with parseEmailFile |
7fca6080 |
*/ |
68be129f |
static message * |
de617e3e |
parseEmailHeaders(const message *m, const table_t *rfc821) |
7fca6080 |
{ |
68be129f |
bool inHeader = TRUE; |
de617e3e |
const text *t; |
f54a8635 |
message *ret; |
f2b068fb |
bool anyHeadersFound = FALSE; |
4fc38d69 |
int commandNumber = -1; |
d768ac5a |
char *fullline = NULL; |
ad642304 |
size_t fulllinelength = 0; |
f54a8635 |
|
98685ac1 |
cli_dbgmsg("parseEmailHeaders\n");
|
f54a8635 |
if(m == NULL)
return NULL;
ret = messageCreate(); |
7fca6080 |
|
de617e3e |
for(t = messageGetBody(m); t; t = t->t_next) {
const char *buffer; |
7fca6080 |
|
de617e3e |
if(t->t_line)
buffer = lineGetData(t->t_line);
else |
98685ac1 |
buffer = NULL; |
7fca6080 |
|
b4cb4486 |
if(inHeader) { |
663e5963 |
cli_dbgmsg("parseEmailHeaders: check '%s'\n",
buffer ? buffer : ""); |
82933497 |
if(buffer == NULL) { |
663e5963 |
/*
* A blank line signifies the end of
* the header and the start of the text
*/
cli_dbgmsg("End of header information\n");
inHeader = FALSE; |
f52d7358 |
if(!anyHeadersFound) {
cli_dbgmsg("Nothing interesting in the header\n");
break;
} |
ad642304 |
} else { |
0856891e |
char *ptr; |
ad642304 |
const char *qptr;
int quotes; |
0856891e |
|
ad642304 |
if(fullline == NULL) { |
e82a5185 |
char cmd[RFC2821LENGTH + 1]; |
9180b8bb |
/*
* Continuation of line we're ignoring?
*/ |
e82a5185 |
if(isblank(buffer[0])) |
9180b8bb |
continue;
/*
* Is this a header we're interested in?
*/ |
0856891e |
if((strchr(buffer, ':') == NULL) ||
(cli_strtokbuf(buffer, 0, ":", cmd) == NULL)) {
if(strncmp(buffer, "From ", 5) == 0)
anyHeadersFound = TRUE; |
9180b8bb |
continue; |
0856891e |
} |
9180b8bb |
|
273cd2bb |
ptr = rfc822comments(cmd, NULL); |
a1c924f9 |
commandNumber = tableFind(rfc821, ptr ? ptr : cmd);
if(ptr)
free(ptr); |
9180b8bb |
switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE: |
0856891e |
anyHeadersFound = TRUE; |
9180b8bb |
break;
default: |
82933497 |
if(!anyHeadersFound)
anyHeadersFound = usefulHeader(commandNumber, cmd); |
9180b8bb |
continue;
}
fullline = strdup(buffer);
fulllinelength = strlen(buffer) + 1;
} else if(buffer) {
fulllinelength += strlen(buffer); |
e82a5185 |
ptr = cli_realloc(fullline, fulllinelength);
if(ptr == NULL)
continue;
fullline = ptr; |
9180b8bb |
strcat(fullline, buffer); |
ad642304 |
} |
3a0946f5 |
|
82933497 |
assert(fullline != NULL); |
ad642304 |
|
82933497 |
if(t->t_next && (t->t_next->t_line != NULL)) |
ad642304 |
/* |
0856891e |
* Section B.2 of RFC822 says TAB or
* SPACE means a continuation of the
* previous entry. |
ad642304 |
*
* Add all the arguments on the line
*/ |
e82a5185 |
if(isblank(lineGetData(t->t_next->t_line)[0]))
continue; |
ad642304 |
quotes = 0; |
cdfc0f05 |
for(qptr = fullline; *qptr; qptr++) |
ad642304 |
if(*qptr == '\"')
quotes++;
|
9180b8bb |
if(quotes & 1) |
ad642304 |
continue;
|
273cd2bb |
ptr = rfc822comments(fullline, NULL); |
ad642304 |
if(ptr) {
free(fullline);
fullline = ptr;
} |
37819555 |
|
82933497 |
if(parseEmailHeader(ret, fullline, rfc821) < 0)
continue; |
b4cb4486 |
|
82933497 |
free(fullline);
fullline = NULL; |
d32343c3 |
} |
9180b8bb |
} else |
09ccd6e0 |
/*cli_dbgmsg("Add line to body '%s'\n", buffer);*/ |
de617e3e |
if(messageAddLine(ret, t->t_line) < 0) |
80a8c7d8 |
break; |
ffd59a3e |
} |
68be129f |
|
d768ac5a |
if(fullline) { |
ad642304 |
if(*fullline) switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE: |
17bd4b65 |
cli_dbgmsg("parseEmailHeaders: Fullline unparsed '%s'\n", fullline); |
ad642304 |
} |
d768ac5a |
free(fullline);
}
|
f2b068fb |
if(!anyHeadersFound) {
/*
* False positive in believing we have an e-mail when we don't
*/
messageDestroy(ret);
cli_dbgmsg("parseEmailHeaders: no headers found, assuming it isn't an email\n");
return NULL;
}
|
4465fb04 |
messageClean(ret);
|
09ccd6e0 |
cli_dbgmsg("parseEmailHeaders: return\n");
|
68be129f |
return ret; |
7fca6080 |
}
/* |
8c0250d5 |
* Handle a header line of an email message
*/
static int |
49674596 |
parseEmailHeader(message *m, const char *line, const table_t *rfc821) |
8c0250d5 |
{ |
de509b8e |
char *cmd; |
8c0250d5 |
int ret = -1;
#ifdef CL_THREAD_SAFE
char *strptr;
#endif |
31b05bcb |
const char *separater; |
0674e2af |
char *copy, tokenseparater[2]; |
8c0250d5 |
|
0704dad8 |
cli_dbgmsg("parseEmailHeader '%s'\n", line);
|
31b05bcb |
/*
* In RFC822 the separater between the key a value is a colon,
* e.g. Content-Transfer-Encoding: base64
* However some MUA's are lapse about this and virus writers exploit
* this hole, so we need to check all known possiblities
*/
for(separater = ":= "; *separater; separater++)
if(strchr(line, *separater) != NULL)
break;
if(*separater == '\0') |
74b5c349 |
return -1;
|
0674e2af |
copy = rfc2047(line);
if(copy == NULL) |
bd6146af |
/* an RFC checker would return -1 here */
copy = strdup(line); |
d1382234 |
|
31b05bcb |
tokenseparater[0] = *separater;
tokenseparater[1] = '\0';
|
897fd9c7 |
#ifdef CL_THREAD_SAFE |
31b05bcb |
cmd = strtok_r(copy, tokenseparater, &strptr); |
897fd9c7 |
#else
cmd = strtok(copy, tokenseparater);
#endif |
8c0250d5 |
|
3a0ef2ee |
if(cmd && (strstrip(cmd) > 0)) { |
897fd9c7 |
#ifdef CL_THREAD_SAFE |
8c0250d5 |
char *arg = strtok_r(NULL, "", &strptr); |
897fd9c7 |
#else
char *arg = strtok(NULL, "");
#endif |
8c0250d5 |
if(arg)
/*
* Found a header such as
* Content-Type: multipart/mixed;
* set arg to be
* "multipart/mixed" and cmd to |
39ff42ee |
* be "Content-Type" |
8c0250d5 |
*/ |
49674596 |
ret = parseMimeHeader(m, cmd, rfc821, arg); |
8c0250d5 |
} |
0674e2af |
free(copy); |
8c0250d5 |
return ret;
}
/* |
b151ef55 |
* This is a recursive routine. |
9c8806fb |
* FIXME: We are not passed &mrec so we can't check against MAX_MAIL_RECURSION |
b151ef55 |
* |
7fca6080 |
* This function parses the body of mainMessage and saves its attachments in dir
* |
68be129f |
* mainMessage is the buffer to be parsed, it contains an e-mail's body, without |
d32343c3 |
* any headers. First time of calling it'll be
* the whole message. Later it'll be parts of a multipart message |
b151ef55 |
* textIn is the plain text message being built up so far
* |
0bcad2b1 |
* Returns: |
b151ef55 |
* 0 for fail |
852e3ce4 |
* 1 for success, attachments saved
* 2 for success, attachments not saved |
b151ef55 |
*/
static int /* success or fail */ |
565c449d |
parseEmailBody(message *messageIn, text *textIn, const char *dir, const table_t *rfc821Table, const table_t *subtypeTable, unsigned int options) |
b151ef55 |
{ |
6613d595 |
message **messages; /* parts of a multipart message */ |
30fb8a0b |
int inMimeHead, i, rc = 1, htmltextPart, multiparts = 0; |
b151ef55 |
text *aText;
const char *cptr; |
2250ea69 |
message *mainMessage; |
565c449d |
fileblob *fb; |
b151ef55 |
|
565c449d |
cli_dbgmsg("in parseEmailBody\n"); |
b151ef55 |
aText = textIn; |
6613d595 |
messages = NULL; |
2250ea69 |
mainMessage = messageIn; |
b151ef55 |
/* Anything left to be parsed? */ |
0bcad2b1 |
if(mainMessage && (messageGetBody(mainMessage) != NULL)) { |
b151ef55 |
mime_type mimeType; |
30fb8a0b |
int subtype, inhead; |
b62a19da |
const char *mimeSubtype, *boundary;
char *protocol; |
b151ef55 |
const text *t_line; |
f5e9abc8 |
/*bool isAlternative;*/ |
b151ef55 |
message *aMessage;
|
c6259ac5 |
cli_dbgmsg("Parsing mail file\n");
|
b151ef55 |
mimeType = messageGetMimeType(mainMessage);
mimeSubtype = messageGetMimeSubtype(mainMessage);
|
20917083 |
/* pre-process */ |
3497daca |
subtype = tableFind(subtypeTable, mimeSubtype); |
5eeffbb9 |
if((mimeType == TEXT) && (subtype == PLAIN)) { |
b151ef55 |
/*
* This is effectively no encoding, notice that we
* don't check that charset is us-ascii
*/
cli_dbgmsg("assume no encoding\n");
mimeType = NOMIME; |
b3a5cdd8 |
messageSetMimeSubtype(mainMessage, ""); |
20917083 |
} else if((mimeType == MESSAGE) &&
(strcasecmp(mimeSubtype, "rfc822-headers") == 0)) {
/*
* RFC1892/RFC3462: section 2 text/rfc822-headers
* incorrectly sent as message/rfc822-headers |
ef5eba29 |
*
* Parse as text/plain, i.e. no mime |
20917083 |
*/
cli_dbgmsg("Changing message/rfc822-headers to text/rfc822-headers\n"); |
ef5eba29 |
mimeType = NOMIME; |
b3a5cdd8 |
messageSetMimeSubtype(mainMessage, ""); |
b151ef55 |
}
|
c6259ac5 |
cli_dbgmsg("mimeType = %d\n", mimeType);
|
b151ef55 |
switch(mimeType) {
case NOMIME: |
22f3b19b |
cli_dbgmsg("Not a mime encoded message\n"); |
b151ef55 |
aText = textAddMessage(aText, mainMessage);
break;
case TEXT: |
3497daca |
/* text/plain has been preprocessed as no encoding */
if((options&CL_SCAN_MAILURL) && (subtype == HTML)) |
5eeffbb9 |
checkURLs(mainMessage, dir); |
b151ef55 |
break;
case MULTIPART: |
10c3ed55 |
cli_dbgmsg("Content-type 'multipart' handler\n"); |
b151ef55 |
boundary = messageFindArgument(mainMessage, "boundary");
if(boundary == NULL) {
cli_warnmsg("Multipart MIME message contains no boundaries\n"); |
2227f20e |
/* Broken e-mail message */
mimeType = NOMIME;
/*
* The break means that we will still
* check if the file contains a uuencoded file
*/
break; |
b151ef55 |
}
|
cbc2eaa9 |
/* Perhaps it should assume mixed? */ |
93002b48 |
if(mimeSubtype[0] == '\0') {
cli_warnmsg("Multipart has no subtype assuming alternative\n");
mimeSubtype = "alternative";
messageSetMimeSubtype(mainMessage, "alternative");
}
|
b151ef55 |
/*
* Get to the start of the first message
*/ |
0704dad8 |
t_line = messageGetBody(mainMessage);
if(t_line == NULL) {
cli_warnmsg("Multipart MIME message has no body\n");
free((char *)boundary);
mimeType = NOMIME;
break;
}
do |
74ca33e9 |
if(t_line->t_line) {
if(boundaryStart(lineGetData(t_line->t_line), boundary))
break;
/* |
e82a5185 |
* Found a binhex file before |
4b54f2e0 |
* the first multipart |
0856891e |
* TODO: check yEnc |
74ca33e9 |
*/ |
e82a5185 |
if(binhexBegin(mainMessage) == t_line) { |
0856891e |
if(messageGetEncoding(mainMessage) == NOENCODING) {
messageSetEncoding(mainMessage, "x-binhex");
fb = messageToFileblob(mainMessage, dir);
if(fb)
fileblobDestroy(fb);
} |
9f43cc75 |
} else if(encodingLine(mainMessage) == t_line->t_next) {
/*
* We look for the next line
* since later on we'll skip
* over the important line when
* we think it's a blank line
* at the top of the message -
* which it would have been in
* an RFC compliant world
*/
cli_dbgmsg("Found MIME attachment before the first MIME section\n");
if(messageGetEncoding(mainMessage) == NOENCODING)
break; |
0856891e |
} |
74ca33e9 |
} |
0704dad8 |
while((t_line = t_line->t_next) != NULL); |
b151ef55 |
if(t_line == NULL) { |
b4cb4486 |
cli_dbgmsg("Multipart MIME message contains no boundary lines\n"); |
bf8ea488 |
/*
* Free added by Thomas Lamy
* <Thomas.Lamy@in-online.net>
*/
free((char *)boundary); |
2227f20e |
mimeType = NOMIME;
/*
* The break means that we will still |
e82a5185 |
* check if the file contains a yEnc/binhex file |
2227f20e |
*/
break; |
b151ef55 |
}
/*
* Build up a table of all of the parts of this
* multipart message. Remember, each part may itself
* be a multipart message.
*/
inhead = 1;
inMimeHead = 0;
|
68be129f |
/* |
b62a19da |
* Parse the mainMessage object and create an array
* of objects called messages, one for each of the
* multiparts that mainMessage contains
* |
68be129f |
* This looks like parseEmailHeaders() - maybe there's
* some duplication of code to be cleaned up
*/ |
6613d595 |
for(multiparts = 0; t_line; multiparts++) { |
26564cf5 |
int lines = 0; |
79e432d2 |
message **m; |
26564cf5 |
|
79e432d2 |
m = cli_realloc(messages, ((multiparts + 1) * sizeof(message *))); |
d32343c3 |
if(m == NULL) |
79e432d2 |
break;
messages = m; |
6613d595 |
|
b151ef55 |
aMessage = messages[multiparts] = messageCreate(); |
89e9a596 |
if(aMessage == NULL) {
multiparts--;
continue;
} |
b151ef55 |
cli_dbgmsg("Now read in part %d\n", multiparts);
|
0bf1353d |
/*
* Ignore blank lines. There shouldn't be ANY
* but some viruses insert them
*/ |
98685ac1 |
while((t_line = t_line->t_next) != NULL) |
de617e3e |
if(t_line->t_line &&
/*(cli_chomp(t_line->t_text) > 0))*/
(strlen(lineGetData(t_line->t_line)) > 0)) |
784e2335 |
break; |
0bf1353d |
if(t_line == NULL) {
cli_dbgmsg("Empty part\n"); |
61db35a1 |
/*
* Remove this part unless there's |
e82a5185 |
* a binhex portion somewhere in |
61db35a1 |
* the complete message that we may
* throw away by mistake if the MIME
* encoding information is incorrect
*/ |
e82a5185 |
if(binhexBegin(mainMessage) == NULL) { |
61db35a1 |
messageDestroy(aMessage);
--multiparts;
} |
0bf1353d |
continue;
}
do { |
de617e3e |
const char *line = lineGetData(t_line->t_line); |
b151ef55 |
|
e82a5185 |
/*cli_dbgmsg("multipart %d: inMimeHead %d inhead %d boundary '%s' line '%s' next '%s'\n",
multiparts, inMimeHead, inhead, boundary, line, |
30fb8a0b |
t_line->t_next && t_line->t_next->t_line ? lineGetData(t_line->t_next->t_line) : "(null)");*/ |
b151ef55 |
|
e0377124 |
if(inMimeHead) { /* continuation line */ |
98685ac1 |
if(line == NULL) { |
699fafc3 |
/*inhead =*/ inMimeHead = 0; |
98685ac1 |
continue;
} |
7baeb4a6 |
/*
* Handle continuation lines
* because the previous line |
21cd233d |
* ended with a ; or this line
* starts with a white space |
7baeb4a6 |
*/ |
21cd233d |
cli_dbgmsg("Multipart %d: About to add mime Argument '%s'\n",
multiparts, line); |
7baeb4a6 |
/*
* Handle the case when it
* isn't really a continuation
* line:
* Content-Type: application/octet-stream;
* Content-Transfer-Encoding: base64
*/
parseEmailHeader(aMessage, line, rfc821Table);
|
b151ef55 |
while(isspace((int)*line))
line++;
if(*line == '\0') {
inhead = inMimeHead = 0;
continue;
}
/*
* This may cause a trailing ';'
* to be added if this test
* fails - TODO: verify this
*/
inMimeHead = continuationMarker(line);
messageAddArgument(aMessage, line); |
e0377124 |
} else if(inhead) { /* handling normal headers */ |
10c3ed55 |
int quotes;
char *fullline, *ptr;
const char *qptr;
const text *next; |
ad642304 |
|
98685ac1 |
if(line == NULL) { |
9cb47b80 |
/*
* empty line, should the end of the headers,
* but some base64 decoders, e.g. uudeview, are broken
* and will handle this type of entry, decoding the
* base64 content...
* Content-Type: application/octet-stream; name=text.zip
* Content-Transfer-Encoding: base64
* Content-Disposition: attachment; filename="text.zip" |
64ff0d49 |
* |
9cb47b80 |
* Content-Disposition: attachment;
* filename=text.zip
* Content-Type: application/octet-stream;
* name=text.zip
* Content-Transfer-Encoding: base64 |
64ff0d49 |
* |
9cb47b80 |
* UEsDBAoAAAAAAACgPjJ2RHw676gAAO+oAABEAAAAbWFpbF90ZXh0LWluZm8udHh0ICAgICAgICAg
*/
next = t_line->t_next;
if(next && next->t_line) {
const char *data = lineGetData(next->t_line); |
6c4485f9 |
if((messageGetEncoding(aMessage) == NOENCODING) &&
(messageGetMimeType(aMessage) == APPLICATION)) |
273cd2bb |
/*
* Handle this nightmare (note the blank
* line in the header and the incorrect
* content-transfer-encoding header)
*
* Content-Type: application/octet-stream; name="zipped_files.EXEX-Spanska: Yes
*
* r-Encoding: base64
* Content-Disposition: attachment; filename="zipped_files.EXE"
*/
if(strstr(data, "base64")) { |
6c4485f9 |
messageSetEncoding(aMessage, "base64");
cli_dbgmsg("Ignoring fake end of headers\n");
continue;
} |
11466e82 |
if((strncmp(data, "Content", 7) == 0) ||
(strncmp(data, "filename=", 9) == 0)) { |
9cb47b80 |
cli_dbgmsg("Ignoring fake end of headers\n");
continue;
}
} |
699fafc3 |
cli_dbgmsg("Multipart %d: End of header information\n",
multiparts); |
b151ef55 |
inhead = 0;
continue;
} |
a64bf87e |
if(isspace((int)*line)) {
/*
* The first line is
* continuation line.
* This is tricky
* to handle, but
* all we can do is our
* best
*/
cli_dbgmsg("Part %d starts with a continuation line\n",
multiparts);
messageAddArgument(aMessage, line);
/*
* Give it a default
* MIME type since
* that may be the
* missing line
*
* Choose application to
* force a save
*/
if(messageGetMimeType(aMessage) == NOMIME)
messageSetMimeType(aMessage, "application");
continue;
}
|
10c3ed55 |
inMimeHead = FALSE; |
de617e3e |
|
e82a5185 |
assert(strlen(line) <= RFC2821LENGTH); |
e0377124 |
|
273cd2bb |
fullline = rfc822comments(line, NULL); |
10c3ed55 |
if(fullline == NULL)
fullline = strdup(line); |
ad642304 |
|
10c3ed55 |
quotes = 0;
for(qptr = fullline; *qptr; qptr++)
if(*qptr == '\"')
quotes++; |
e0377124 |
|
10c3ed55 |
/*
* Fold next lines to the end of this
* if they start with a white space
* or if this line has an odd number of quotes:
* Content-Type: application/octet-stream; name="foo
* "
*/
next = t_line->t_next;
while(next && next->t_line) {
const char *data = lineGetData(next->t_line); |
21cd233d |
|
60aec445 |
/*if((!isspace(data[0])) && |
10c3ed55 |
((quotes & 1) == 0)) |
60aec445 |
break;*/
if(!isspace(data[0])) |
10c3ed55 |
break; |
e0377124 |
|
d99b1840 |
if(data[1] == '\0') {
/*
* Broken message: the
* blank line at the end
* of the headers isn't blank -
* it contains a space
*/
cli_dbgmsg("Multipart %d: headers not terminated by blank line\n",
multiparts);
inhead = FALSE;
break;
}
|
10c3ed55 |
ptr = cli_realloc(fullline,
strlen(fullline) + strlen(data) + 1); |
68badbc1 |
|
10c3ed55 |
if(ptr == NULL)
break; |
21cd233d |
|
10c3ed55 |
fullline = ptr;
strcat(fullline, data); |
ad642304 |
|
60aec445 |
/*for(qptr = data; *qptr; qptr++) |
10c3ed55 |
if(*qptr == '\"') |
60aec445 |
quotes++;*/ |
ad642304 |
|
10c3ed55 |
t_line = next;
next = next->t_next; |
21cd233d |
} |
10c3ed55 |
cli_dbgmsg("Multipart %d: About to parse folded header '%s'\n",
multiparts, fullline);
parseEmailHeader(aMessage, fullline, rfc821Table);
free(fullline); |
b151ef55 |
} else if(endOfMessage(line, boundary)) {
/*
* Some viruses put information
* *after* the end of message,
* which presumably some broken
* mail clients find, so we
* can't assume that this
* is the end of the message
*/
/* t_line = NULL;*/
break; |
30fb8a0b |
} else if(boundaryStart(line, boundary)) {
inhead = 1;
break; |
26564cf5 |
} else { |
de617e3e |
if(messageAddLine(aMessage, t_line->t_line) < 0) |
79e432d2 |
break; |
26564cf5 |
lines++;
} |
0bf1353d |
} while((t_line = t_line->t_next) != NULL);
|
b151ef55 |
messageClean(aMessage); |
26564cf5 |
cli_dbgmsg("Part %d has %d lines\n",
multiparts, lines); |
b151ef55 |
}
free((char *)boundary);
|
6638be41 |
/* |
cbc2eaa9 |
* Preprocess. Anything special to be done before
* we handle the multiparts? |
b62a19da |
*/ |
cbc2eaa9 |
switch(tableFind(subtypeTable, mimeSubtype)) {
case KNOWBOT:
/* TODO */
cli_dbgmsg("multipart/knowbot parsed as multipart/mixed for now\n");
mimeSubtype = "mixed";
break; |
e279f3ea |
case -1:
/*
* According to section 7.2.6 of
* RFC1521, unrecognised multiparts
* should be treated as multipart/mixed.
*/ |
0cb54827 |
cli_dbgmsg("Unsupported multipart format `%s', parsed as mixed\n", mimeSubtype); |
e279f3ea |
mimeSubtype = "mixed";
break; |
cbc2eaa9 |
} |
b62a19da |
/* |
6638be41 |
* We've finished message we're parsing
*/
if(mainMessage && (mainMessage != messageIn)) {
messageDestroy(mainMessage);
mainMessage = NULL; |
2250ea69 |
} |
b151ef55 |
|
6613d595 |
if(multiparts == 0) {
if(messages)
free(messages); |
6638be41 |
return 2; /* Nothing to do */ |
6613d595 |
} |
6638be41 |
|
b151ef55 |
cli_dbgmsg("The message has %d parts\n", multiparts); |
49674596 |
cli_dbgmsg("Find out the multipart type (%s)\n", mimeSubtype); |
b151ef55 |
|
b62a19da |
/*
* We now have all the parts of the multipart message
* in the messages array:
* message *messages[multiparts]
* Let's decide what to do with them all
*/ |
b151ef55 |
switch(tableFind(subtypeTable, mimeSubtype)) {
case RELATED: |
68be129f |
cli_dbgmsg("Multipart related handler\n"); |
b151ef55 |
/* |
295e425f |
* Have a look to see if there's HTML code
* which will need scanning |
b151ef55 |
*/
aMessage = NULL;
assert(multiparts > 0);
|
0bcad2b1 |
htmltextPart = getTextPart(messages, multiparts); |
b151ef55 |
|
0bcad2b1 |
if(htmltextPart >= 0)
aText = textAddMessage(aText, messages[htmltextPart]); |
b151ef55 |
else
/* |
295e425f |
* There isn't an HTML bit. If there's a
* multipart bit, it'll may be in there
* somewhere |
b151ef55 |
*/
for(i = 0; i < multiparts; i++)
if(messageGetMimeType(messages[i]) == MULTIPART) {
aMessage = messages[i]; |
0bcad2b1 |
htmltextPart = i; |
b151ef55 |
break;
}
|
74c6f514 |
if(htmltextPart == -1) |
295e425f |
cli_dbgmsg("No HTML code found to be scanned"); |
74c6f514 |
else { |
565c449d |
rc = parseEmailBody(aMessage, aText, dir, rfc821Table, subtypeTable, options); |
74c6f514 |
if(rc == 1) {
assert(aMessage == messages[htmltextPart]);
messageDestroy(aMessage);
messages[htmltextPart] = NULL;
}
} |
b151ef55 |
/*
* Fixed based on an idea from Stephen White <stephen@earth.li>
* The message is confused about the difference
* between alternative and related. Badtrans.B
* suffers from this problem.
*
* Fall through in this case:
* Content-Type: multipart/related;
* type="multipart/alternative"
*/ |
f5e9abc8 |
/*
* Changed to always fall through based on
* an idea from Michael Dankov <misha@btrc.ru>
* that some viruses are completely confused
* about the difference between related
* and mixed
*/
/*cptr = messageFindArgument(mainMessage, "type"); |
b151ef55 |
if(cptr == NULL)
break;
isAlternative = (bool)(strcasecmp(cptr, "multipart/alternative") == 0);
free((char *)cptr);
if(!isAlternative) |
f5e9abc8 |
break;*/ |
d28e1902 |
case DIGEST:
/*
* According to section 5.1.5 RFC2046, the
* default mime type of multipart/digest parts
* is message/rfc822
*
* We consider them as alternative, wrong in
* the strictest sense since they aren't
* alternatives - all parts a valid - but it's
* OK for our needs since it means each part
* will be scanned
*/ |
b151ef55 |
case ALTERNATIVE:
cli_dbgmsg("Multipart alternative handler\n");
/*
* Fall through - some clients are broken and
* say alternative instead of mixed. The Klez |
e279f3ea |
* virus is broken that way, and anyway we
* wish to scan all of the alternatives |
b151ef55 |
*/
case REPORT:
/*
* According to section 1 of RFC1892, the
* syntax of multipart/report is the same
* as multipart/mixed. There are some required
* parameters, but there's no need for us to
* verify that they exist
*/
case MIXED: |
fdc8a467 |
case APPLEDOUBLE: /* not really supported */ |
b151ef55 |
/*
* Look for attachments
*
* Not all formats are supported. If an
* unsupported format turns out to be
* common enough to implement, it is a simple
* matter to add it
*/ |
2250ea69 |
if(aText) {
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage); |
b151ef55 |
mainMessage = NULL; |
2250ea69 |
} |
b151ef55 |
cli_dbgmsg("Mixed message with %d parts\n", multiparts);
for(i = 0; i < multiparts; i++) {
bool addToText = FALSE;
const char *dtype; |
565c449d |
#ifndef SAVE_TO_DISC |
f54a8635 |
message *body; |
d32343c3 |
#endif |
b151ef55 |
aMessage = messages[i];
|
74c6f514 |
if(aMessage == NULL)
continue; |
b151ef55 |
cli_dbgmsg("Mixed message part %d is of type %d\n",
i, messageGetMimeType(aMessage));
switch(messageGetMimeType(aMessage)) {
case APPLICATION: |
9c8806fb |
case AUDIO:
case IMAGE:
case VIDEO: |
b151ef55 |
break;
case NOMIME: |
c29ebe66 |
cli_dbgmsg("No mime headers found in multipart part %d\n", i); |
7b8fb055 |
if(mainMessage) { |
e82a5185 |
if(binhexBegin(aMessage)) {
cli_dbgmsg("Found binhex message in multipart/mixed mainMessage\n");
messageSetEncoding(mainMessage, "x-bunhex"); |
1e06e1ab |
fb = messageToFileblob(mainMessage, dir); |
7b8fb055 |
|
1e06e1ab |
if(fb)
fileblobDestroy(fb); |
7b8fb055 |
}
if(mainMessage != messageIn)
messageDestroy(mainMessage);
mainMessage = NULL; |
30fb8a0b |
} else if(aMessage) { |
e82a5185 |
if(binhexBegin(aMessage)) { |
39b5a552 |
cli_dbgmsg("Found binhex message in multipart/mixed non mime part\n");
messageSetEncoding(aMessage, "x-binhex");
fb = messageToFileblob(aMessage, dir);
if(fb)
fileblobDestroy(fb);
assert(aMessage == messages[i]);
messageReset(messages[i]); |
30fb8a0b |
} |
7b8fb055 |
} |
b151ef55 |
addToText = TRUE;
if(messageGetBody(aMessage) == NULL)
/*
* No plain text version
*/ |
e82a5185 |
cli_dbgmsg("No plain text alternative"); |
b151ef55 |
break;
case TEXT: |
ef3cf57d |
dtype = messageGetDispositionType(aMessage); |
852e3ce4 |
cli_dbgmsg("Mixed message text part disposition \"%s\"\n",
dtype); |
b151ef55 |
if(strcasecmp(dtype, "attachment") == 0) |
23e1c37c |
break;
if((*dtype == '\0') || (strcasecmp(dtype, "inline") == 0)) { |
2250ea69 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage); |
b151ef55 |
mainMessage = NULL; |
ef3cf57d |
cptr = messageGetMimeSubtype(aMessage); |
e6b25cd3 |
cli_dbgmsg("Mime subtype \"%s\"\n", cptr); |
e82a5185 |
if((tableFind(subtypeTable, cptr) == PLAIN) && |
10c3ed55 |
(messageGetEncoding(aMessage) == NOENCODING)) { |
7584963d |
char *filename; |
852e3ce4 |
/*
* Strictly speaking |
c9ae17be |
* a text/plain part is |
852e3ce4 |
* not an attachment. We
* pretend it is so that
* we can decode and
* scan it
*/ |
7584963d |
filename = (char *)messageFindArgument(aMessage, "filename");
if(filename == NULL)
filename = (char *)messageFindArgument(aMessage, "name");
if(filename == NULL) {
cli_dbgmsg("Adding part to main message\n");
addToText = TRUE;
} else {
cli_dbgmsg("Treating %s as attachment\n",
filename);
free(filename);
} |
852e3ce4 |
} else { |
06d4e856 |
if(options&CL_SCAN_MAILURL) |
c9ae17be |
if(tableFind(subtypeTable, cptr) == HTML)
checkURLs(aMessage, dir); |
f52d7358 |
messageAddArgument(aMessage, "filename=mixedtextportion"); |
b151ef55 |
}
} else { |
b4cb4486 |
cli_dbgmsg("Text type %s is not supported\n", dtype); |
b151ef55 |
continue;
}
break;
case MESSAGE: |
2e0f78a6 |
/* Content-Type: message/rfc822 */ |
27116a59 |
cli_dbgmsg("Found message inside multipart (encoding type %d)\n",
messageGetEncoding(aMessage)); |
388072d8 |
switch(messageGetEncoding(aMessage)) {
case NOENCODING:
case EIGHTBIT:
case BINARY:
if(encodingLine(aMessage) == NULL) {
/*
* This means that the message has no attachments
* The test for messageGetEncoding is needed since
* encodingLine won't have been set if the message
* itself has been encoded
*/
cli_dbgmsg("No encoding line found in the multipart/message\n");
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL;
continue;
} |
ffd59a3e |
} |
388072d8 |
#if 0 |
de617e3e |
messageAddStrAtTop(aMessage, |
0856891e |
"Received: by clamd (message/rfc822)"); |
27116a59 |
#endif |
4465fb04 |
#ifdef SAVE_TO_DISC
/*
* Save this embedded message
* to a temporary file
*/
saveTextPart(aMessage, dir);
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL;
#else
/*
* Scan in memory, faster but
* is open to DoS attacks when
* many nested levels are
* involved.
*/ |
735377bc |
body = parseEmailHeaders(aMessage, rfc821Table, TRUE); |
bad123c6 |
/*
* We've fininished with the
* original copy of the message,
* so throw that away and
* deal with the encapsulated
* message as a message.
* This can save a lot of memory
*/
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL; |
f54a8635 |
if(body) { |
565c449d |
rc = parseEmailBody(body, NULL, dir, rfc821Table, subtypeTable, options); |
f54a8635 |
messageDestroy(body);
} |
4465fb04 |
#endif |
b151ef55 |
continue;
case MULTIPART:
/*
* It's a multi part within a multi part
* Run the message parser on this bit, it won't
* be an attachment
*/
cli_dbgmsg("Found multipart inside multipart\n"); |
d32343c3 |
if(aMessage) {
/*
* The headers were parsed when reading in the
* whole multipart section
*/ |
565c449d |
rc = parseEmailBody(aMessage, aText, dir, rfc821Table, subtypeTable, options); |
d32343c3 |
cli_dbgmsg("Finished recursion\n");
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL; |
f54a8635 |
} else { |
565c449d |
rc = parseEmailBody(NULL, NULL, dir, rfc821Table, subtypeTable, options); |
2250ea69 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage); |
f54a8635 |
mainMessage = NULL;
} |
b151ef55 |
continue;
default: |
c7256385 |
cli_warnmsg("Only text and application attachments are supported, type = %d\n", |
b151ef55 |
messageGetMimeType(aMessage));
continue;
}
|
c29ebe66 |
if(addToText) {
cli_dbgmsg("Adding to non mime-part\n"); |
b151ef55 |
aText = textAdd(aText, messageGetBody(aMessage)); |
c29ebe66 |
} else { |
565c449d |
fb = messageToFileblob(aMessage, dir); |
b151ef55 |
|
1e06e1ab |
if(fb)
fileblobDestroy(fb); |
b151ef55 |
} |
6638be41 |
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL; |
b151ef55 |
}
|
565c449d |
/* rc = parseEmailBody(NULL, NULL, dir, rfc821Table, subtypeTable, options); */ |
b151ef55 |
break;
case SIGNED:
case PARALLEL:
/*
* If we're here it could be because we have a
* multipart/mixed message, consisting of a
* message followed by an attachment. That
* message itself is a multipart/alternative
* message and we need to dig out the plain
* text part of that alternative
*/ |
0bcad2b1 |
htmltextPart = getTextPart(messages, multiparts);
if(htmltextPart == -1)
htmltextPart = 0; |
b151ef55 |
|
565c449d |
rc = parseEmailBody(messages[htmltextPart], aText, dir, rfc821Table, subtypeTable, options); |
b151ef55 |
break; |
b62a19da |
case ENCRYPTED:
rc = 0; |
cbc2eaa9 |
protocol = (char *)messageFindArgument(mainMessage, "protocol"); |
b62a19da |
if(protocol) {
if(strcasecmp(protocol, "application/pgp-encrypted") == 0) {
/* RFC2015 */
cli_warnmsg("PGP encoded attachment not scanned\n");
rc = 2;
} else |
e82a5185 |
cli_warnmsg("Unknown encryption protocol '%s' - if you believe this file contains a virus, submit it to www.clamav.net\n", protocol); |
b62a19da |
free(protocol);
} else |
23e1c37c |
cli_dbgmsg("Encryption method missing protocol name\n"); |
b62a19da |
break; |
b151ef55 |
default: |
e279f3ea |
assert(0); |
b151ef55 |
}
|
2250ea69 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage);
|
c29ebe66 |
if(aText && (textIn == NULL)) {
if((fb = fileblobCreate()) != NULL) { |
10c3ed55 |
cli_dbgmsg("Save non mime and/or text/plain part\n"); |
c29ebe66 |
fileblobSetFilename(fb, dir, "textpart"); |
b5053e3f |
/*fileblobAddData(fb, "Received: by clamd (textpart)\n", 30);*/ |
60aec445 |
(void)textToFileblob(aText, fb); |
c29ebe66 |
fileblobDestroy(fb);
} |
c6259ac5 |
textDestroy(aText); |
c29ebe66 |
} |
c6259ac5 |
|
0856891e |
for(i = 0; i < multiparts; i++)
if(messages[i])
messageDestroy(messages[i]);
|
6613d595 |
if(messages)
free(messages);
|
b151ef55 |
return rc;
case MESSAGE:
/*
* Check for forbidden encodings
*/
switch(messageGetEncoding(mainMessage)) {
case NOENCODING:
case EIGHTBIT:
case BINARY:
break;
default: |
c6259ac5 |
cli_warnmsg("MIME type 'message' cannot be decoded\n"); |
b151ef55 |
break;
} |
9a7398ee |
rc = 0; |
c6259ac5 |
if((strcasecmp(mimeSubtype, "rfc822") == 0) ||
(strcasecmp(mimeSubtype, "delivery-status") == 0)) { |
de617e3e |
message *m = parseEmailHeaders(mainMessage, rfc821Table); |
bad123c6 |
if(m) {
cli_dbgmsg("Decode rfc822");
|
4465fb04 |
if(mainMessage && (mainMessage != messageIn)) {
messageDestroy(mainMessage);
mainMessage = NULL; |
74c6f514 |
} else
messageReset(mainMessage); |
bad123c6 |
if(messageGetBody(m)) |
565c449d |
rc = parseEmailBody(m, NULL, dir, rfc821Table, subtypeTable, options); |
bad123c6 |
messageDestroy(m);
} |
b151ef55 |
break; |
f8c25c7a |
} else if(strcasecmp(mimeSubtype, "disposition-notification") == 0) { |
d6e30cce |
/* RFC 2298 - handle like a normal email */ |
f8c25c7a |
rc = 1; |
d6e30cce |
break; |
f8c25c7a |
} else if(strcasecmp(mimeSubtype, "partial") == 0) { |
9a7398ee |
#ifdef PARTIAL_DIR
/* RFC1341 message split over many emails */
if(rfc1341(mainMessage, dir) >= 0)
rc = 1;
#else |
e94471f4 |
cli_warnmsg("Partial message received from MUA/MTA - message cannot be scanned\n"); |
9a7398ee |
rc = 0;
#endif
} else if(strcasecmp(mimeSubtype, "external-body") == 0) |
e94471f4 |
/* TODO */ |
b151ef55 |
cli_warnmsg("Attempt to send Content-type message/external-body trapped"); |
bf8ea488 |
else |
e82a5185 |
cli_warnmsg("Unsupported message format `%s' - if you believe this file contains a virus, submit it to www.clamav.net\n", mimeSubtype); |
b151ef55 |
|
9a7398ee |
|
2250ea69 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage); |
6613d595 |
if(messages)
free(messages); |
9a7398ee |
return rc; |
b151ef55 |
case APPLICATION: |
23e1c37c |
/*cptr = messageGetMimeSubtype(mainMessage); |
0bcad2b1 |
|
23e1c37c |
if((strcasecmp(cptr, "octet-stream") == 0) || |
04421a14 |
(strcasecmp(cptr, "x-msdownload") == 0)) {*/
{ |
565c449d |
fb = messageToFileblob(mainMessage, dir); |
b151ef55 |
|
1e06e1ab |
if(fb) {
cli_dbgmsg("Saving main message as attachment\n");
fileblobDestroy(fb); |
43756987 |
if(mainMessage != messageIn) {
messageDestroy(mainMessage);
mainMessage = NULL;
} else
messageReset(mainMessage); |
b151ef55 |
} |
04421a14 |
} /*else
cli_warnmsg("Discarded application not sent as attachment\n");*/ |
b151ef55 |
break;
case AUDIO:
case VIDEO:
case IMAGE:
break;
default:
cli_warnmsg("Message received with unknown mime encoding");
break;
}
}
|
d32343c3 |
if(aText && (textIn == NULL)) { |
22f3b19b |
/* Look for a bounce in the text (non mime encoded) portion */
const text *t; |
0856891e |
|
22f3b19b |
for(t = aText; t; t = t->t_next) {
const line_t *l = t->t_line; |
4b54f2e0 |
const text *lookahead, *topofbounce; |
22f3b19b |
const char *s; |
4b54f2e0 |
bool inheader; |
0856891e |
|
22f3b19b |
if(l == NULL)
continue;
s = lineGetData(l);
if(cli_filetype(s, strlen(s)) != CL_TYPE_MAIL)
continue;
/*
* We've found what looks like the start of a bounce
* message. Only bother saving if it really is a bounce
* message, this helps to speed up scanning of ping-pong
* messages that have lots of bounces within bounces in
* them
*/
for(lookahead = t->t_next; lookahead; lookahead = lookahead->t_next) {
l = lookahead->t_line;
if(l == NULL)
break;
s = lineGetData(l);
if(strncasecmp(s, "Content-Type:", 13) == 0)
/*
* Don't bother with plain/text or
* plain/html
*/
if(strstr(s, "text/") == NULL) |
0d9e07a9 |
/*
* Don't bother to save the unuseful
* part
*/ |
22f3b19b |
break;
}
if(lookahead && (lookahead->t_line == NULL)) {
cli_dbgmsg("Non mime part bounce message is not mime encoded, so it will not be scanned\n");
t = lookahead;
/* look for next bounce message */
continue;
}
|
0d9e07a9 |
/*
* Prescan the bounce message to see if there's likely
* to be anything nasty.
* This algorithm is hand crafted and may be breakable
* so all submissions are welcome. It's best NOT to
* remove this however you may be tempted, because it
* significantly speeds up the scanning of multiple
* bounces (i.e. bounces within many bounces)
*/
for(; lookahead; lookahead = lookahead->t_next) {
l = lookahead->t_line;
if(l) {
s = lineGetData(l);
if((strncasecmp(s, "Content-Type:", 13) == 0) &&
(strstr(s, "multipart/") == NULL) &&
(strstr(s, "message/rfc822") == NULL) &&
(strstr(s, "text/plain") == NULL))
break;
}
}
if(lookahead == NULL) { |
f60a8d41 |
cli_dbgmsg("cli_mbox: I believe it's plain text which must be clean\n"); |
0d9e07a9 |
/* nothing here, move along please */
break;
} |
4b54f2e0 |
if((fb = fileblobCreate()) == NULL)
break;
cli_dbgmsg("Save non mime part bounce message\n");
fileblobSetFilename(fb, dir, "bounce");
fileblobAddData(fb, (unsigned char *)"Received: by clamd (bounce)\n", 28);
inheader = TRUE;
topofbounce = NULL;
for(;;) {
l = t->t_line;
if(l == NULL) {
if(inheader) {
inheader = FALSE;
topofbounce = t;
}
} else {
s = lineGetData(l);
fileblobAddData(fb, (unsigned char *)s, strlen(s));
}
fileblobAddData(fb, (unsigned char *)"\n", 1);
lookahead = t->t_next;
if(lookahead == NULL)
break;
t = lookahead;
l = t->t_line;
if((!inheader) && l) {
s = lineGetData(l);
if(cli_filetype(s, strlen(s)) == CL_TYPE_MAIL) {
cli_dbgmsg("Found the start of another bounce candidate\n");
break;
}
} |
22f3b19b |
} |
4b54f2e0 |
fileblobDestroy(fb);
if(topofbounce)
t = topofbounce;
/*
* Don't do this - it slows bugs.txt
*/
/*if(mainMessage)
mainMessage->bounce = NULL;*/ |
22f3b19b |
} |
d32343c3 |
textDestroy(aText);
aText = NULL;
}
|
565c449d |
/*
* No attachments - scan the text portions, often files
* are hidden in HTML code
*/
cli_dbgmsg("%d multiparts found\n", multiparts);
for(i = 0; i < multiparts; i++) {
fb = messageToFileblob(messages[i], dir); |
b151ef55 |
|
565c449d |
if(fb) { |
0e3b08fc |
cli_dbgmsg("Saving multipart %d\n", i); |
0bcad2b1 |
|
565c449d |
fileblobDestroy(fb);
}
}
if(mainMessage) { |
b151ef55 |
/* |
565c449d |
* Look for uu-encoded main file |
b151ef55 |
*/ |
565c449d |
const text *t_line;
|
e82a5185 |
if((encodingLine(mainMessage) != NULL) && |
f1c33aa0 |
((t_line = bounceBegin(mainMessage)) != NULL)) { |
0856891e |
const text *t, *start; |
0bcad2b1 |
/* |
565c449d |
* Attempt to save the original (unbounced)
* message - clamscan will find that in the
* directory and call us again (with any luck) |
8386482b |
* having found an e-mail message to handle. |
565c449d |
*
* This finds a lot of false positives, the |
8386482b |
* search that a content type is in the |
565c449d |
* bounce (i.e. it's after the bounce header) |
8386482b |
* helps a bit.
*
* messageAddLine |
565c449d |
* optimisation could help here, but needs
* careful thought, do it with line numbers
* would be best, since the current method in
* messageAddLine of checking encoding first
* must remain otherwise non bounce messages
* won't be scanned |
0bcad2b1 |
*/ |
0856891e |
for(t = start = t_line; t; t = t->t_next) { |
e82a5185 |
char cmd[RFC2821LENGTH + 1]; |
565c449d |
const char *txt = lineGetData(t->t_line);
|
0856891e |
if(txt == NULL)
continue;
if(cli_strtokbuf(txt, 0, ":", cmd) == NULL)
continue;
switch(tableFind(rfc821Table, cmd)) {
case CONTENT_TRANSFER_ENCODING:
if((strstr(txt, "7bit") == NULL) &&
(strstr(txt, "8bit") == NULL))
break;
continue;
case CONTENT_DISPOSITION:
break;
case CONTENT_TYPE:
if(strstr(txt, "text/plain") != NULL)
t = NULL;
break;
default:
if(strcasecmp(cmd, "From") == 0)
start = t_line;
else if(strcasecmp(cmd, "Received") == 0)
start = t_line;
continue; |
8386482b |
} |
0856891e |
break; |
565c449d |
}
if(t && ((fb = fileblobCreate()) != NULL)) {
cli_dbgmsg("Found a bounce message\n");
fileblobSetFilename(fb, dir, "bounce"); |
cbe29191 |
if(textToFileblob(start, fb) == NULL)
cli_dbgmsg("Nothing new to save in the bounce message");
else
rc = 1; |
565c449d |
fileblobDestroy(fb); |
e745ac7e |
} else
cli_dbgmsg("Not found a bounce message\n"); |
565c449d |
} else {
bool saveIt; |
0bcad2b1 |
|
565c449d |
if(messageGetMimeType(mainMessage) == MESSAGE) |
15c8cace |
/* |
565c449d |
* Quick peek, if the encapsulated
* message has no
* content encoding statement don't
* bother saving to scan, it's safe |
15c8cace |
*/ |
565c449d |
saveIt = (encodingLine(mainMessage) != NULL);
else if((t_line = encodingLine(mainMessage)) != NULL) { |
92915cee |
/* |
565c449d |
* Some bounces include the message
* body without the headers. |
4b54f2e0 |
* FIXME: Unfortunately this generates a |
565c449d |
* lot of false positives that a bounce
* has been found when it hasn't. |
92915cee |
*/ |
565c449d |
if((fb = fileblobCreate()) != NULL) { |
b3a5cdd8 |
cli_dbgmsg("Found a bounce message with no header at '%s'\n",
lineGetData(t_line->t_line)); |
1e06e1ab |
fileblobSetFilename(fb, dir, "bounce"); |
00615ec9 |
fileblobAddData(fb,
(const unsigned char *)"Received: by clamd (bounce)\n",
28); |
5a01973c |
|
565c449d |
fb = textToFileblob(t_line, fb); |
b759d5eb |
|
565c449d |
fileblobDestroy(fb); |
b759d5eb |
} |
565c449d |
saveIt = FALSE; |
74c6f514 |
} else if(multiparts == 0) |
565c449d |
/*
* Save the entire text portion,
* since it it may be an HTML file with
* a JavaScript virus
*/
saveIt = TRUE; |
74c6f514 |
else
saveIt = FALSE; |
b151ef55 |
|
565c449d |
if(saveIt) {
cli_dbgmsg("Saving text part to scan\n");
/*
* TODO: May be better to save aText
*/
saveTextPart(mainMessage, dir); |
74c6f514 |
if(mainMessage != messageIn) {
messageDestroy(mainMessage);
mainMessage = NULL;
} else
messageReset(mainMessage);
rc = 1; |
c6259ac5 |
} |
b151ef55 |
} |
565c449d |
} else
rc = (multiparts) ? 1 : 2; /* anything saved? */ |
b151ef55 |
|
2250ea69 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage);
|
6613d595 |
if(messages)
free(messages);
|
68be129f |
cli_dbgmsg("parseEmailBody() returning %d\n", rc); |
b151ef55 |
|
68be129f |
return rc; |
b151ef55 |
}
/*
* Is the current line the start of a new section?
*
* New sections start with --boundary
*/
static int
boundaryStart(const char *line, const char *boundary)
{ |
273cd2bb |
char *ptr, *out; |
2ed1bc5a |
int rc; |
e82a5185 |
char buf[RFC2821LENGTH + 1]; |
ad642304 |
|
80a8c7d8 |
if(line == NULL)
return 0; /* empty line */
|
c29ebe66 |
/*cli_dbgmsg("boundaryStart: line = '%s' boundary = '%s'\n", line, boundary);*/ |
ad642304 |
|
4b54f2e0 |
if((*line != '-') && (*line != '('))
return 0;
if(strchr(line, '-') == NULL)
return 0;
|
273cd2bb |
if(strlen(line) <= sizeof(buf)) {
out = NULL;
ptr = rfc822comments(line, buf);
} else
out = ptr = rfc822comments(line, NULL);
|
ad642304 |
if(ptr == NULL) |
aa479b7d |
ptr = (char *)line; |
ad642304 |
if(*ptr++ != '-') { |
273cd2bb |
if(out)
free(out); |
80a8c7d8 |
return 0; |
ad642304 |
} |
80a8c7d8 |
|
b151ef55 |
/* |
80a8c7d8 |
* Gibe.B3 is broken, it has: |
b151ef55 |
* boundary="---- =_NextPart_000_01C31177.9DC7C000"
* but it's boundaries look like
* ------ =_NextPart_000_01C31177.9DC7C000 |
80a8c7d8 |
* notice the one too few '-'.
* Presumably this is a deliberate exploitation of a bug in some mail
* clients.
*
* The trouble is that this creates a lot of false positives for
* boundary conditions, if we're too lax about matches. We do our level
* best to avoid these false positives. For example if we have
* boundary="1" we want to ensure that we don't break out of every line
* that has -1 in it instead of starting --1. This needs some more work. |
f60a8d41 |
*
* Look with and without RFC822 comments stripped, I've seen some
* samples where () are taken as comments in boundaries and some where
* they're not. Irrespective of whatever RFC2822 says we need to find
* viruses in both types of mails |
b151ef55 |
*/ |
f60a8d41 |
if((strstr(ptr, boundary) != NULL) || (strstr(line, boundary) != NULL)) |
2ed1bc5a |
rc = 1;
else if(*ptr++ != '-')
rc = 0;
else |
4ce9c996 |
rc = (strcasecmp(ptr, boundary) == 0); |
2ed1bc5a |
|
273cd2bb |
if(out)
free(out); |
2ed1bc5a |
if(rc == 1)
cli_dbgmsg("boundaryStart: found %s in %s\n", boundary, line);
return rc; |
b151ef55 |
}
/*
* Is the current line the end?
*
* The message ends with with --boundary--
*/
static int
endOfMessage(const char *line, const char *boundary)
{
size_t len;
|
98685ac1 |
if(line == NULL)
return 0; |
c29ebe66 |
/*cli_dbgmsg("endOfMessage: line = '%s' boundary = '%s'\n", line, boundary);*/ |
b151ef55 |
if(*line++ != '-')
return 0;
if(*line++ != '-')
return 0;
len = strlen(boundary); |
c6259ac5 |
if(strncasecmp(line, boundary, len) != 0)
return 0; |
ef3cf57d |
/*
* Use < rather than == because some broken mails have white
* space after the boundary
*/ |
39b5a552 |
if(strlen(line) < (len + 2)) |
b151ef55 |
return 0;
line = &line[len];
if(*line++ != '-')
return 0;
return *line == '-';
}
/*
* Initialise the various lookup tables
*/
static int
initialiseTables(table_t **rfc821Table, table_t **subtypeTable)
{
const struct tableinit *tableinit;
/*
* Initialise the various look up tables
*/
*rfc821Table = tableCreate();
assert(*rfc821Table != NULL);
for(tableinit = rfc821headers; tableinit->key; tableinit++) |
7b8fb055 |
if(tableInsert(*rfc821Table, tableinit->key, tableinit->value) < 0) {
tableDestroy(*rfc821Table); |
4d9c0ca8 |
*rfc821Table = NULL; |
b151ef55 |
return -1; |
7b8fb055 |
} |
b151ef55 |
*subtypeTable = tableCreate();
assert(*subtypeTable != NULL);
for(tableinit = mimeSubtypes; tableinit->key; tableinit++)
if(tableInsert(*subtypeTable, tableinit->key, tableinit->value) < 0) {
tableDestroy(*rfc821Table); |
7b8fb055 |
tableDestroy(*subtypeTable); |
4d9c0ca8 |
*rfc821Table = NULL;
*subtypeTable = NULL; |
b151ef55 |
return -1;
}
return 0;
}
/* |
0bcad2b1 |
* If there's a HTML text version use that, otherwise |
b151ef55 |
* use the first text part, otherwise just use the |
0bcad2b1 |
* first one around. HTML text is most likely to include
* a scripting worm |
b151ef55 |
*
* If we can't find one, return -1
*/
static int
getTextPart(message *const messages[], size_t size)
{
size_t i; |
b4cb4486 |
int textpart = -1; |
b151ef55 |
for(i = 0; i < size; i++) {
assert(messages[i] != NULL); |
b4cb4486 |
if(messageGetMimeType(messages[i]) == TEXT) {
if(strcasecmp(messageGetMimeSubtype(messages[i]), "html") == 0)
return (int)i;
textpart = (int)i;
} |
b151ef55 |
} |
b4cb4486 |
return textpart; |
b151ef55 |
}
/*
* strip - |
4d9c0ca8 |
* Remove the trailing spaces from a buffer. Don't call this directly,
* always call strstrip() which is a wrapper to this routine to be used with
* NUL terminated strings. This code looks a bit strange because of it's
* heritage from code that worked on strings that weren't necessarily NUL
* terminated.
* TODO: rewrite for clamAV
* |
b151ef55 |
* Returns it's new length (a la strlen)
*
* len must be int not size_t because of the >= 0 test, it is sizeof(buf)
* not strlen(buf)
*/
static size_t
strip(char *buf, int len)
{
register char *ptr;
register size_t i;
if((buf == NULL) || (len <= 0)) |
4d9c0ca8 |
return 0; |
b151ef55 |
i = strlen(buf);
if(len > (int)(i + 1)) |
4d9c0ca8 |
return i; |
b151ef55 |
ptr = &buf[--len];
|
f0635204 |
#if defined(UNIX) || defined(C_LINUX) || defined(C_DARWIN) || defined(C_KFREEBSD_GNU) /* watch - it may be in shared text area */ |
b151ef55 |
do
if(*ptr)
*ptr = '\0'; |
6b93ea0c |
while((--len >= 0) && (!isgraph(*--ptr)) && (*ptr != '\n') && (*ptr != '\r')); |
b151ef55 |
#else /* more characters can be displayed on DOS */
do
#ifndef REAL_MODE_DOS
if(*ptr) /* C8.0 puts into a text area */
#endif
*ptr = '\0';
while((--len >= 0) && ((*--ptr == '\0') || (isspace((int)*ptr))));
#endif
return((size_t)(len + 1));
}
/*
* strstrip:
* Strip a given string
*/ |
3db105a2 |
size_t |
b151ef55 |
strstrip(char *s)
{
if(s == (char *)NULL)
return(0); |
98685ac1 |
|
b151ef55 |
return(strip(s, strlen(s) + 1));
}
/* |
10c3ed55 |
* Some broken email headers use ';' at the end of a line to continue
* to the next line and don't add a leading white space on the next line |
b151ef55 |
*/
static bool
continuationMarker(const char *line)
{ |
10c3ed55 |
const char *ptr;
if(line == NULL)
return FALSE;
#ifdef CL_DEBUG
cli_dbgmsg("continuationMarker(%s)\n", line);
#endif
if(strlen(line) == 0)
return FALSE;
ptr = strchr(line, '\0');
assert(ptr != NULL);
while(ptr > line)
switch(*--ptr) {
case '\n':
case '\r':
case ' ':
case '\t':
continue;
case ';':
return TRUE;
default:
return FALSE;
}
|
b151ef55 |
return FALSE;
}
static int
parseMimeHeader(message *m, const char *cmd, const table_t *rfc821Table, const char *arg)
{ |
0cb54827 |
char *copy, *p;
const char *ptr; |
56d8328d |
int commandNumber; |
f2b068fb |
|
b151ef55 |
cli_dbgmsg("parseMimeHeader: cmd='%s', arg='%s'\n", cmd, arg); |
56d8328d |
|
273cd2bb |
copy = rfc822comments(cmd, NULL); |
0cb54827 |
if(copy) {
commandNumber = tableFind(rfc821Table, copy);
free(copy); |
e9bdeb72 |
} else
commandNumber = tableFind(rfc821Table, cmd); |
56d8328d |
|
273cd2bb |
copy = rfc822comments(arg, NULL); |
b151ef55 |
|
0cb54827 |
if(copy)
ptr = copy;
else
ptr = arg; |
ad3d1172 |
|
56d8328d |
switch(commandNumber) { |
b151ef55 |
case CONTENT_TYPE:
/*
* Fix for non RFC1521 compliant mailers
* that send content-type: Text instead
* of content-type: Text/Plain, or
* just simply "Content-Type:"
*/ |
a8c7e017 |
if(arg == NULL) |
f1c33aa0 |
/*
* According to section 4 of RFC1521:
* "Note also that a subtype specification is
* MANDATORY. There are no default subtypes"
* |
21cd233d |
* We have to break this and make an assumption |
f1c33aa0 |
* for the subtype because virus writers and
* email client writers don't get it right
*/
cli_warnmsg("Empty content-type received, no subtype specified, assuming text/plain; charset=us-ascii\n"); |
0cb54827 |
else if(strchr(ptr, '/') == NULL) |
f1c33aa0 |
/*
* Empty field, such as
* Content-Type:
* which I believe is illegal according to
* RFC1521
*/ |
0cb54827 |
cli_dbgmsg("Invalid content-type '%s' received, no subtype specified, assuming text/plain; charset=us-ascii\n", ptr); |
b151ef55 |
else { |
137740e1 |
int i; |
6fd6d771 |
char *mimeArgs; /* RHS of the ; */
|
b151ef55 |
/*
* Some clients are broken and
* put white space after the ;
*/ |
7e572372 |
if(*arg == '/') {
cli_warnmsg("Content-type '/' received, assuming application/octet-stream\n");
messageSetMimeType(m, "application");
messageSetMimeSubtype(m, "octet-stream");
} else {
/* |
de509b8e |
* The content type could be in quotes:
* Content-Type: "multipart/mixed"
* FIXME: this is a hack in that ignores
* the quotes, it doesn't handle
* them properly |
7e572372 |
*/ |
0cb54827 |
while(isspace(*ptr))
ptr++;
if(ptr[0] == '\"')
ptr++; |
de509b8e |
|
0cb54827 |
if(ptr[0] != '/') { |
e9bdeb72 |
char *s;
char *mimeType; /* LHS of the ; */ |
0cb54827 |
#ifdef CL_THREAD_SAFE
char *strptr;
#endif |
e9bdeb72 |
|
0cb54827 |
s = mimeType = cli_strtok(ptr, 0, ";"); |
de509b8e |
/* |
6fd6d771 |
* Handle
* Content-Type: foo/bar multipart/mixed
* and
* Content-Type: multipart/mixed foo/bar |
de509b8e |
*/ |
803055ea |
if(s && *s) for(;;) { |
897fd9c7 |
#ifdef CL_THREAD_SAFE |
6fd6d771 |
int set = messageSetMimeType(m, strtok_r(s, "/", &strptr)); |
897fd9c7 |
#else
int set = messageSetMimeType(m, strtok(s, "/"));
#endif |
6fd6d771 |
/*
* Stephen White <stephen@earth.li>
* Some clients put space after
* the mime type but before
* the ;
*/ |
897fd9c7 |
#ifdef CL_THREAD_SAFE |
6fd6d771 |
s = strtok_r(NULL, ";", &strptr); |
897fd9c7 |
#else
s = strtok(NULL, ";");
#endif |
6fd6d771 |
if(s == NULL)
break;
if(set) { |
e9bdeb72 |
size_t len = strstrip(s) - 1; |
6fd6d771 |
if(s[len] == '\"') {
s[len] = '\0';
len = strstrip(s);
}
if(len) { |
e9bdeb72 |
if(strchr(s, ' ')) {
char *t = cli_strtok(s, 0, " "); |
6fd6d771 |
|
e9bdeb72 |
messageSetMimeSubtype(m, t);
free(t);
} else
messageSetMimeSubtype(m, s); |
6fd6d771 |
} |
de509b8e |
} |
6fd6d771 |
while(*s && !isspace(*s))
s++;
if(*s++ == '\0')
break;
if(*s == '\0')
break; |
de509b8e |
} |
803055ea |
if(mimeType)
free(mimeType); |
e9bdeb72 |
} |
7e572372 |
} |
b151ef55 |
/* |
0704dad8 |
* Add in all rest of the the arguments.
* e.g. if the header is this:
* Content-Type:', arg='multipart/mixed; boundary=foo
* we find the boundary argument set it |
b151ef55 |
*/ |
137740e1 |
i = 1; |
0cb54827 |
while((mimeArgs = cli_strtok(ptr, i++, ";")) != NULL) { |
137740e1 |
cli_dbgmsg("mimeArgs = '%s'\n", mimeArgs);
|
6fd6d771 |
messageAddArguments(m, mimeArgs);
free(mimeArgs);
} |
b151ef55 |
}
break;
case CONTENT_TRANSFER_ENCODING: |
0cb54827 |
messageSetEncoding(m, ptr); |
b151ef55 |
break;
case CONTENT_DISPOSITION: |
0cb54827 |
p = cli_strtok(ptr, 0, ";");
if(p) {
if(*p) {
messageSetDispositionType(m, p);
free(p);
p = cli_strtok(ptr, 1, ";");
messageAddArgument(m, p);
}
free(p); |
897fd9c7 |
} |
55b1a5f8 |
if((p = (char *)messageFindArgument(m, "filename")) == NULL)
/*
* Handle this type of header, without
* a filename (e.g. some Worm.Torvil.D)
* Content-ID: <nRfkHdrKsAxRU>
* Content-Transfer-Encoding: base64
* Content-Disposition: attachment
*/
messageAddArgument(m, "filename=unknown");
else
free(p); |
b151ef55 |
} |
0cb54827 |
if(copy)
free(copy); |
b151ef55 |
|
f2b068fb |
return 0; |
b151ef55 |
}
|
68be129f |
/* |
5a01973c |
* Save the text portion of the message
*/
static void
saveTextPart(message *m, const char *dir)
{ |
1e06e1ab |
fileblob *fb; |
5a01973c |
messageAddArgument(m, "filename=textportion"); |
1e06e1ab |
if((fb = messageToFileblob(m, dir)) != NULL) { |
5a01973c |
/*
* Save main part to scan that
*/ |
37819555 |
cli_dbgmsg("Saving main message\n"); |
5a01973c |
|
1e06e1ab |
fileblobDestroy(fb); |
5a01973c |
}
}
|
90905415 |
/* |
e9bdeb72 |
* Handle RFC822 comments in headers. |
273cd2bb |
* If out == NULL, return a buffer without the comments, the caller must free
* the returned buffer
* Return NULL on error or if the input * has no comments. |
e9bdeb72 |
* See secion 3.4.3 of RFC822 |
90905415 |
* TODO: handle comments that go on to more than one line
*/
static char * |
273cd2bb |
rfc822comments(const char *in, char *out) |
90905415 |
{
const char *iptr; |
273cd2bb |
char *optr; |
90905415 |
int backslash, inquote, commentlevel;
if(in == NULL) |
e9bdeb72 |
return NULL; |
90905415 |
if(strchr(in, '(') == NULL) |
e9bdeb72 |
return NULL; |
90905415 |
|
273cd2bb |
assert(out != in);
if(out == NULL) {
out = cli_malloc(strlen(in) + 1);
if(out == NULL)
return NULL;
} |
90905415 |
backslash = commentlevel = inquote = 0;
optr = out;
cli_dbgmsg("rfc822comments: contains a comment\n");
for(iptr = in; *iptr; iptr++)
if(backslash) { |
0a94ffaf |
if(commentlevel == 0)
*optr++ = *iptr; |
90905415 |
backslash = 0;
} else switch(*iptr) {
case '\\':
backslash = 1;
break;
case '\"': |
f60a8d41 |
*optr++ = '\"'; |
90905415 |
inquote = !inquote;
break;
case '(': |
f60a8d41 |
if(inquote)
*optr++ = '(';
else
commentlevel++; |
90905415 |
break;
case ')': |
f60a8d41 |
if(inquote)
*optr++ = ')';
else if(commentlevel > 0) |
90905415 |
commentlevel--;
break;
default:
if(commentlevel == 0)
*optr++ = *iptr;
}
if(backslash) /* last character was a single backslash */
*optr++ = '\\';
*optr = '\0';
|
273cd2bb |
/*strstrip(out);*/ |
90905415 |
cli_dbgmsg("rfc822comments '%s'=>'%s'\n", in, out);
return out;
} |
0674e2af |
/*
* Handle RFC2047 encoding. Returns a malloc'd buffer that the caller must
* free, or NULL on error
*/
static char *
rfc2047(const char *in)
{
char *out, *pout;
size_t len;
|
95f98162 |
if((strstr(in, "=?") == NULL) || (strstr(in, "?=") == NULL)) |
0674e2af |
return strdup(in);
cli_dbgmsg("rfc2047 '%s'\n", in);
out = cli_malloc(strlen(in) + 1);
if(out == NULL)
return NULL;
pout = out;
/* For each RFC2047 string */
while(*in) { |
291ac47f |
char encoding, *ptr, *enctext; |
0674e2af |
message *m;
blob *b;
/* Find next RFC2047 string */
while(*in) {
if((*in == '=') && (in[1] == '?')) {
in += 2;
break;
}
*pout++ = *in++;
}
/* Skip over charset, find encoding */
while((*in != '?') && *in)
in++;
if(*in == '\0')
break;
encoding = *++in;
encoding = tolower(encoding);
if((encoding != 'q') && (encoding != 'b')) { |
e82a5185 |
cli_warnmsg("Unsupported RFC2047 encoding type '%c' - if you believe this file contains a virus, submit it to www.clamav.net\n", encoding); |
c3400886 |
free(out);
out = NULL; |
0674e2af |
break;
}
/* Skip to encoded text */
if(*++in != '?')
break;
if(*++in == '\0')
break;
|
291ac47f |
enctext = strdup(in);
if(enctext == NULL) {
free(out);
out = NULL;
break;
} |
0674e2af |
in = strstr(in, "?="); |
291ac47f |
if(in == NULL) {
free(enctext); |
0674e2af |
break; |
291ac47f |
} |
0674e2af |
in += 2;
ptr = strstr(enctext, "?=");
assert(ptr != NULL);
*ptr = '\0';
/*cli_dbgmsg("Need to decode '%s' with method '%c'\n", enctext, encoding);*/
m = messageCreate(); |
37819555 |
if(m == NULL) |
0674e2af |
break; |
321d5c00 |
messageAddStr(m, enctext); |
291ac47f |
free(enctext); |
4d9c0ca8 |
switch(encoding) { |
0674e2af |
case 'q':
messageSetEncoding(m, "quoted-printable");
break;
case 'b':
messageSetEncoding(m, "base64");
break;
}
b = messageToBlob(m);
len = blobGetDataSize(b);
cli_dbgmsg("Decoded as '%*.*s'\n", len, len, blobGetData(b));
memcpy(pout, blobGetData(b), len);
blobDestroy(b);
messageDestroy(m);
if(pout[len - 1] == '\n')
pout += len - 1;
else
pout += len;
} |
5e5a162c |
if(out == NULL)
return NULL;
*pout = '\0'; |
0674e2af |
|
5e5a162c |
cli_dbgmsg("rfc2047 returns '%s'\n", out); |
0674e2af |
return out;
}
|
9a7398ee |
#ifdef PARTIAL_DIR
/*
* Handle partial messages
*/
static int
rfc1341(message *m, const char *dir)
{
fileblob *fb; |
3a0946f5 |
char *arg, *id, *number, *total, *oldfilename; |
d85c1fad |
const char *tmpdir; |
a7a6fff4 |
char pdir[NAME_MAX + 1]; |
d85c1fad |
|
53ee0b60 |
id = (char *)messageFindArgument(m, "id");
if(id == NULL)
return -1;
|
f0635204 |
#ifdef C_CYGWIN |
9180b8bb |
if((tmpdir = getenv("TEMP")) == (char *)NULL)
if((tmpdir = getenv("TMP")) == (char *)NULL)
if((tmpdir = getenv("TMPDIR")) == (char *)NULL)
tmpdir = "C:\\"; |
d85c1fad |
#else |
9180b8bb |
if((tmpdir = getenv("TMPDIR")) == (char *)NULL)
if((tmpdir = getenv("TMP")) == (char *)NULL)
if((tmpdir = getenv("TEMP")) == (char *)NULL) |
d85c1fad |
#ifdef P_tmpdir |
9180b8bb |
tmpdir = P_tmpdir; |
d85c1fad |
#else |
9180b8bb |
tmpdir = "/tmp"; |
d85c1fad |
#endif
#endif |
9a7398ee |
|
a7a6fff4 |
snprintf(pdir, sizeof(pdir) - 1, "%s/clamav-partial", tmpdir); |
d85c1fad |
if((mkdir(pdir, 0700) < 0) && (errno != EEXIST)) {
cli_errmsg("Can't create the directory '%s'\n", pdir); |
9a7398ee |
return -1; |
3a0946f5 |
} else {
struct stat statb;
|
d85c1fad |
if(stat(pdir, &statb) < 0) {
cli_errmsg("Can't stat the directory '%s'\n", pdir); |
3a0946f5 |
return -1;
}
if(statb.st_mode & 077)
cli_warnmsg("Insecure partial directory %s (mode 0%o)\n", |
d85c1fad |
pdir, statb.st_mode & 0777); |
9a7398ee |
}
number = (char *)messageFindArgument(m, "number");
if(number == NULL) {
free(id);
return -1;
}
oldfilename = (char *)messageFindArgument(m, "filename");
if(oldfilename == NULL)
oldfilename = (char *)messageFindArgument(m, "name");
arg = cli_malloc(10 + strlen(id) + strlen(number)); |
e82a5185 |
if(arg) {
sprintf(arg, "filename=%s%s", id, number);
messageAddArgument(m, arg);
free(arg);
} |
9a7398ee |
if(oldfilename) {
cli_warnmsg("Must reset to %s\n", oldfilename);
free(oldfilename);
}
|
d85c1fad |
if((fb = messageToFileblob(m, pdir)) == NULL) { |
9a7398ee |
free(id);
free(number);
return -1;
}
fileblobDestroy(fb);
total = (char *)messageFindArgument(m, "total");
cli_dbgmsg("rfc1341: %s, %s of %s\n", id, number, (total) ? total : "?");
if(total) {
int n = atoi(number);
int t = atoi(total);
DIR *dd = NULL;
|
53ee0b60 |
free(total); |
9a7398ee |
/*
* If it's the last one - reassemble it |
138b73f6 |
* FIXME: this assumes that we receive the parts in order |
9a7398ee |
*/ |
d85c1fad |
if((n == t) && ((dd = opendir(pdir)) != NULL)) { |
9a7398ee |
FILE *fout;
char outname[NAME_MAX + 1]; |
e82a5185 |
time_t now; |
9a7398ee |
snprintf(outname, sizeof(outname) - 1, "%s/%s", dir, id);
cli_dbgmsg("outname: %s\n", outname);
fout = fopen(outname, "wb");
if(fout == NULL) { |
138b73f6 |
cli_errmsg("Can't open '%s' for writing", outname); |
9a7398ee |
free(id);
free(number);
closedir(dd);
return -1;
}
|
e82a5185 |
time(&now); |
9a7398ee |
for(n = 1; n <= t; n++) {
char filename[NAME_MAX + 1]; |
3a0946f5 |
const struct dirent *dent;
#if defined(HAVE_READDIR_R_3) || defined(HAVE_READDIR_R_2) |
a77dc192 |
union {
struct dirent d;
char b[offsetof(struct dirent, d_name) + NAME_MAX + 1];
} result; |
3a0946f5 |
#endif |
9a7398ee |
snprintf(filename, sizeof(filename), "%s%d", id, n); |
2c7b958d |
|
a77dc192 |
#ifdef HAVE_READDIR_R_3
while((readdir_r(dd, &result.d, &dent) == 0) && dent) { |
9a7398ee |
#elif defined(HAVE_READDIR_R_2) |
a77dc192 |
while((dent = (struct dirent *)readdir_r(dd, &result.d))) { |
2c7b958d |
#else /*!HAVE_READDIR_R*/ |
9a7398ee |
while((dent = readdir(dd))) {
#endif
FILE *fin; |
e82a5185 |
char buffer[BUFSIZ], fullname[NAME_MAX + 1]; |
9a7398ee |
int nblanks; |
d85c1fad |
extern short cli_leavetemps_flag; |
e82a5185 |
struct stat statb; |
9a7398ee |
|
a6d49269 |
#ifndef C_CYGWIN |
9a7398ee |
if(dent->d_ino == 0)
continue; |
a6d49269 |
#endif |
9a7398ee |
|
e82a5185 |
snprintf(fullname, sizeof(fullname) - 1,
"%s/%s", pdir, dent->d_name);
if(strncmp(filename, dent->d_name, strlen(filename)) != 0) {
if(!cli_leavetemps_flag)
continue;
if(stat(fullname, &statb) < 0)
continue;
if(now - statb.st_mtime > (time_t)(7 * 24 * 3600))
if(unlink(fullname) >= 0)
cli_warnmsg("removed old RFC1341 file %s\n", fullname); |
9a7398ee |
continue; |
e82a5185 |
} |
9a7398ee |
|
e82a5185 |
fin = fopen(fullname, "rb"); |
9a7398ee |
if(fin == NULL) { |
e82a5185 |
cli_errmsg("Can't open '%s' for reading", fullname); |
9a7398ee |
fclose(fout);
unlink(outname);
free(id);
free(number);
closedir(dd);
return -1;
}
nblanks = 0; |
d99b1840 |
while(fgets(buffer, sizeof(buffer) - 1, fin) != NULL) |
9a7398ee |
/*
* Ensure that trailing newlines
* aren't copied
*/ |
53ee0b60 |
if(buffer[0] == '\n') |
9a7398ee |
nblanks++; |
53ee0b60 |
else { |
9a7398ee |
if(nblanks)
do
putc('\n', fout);
while(--nblanks > 0);
fputs(buffer, fout);
}
fclose(fin); |
d85c1fad |
/* don't unlink if leave temps */
if(!cli_leavetemps_flag) |
e82a5185 |
unlink(fullname); |
9a7398ee |
break;
}
rewinddir(dd);
}
closedir(dd);
fclose(fout);
}
} |
53ee0b60 |
free(number); |
9a7398ee |
free(id);
return 0;
}
#endif
|
f52d7358 |
#if defined(FOLLOWURLS) && (FOLLOWURLS > 0) |
c5ed8336 |
static void
checkURLs(message *m, const char *dir)
{
blob *b = messageToBlob(m);
size_t len; |
6b93ea0c |
table_t *t; |
6da40aa1 |
int i, n; |
314ff77b |
#if defined(WITH_CURL) && defined(CL_THREAD_SAFE) |
f52d7358 |
pthread_t tid[FOLLOWURLS];
struct arg args[FOLLOWURLS]; |
314ff77b |
#endif |
6da40aa1 |
tag_arguments_t hrefs; |
c5ed8336 |
if(b == NULL)
return;
len = blobGetDataSize(b);
|
e94471f4 |
if(len == 0) {
blobDestroy(b); |
3eb12bae |
return; |
e94471f4 |
} |
3eb12bae |
|
6b93ea0c |
/* TODO: make this size customisable */
if(len > 100*1024) {
cli_warnmsg("Viruses pointed to by URL not scanned in large message\n");
blobDestroy(b); |
e94471f4 |
return; |
6b93ea0c |
}
|
e94471f4 |
blobClose(b); |
6b93ea0c |
t = tableCreate(); |
3a0ef2ee |
if(t == NULL) {
blobDestroy(b);
return;
} |
3eb12bae |
|
a2d786fc |
hrefs.count = 0;
hrefs.tag = hrefs.value = NULL; |
6da40aa1 |
cli_dbgmsg("checkURLs: calling html_normalise_mem\n"); |
3a0ef2ee |
if(!html_normalise_mem(blobGetData(b), len, NULL, &hrefs)) { |
7d3d11d0 |
blobDestroy(b);
tableDestroy(t);
return; |
3a0ef2ee |
}
cli_dbgmsg("checkURLs: html_normalise_mem returned\n");
|
e745ac7e |
/* TODO: Do we need to call remove_html_comments? */ |
6b93ea0c |
|
6da40aa1 |
n = 0;
for(i = 0; i < hrefs.count; i++) { |
a77dc192 |
const char *url = (const char *)hrefs.value[i]; |
6da40aa1 |
|
e82a5185 |
/*
* TODO: If it's an image source, it'd be nice to note beacons
* where width="0" height="0", which needs support from
* the HTML normalise code
*/ |
6da40aa1 |
if(strncasecmp("http://", url, 7) == 0) {
char *ptr; |
314ff77b |
#ifdef WITH_CURL
#ifndef CL_THREAD_SAFE
struct arg arg;
#endif
#else /*!WITH_CURL*/ |
bf6f653d |
#ifdef CL_THREAD_SAFE
static pthread_mutex_t system_mutex = PTHREAD_MUTEX_INITIALIZER;
#endif |
6b93ea0c |
struct stat statb;
char cmd[512]; |
314ff77b |
#endif /*WITH_CURL*/ |
a95c894a |
char name[NAME_MAX + 1]; |
6da40aa1 |
if(tableFind(t, url) == 1) {
cli_dbgmsg("URL %s already downloaded\n", url); |
f2b068fb |
continue;
} |
e82a5185 |
/*
* What about foreign character spoofing?
* It would be useful be able to check if url
* is the same as the text displayed, e.g.
* <a href="http://dodgy.biz">www.paypal.com</a>
* but that needs support from HTML normalise
*/
if(strchr(url, '%') && strchr(url, '@'))
cli_warnmsg("Possible URL spoofing attempt noticed, but not yet handled (%s)\n", url);
|
f52d7358 |
if(n == FOLLOWURLS) { |
e82a5185 |
cli_warnmsg("URL %s will not be scanned\n", url); |
de617e3e |
break;
} |
e82a5185 |
|
6da40aa1 |
(void)tableInsert(t, url, 1);
cli_dbgmsg("Downloading URL %s to be scanned\n", url); |
2176c0e5 |
strncpy(name, url, sizeof(name) - 1); |
39d09964 |
name[sizeof(name) - 1] = '\0'; |
6da40aa1 |
for(ptr = name; *ptr; ptr++)
if(*ptr == '/')
*ptr = '_'; |
c5ed8336 |
|
da812a6a |
#ifdef WITH_CURL |
314ff77b |
#ifdef CL_THREAD_SAFE |
a95c894a |
args[n].dir = dir;
args[n].url = url; |
314ff77b |
args[n].filename = strdup(name);
pthread_create(&tid[n], NULL, getURL, &args[n]);
#else |
6da40aa1 |
arg.url = url; |
314ff77b |
arg.dir = dir;
arg.filename = name;
getURL(&arg);
#endif
|
3fa72383 |
#else
/*
* TODO: maximum size and timeouts
*/ |
2176c0e5 |
len = sizeof(cmd) - 26 - strlen(dir) - strlen(name);
#ifdef CL_DEBUG |
e82a5185 |
snprintf(cmd, sizeof(cmd) - 1, "GET -t10 \"%.*s\" >%s/%s", len, url, dir, name); |
2176c0e5 |
#else |
e82a5185 |
snprintf(cmd, sizeof(cmd) - 1, "GET -t10 \"%.*s\" >%s/%s 2>/dev/null", len, url, dir, name); |
2176c0e5 |
#endif
cmd[sizeof(cmd) - 1] = '\0';
|
c5ed8336 |
cli_dbgmsg("%s\n", cmd); |
bf6f653d |
#ifdef CL_THREAD_SAFE
pthread_mutex_lock(&system_mutex);
#endif |
c5ed8336 |
system(cmd); |
bf6f653d |
#ifdef CL_THREAD_SAFE
pthread_mutex_unlock(&system_mutex);
#endif
snprintf(cmd, sizeof(cmd), "%s/%s", dir, name);
if(stat(cmd, &statb) >= 0)
if(statb.st_size == 0) { |
6da40aa1 |
cli_warnmsg("URL %s failed to download\n", url); |
bf6f653d |
/*
* Don't bother scanning an empty file
*/
(void)unlink(cmd);
} |
3fa72383 |
#endif |
314ff77b |
++n; |
c5ed8336 |
}
}
blobDestroy(b); |
f2b068fb |
tableDestroy(t); |
314ff77b |
#if defined(WITH_CURL) && defined(CL_THREAD_SAFE) |
f52d7358 |
assert(n <= FOLLOWURLS); |
314ff77b |
cli_dbgmsg("checkURLs: waiting for %d thread(s) to finish\n", n);
while(--n >= 0) {
pthread_join(tid[n], NULL);
free(args[n].filename);
}
#endif |
a95c894a |
html_tag_arg_free(&hrefs); |
c5ed8336 |
}
|
e82a5185 |
/*
* Includes some Win32 patches by Gianluigi Tiesi <sherpya@netfarm.it>
*
* FIXME: Often WMF exploits work by sending people an email directing them
* to a page which displays a picture containing the exploit. This is not
* currently found, since only the HTML on the referred page is downloaded.
* It would be useful to scan the HTML for references to pictures and
* download them for scanning. But that will hit performance so there is
* an issue here.
*/ |
da812a6a |
#ifdef WITH_CURL |
314ff77b |
static void *
#ifdef CL_THREAD_SAFE
getURL(void *a)
#else
getURL(struct arg *arg)
#endif |
3fa72383 |
{ |
6b93ea0c |
CURL *curl; |
3fa72383 |
FILE *fp; |
6b93ea0c |
struct curl_slist *headers;
static int initialised = 0; |
314ff77b |
#ifdef CL_THREAD_SAFE
static pthread_mutex_t init_mutex = PTHREAD_MUTEX_INITIALIZER;
struct arg *arg = (struct arg *)a;
#endif
const char *url = arg->url;
const char *dir = arg->dir;
const char *filename = arg->filename; |
0cb54827 |
char fout[NAME_MAX + 1]; |
9c8806fb |
#ifdef CURLOPT_ERRORBUFFER |
e82a5185 |
char errorbuffer[CURL_ERROR_SIZE];
#elif (LIBCURL_VERSION_NUM >= 0x070C00)
CURLcode res = CURLE_OK; |
9c8806fb |
#endif |
3fa72383 |
|
314ff77b |
#ifdef CL_THREAD_SAFE
pthread_mutex_lock(&init_mutex);
#endif |
6b93ea0c |
if(!initialised) { |
e82a5185 |
if(curl_global_init(CURL_GLOBAL_ALL) != 0) { |
314ff77b |
#ifdef CL_THREAD_SAFE
pthread_mutex_unlock(&init_mutex);
#endif |
e82a5185 |
cli_errmsg("curl_global_init failed"); |
314ff77b |
return NULL;
} |
6b93ea0c |
initialised = 1; |
3fa72383 |
} |
314ff77b |
#ifdef CL_THREAD_SAFE
pthread_mutex_unlock(&init_mutex);
#endif
|
6b93ea0c |
/* easy isn't the word I'd use... */
curl = curl_easy_init(); |
e82a5185 |
if(curl == NULL) {
cli_errmsg("curl_easy_init failed"); |
314ff77b |
return NULL; |
e82a5185 |
} |
da812a6a |
|
6b93ea0c |
(void)curl_easy_setopt(curl, CURLOPT_USERAGENT, "www.clamav.net");
|
e82a5185 |
if(curl_easy_setopt(curl, CURLOPT_URL, url) != 0) {
cli_errmsg("%s: curl_easy_setopt failed", url);
curl_easy_cleanup(curl); |
314ff77b |
return NULL; |
e82a5185 |
} |
6b93ea0c |
|
a95c894a |
snprintf(fout, NAME_MAX, "%s/%s", dir, filename); |
3fa72383 |
|
e82a5185 |
fp = fopen(fout, "wb"); |
3fa72383 |
if(fp == NULL) { |
138b73f6 |
cli_errmsg("Can't open '%s' for writing", fout); |
da812a6a |
curl_easy_cleanup(curl); |
314ff77b |
return NULL; |
3fa72383 |
} |
05ea2522 |
#ifdef CURLOPT_WRITEDATA |
314ff77b |
if(curl_easy_setopt(curl, CURLOPT_WRITEDATA, fp) != 0) {
fclose(fp);
curl_easy_cleanup(curl);
return NULL;
} |
05ea2522 |
#else
if(curl_easy_setopt(curl, CURLOPT_FILE, fp) != 0) {
fclose(fp);
curl_easy_cleanup(curl);
return NULL;
}
#endif |
314ff77b |
|
6b93ea0c |
/* |
3eb12bae |
* If an item is in squid's cache get it from there (TCP_HIT/200) |
6b93ea0c |
* by default curl doesn't (TCP_CLIENT_REFRESH_MISS/200)
*/
headers = curl_slist_append(NULL, "Pragma:");
curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers); |
3fa72383 |
|
6b93ea0c |
/* These should be customisable */
curl_easy_setopt(curl, CURLOPT_TIMEOUT, 30);
curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, 10); |
49674596 |
#ifdef CURLOPT_MAXFILESIZE
curl_easy_setopt(curl, CURLOPT_MAXFILESIZE, 50*1024);
#endif |
3fa72383 |
|
314ff77b |
#ifdef CL_THREAD_SAFE |
c07de365 |
#ifdef CURLOPT_DNS_USE_GLOBAL_CACHE |
314ff77b |
curl_easy_setopt(curl, CURLOPT_DNS_USE_GLOBAL_CACHE, 0);
#endif |
c07de365 |
#endif |
02406150 |
/*
* Prevent password: prompting with older versions
* FIXME: a better username?
*/ |
66df01fa |
curl_easy_setopt(curl, CURLOPT_USERPWD, "username:password"); |
02406150 |
|
314ff77b |
/*
* FIXME: valgrind reports "pthread_mutex_unlock: mutex is not locked"
* from gethostbyaddr_r within this. It may be a bug in libcurl
* rather than this code, but I need to check, see Curl_resolv()
* If pushed really hard it will sometimes say
* Conditional jump or move depends on uninitialised value(s) and
* quit. But the program seems to work OK without valgrind...
* Perhaps Curl_resolv() isn't thread safe? |
e82a5185 |
*
* I have seen segfaults in version 7.12.3. Version 7.14 seems OK. |
314ff77b |
*/ |
8386482b |
/* |
aa479b7d |
* On some C libraries (notably with FC3, glibc-2.3.3-74) you get a |
e82a5185 |
* memory leak here in getaddrinfo(), see |
0856891e |
* https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=139559 |
8386482b |
*/ |
9c8806fb |
#ifdef CURLOPT_ERRORBUFFER |
e82a5185 |
curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errorbuffer);
if(curl_easy_perform(curl) != CURLE_OK) |
9c8806fb |
cli_warnmsg("URL %s failed to download: %s\n", url, errorbuffer); |
e82a5185 |
#elif (LIBCURL_VERSION_NUM >= 0x070C00)
if((res = curl_easy_perform(curl)) != CURLE_OK)
cli_warnmsg("URL %s failed to download: %s\n", url,
curl_easy_strerror(res)); |
9c8806fb |
#else |
e82a5185 |
if(curl_easy_perform(curl) != CURLE_OK) |
6b93ea0c |
cli_warnmsg("URL %s failed to download\n", url); |
9c8806fb |
#endif |
6b93ea0c |
fclose(fp);
curl_easy_cleanup(curl); |
e82a5185 |
curl_slist_free_all(headers); |
314ff77b |
return NULL; |
3fa72383 |
}
#endif
#else
static void
checkURLs(message *m, const char *dir)
{
}
#endif
|
d1382234 |
#ifdef HAVE_BACKTRACE |
f2b068fb |
static void |
98685ac1 |
sigsegv(int sig)
{
signal(SIGSEGV, SIG_DFL); |
d1382234 |
print_trace(1); |
98685ac1 |
exit(SIGSEGV);
}
|
f2b068fb |
static void |
98685ac1 |
print_trace(int use_syslog)
{
void *array[10];
size_t size;
char **strings;
size_t i;
pid_t pid = getpid();
size = backtrace(array, 10);
strings = backtrace_symbols(array, size);
if(use_syslog == 0)
cli_dbgmsg("Backtrace of pid %d:\n", pid); |
735377bc |
else |
98685ac1 |
syslog(LOG_ERR, "Backtrace of pid %d:", pid);
for(i = 0; i < size; i++)
if(use_syslog) |
60aec445 |
syslog(LOG_ERR, "bt[%u]: %s", i, strings[i]); |
98685ac1 |
else
cli_dbgmsg("%s\n", strings[i]);
|
7d3d11d0 |
/* TODO: dump the current email */
|
98685ac1 |
free(strings);
}
#endif |
00615ec9 |
|
82933497 |
static bool
usefulHeader(int commandNumber, const char *cmd)
{
switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE:
return TRUE;
default:
if(strcasecmp(cmd, "From") == 0)
return TRUE;
else if(strcasecmp(cmd, "Received") == 0)
return TRUE;
else if(strcasecmp(cmd, "De") == 0)
return TRUE;
}
return FALSE;
}
|
b3a5cdd8 |
/* |
9c8806fb |
* Like fgets but cope with end of line by "\n", "\r\n", "\n\r", "\r"
*/
static char * |
c840fa41 |
getline_from_mbox(char *buffer, size_t len, FILE *fin) |
9c8806fb |
{
char *ret;
if(feof(fin))
return NULL;
if((len == 0) || (buffer == NULL)) { |
c840fa41 |
cli_errmsg("Invalid call to getline_from_mbox(). Report to bugs@clamav.net\n"); |
9c8806fb |
return NULL;
}
ret = buffer;
do {
int c = getc(fin);
if(ferror(fin))
return NULL;
switch(c) {
case '\n':
*buffer++ = '\n';
c = getc(fin);
if((c != '\r') && !feof(fin))
ungetc(c, fin);
break;
default: |
e82a5185 |
*buffer++ = (char)c; |
9c8806fb |
continue;
case EOF:
break;
case '\r':
*buffer++ = '\n';
c = getc(fin);
if((c != '\n') && !feof(fin))
ungetc(c, fin);
break;
}
break; |
e82a5185 |
} while(--len > 1); |
9c8806fb |
if(len == 0) { |
e82a5185 |
/* the email probably breaks RFC821 */
cli_warnmsg("getline_from_mbox: buffer overflow stopped - line lost\n"); |
9c8806fb |
return NULL;
} |
e82a5185 |
if(len == 1)
/* over flows will have appear on separate lines */
cli_dbgmsg("getline_from_mbox: buffer overflow stopped - line recovered\n"); |
9c8806fb |
*buffer = '\0';
return ret;
} |