e3aaff8e |
/* |
c442ca9c |
* Copyright (C) 2013-2019 Cisco Systems, Inc. and/or its affiliates. All rights reserved.
* Copyright (C) 2007-2013 Sourcefire, Inc. |
2023340a |
*
* Authors: Nigel Horne |
6289eda8 |
*
* Acknowledgements: Some ideas came from Stephen White <stephen@earth.li>,
* Michael Dankov <misha@btrc.ru>, Gianluigi Tiesi <sherpya@netfarm.it>,
* Everton da Silva Marques, Thomas Lamy <Thomas.Lamy@in-online.net>,
* James Stevens <James@kyzo.com> |
e3aaff8e |
*
* This program is free software; you can redistribute it and/or modify |
2023340a |
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation. |
e3aaff8e |
*
* 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 |
48b7b4a7 |
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
* MA 02110-1301, USA. |
e3aaff8e |
*/ |
2023340a |
|
6d6e8271 |
#if HAVE_CONFIG_H
#include "clamav-config.h"
#endif |
e3aaff8e |
#ifdef CL_THREAD_SAFE |
98cb5cba |
#ifndef _REENTRANT |
e3aaff8e |
#define _REENTRANT /* for Solaris 2.8 */
#endif |
98cb5cba |
#endif |
e3aaff8e |
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <assert.h>
#include <string.h> |
bc6bbeff |
#ifdef HAVE_STRINGS_H |
e3aaff8e |
#include <strings.h> |
bc6bbeff |
#endif |
e906fef3 |
#ifdef HAVE_STRING_H
#include <string.h>
#endif |
e3aaff8e |
#include <ctype.h>
#include <time.h>
#include <fcntl.h> |
bc6bbeff |
#ifdef HAVE_SYS_PARAM_H |
d4d14218 |
#include <sys/param.h> |
bc6bbeff |
#endif |
f10460ed |
#include <dirent.h> |
a0b21816 |
#include <limits.h> |
093e013c |
#include <signal.h> |
e3aaff8e |
|
cd153266 |
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
|
242bfde8 |
#if defined(HAVE_READDIR_R_3) || defined(HAVE_READDIR_R_2)
#include <stddef.h>
#endif
|
e2875303 |
#ifdef CL_THREAD_SAFE
#include <pthread.h>
#endif
|
60d8d2c3 |
#include "clamav.h" |
0f7f7682 |
#include "others.h"
#include "str.h"
#include "filetypes.h" |
e3aaff8e |
#include "mbox.h" |
d77ac7de |
#include "dconf.h" |
ee1b2a6c |
#include "fmap.h" |
ede9939c |
#include "json_api.h" |
c2df9f79 |
#include "msxml_parser.h"
#if HAVE_LIBXML2 |
f5412cc9 |
#include <libxml/xmlversion.h> |
c2df9f79 |
#include <libxml/HTMLtree.h> |
f5412cc9 |
#include <libxml/HTMLparser.h> |
c2df9f79 |
#include <libxml/xmlreader.h>
#endif |
d77ac7de |
#define DCONF_PHISHING mctx->ctx->dconf->phishing |
e3aaff8e |
|
02927896 |
#ifdef CL_DEBUG |
093e013c |
|
6670d61d |
#if defined(C_LINUX) |
093e013c |
#include <features.h> |
92dbfae7 |
#endif |
093e013c |
|
02927896 |
#if __GLIBC__ == 2 && __GLIBC_MINOR__ >= 1
#define HAVE_BACKTRACE
#endif |
3f3f9085 |
#endif |
02927896 |
#ifdef HAVE_BACKTRACE
#include <execinfo.h>
#include <syslog.h>
static void sigsegv(int sig);
static void print_trace(int use_syslog); |
a9d251e0 |
|
96435bdc |
/*#define SAVE_TMP */ /* Save the file being worked on in tmp */ |
02927896 |
#endif
|
c2b2d8af |
#if defined(NO_STRTOK_R) || !defined(CL_THREAD_SAFE) |
e3aaff8e |
#undef strtok_r
#undef __strtok_r
#define strtok_r(a,b,c) strtok(a,b)
#endif
|
0cf4cea7 |
#ifdef HAVE_STDBOOL_H |
89d4073d |
#ifdef C_BEOS
#include "SupportDefs.h"
#else |
edee0700 |
#include <stdbool.h> |
89d4073d |
#endif |
edee0700 |
#else
#ifdef FALSE
typedef unsigned char bool;
#else
typedef enum { FALSE = 0, TRUE = 1 } bool; |
e3aaff8e |
#endif
#endif
|
ecc3d638 |
typedef enum {
FAIL,
OK,
OK_ATTACHMENTS_NOT_SAVED, |
69c62847 |
VIRUS, |
001ad879 |
MAXREC,
MAXFILES |
ecc3d638 |
} mbox_status;
|
9f2024cc |
#ifndef isblank
#define isblank(c) (((c) == ' ') || ((c) == '\t'))
#endif
|
9b4bb8b7 |
#define SAVE_TO_DISC /* multipart/message are saved in a temporary file */ |
393a6d67 |
|
c52d991e |
#include "htmlnorm.h"
#include "phishcheck.h"
|
be4bf7f4 |
#ifndef _WIN32 |
081f6473 |
#include <sys/time.h> |
ea541184 |
#include <netdb.h>
#include <sys/socket.h>
#include <netinet/in.h> |
29b92a2d |
#if !defined(C_BEOS) && !defined(C_INTERIX) |
ea541184 |
#include <net/if.h>
#include <arpa/inet.h>
#endif |
89d4073d |
#endif |
f4a02249 |
|
ea541184 |
#include <fcntl.h>
|
f10460ed |
/* |
d7979d4f |
* Use CL_SCAN_MAIL_PARTIAL_MESSAGE to handle messages covered by section 7.3.2 of RFC1341. |
f10460ed |
* This is experimental code so it is up to YOU to (1) ensure it's secure |
cf569541 |
* (2) periodically trim the directory of old files
*
* If you use the load balancing feature of clamav-milter to run clamd on |
fb79b576 |
* more than one machine you must make sure that .../partial is on a shared |
cf569541 |
* network filesystem |
f10460ed |
*/ |
ac9b941b |
/*#define NEW_WORLD*/ |
d72749e0 |
|
8c68fcc1 |
/*#define SCAN_UNENCODED_BOUNCES *//* |
12bd9764 |
* Slows things down a lot and only catches unencoded copies |
214621f2 |
* of EICAR within bounces, which don't matter |
12bd9764 |
*/
|
c1fce7f7 |
typedef struct mbox_ctx {
const char *dir;
const table_t *rfc821Table;
const table_t *subtypeTable;
cli_ctx *ctx; |
b5231f5f |
unsigned int files; /* number of files extracted */ |
ede9939c |
#if HAVE_JSON
json_object *wrkobj;
#endif |
c1fce7f7 |
} mbox_ctx;
|
f87a92ca |
/* if supported by the system, use the optimized
* version of getc, that doesn't do locking,
* and is possibly implemented entirely as a macro */
#if defined(_POSIX_C_SOURCE) && _POSIX_C_SOURCE >= 200112L
#define GETC(fp) getc_unlocked(fp)
#define LOCKFILE(fp) flockfile(fp)
#define UNLOCKFILE(fp) funlockfile(fp)
#else
#define GETC(fp) getc(fp)
#define LOCKFILE(fp)
#define UNLOCKFILE(fp)
#endif
|
2df29bde |
static int cli_parse_mbox(const char *dir, cli_ctx *ctx); |
49cc1e3c |
static message *parseEmailFile(fmap_t *map, size_t *at, const table_t *rfc821Table, const char *firstLine, const char *dir); |
2673dc74 |
static message *parseEmailHeaders(message *m, const table_t *rfc821Table); |
4c60b74f |
static int parseEmailHeader(message *m, const char *line, const table_t *rfc821Table); |
a6369bc8 |
static int parseMHTMLComment(const char *comment, cli_ctx *ctx, void *wrkjobj, void *cbdata); |
c2df9f79 |
static mbox_status parseRootMHTML(mbox_ctx *mctx, message *m, text *t); |
ecc3d638 |
static mbox_status parseEmailBody(message *messageIn, text *textIn, mbox_ctx *mctx, unsigned int recursion_level); |
e3aaff8e |
static int boundaryStart(const char *line, const char *boundary); |
69c62847 |
static int boundaryEnd(const char *line, const char *boundary); |
e3aaff8e |
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 int parseMimeHeader(message *m, const char *cmd, const table_t *rfc821Table, const char *arg); |
001ad879 |
static int saveTextPart(mbox_ctx *mctx, message *m, int destroy_text); |
50df4118 |
static char *rfc2047(const char *in); |
d72749e0 |
static char *rfc822comments(const char *in, char *out); |
f10460ed |
static int rfc1341(message *m, const char *dir); |
ddea752e |
static bool usefulHeader(int commandNumber, const char *cmd); |
49cc1e3c |
static char *getline_from_mbox(char *buffer, size_t len, fmap_t *map, size_t *at); |
7021b545 |
static bool isBounceStart(mbox_ctx *mctx, const char *line); |
001ad879 |
static bool exportBinhexMessage(mbox_ctx *mctx, message *m);
static int exportBounceMessage(mbox_ctx *ctx, text *start); |
ede9939c |
static const char *getMimeTypeStr(mime_type mimetype);
static const char *getEncTypeStr(encoding_type enctype); |
ecc3d638 |
static message *do_multipart(message *mainMessage, message **messages, int i, mbox_status *rc, mbox_ctx *mctx, message *messageIn, text **tptr, unsigned int recursion_level); |
4f4a8f4a |
static int count_quotes(const char *buf); |
842c7d49 |
static bool next_is_folded_header(const text *t); |
0cf4cea7 |
static bool newline_in_header(const char *line); |
9b4bb8b7 |
|
ecc3d638 |
static blob *getHrefs(message *m, tag_arguments_t *hrefs);
static void hrefs_done(blob *b, tag_arguments_t *hrefs); |
ad422cc9 |
static void checkURLs(message *m, mbox_ctx *mctx, mbox_status *rc, int is_html); |
9b4bb8b7 |
|
9fe789f8 |
/* Maximum line length according to RFC2821 */ |
85bb253e |
#define RFC2821LENGTH 1000 |
e3aaff8e |
/* 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 |
946a0ad3 |
#define ALTERNATIVE 6 /* RFC1521*/ |
e3aaff8e |
#define DIGEST 7
#define SIGNED 8
#define PARALLEL 9
#define RELATED 10 /* RFC2387 */
#define REPORT 11 /* RFC1892 */ |
c9b8f252 |
#define APPLEDOUBLE 12 /* Handling of this in only noddy for now */ |
393a6d67 |
#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)
*/ |
9a729c80 |
#define ENCRYPTED 13 /*
* e.g. RFC2015
* Content-Type: multipart/encrypted;
* boundary="nextPart1383049.XCRrrar2yq";
* protocol="application/pgp-encrypted"
*/ |
6e5d95eb |
#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
*/ |
c79a2273 |
#define KNOWBOT 14 /* Unknown and undocumented format? */ |
e3aaff8e |
static const struct tableinit {
const char *key;
int value;
} rfc821headers[] = { |
303f9be9 |
/* TODO: make these regular expressions */ |
5c1150ac |
{ "Content-Type", CONTENT_TYPE }, |
a9f386ed |
{ "Content-Transfer-Encoding", CONTENT_TRANSFER_ENCODING },
{ "Content-Disposition", CONTENT_DISPOSITION }, |
e3aaff8e |
{ NULL, 0 } |
15033cb6 |
}, mimeSubtypes[] = { /* see RFC2045 */ |
e3aaff8e |
/* 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 }, |
c9b8f252 |
{ "appledouble", APPLEDOUBLE }, |
393a6d67 |
{ "fax-message", FAX }, |
9a729c80 |
{ "encrypted", ENCRYPTED }, |
6e5d95eb |
{ "x-bfile", X_BFILE }, /* BeOS */ |
c79a2273 |
{ "knowbot", KNOWBOT }, /* ??? */
{ "knowbot-metadata", KNOWBOT }, /* ??? */
{ "knowbot-code", KNOWBOT }, /* ??? */
{ "knowbot-state", KNOWBOT }, /* ??? */ |
e3aaff8e |
{ NULL, 0 } |
ede9939c |
}, mimeTypeStr[] = {
{ "NOMIME", NOMIME },
{ "APPLICATION", APPLICATION },
{ "AUDIO", AUDIO },
{ "IMAGE", IMAGE },
{ "MESSAGE", MESSAGE },
{ "MULTIPART", MULTIPART },
{ "TEXT", TEXT },
{ "VIDEO", VIDEO },
{ "MEXTENSION", MEXTENSION },
{ NULL, 0 }
}, encTypeStr[] = {
{ "NOENCODING", NOENCODING },
{ "QUOTEDPRINTABLE", QUOTEDPRINTABLE },
{ "BASE64", BASE64 },
{ "EIGHTBIT", EIGHTBIT },
{ "BINARY", BINARY },
{ "UUENCODE", UUENCODE },
{ "YENCODE", YENCODE },
{ "EEXTENSION", EEXTENSION },
{ "BINHEX", BINHEX },
{ NULL, 0 } |
e3aaff8e |
}; |
e2875303 |
#ifdef CL_THREAD_SAFE
static pthread_mutex_t tables_mutex = PTHREAD_MUTEX_INITIALIZER;
#endif |
61f01c95 |
static table_t *rfc821 = NULL;
static table_t *subtype = NULL; |
e3aaff8e |
|
f24bf390 |
int |
2df29bde |
cli_mbox(const char *dir, cli_ctx *ctx) |
f24bf390 |
{ |
7c56033f |
if(dir == NULL) { |
6351aa86 |
cli_dbgmsg("cli_mbox called with NULL dir\n"); |
7c56033f |
return CL_ENULLARG;
} |
2df29bde |
return cli_parse_mbox(dir, ctx); |
f24bf390 |
}
|
e3aaff8e |
/*
* TODO: when signal handling is added, need to remove temp files when a |
ef822cfc |
* signal is received |
e3aaff8e |
* TODO: add option to scan in memory not via temp files, perhaps with a |
1bfbedd4 |
* 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 |
049a18b9 |
* TODO: parse .msg format files |
c9b8f252 |
* TODO: fully handle AppleDouble format, see |
ef822cfc |
* http://www.lazerware.com/formats/Specs/AppleSingle_AppleDouble.pdf |
89670d69 |
* TODO: ensure parseEmailHeaders is always called before parseEmailBody
* TODO: create parseEmail which calls parseEmailHeaders then parseEmailBody |
7cd9337a |
* TODO: Handle unexpected NUL bytes in header lines which stop strcmp()s: |
9f2024cc |
* e.g. \0Content-Type: application/binary; |
e3aaff8e |
*/ |
f24bf390 |
static int |
2df29bde |
cli_parse_mbox(const char *dir, cli_ctx *ctx) |
e3aaff8e |
{ |
bb1e844c |
int retcode; |
ddea752e |
message *body; |
85bb253e |
char buffer[RFC2821LENGTH + 1]; |
c1fce7f7 |
mbox_ctx mctx; |
ee1b2a6c |
size_t at = 0; |
49cc1e3c |
fmap_t *map = *ctx->fmap; |
e3aaff8e |
cli_dbgmsg("in mbox()\n"); |
5cdb01fc |
|
ee1b2a6c |
if(!fmap_gets(map, buffer, &at, sizeof(buffer) - 1)) { |
049a18b9 |
/* empty message */ |
ef822cfc |
return CL_CLEAN; |
049a18b9 |
} |
e2875303 |
#ifdef CL_THREAD_SAFE
pthread_mutex_lock(&tables_mutex);
#endif |
393a6d67 |
if(rfc821 == NULL) {
assert(subtype == NULL); |
51fc2aa8 |
|
393a6d67 |
if(initialiseTables(&rfc821, &subtype) < 0) {
rfc821 = NULL;
subtype = NULL; |
e2875303 |
#ifdef CL_THREAD_SAFE
pthread_mutex_unlock(&tables_mutex);
#endif |
ef822cfc |
return CL_EMEM; |
51fc2aa8 |
} |
e3aaff8e |
} |
e2875303 |
#ifdef CL_THREAD_SAFE
pthread_mutex_unlock(&tables_mutex);
#endif |
e3aaff8e |
|
a603478f |
retcode = CL_SUCCESS; |
e791b5ac |
body = NULL;
|
c1fce7f7 |
mctx.dir = dir;
mctx.rfc821Table = rfc821;
mctx.subtypeTable = subtype;
mctx.ctx = ctx; |
001ad879 |
mctx.files = 0; |
ede9939c |
#if HAVE_JSON
mctx.wrkobj = ctx->wrkproperty;
#endif |
c1fce7f7 |
|
89670d69 |
/* |
45dc1456 |
* Is it a UNIX style mbox with more than one |
89670d69 |
* mail message, or just a single mail message? |
45dc1456 |
*
* 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 |
001ad879 |
* than one message is handled, e.g. giving a better indication of
* which message within the mailbox is infected |
89670d69 |
*/ |
25071deb |
/*if((strncmp(buffer, "From ", 5) == 0) && isalnum(buffer[5])) {*/
if(strncmp(buffer, "From ", 5) == 0) { |
e3aaff8e |
/* |
049a18b9 |
* Have been asked to check a UNIX style mbox file, which
* may contain more than one e-mail message to decode |
f003b79e |
*
* 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 |
e3aaff8e |
*/ |
ddea752e |
bool lastLineWasEmpty;
int messagenumber;
message *m = messageCreate();
|
ee1b2a6c |
if(m == NULL) |
ddea752e |
return CL_EMEM;
lastLineWasEmpty = FALSE;
messagenumber = 1; |
a603478f |
messageSetCTX(m, ctx); |
e3aaff8e |
|
049a18b9 |
do { |
89670d69 |
cli_chomp(buffer); |
25071deb |
/*if(lastLineWasEmpty && (strncmp(buffer, "From ", 5) == 0) && isalnum(buffer[5])) {*/
if(lastLineWasEmpty && (strncmp(buffer, "From ", 5) == 0)) { |
7dde984e |
cli_dbgmsg("Deal with message number %d\n", messagenumber++); |
e3aaff8e |
/* |
89670d69 |
* End of a message in the mail box |
e3aaff8e |
*/ |
b2223aad |
body = parseEmailHeaders(m, rfc821); |
4f1d0bfc |
if(body == NULL) {
messageReset(m);
continue;
} |
a603478f |
messageSetCTX(body, ctx); |
89670d69 |
messageDestroy(m); |
a603478f |
if(messageGetBody(body)) { |
ecc3d638 |
mbox_status rc = parseEmailBody(body, NULL, &mctx, 0);
if(rc == FAIL) { |
e17491b2 |
messageReset(body);
m = body;
continue; |
ecc3d638 |
} else if(rc == VIRUS) { |
a603478f |
cli_dbgmsg("Message number %d is infected\n", |
150c9f33 |
messagenumber-1); |
a603478f |
retcode = CL_VIRUS; |
826c9892 |
m = NULL; |
a603478f |
break; |
e17491b2 |
} |
a603478f |
} |
e3aaff8e |
/* |
89670d69 |
* Starting a new message, throw away all the |
f24bf390 |
* 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 |
e3aaff8e |
*/ |
89670d69 |
m = body;
messageReset(body); |
a603478f |
messageSetCTX(body, ctx); |
e3aaff8e |
|
049a18b9 |
cli_dbgmsg("Finished processing message\n"); |
89670d69 |
} else |
547b89de |
lastLineWasEmpty = (bool)(buffer[0] == '\0'); |
4945127a |
|
fa5661be |
if(isuuencodebegin(buffer)) { |
5198de85 |
/* |
4945127a |
* Fast track visa to uudecode.
* TODO: binhex, yenc
*/ |
ee1b2a6c |
if(uudecodeFile(m, buffer, dir, map, &at) < 0) |
fa5661be |
if(messageAddStr(m, buffer) < 0)
break;
} else |
69c62847 |
/* at this point, the \n has been removed */ |
4945127a |
if(messageAddStr(m, buffer) < 0)
break; |
ee1b2a6c |
} while(fmap_gets(map, buffer, &at, sizeof(buffer) - 1)); |
ddea752e |
|
a603478f |
if(retcode == CL_SUCCESS) {
cli_dbgmsg("Extract attachments from email %d\n", messagenumber);
body = parseEmailHeaders(m, rfc821);
}
if(m)
messageDestroy(m); |
4f1d0bfc |
} else { |
7e577f26 |
/*
* It's a single message, parse the headers then the body |
4f1d0bfc |
*/ |
69543a9d |
if(strncmp(buffer, "P I ", 4) == 0)
/*
* CommuniGate Pro format: ignore headers until
* blank line
*/ |
ee1b2a6c |
while(fmap_gets(map, buffer, &at, sizeof(buffer) - 1) && |
69543a9d |
(strchr("\r\n", buffer[0]) == NULL))
; |
f87a92ca |
/* getline_from_mbox could be using unlocked_stdio(3),
* so lock file here */ |
69543a9d |
/*
* Ignore any blank lines at the top of the message
*/ |
4f1d0bfc |
while(strchr("\r\n", buffer[0]) && |
ee1b2a6c |
(getline_from_mbox(buffer, sizeof(buffer) - 1, map, &at) != NULL)) |
87c9313e |
;
|
9ed148a8 |
buffer[sizeof(buffer) - 1] = '\0'; |
4b187745 |
|
ee1b2a6c |
body = parseEmailFile(map, &at, rfc821, buffer, dir); |
4f1d0bfc |
} |
7e577f26 |
|
4f1d0bfc |
if(body) {
/*
* Write out the last entry in the mailbox
*/ |
a603478f |
if((retcode == CL_SUCCESS) && messageGetBody(body)) {
messageSetCTX(body, ctx); |
242ffd7a |
switch(parseEmailBody(body, NULL, &mctx, 0)) { |
96435bdc |
case OK:
case OK_ATTACHMENTS_NOT_SAVED:
break; |
ecc3d638 |
case FAIL: |
69c62847 |
/*
* beware: cli_magic_scandesc(),
* changes this into CL_CLEAN, so only
* use it to inform the higher levels
* that we couldn't decode it because
* it isn't an mbox, not to signal
* decoding errors on what *is* a valid
* mbox
*/ |
a603478f |
retcode = CL_EFORMAT;
break; |
69c62847 |
case MAXREC:
retcode = CL_EMAXREC;
break; |
001ad879 |
case MAXFILES:
retcode = CL_EMAXFILES;
break; |
ecc3d638 |
case VIRUS: |
a603478f |
retcode = CL_VIRUS;
break;
}
} |
e3aaff8e |
|
92012beb |
if(body->isTruncated && retcode == CL_SUCCESS)
retcode = CL_EMEM; |
4f1d0bfc |
/*
* Tidy up and quit
*/
messageDestroy(body);
} |
6ad45a29 |
if((retcode == CL_CLEAN) && ctx->found_possibly_unwanted && |
d7979d4f |
(*ctx->virname == NULL || SCAN_ALLMATCHES)) { |
cbf5017a |
retcode = cli_append_virus(ctx, "Heuristics.Phishing.Email"); |
6ad45a29 |
ctx->found_possibly_unwanted = 0; |
ef5b10e0 |
}
|
e3aaff8e |
cli_dbgmsg("cli_mbox returning %d\n", retcode);
return retcode;
}
/* |
ddea752e |
* Read in an email message from fin, parse it, and return the message |
7e577f26 |
* |
ddea752e |
* FIXME: files full of new lines and nothing else are
* handled ungracefully...
*/
static message * |
49cc1e3c |
parseEmailFile(fmap_t *map, size_t *at, const table_t *rfc821, const char *firstLine, const char *dir) |
ddea752e |
{
bool inHeader = TRUE; |
6e3d492a |
bool bodyIsEmpty = TRUE; |
06466233 |
bool lastWasBlank = FALSE, lastBodyLineWasBlank = FALSE; |
ddea752e |
message *ret;
bool anyHeadersFound = FALSE;
int commandNumber = -1; |
41b7a56b |
char *fullline = NULL, *boundary = NULL; |
ddea752e |
size_t fulllinelength = 0; |
85bb253e |
char buffer[RFC2821LENGTH + 1]; |
ddea752e |
cli_dbgmsg("parseEmailFile\n");
ret = messageCreate();
if(ret == NULL)
return NULL;
|
69ffb9da |
strncpy(buffer, firstLine, sizeof(buffer)-1); |
ddea752e |
do { |
4f4a8f4a |
const char *line; |
ddea752e |
(void)cli_chomp(buffer);
|
4f4a8f4a |
if(buffer[0] == '\0') |
72cf1461 |
line = NULL; |
4f4a8f4a |
else
line = buffer; |
ddea752e |
/*
* Don't blank lines which are only spaces from headers,
* otherwise they'll be treated as the end of header marker
*/ |
41b7a56b |
if(lastWasBlank) {
lastWasBlank = FALSE;
if(boundaryStart(buffer, boundary)) {
cli_dbgmsg("Found a header line with space that should be blank\n");
inHeader = FALSE;
}
} |
ddea752e |
if(inHeader) { |
0ed29506 |
cli_dbgmsg("parseEmailFile: check '%s' fullline %p\n", |
1f6e52cb |
buffer, fullline); |
2a0041b8 |
/*
* Ensure wide characters are handled where
* sizeof(char) > 1
*/
if(line && isspace(line[0] & 0xFF)) { |
41b7a56b |
char copy[sizeof(buffer)];
strcpy(copy, buffer);
strstrip(copy);
if(copy[0] == '\0') {
/* |
4d4166a9 |
* 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. |
41b7a56b |
*/
if(fullline) {
if(parseEmailHeader(ret, fullline, rfc821) < 0)
continue;
free(fullline);
fullline = NULL;
} |
300a8ae9 |
if(boundary ||
((boundary = (char *)messageFindArgument(ret, "boundary")) != NULL)) { |
41b7a56b |
lastWasBlank = TRUE;
continue;
}
}
} |
72cf1461 |
if((line == NULL) && (fullline == NULL)) { /* empty line */ |
0ed29506 |
/*
* A blank line signifies the end of
* the header and the start of the text
*/
if(!anyHeadersFound)
/* Ignore the junk at the top */
continue; |
5860ae08 |
|
0ed29506 |
cli_dbgmsg("End of header information\n");
inHeader = FALSE;
bodyIsEmpty = TRUE; |
ddea752e |
} else {
char *ptr; |
f304dc68 |
const char *lookahead; |
ddea752e |
if(fullline == NULL) { |
85bb253e |
char cmd[RFC2821LENGTH + 1], out[RFC2821LENGTH + 1]; |
ddea752e |
/*
* Continuation of line we're ignoring?
*/ |
0ed29506 |
if(isblank(line[0])) |
ddea752e |
continue; |
e107e8cf |
|
ddea752e |
/*
* Is this a header we're interested in?
*/ |
72cf1461 |
if((strchr(line, ':') == NULL) ||
(cli_strtokbuf(line, 0, ":", cmd) == NULL)) {
if(strncmp(line, "From ", 5) == 0) |
ddea752e |
anyHeadersFound = TRUE;
continue;
}
|
d72749e0 |
ptr = rfc822comments(cmd, out); |
ddea752e |
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;
} |
0cf4cea7 |
fullline = cli_strdup(line); |
72cf1461 |
fulllinelength = strlen(line) + 1; |
92012beb |
if(!fullline) {
if(ret)
ret->isTruncated = TRUE;
break;
} |
72cf1461 |
} else if(line != NULL) { |
1f271616 |
fulllinelength += strlen(line) + 1; |
468c0f21 |
ptr = cli_realloc(fullline, fulllinelength);
if(ptr == NULL)
continue;
fullline = ptr; |
1f271616 |
cli_strlcat(fullline, line, fulllinelength); |
ddea752e |
}
assert(fullline != NULL);
|
ee1b2a6c |
if((lookahead = fmap_need_off_once(map, *at, 1))) { |
ddea752e |
/*
* Section B.2 of RFC822 says TAB or
* SPACE means a continuation of the
* previous entry.
*
* Add all the arguments on the line
*/ |
ee1b2a6c |
if(isblank(*lookahead)) |
ddea752e |
continue;
}
|
11f253d6 |
/*
* Handle broken headers, where the next
* line isn't indented by whitespace
*/ |
631baca8 |
if(fullline[strlen(fullline) - 1] == ';') |
11f253d6 |
/* Add arguments to this line */
continue;
|
4f4a8f4a |
if(line && (count_quotes(fullline) & 1))
continue; |
ddea752e |
|
d72749e0 |
ptr = rfc822comments(fullline, NULL); |
ddea752e |
if(ptr) {
free(fullline);
fullline = ptr;
}
if(parseEmailHeader(ret, fullline, rfc821) < 0)
continue;
free(fullline);
fullline = NULL;
} |
fa5661be |
} else if(line && isuuencodebegin(line)) { |
ae5c693a |
/*
* Fast track visa to uudecode.
* TODO: binhex, yenc
*/ |
6e3d492a |
bodyIsEmpty = FALSE; |
ee1b2a6c |
if(uudecodeFile(ret, line, dir, map, at) < 0) |
fa5661be |
if(messageAddStr(ret, line) < 0)
break; |
06466233 |
} else {
if(line == NULL) { |
c52d991e |
/*
* Although this would save time and RAM, some
* phish signatures have been built which need
* the blank lines
*/
if(lastBodyLineWasBlank &&
(messageGetMimeType(ret) != TEXT)) { |
06466233 |
cli_dbgmsg("Ignoring consecutive blank lines in the body\n");
continue;
}
lastBodyLineWasBlank = TRUE; |
6e3d492a |
} else {
if(bodyIsEmpty) {
/*
* Broken message: new line in the
* middle of the headers, so the first
* line of the body is in fact
* the last lines of the header
*/ |
0cf4cea7 |
if(newline_in_header(line)) |
6e3d492a |
continue; |
0cf4cea7 |
bodyIsEmpty = FALSE; |
6e3d492a |
} |
06466233 |
lastBodyLineWasBlank = FALSE; |
6e3d492a |
} |
06466233 |
|
72cf1461 |
if(messageAddStr(ret, line) < 0) |
ddea752e |
break; |
06466233 |
} |
ee1b2a6c |
} while(getline_from_mbox(buffer, sizeof(buffer) - 1, map, at) != NULL); |
ddea752e |
|
300a8ae9 |
if(boundary)
free(boundary);
|
ddea752e |
if(fullline) {
if(*fullline) switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE: |
0d35f10f |
cli_dbgmsg("parseEmailFile: Fullline unparsed '%s'\n", fullline); |
ddea752e |
}
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;
}
cli_dbgmsg("parseEmailFile: return\n");
return ret;
}
/*
* The given message contains a raw e-mail. |
e06d34dc |
* |
94f051b0 |
* Returns the message's body with the correct arguments set, empties the
* given message's contents (note that it isn't destroyed) |
ddea752e |
*
* TODO: remove the duplication with parseEmailFile |
7e577f26 |
*/ |
e06d34dc |
static message * |
2673dc74 |
parseEmailHeaders(message *m, const table_t *rfc821) |
7e577f26 |
{ |
e06d34dc |
bool inHeader = TRUE; |
0d35f10f |
bool bodyIsEmpty = TRUE; |
94f051b0 |
text *t; |
89670d69 |
message *ret; |
4f1d0bfc |
bool anyHeadersFound = FALSE; |
15033cb6 |
int commandNumber = -1; |
efb5f16c |
char *fullline = NULL; |
2ad0c86e |
size_t fulllinelength = 0; |
89670d69 |
|
02927896 |
cli_dbgmsg("parseEmailHeaders\n");
|
89670d69 |
if(m == NULL)
return NULL;
ret = messageCreate(); |
7e577f26 |
|
b2223aad |
for(t = messageGetBody(m); t; t = t->t_next) { |
0cf4cea7 |
const char *line; |
7e577f26 |
|
b2223aad |
if(t->t_line) |
0cf4cea7 |
line = lineGetData(t->t_line); |
b2223aad |
else |
0cf4cea7 |
line = NULL; |
7e577f26 |
|
28010d29 |
if(inHeader) { |
4e75d9b6 |
cli_dbgmsg("parseEmailHeaders: check '%s'\n", |
0cf4cea7 |
line ? line : "");
if(line == NULL) { |
4e75d9b6 |
/*
* A blank line signifies the end of
* the header and the start of the text
*/
cli_dbgmsg("End of header information\n"); |
ad091acf |
if(!anyHeadersFound) {
cli_dbgmsg("Nothing interesting in the header\n");
break;
} |
0d35f10f |
inHeader = FALSE;
bodyIsEmpty = TRUE; |
2ad0c86e |
} else { |
b116962d |
char *ptr;
|
2ad0c86e |
if(fullline == NULL) { |
85bb253e |
char cmd[RFC2821LENGTH + 1]; |
45aba293 |
/*
* Continuation of line we're ignoring?
*/ |
0cf4cea7 |
if(isblank(line[0])) |
45aba293 |
continue;
/*
* Is this a header we're interested in?
*/ |
0cf4cea7 |
if((strchr(line, ':') == NULL) ||
(cli_strtokbuf(line, 0, ":", cmd) == NULL)) {
if(strncmp(line, "From ", 5) == 0) |
b116962d |
anyHeadersFound = TRUE; |
45aba293 |
continue; |
b116962d |
} |
45aba293 |
|
d72749e0 |
ptr = rfc822comments(cmd, NULL); |
59921c02 |
commandNumber = tableFind(rfc821, ptr ? ptr : cmd);
if(ptr)
free(ptr); |
45aba293 |
switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE: |
b116962d |
anyHeadersFound = TRUE; |
45aba293 |
break;
default: |
ddea752e |
if(!anyHeadersFound)
anyHeadersFound = usefulHeader(commandNumber, cmd); |
45aba293 |
continue;
} |
0cf4cea7 |
fullline = cli_strdup(line);
fulllinelength = strlen(line) + 1;
} else if(line) { |
1f271616 |
fulllinelength += strlen(line) + 1; |
468c0f21 |
ptr = cli_realloc(fullline, fulllinelength);
if(ptr == NULL)
continue;
fullline = ptr; |
1f271616 |
cli_strlcat(fullline, line, fulllinelength); |
2ad0c86e |
} |
ddea752e |
assert(fullline != NULL); |
2ad0c86e |
|
842c7d49 |
if(next_is_folded_header(t))
/* Add arguments to this line */
continue; |
2ad0c86e |
|
94f051b0 |
lineUnlink(t->t_line);
t->t_line = NULL;
|
4f4a8f4a |
if(count_quotes(fullline) & 1) |
2ad0c86e |
continue;
|
d72749e0 |
ptr = rfc822comments(fullline, NULL); |
2ad0c86e |
if(ptr) {
free(fullline);
fullline = ptr;
} |
c77c8809 |
|
ddea752e |
if(parseEmailHeader(ret, fullline, rfc821) < 0)
continue; |
28010d29 |
|
ddea752e |
free(fullline);
fullline = NULL; |
f12d2498 |
} |
0d35f10f |
} else {
if(bodyIsEmpty) { |
0cf4cea7 |
if(line == NULL) |
0d35f10f |
/* throw away leading blank lines */
continue;
/*
* Broken message: new line in the
* middle of the headers, so the first
* line of the body is in fact
* the last lines of the header
*/ |
0cf4cea7 |
if(newline_in_header(line)) |
0d35f10f |
continue;
bodyIsEmpty = FALSE;
}
/*if(t->t_line && isuuencodebegin(t->t_line))
puts("FIXME: add fast visa here");*/ |
faa0d267 |
cli_dbgmsg("parseEmailHeaders: finished with headers, moving body\n"); |
94f051b0 |
messageMoveText(ret, t, m);
break; |
0d35f10f |
} |
d879a7b0 |
} |
e06d34dc |
|
efb5f16c |
if(fullline) { |
2ad0c86e |
if(*fullline) switch(commandNumber) {
case CONTENT_TRANSFER_ENCODING:
case CONTENT_DISPOSITION:
case CONTENT_TYPE: |
aaaae842 |
cli_dbgmsg("parseEmailHeaders: Fullline unparsed '%s'\n", fullline); |
2ad0c86e |
} |
efb5f16c |
free(fullline);
}
|
4f1d0bfc |
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;
}
|
e17491b2 |
cli_dbgmsg("parseEmailHeaders: return\n");
|
e06d34dc |
return ret; |
7e577f26 |
}
/* |
4c60b74f |
* Handle a header line of an email message
*/
static int |
393a6d67 |
parseEmailHeader(message *m, const char *line, const table_t *rfc821) |
4c60b74f |
{ |
001ad879 |
int ret; |
4c60b74f |
#ifdef CL_THREAD_SAFE
char *strptr;
#endif |
7cd9337a |
const char *separator;
char *cmd, *copy, tokenseparator[2]; |
4c60b74f |
|
20d3dde9 |
cli_dbgmsg("parseEmailHeader '%s'\n", line);
|
97867f21 |
/* |
7cd9337a |
* In RFC822 the separator between the key a value is a colon, |
97867f21 |
* e.g. Content-Transfer-Encoding: base64
* However some MUA's are lapse about this and virus writers exploit |
22cb38ed |
* this hole, so we need to check all known possibilities |
97867f21 |
*/ |
7cd9337a |
for(separator = ":= "; *separator; separator++)
if(strchr(line, *separator) != NULL) |
97867f21 |
break;
|
7cd9337a |
if(*separator == '\0') |
1bfbedd4 |
return -1;
|
50df4118 |
copy = rfc2047(line);
if(copy == NULL) |
8ebe8dbc |
/* an RFC checker would return -1 here */ |
0cf4cea7 |
copy = cli_strdup(line); |
f2f25418 |
|
7cd9337a |
tokenseparator[0] = *separator;
tokenseparator[1] = '\0'; |
97867f21 |
|
001ad879 |
ret = -1;
|
548a5f96 |
#ifdef CL_THREAD_SAFE |
7cd9337a |
cmd = strtok_r(copy, tokenseparator, &strptr); |
548a5f96 |
#else |
7cd9337a |
cmd = strtok(copy, tokenseparator); |
548a5f96 |
#endif |
4c60b74f |
|
3499d81e |
if(cmd && (strstrip(cmd) > 0)) { |
548a5f96 |
#ifdef CL_THREAD_SAFE |
4c60b74f |
char *arg = strtok_r(NULL, "", &strptr); |
548a5f96 |
#else
char *arg = strtok(NULL, "");
#endif |
4c60b74f |
if(arg)
/*
* Found a header such as
* Content-Type: multipart/mixed;
* set arg to be
* "multipart/mixed" and cmd to |
a9f386ed |
* be "Content-Type" |
4c60b74f |
*/ |
393a6d67 |
ret = parseMimeHeader(m, cmd, rfc821, arg); |
4c60b74f |
} |
50df4118 |
free(copy); |
4c60b74f |
return ret;
}
|
c2df9f79 |
#if HAVE_LIBXML2
static const struct key_entry mhtml_keys[] = { |
a6369bc8 |
/* root html tags for microsoft office document */
{ "html", "RootHTML", MSXML_JSON_ROOT | MSXML_JSON_ATTRIB }, |
c2df9f79 |
|
a6369bc8 |
{ "head", "Head", MSXML_JSON_WRKPTR | MSXML_COMMENT_CB },
{ "meta", "Meta", MSXML_JSON_WRKPTR | MSXML_JSON_MULTI | MSXML_JSON_ATTRIB },
{ "link", "Link", MSXML_JSON_WRKPTR | MSXML_JSON_MULTI | MSXML_JSON_ATTRIB },
{ "script", "Script", MSXML_JSON_WRKPTR | MSXML_JSON_MULTI | MSXML_JSON_VALUE } |
c2df9f79 |
};
static size_t num_mhtml_keys = sizeof(mhtml_keys) / sizeof(struct key_entry); |
a6369bc8 |
static const struct key_entry mhtml_comment_keys[] = {
/* embedded xml tags (comment) for microsoft office document */
{ "o:documentproperties", "DocumentProperties", MSXML_JSON_ROOT | MSXML_JSON_ATTRIB },
{ "o:author", "Author", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:lastauthor", "LastAuthor", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:revision", "Revision", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:totaltime", "TotalTime", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:created", "Created", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:lastsaved", "LastSaved", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:pages", "Pages", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:words", "Words", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:characters", "Characters", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:company", "Company", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:lines", "Lines", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:paragraphs", "Paragraphs", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:characterswithspaces", "CharactersWithSpaces", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:version", "Version", MSXML_JSON_WRKPTR | MSXML_JSON_VALUE },
{ "o:officedocumentsettings", "DocumentSettings", MSXML_IGNORE_ELEM },
{ "w:worddocument", "WordDocument", MSXML_IGNORE_ELEM },
{ "w:latentstyles", "LatentStyles", MSXML_IGNORE_ELEM }
};
static size_t num_mhtml_comment_keys = sizeof(mhtml_comment_keys) / sizeof(struct key_entry); |
c2df9f79 |
#endif
/* |
a6369bc8 |
* The related multipart root HTML file comment parsing wrapper.
*
* Attempts to leverage msxml parser, cannot operate without LIBXML2.
* This function is only used for Preclassification JSON.
*/
static int
parseMHTMLComment(const char *comment, cli_ctx *ctx, void *wrkjobj, void *cbdata)
{
#if HAVE_LIBXML2
const char *xmlsrt, *xmlend;
xmlTextReaderPtr reader;
#if HAVE_JSON
json_object *thisjobj = (json_object *)wrkjobj;
#endif
int ret = CL_SUCCESS;
UNUSEDPARAM(cbdata); |
4cd97da4 |
UNUSEDPARAM(wrkjobj); |
a6369bc8 |
xmlend = comment;
while ((xmlsrt = strstr(xmlend, "<xml>"))) {
xmlend = strstr(xmlsrt, "</xml>");
if (xmlend == NULL) {
cli_dbgmsg("parseMHTMLComment: unbounded xml tag\n");
break;
}
reader = xmlReaderForMemory(xmlsrt, xmlend-xmlsrt+6, "comment.xml", NULL, CLAMAV_MIN_XMLREADER_FLAGS);
if (!reader) { |
7cd9337a |
cli_dbgmsg("parseMHTMLComment: cannot initialize xmlReader\n"); |
a6369bc8 |
#if HAVE_JSON |
ee6fb5fb |
if (ctx->wrkproperty != NULL)
ret = cli_json_parse_error(ctx->wrkproperty, "MHTML_ERROR_XML_READER_MEM"); |
a6369bc8 |
#endif
return ret; // libxml2 failed!
}
/* comment callback is not set to prevent recursion */
/* TODO: should we separate the key dictionaries? */
/* TODO: should we use the json object pointer? */
ret = cli_msxml_parse_document(ctx, reader, mhtml_comment_keys, num_mhtml_comment_keys, MSXML_FLAG_JSON, NULL);
xmlTextReaderClose(reader);
xmlFreeTextReader(reader);
if (ret != CL_SUCCESS)
return ret;
}
#else
UNUSEDPARAM(comment);
UNUSEDPARAM(ctx);
UNUSEDPARAM(wrkjobj);
UNUSEDPARAM(cbdata);
cli_dbgmsg("in parseMHTMLComment\n");
cli_dbgmsg("parseMHTMLComment: parsing html xml-comments requires libxml2!\n");
#endif
return CL_SUCCESS;
}
/* |
c2df9f79 |
* The related multipart root HTML file parsing wrapper.
*
* Attempts to leverage msxml parser, cannot operate without LIBXML2.
* This function is only used for Preclassification JSON.
*/
static mbox_status
parseRootMHTML(mbox_ctx *mctx, message *m, text *t)
{ |
a6369bc8 |
cli_ctx *ctx = mctx->ctx; |
c2df9f79 |
#if HAVE_LIBXML2 |
f5412cc9 |
#ifdef LIBXML_HTML_ENABLED |
a6369bc8 |
struct msxml_ctx mxctx; |
4cd97da4 |
blob *input = NULL; |
a6369bc8 |
htmlDocPtr htmlDoc;
xmlTextReaderPtr reader;
int ret = CL_SUCCESS;
mbox_status rc = OK; |
c2df9f79 |
#if HAVE_JSON |
a6369bc8 |
json_object *rhtml; |
c2df9f79 |
#endif
|
a6369bc8 |
cli_dbgmsg("in parseRootMHTML\n"); |
c2df9f79 |
|
a6369bc8 |
if (ctx == NULL)
return OK; |
c2df9f79 |
|
a6369bc8 |
if (m == NULL && t == NULL)
return OK; |
c2df9f79 |
|
a6369bc8 |
if (m != NULL)
input = messageToBlob(m, 0); |
4cd97da4 |
else /* t != NULL */ |
a6369bc8 |
input = textToBlob(t, NULL, 0); |
4cd97da4 |
|
a6369bc8 |
if (input == NULL)
return OK; |
c2df9f79 |
|
26427291 |
htmlDoc = htmlReadMemory((char*)input->data, input->len, "mhtml.html", NULL, CLAMAV_MIN_XMLREADER_FLAGS); |
a6369bc8 |
if (htmlDoc == NULL) { |
7cd9337a |
cli_dbgmsg("parseRootMHTML: cannot initialize read html document\n"); |
c2df9f79 |
#if HAVE_JSON |
ee6fb5fb |
if (ctx->wrkproperty != NULL)
ret = cli_json_parse_error(ctx->wrkproperty, "MHTML_ERROR_HTML_READ"); |
a6369bc8 |
if (ret != CL_SUCCESS)
rc = FAIL; |
c2df9f79 |
#endif |
a6369bc8 |
blobDestroy(input);
return rc;
} |
c2df9f79 |
#if HAVE_JSON |
a6369bc8 |
if (mctx->wrkobj) {
rhtml = cli_jsonobj(mctx->wrkobj, "RootHTML");
if (rhtml != NULL) {
/* MHTML-specific properties */ |
dd2ed14d |
cli_jsonstr(rhtml, "Encoding", (const char*)htmlGetMetaEncoding(htmlDoc)); |
a6369bc8 |
cli_jsonint(rhtml, "CompressMode", xmlGetDocCompressMode(htmlDoc));
} |
c2df9f79 |
}
#endif
|
a6369bc8 |
reader = xmlReaderWalker(htmlDoc);
if (reader == NULL) { |
7cd9337a |
cli_dbgmsg("parseRootMHTML: cannot initialize xmlTextReader\n"); |
c2df9f79 |
#if HAVE_JSON |
ee6fb5fb |
if (ctx->wrkproperty != NULL)
ret = cli_json_parse_error(ctx->wrkproperty, "MHTML_ERROR_XML_READER_IO"); |
a6369bc8 |
if (ret != CL_SUCCESS)
rc = FAIL; |
c2df9f79 |
#endif |
a6369bc8 |
blobDestroy(input);
return rc;
}
memset(&mxctx, 0, sizeof(mxctx));
/* no scanning callback set */
mxctx.comment_cb = parseMHTMLComment;
ret = cli_msxml_parse_document(ctx, reader, mhtml_keys, num_mhtml_keys, MSXML_FLAG_JSON | MSXML_FLAG_WALK, &mxctx);
switch (ret) {
case CL_SUCCESS:
case CL_ETIMEOUT:
case CL_BREAK:
rc = OK;
break;
case CL_EMAXREC:
rc = MAXREC;
break;
case CL_EMAXFILES:
rc = MAXFILES;
break;
case CL_VIRUS: |
26427291 |
rc = VIRUS; |
a6369bc8 |
break;
default:
rc = FAIL;
}
|
c44d7e7a |
xmlTextReaderClose(reader);
xmlFreeTextReader(reader); |
a6369bc8 |
xmlFreeDoc(htmlDoc); |
c2df9f79 |
blobDestroy(input);
return rc; |
f5412cc9 |
#else /* LIBXML_HTML_ENABLED */
UNUSEDPARAM(m);
UNUSEDPARAM(t);
cli_dbgmsg("in parseRootMHTML\n");
cli_dbgmsg("parseRootMHTML: parsing html documents disabled in libxml2!\n");
#endif /* LIBXML_HTML_ENABLED */
#else /* HAVE_LIBXML2 */ |
a6369bc8 |
UNUSEDPARAM(m);
UNUSEDPARAM(t); |
c2df9f79 |
cli_dbgmsg("in parseRootMHTML\n");
cli_dbgmsg("parseRootMHTML: parsing html documents requires libxml2!\n");
return OK; |
f5412cc9 |
#endif /* HAVE_LIBXML2 */ |
c2df9f79 |
}
|
4c60b74f |
/* |
e3aaff8e |
* This is a recursive routine.
* |
7e577f26 |
* This function parses the body of mainMessage and saves its attachments in dir
* |
e06d34dc |
* mainMessage is the buffer to be parsed, it contains an e-mail's body, without |
f12d2498 |
* any headers. First time of calling it'll be
* the whole message. Later it'll be parts of a multipart message |
e3aaff8e |
* textIn is the plain text message being built up so far
*/ |
ecc3d638 |
static mbox_status |
242ffd7a |
parseEmailBody(message *messageIn, text *textIn, mbox_ctx *mctx, unsigned int recursion_level) |
e3aaff8e |
{ |
42eebd87 |
mbox_status rc; |
2673dc74 |
text *aText = textIn;
message *mainMessage = messageIn; |
b0b860f1 |
fileblob *fb; |
a603478f |
bool infected = FALSE; |
724b2bf7 |
const struct cl_engine *engine = mctx->ctx->engine;
const int doPhishingScan = engine->dboptions&CL_DB_PHISHING_URLS && (DCONF_PHISHING & PHISHING_CONF_ENGINE); |
ede9939c |
#if HAVE_JSON
json_object *saveobj = mctx->wrkobj;
#endif |
8386c723 |
|
001ad879 |
cli_dbgmsg("in parseEmailBody, %u files saved so far\n",
mctx->files); |
e3aaff8e |
|
724b2bf7 |
/* FIXMELIMITS: this should be better integrated */
if(engine->maxreclevel)
/*
* This is approximate
*/
if(recursion_level > engine->maxreclevel) { |
a5faab66 |
|
6351aa86 |
cli_dbgmsg("parseEmailBody: hit maximum recursion level (%u)\n", recursion_level); |
d91ab809 |
return MAXREC; |
001ad879 |
} |
724b2bf7 |
if(engine->maxfiles && (mctx->files >= engine->maxfiles)) {
/*
* FIXME: This is only approx - it may have already
* been exceeded
*/
cli_dbgmsg("parseEmailBody: number of files exceeded %u\n", engine->maxfiles);
return MAXFILES; |
72910996 |
} |
242ffd7a |
|
f7bf6fd2 |
rc = OK;
|
e3aaff8e |
/* Anything left to be parsed? */ |
d4d14218 |
if(mainMessage && (messageGetBody(mainMessage) != NULL)) { |
e3aaff8e |
mime_type mimeType; |
2673dc74 |
int subtype, inhead, htmltextPart, inMimeHead, i; |
95e11e5a |
const char *mimeSubtype; |
6351aa86 |
char *boundary; |
e3aaff8e |
const text *t_line; |
98cb5cba |
/*bool isAlternative;*/ |
e3aaff8e |
message *aMessage; |
2673dc74 |
int multiparts = 0;
message **messages = NULL; /* parts of a multipart message */ |
e3aaff8e |
|
049a18b9 |
cli_dbgmsg("Parsing mail file\n");
|
e3aaff8e |
mimeType = messageGetMimeType(mainMessage);
mimeSubtype = messageGetMimeSubtype(mainMessage); |
ede9939c |
#if HAVE_JSON
if (mctx->wrkobj != NULL) {
mctx->wrkobj = cli_jsonobj(mctx->wrkobj, "Body");
cli_jsonstr(mctx->wrkobj, "MimeType", getMimeTypeStr(mimeType));
cli_jsonstr(mctx->wrkobj, "MimeSubtype", mimeSubtype);
cli_jsonstr(mctx->wrkobj, "EncodingType", getEncTypeStr(messageGetEncoding(mainMessage)));
cli_jsonstr(mctx->wrkobj, "Disposition", messageGetDispositionType(mainMessage));
cli_jsonstr(mctx->wrkobj, "Filename", messageHasFilename(mainMessage) ?
messageGetFilename(mainMessage): "(inline)");
}
#endif |
e3aaff8e |
|
528c8a2d |
/* pre-process */ |
c1fce7f7 |
subtype = tableFind(mctx->subtypeTable, mimeSubtype); |
7c5a7a47 |
if((mimeType == TEXT) && (subtype == PLAIN)) { |
e3aaff8e |
/*
* This is effectively no encoding, notice that we
* don't check that charset is us-ascii
*/ |
7cd9337a |
cli_dbgmsg("text/plain: Assume no attachments\n"); |
e3aaff8e |
mimeType = NOMIME; |
ae5c693a |
messageSetMimeSubtype(mainMessage, ""); |
528c8a2d |
} else if((mimeType == MESSAGE) &&
(strcasecmp(mimeSubtype, "rfc822-headers") == 0)) {
/*
* RFC1892/RFC3462: section 2 text/rfc822-headers
* incorrectly sent as message/rfc822-headers |
46d375fe |
*
* Parse as text/plain, i.e. no mime |
528c8a2d |
*/
cli_dbgmsg("Changing message/rfc822-headers to text/rfc822-headers\n"); |
46d375fe |
mimeType = NOMIME; |
ae5c693a |
messageSetMimeSubtype(mainMessage, ""); |
a05e6d45 |
} else |
8affc406 |
cli_dbgmsg("mimeType = %d\n", (int)mimeType); |
049a18b9 |
|
e3aaff8e |
switch(mimeType) {
case NOMIME: |
6fd711b2 |
cli_dbgmsg("Not a mime encoded message\n"); |
e3aaff8e |
aText = textAddMessage(aText, mainMessage); |
faa0d267 |
|
647a4f8d |
if(!doPhishingScan)
break; |
94aea271 |
/*
* Fall through: some phishing mails claim they are
* text/plain, when they are in fact html
*/ |
e3aaff8e |
case TEXT: |
8a892c3b |
/* text/plain has been preprocessed as no encoding */ |
6a4dd9dc |
if(doPhishingScan) { |
47d9cc65 |
/*
* It would be better to save and scan the
* file and only checkURLs if it's found to be
* clean
*/ |
8b899010 |
checkURLs(mainMessage, mctx, &rc, (subtype == HTML));
/*
* There might be html sent without subtype
* html too, so scan them for phishing
*/ |
ecc3d638 |
if(rc == VIRUS) |
8b899010 |
infected = TRUE; |
c52d991e |
} |
e3aaff8e |
break;
case MULTIPART: |
c8a7cef0 |
cli_dbgmsg("Content-type 'multipart' handler\n"); |
e3aaff8e |
boundary = messageFindArgument(mainMessage, "boundary");
|
ede9939c |
#if HAVE_JSON |
ee6fb5fb |
if (mctx->wrkobj != NULL)
cli_jsonstr(mctx->wrkobj, "Boundary", boundary); |
ede9939c |
#endif
|
e3aaff8e |
if(boundary == NULL) { |
e25960d2 |
cli_dbgmsg("Multipart/%s MIME message contains no boundary header\n", |
842c7d49 |
mimeSubtype); |
e2e7ebf5 |
/* Broken e-mail message */
mimeType = NOMIME;
/*
* The break means that we will still
* check if the file contains a uuencoded file
*/
break; |
e3aaff8e |
}
|
ede9939c |
cli_chomp(boundary); |
23889d27 |
|
c79a2273 |
/* Perhaps it should assume mixed? */ |
cb5a87e0 |
if(mimeSubtype[0] == '\0') { |
6351aa86 |
cli_dbgmsg("Multipart has no subtype assuming alternative\n"); |
cb5a87e0 |
mimeSubtype = "alternative";
messageSetMimeSubtype(mainMessage, "alternative");
}
|
e3aaff8e |
/*
* Get to the start of the first message
*/ |
20d3dde9 |
t_line = messageGetBody(mainMessage);
if(t_line == NULL) { |
6351aa86 |
cli_dbgmsg("Multipart MIME message has no body\n"); |
20d3dde9 |
free((char *)boundary);
mimeType = NOMIME;
break;
}
do |
bae9c53f |
if(t_line->t_line) {
if(boundaryStart(lineGetData(t_line->t_line), boundary))
break;
/* |
182bbcc8 |
* Found a binhex file before |
ebe57840 |
* the first multipart |
b116962d |
* TODO: check yEnc |
bae9c53f |
*/ |
182bbcc8 |
if(binhexBegin(mainMessage) == t_line) { |
001ad879 |
if(exportBinhexMessage(mctx, mainMessage)) { |
a05e6d45 |
/* virus found */ |
ecc3d638 |
rc = VIRUS; |
47d9cc65 |
infected = TRUE; |
a05e6d45 |
break; |
b116962d |
} |
42eebd87 |
} else if(t_line->t_next && |
a5faab66 |
(encodingLine(mainMessage) == t_line->t_next)) { |
99c2299d |
/*
* 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
*/ |
42eebd87 |
cli_dbgmsg("Found MIME attachment before the first MIME section \"%s\"\n",
lineGetData(t_line->t_next->t_line)); |
99c2299d |
if(messageGetEncoding(mainMessage) == NOENCODING)
break; |
b116962d |
} |
bae9c53f |
} |
20d3dde9 |
while((t_line = t_line->t_next) != NULL); |
e3aaff8e |
if(t_line == NULL) { |
47d9cc65 |
cli_dbgmsg("Multipart MIME message contains no boundary lines (%s)\n",
boundary); |
5a642650 |
free((char *)boundary); |
e2e7ebf5 |
mimeType = NOMIME;
/*
* The break means that we will still |
182bbcc8 |
* check if the file contains a yEnc/binhex file |
e2e7ebf5 |
*/
break; |
e3aaff8e |
}
/*
* 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;
|
e06d34dc |
/* |
9bccc1e5 |
* Re-read this variable in case mimeSubtype has changed
*/
subtype = tableFind(mctx->subtypeTable, mimeSubtype);
/* |
9a729c80 |
* Parse the mainMessage object and create an array
* of objects called messages, one for each of the |
9bccc1e5 |
* multiparts that mainMessage contains. |
3f46285b |
* |
e06d34dc |
* This looks like parseEmailHeaders() - maybe there's
* some duplication of code to be cleaned up |
9bccc1e5 |
* |
842c7d49 |
* We may need to create an array rather than just |
9bccc1e5 |
* save each part as it is found because not all
* elements will need scanning, and we don't yet know
* which of those elements it will be, except in
* the case of mixed, when all parts need to be scanned. |
e06d34dc |
*/ |
9bccc1e5 |
for(multiparts = 0; t_line && !infected; multiparts++) { |
56ae62e2 |
int lines = 0; |
1a74d4df |
message **m; |
69c62847 |
mbox_status old_rc; |
56ae62e2 |
|
1a74d4df |
m = cli_realloc(messages, ((multiparts + 1) * sizeof(message *))); |
f12d2498 |
if(m == NULL) |
1a74d4df |
break;
messages = m; |
b726511f |
|
e3aaff8e |
aMessage = messages[multiparts] = messageCreate(); |
3f3f9085 |
if(aMessage == NULL) {
multiparts--; |
f7f59c08 |
/* if allocation failed the first time,
* there's no point in retrying, just
* break out */
break; |
3f3f9085 |
} |
c1fce7f7 |
messageSetCTX(aMessage, mctx->ctx); |
e3aaff8e |
cli_dbgmsg("Now read in part %d\n", multiparts);
|
8ba634a9 |
/*
* Ignore blank lines. There shouldn't be ANY
* but some viruses insert them
*/ |
02927896 |
while((t_line = t_line->t_next) != NULL) |
b2223aad |
if(t_line->t_line &&
/*(cli_chomp(t_line->t_text) > 0))*/
(strlen(lineGetData(t_line->t_line)) > 0)) |
d79597e3 |
break; |
8ba634a9 |
if(t_line == NULL) {
cli_dbgmsg("Empty part\n"); |
b9ce9639 |
/*
* Remove this part unless there's |
182bbcc8 |
* a binhex portion somewhere in |
b9ce9639 |
* the complete message that we may
* throw away by mistake if the MIME
* encoding information is incorrect
*/ |
9bccc1e5 |
if(mainMessage &&
(binhexBegin(mainMessage) == NULL)) { |
b9ce9639 |
messageDestroy(aMessage);
--multiparts;
} |
8ba634a9 |
continue;
}
do { |
b2223aad |
const char *line = lineGetData(t_line->t_line); |
e3aaff8e |
|
fa5661be |
/*cli_dbgmsg("multipart %d: inMimeHead %d inhead %d boundary '%s' line '%s' next '%s'\n",
multiparts, inMimeHead, inhead, boundary, line, |
391f7bb3 |
t_line->t_next && t_line->t_next->t_line ? lineGetData(t_line->t_next->t_line) : "(null)");*/ |
e3aaff8e |
|
f1c1300c |
if(inMimeHead) { /* continuation line */ |
02927896 |
if(line == NULL) { |
59b99810 |
/*inhead =*/ inMimeHead = 0; |
02927896 |
continue;
} |
3a978f7d |
/*
* Handle continuation lines
* because the previous line |
1eec55a6 |
* ended with a ; or this line
* starts with a white space |
3a978f7d |
*/ |
1eec55a6 |
cli_dbgmsg("Multipart %d: About to add mime Argument '%s'\n",
multiparts, line); |
3a978f7d |
/*
* Handle the case when it
* isn't really a continuation
* line:
* Content-Type: application/octet-stream;
* Content-Transfer-Encoding: base64
*/ |
c1fce7f7 |
parseEmailHeader(aMessage, line, mctx->rfc821Table); |
3a978f7d |
|
e3aaff8e |
while(isspace((int)*line))
line++;
if(*line == '\0') {
inhead = inMimeHead = 0;
continue;
} |
0ed29506 |
inMimeHead = FALSE; |
e3aaff8e |
messageAddArgument(aMessage, line); |
f1c1300c |
} else if(inhead) { /* handling normal headers */ |
4f4a8f4a |
/*int quotes;*/ |
c8a7cef0 |
char *fullline, *ptr; |
2ad0c86e |
|
02927896 |
if(line == NULL) { |
7e67e382 |
/*
* 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" |
5198de85 |
* |
7e67e382 |
* Content-Disposition: attachment;
* filename=text.zip
* Content-Type: application/octet-stream;
* name=text.zip
* Content-Transfer-Encoding: base64 |
5198de85 |
* |
7e67e382 |
* UEsDBAoAAAAAAACgPjJ2RHw676gAAO+oAABEAAAAbWFpbF90ZXh0LWluZm8udHh0ICAgICAgICAg
*/ |
842c7d49 |
const text *next = t_line->t_next;
|
7e67e382 |
if(next && next->t_line) {
const char *data = lineGetData(next->t_line); |
5b76248c |
if((messageGetEncoding(aMessage) == NOENCODING) && |
842c7d49 |
(messageGetMimeType(aMessage) == APPLICATION) && |
faa0d267 |
data && strstr(data, "base64")) { |
d72749e0 |
/*
* 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"
*/ |
842c7d49 |
messageSetEncoding(aMessage, "base64");
cli_dbgmsg("Ignoring fake end of headers\n");
continue;
} |
5cdb01fc |
if((strncmp(data, "Content", 7) == 0) ||
(strncmp(data, "filename=", 9) == 0)) { |
7e67e382 |
cli_dbgmsg("Ignoring fake end of headers\n");
continue;
}
} |
59b99810 |
cli_dbgmsg("Multipart %d: End of header information\n",
multiparts); |
e3aaff8e |
inhead = 0;
continue;
} |
c76810dc |
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;
}
|
c8a7cef0 |
inMimeHead = FALSE; |
b2223aad |
|
85bb253e |
assert(strlen(line) <= RFC2821LENGTH); |
f1c1300c |
|
d72749e0 |
fullline = rfc822comments(line, NULL); |
c8a7cef0 |
if(fullline == NULL) |
0cf4cea7 |
fullline = cli_strdup(line); |
2ad0c86e |
|
4f4a8f4a |
/*quotes = count_quotes(fullline);*/ |
f1c1300c |
|
c8a7cef0 |
/*
* 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
* "
*/ |
842c7d49 |
while(t_line && next_is_folded_header(t_line)) {
const char *data; |
ede9939c |
size_t datasz; |
842c7d49 |
t_line = t_line->t_next;
data = lineGetData(t_line->t_line); |
f1c1300c |
|
f1d57230 |
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;
}
|
ede9939c |
datasz = strlen(fullline) + strlen(data) + 1; |
1f271616 |
ptr = cli_realloc(fullline, datasz); |
303f9be9 |
|
c8a7cef0 |
if(ptr == NULL)
break; |
1eec55a6 |
|
c8a7cef0 |
fullline = ptr; |
1f271616 |
cli_strlcat(fullline, data, datasz); |
2ad0c86e |
|
4f4a8f4a |
/*quotes = count_quotes(data);*/ |
1eec55a6 |
} |
842c7d49 |
|
c8a7cef0 |
cli_dbgmsg("Multipart %d: About to parse folded header '%s'\n",
multiparts, fullline);
|
c1fce7f7 |
parseEmailHeader(aMessage, fullline, mctx->rfc821Table); |
c8a7cef0 |
free(fullline); |
69c62847 |
} else if(boundaryEnd(line, boundary)) { |
e3aaff8e |
/*
* 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; |
391f7bb3 |
} else if(boundaryStart(line, boundary)) {
inhead = 1;
break; |
56ae62e2 |
} else { |
b2223aad |
if(messageAddLine(aMessage, t_line->t_line) < 0) |
1a74d4df |
break; |
56ae62e2 |
lines++;
} |
8ba634a9 |
} while((t_line = t_line->t_next) != NULL);
|
69c62847 |
cli_dbgmsg("Part %d has %d lines, rc = %d\n", |
8affc406 |
multiparts, lines, (int)rc); |
9bccc1e5 |
/*
* Only save in the array of messages if some
* decision will be taken on whether to scan.
* If all parts will be scanned then save to
* file straight away
*/
switch(subtype) {
case MIXED:
case ALTERNATIVE:
case REPORT:
case DIGEST:
case APPLEDOUBLE:
case KNOWBOT:
case -1: |
69c62847 |
old_rc = rc; |
9bccc1e5 |
mainMessage = do_multipart(mainMessage,
messages, multiparts,
&rc, mctx, messageIn, |
5684fccf |
&aText, recursion_level); |
69c62847 |
if((rc == OK_ATTACHMENTS_NOT_SAVED) && (old_rc == OK))
rc = OK; |
a7a2e2d4 |
if(messages[multiparts]) {
messageDestroy(messages[multiparts]);
messages[multiparts] = NULL;
} |
9bccc1e5 |
--multiparts; |
ecc3d638 |
if(rc == VIRUS) |
9bccc1e5 |
infected = TRUE;
break; |
f7f59c08 |
case RELATED:
case ENCRYPTED:
case SIGNED:
case PARALLEL:
/* all the subtypes that we handle
* (all from the switch(tableFind...) below)
* must be listed here */
break;
default:
/* this is a subtype that we
* don't handle anyway,
* don't store */
if(messages[multiparts]) {
messageDestroy(messages[multiparts]);
messages[multiparts] = NULL;
}
--multiparts; |
9bccc1e5 |
} |
e3aaff8e |
}
free((char *)boundary);
|
7c1eb3bf |
/* |
c79a2273 |
* Preprocess. Anything special to be done before
* we handle the multiparts? |
9a729c80 |
*/ |
9bccc1e5 |
switch(subtype) { |
c79a2273 |
case KNOWBOT:
/* TODO */
cli_dbgmsg("multipart/knowbot parsed as multipart/mixed for now\n");
mimeSubtype = "mixed";
break; |
5d8100cb |
case -1:
/*
* According to section 7.2.6 of |
7cd9337a |
* RFC1521, unrecognized multiparts |
5d8100cb |
* should be treated as multipart/mixed.
*/ |
38d07186 |
cli_dbgmsg("Unsupported multipart format `%s', parsed as mixed\n", mimeSubtype); |
5d8100cb |
mimeSubtype = "mixed";
break; |
c79a2273 |
} |
9a729c80 |
/* |
7c1eb3bf |
* We've finished message we're parsing
*/
if(mainMessage && (mainMessage != messageIn)) {
messageDestroy(mainMessage);
mainMessage = NULL; |
ad9c6836 |
} |
e3aaff8e |
|
9bccc1e5 |
cli_dbgmsg("The message has %d parts\n", multiparts);
|
a585329e |
if(infected || ((multiparts == 0) && (aText == NULL))) { |
b912eaf2 |
if(messages) {
for(i = 0; i < multiparts; i++)
if(messages[i])
messageDestroy(messages[i]); |
b726511f |
free(messages); |
b912eaf2 |
} |
584aadd5 |
if(aText && (textIn == NULL))
textDestroy(aText); |
b912eaf2 |
|
ede9939c |
#if HAVE_JSON
mctx->wrkobj = saveobj;
#endif |
9bccc1e5 |
/* |
ecc3d638 |
* Nothing to do |
9bccc1e5 |
*/ |
69c62847 |
switch(rc) {
case VIRUS: return VIRUS;
case MAXREC: return MAXREC;
default: return OK_ATTACHMENTS_NOT_SAVED;
} |
b726511f |
} |
7c1eb3bf |
|
393a6d67 |
cli_dbgmsg("Find out the multipart type (%s)\n", mimeSubtype); |
e3aaff8e |
|
9a729c80 |
/*
* 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
*/ |
c1fce7f7 |
switch(tableFind(mctx->subtypeTable, mimeSubtype)) { |
e3aaff8e |
case RELATED: |
e06d34dc |
cli_dbgmsg("Multipart related handler\n"); |
e3aaff8e |
/* |
294d0774 |
* Have a look to see if there's HTML code
* which will need scanning |
e3aaff8e |
*/
aMessage = NULL;
assert(multiparts > 0);
|
d4d14218 |
htmltextPart = getTextPart(messages, multiparts); |
e3aaff8e |
|
f7f59c08 |
if(htmltextPart >= 0 && messages) { |
73ddf91f |
if(messageGetBody(messages[htmltextPart])) |
faa0d267 |
|
73ddf91f |
aText = textAddMessage(aText, messages[htmltextPart]);
} else |
e3aaff8e |
/* |
294d0774 |
* There isn't an HTML bit. If there's a
* multipart bit, it'll may be in there
* somewhere |
e3aaff8e |
*/
for(i = 0; i < multiparts; i++)
if(messageGetMimeType(messages[i]) == MULTIPART) {
aMessage = messages[i]; |
d4d14218 |
htmltextPart = i; |
e3aaff8e |
break;
}
|
59da5a4f |
if(htmltextPart == -1) |
16037392 |
cli_dbgmsg("No HTML code found to be scanned\n"); |
59da5a4f |
else { |
c2df9f79 |
#if HAVE_JSON
/* Send root HTML file for preclassification */
if (mctx->ctx->wrkproperty)
parseRootMHTML(mctx, aMessage, aText);
#endif |
242ffd7a |
rc = parseEmailBody(aMessage, aText, mctx, recursion_level + 1); |
99f7771b |
if((rc == OK) && aMessage) { |
59da5a4f |
assert(aMessage == messages[htmltextPart]); |
99f7771b |
messageDestroy(aMessage); |
59da5a4f |
messages[htmltextPart] = NULL; |
94f051b0 |
} else if(rc == VIRUS) { |
a585329e |
infected = TRUE;
break;
} |
59da5a4f |
} |
e3aaff8e |
/*
* 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"
*/ |
ba867aed |
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
*/ |
e3aaff8e |
case ALTERNATIVE:
cli_dbgmsg("Multipart alternative handler\n");
/*
* Fall through - some clients are broken and
* say alternative instead of mixed. The Klez |
5d8100cb |
* virus is broken that way, and anyway we
* wish to scan all of the alternatives |
e3aaff8e |
*/
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
*/ |
6351aa86 |
case ENCRYPTED:
/* MUAs without encryption plugins can display as multipart/mixed,
* just scan it*/ |
e3aaff8e |
case MIXED: |
c9b8f252 |
case APPLEDOUBLE: /* not really supported */ |
e3aaff8e |
/*
* 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
*/ |
ad9c6836 |
if(aText) {
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage); |
e3aaff8e |
mainMessage = NULL; |
ad9c6836 |
} |
e3aaff8e |
cli_dbgmsg("Mixed message with %d parts\n", multiparts);
for(i = 0; i < multiparts; i++) { |
c1fce7f7 |
mainMessage = do_multipart(mainMessage,
messages, i, &rc, mctx, |
242ffd7a |
messageIn, &aText, recursion_level + 1); |
ecc3d638 |
if(rc == VIRUS) { |
a603478f |
infected = TRUE;
break;
} |
69c62847 |
if(rc == MAXREC)
break; |
c89cebe8 |
if (rc == OK_ATTACHMENTS_NOT_SAVED)
rc = OK; |
e3aaff8e |
}
|
242ffd7a |
/* rc = parseEmailBody(NULL, NULL, mctx, recursion_level + 1); */ |
e3aaff8e |
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
*/ |
f7f59c08 |
if(messages) {
htmltextPart = getTextPart(messages, multiparts);
if(htmltextPart == -1)
htmltextPart = 0;
rc = parseEmailBody(messages[htmltextPart], aText, mctx, recursion_level + 1);
} |
e3aaff8e |
break;
default: |
5d8100cb |
assert(0); |
e3aaff8e |
}
|
ad9c6836 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage);
|
6e84cebb |
if(aText && (textIn == NULL)) { |
a603478f |
if((!infected) && (fb = fileblobCreate()) != NULL) { |
c8a7cef0 |
cli_dbgmsg("Save non mime and/or text/plain part\n"); |
c1fce7f7 |
fileblobSetFilename(fb, mctx->dir, "textpart"); |
e097c0dd |
/*fileblobAddData(fb, "Received: by clamd (textpart)\n", 30);*/ |
c1fce7f7 |
fileblobSetCTX(fb, mctx->ctx); |
3f46285b |
(void)textToFileblob(aText, fb, 1); |
6e84cebb |
fileblobDestroy(fb); |
001ad879 |
mctx->files++; |
6e84cebb |
} |
049a18b9 |
textDestroy(aText); |
6e84cebb |
} |
049a18b9 |
|
b116962d |
for(i = 0; i < multiparts; i++)
if(messages[i])
messageDestroy(messages[i]);
|
b726511f |
if(messages)
free(messages);
|
ede9939c |
#if HAVE_JSON
mctx->wrkobj = saveobj;
#endif |
e3aaff8e |
return rc;
case MESSAGE:
/*
* Check for forbidden encodings
*/
switch(messageGetEncoding(mainMessage)) {
case NOENCODING:
case EIGHTBIT:
case BINARY:
break;
default: |
6351aa86 |
cli_dbgmsg("MIME type 'message' cannot be decoded\n"); |
e3aaff8e |
break;
} |
ecc3d638 |
rc = FAIL; |
049a18b9 |
if((strcasecmp(mimeSubtype, "rfc822") == 0) ||
(strcasecmp(mimeSubtype, "delivery-status") == 0)) { |
c1fce7f7 |
message *m = parseEmailHeaders(mainMessage, mctx->rfc821Table); |
c693116d |
if(m) { |
af3c6acb |
cli_dbgmsg("Decode rfc822\n"); |
c693116d |
|
c1fce7f7 |
messageSetCTX(m, mctx->ctx); |
a603478f |
|
93d41ee4 |
if(mainMessage && (mainMessage != messageIn)) {
messageDestroy(mainMessage);
mainMessage = NULL; |
59da5a4f |
} else
messageReset(mainMessage); |
c693116d |
if(messageGetBody(m)) |
242ffd7a |
rc = parseEmailBody(m, NULL, mctx, recursion_level + 1); |
c693116d |
messageDestroy(m);
} |
e3aaff8e |
break; |
5a15955b |
} else if(strcasecmp(mimeSubtype, "disposition-notification") == 0) { |
12f3689d |
/* RFC 2298 - handle like a normal email */ |
ecc3d638 |
rc = OK; |
12f3689d |
break; |
5a15955b |
} else if(strcasecmp(mimeSubtype, "partial") == 0) { |
d7979d4f |
if(mctx->ctx->options->mail & CL_SCAN_MAIL_PARTIAL_MESSAGE) { |
4270f93b |
/* RFC1341 message split over many emails */
if(rfc1341(mainMessage, mctx->dir) >= 0)
rc = OK;
} else {
cli_warnmsg("Partial message received from MUA/MTA - message cannot be scanned\n");
} |
f10460ed |
} else if(strcasecmp(mimeSubtype, "external-body") == 0) |
22080fa5 |
/* TODO */ |
1405207a |
cli_warnmsg("Attempt to send Content-type message/external-body trapped\n"); |
5a642650 |
else |
6fcf5624 |
cli_warnmsg("Unsupported message format `%s' - if you believe this file contains a virus, submit it to www.clamav.net\n", mimeSubtype); |
e3aaff8e |
|
f10460ed |
|
ad9c6836 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage); |
b726511f |
if(messages)
free(messages); |
ede9939c |
#if HAVE_JSON
mctx->wrkobj = saveobj;
#endif |
f10460ed |
return rc; |
e3aaff8e |
|
5b2bcb65 |
default: |
1405207a |
cli_dbgmsg("Message received with unknown mime encoding - assume application\n"); |
5b2bcb65 |
/*
* Some Yahoo emails attach as
* Content-Type: X-unknown/unknown;
* instead of
* Content-Type: application/unknown;
* so let's try our best to salvage something
*/ |
e3aaff8e |
case APPLICATION: |
4ab382c3 |
/*cptr = messageGetMimeSubtype(mainMessage); |
d4d14218 |
|
4ab382c3 |
if((strcasecmp(cptr, "octet-stream") == 0) || |
aa0210b6 |
(strcasecmp(cptr, "x-msdownload") == 0)) {*/
{ |
2673dc74 |
fb = messageToFileblob(mainMessage, mctx->dir, 1); |
e3aaff8e |
|
0e5a0129 |
if(fb) {
cli_dbgmsg("Saving main message as attachment\n"); |
a585329e |
if(fileblobScanAndDestroy(fb) == CL_VIRUS)
rc = VIRUS; |
001ad879 |
mctx->files++; |
fbb7262e |
if(mainMessage != messageIn) {
messageDestroy(mainMessage);
mainMessage = NULL;
} else
messageReset(mainMessage); |
e3aaff8e |
} |
aa0210b6 |
} /*else
cli_warnmsg("Discarded application not sent as attachment\n");*/ |
e3aaff8e |
break;
case AUDIO:
case VIDEO:
case IMAGE:
break;
} |
2673dc74 |
|
b912eaf2 |
if(messages) {
/* "can't happen" */ |
fc809c9e |
cli_warnmsg("messages != NULL\n"); |
2673dc74 |
free(messages); |
b912eaf2 |
} |
e3aaff8e |
}
|
f12d2498 |
if(aText && (textIn == NULL)) { |
6fd711b2 |
/* Look for a bounce in the text (non mime encoded) portion */
const text *t; |
93b4a7b1 |
/* isBounceStart() is expensive, reduce the number of calls */
bool lookahead_definately_is_bounce = FALSE; |
b116962d |
|
a585329e |
for(t = aText; t && (rc != VIRUS); t = t->t_next) { |
6fd711b2 |
const line_t *l = t->t_line; |
ebe57840 |
const text *lookahead, *topofbounce; |
6fd711b2 |
const char *s; |
ebe57840 |
bool inheader; |
b116962d |
|
93b4a7b1 |
if(l == NULL) {
/* assert(lookahead_definately_is_bounce == FALSE) */ |
6fd711b2 |
continue; |
93b4a7b1 |
} |
6fd711b2 |
|
93b4a7b1 |
if(lookahead_definately_is_bounce)
lookahead_definately_is_bounce = FALSE; |
7021b545 |
else if(!isBounceStart(mctx, lineGetData(l))) |
6fd711b2 |
continue;
|
a7a2e2d4 |
lookahead = t->t_next;
if(lookahead) { |
7021b545 |
if(isBounceStart(mctx, lineGetData(lookahead->t_line))) { |
93b4a7b1 |
lookahead_definately_is_bounce = TRUE; |
a7a2e2d4 |
/* don't save worthless header lines */
continue; |
93b4a7b1 |
} |
a7a2e2d4 |
} else /* don't save a single liner */
break;
|
6fd711b2 |
/*
* 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
*/ |
a7a2e2d4 |
for(; lookahead; lookahead = lookahead->t_next) { |
6fd711b2 |
l = lookahead->t_line;
if(l == NULL)
break;
s = lineGetData(l); |
584aadd5 |
if(strncasecmp(s, "Content-Type:", 13) == 0) { |
6fd711b2 |
/* |
584aadd5 |
* Don't bother with text/plain or
* text/html |
6fd711b2 |
*/ |
e906fef3 |
if(cli_strcasestr(s, "text/plain") != NULL) |
6862efc7 |
/* |
584aadd5 |
* Don't bother to save the
* unuseful part, read past
* the headers then we'll go
* on to look for the next
* bounce message |
6862efc7 |
*/ |
584aadd5 |
continue;
if((!doPhishingScan) && |
e906fef3 |
(cli_strcasestr(s, "text/html") != NULL)) |
584aadd5 |
continue;
break;
} |
6fd711b2 |
}
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;
}
|
6862efc7 |
/*
* 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) { |
7c56033f |
cli_dbgmsg("cli_mbox: I believe it's plain text which must be clean\n"); |
6862efc7 |
/* nothing here, move along please */
break;
} |
ebe57840 |
if((fb = fileblobCreate()) == NULL)
break;
cli_dbgmsg("Save non mime part bounce message\n"); |
c1fce7f7 |
fileblobSetFilename(fb, mctx->dir, "bounce"); |
95e11e5a |
fileblobAddData(fb, (const unsigned char *)"Received: by clamd (bounce)\n", 28); |
c1fce7f7 |
fileblobSetCTX(fb, mctx->ctx); |
ebe57840 |
inheader = TRUE;
topofbounce = NULL; |
01c99f53 |
do { |
ebe57840 |
l = t->t_line;
if(l == NULL) {
if(inheader) {
inheader = FALSE;
topofbounce = t;
}
} else {
s = lineGetData(l); |
95e11e5a |
fileblobAddData(fb, (const unsigned char *)s, strlen(s)); |
ebe57840 |
} |
95e11e5a |
fileblobAddData(fb, (const unsigned char *)"\n", 1); |
ebe57840 |
lookahead = t->t_next;
if(lookahead == NULL)
break;
t = lookahead;
l = t->t_line;
if((!inheader) && l) {
s = lineGetData(l); |
7021b545 |
if(isBounceStart(mctx, s)) { |
b2ba24f5 |
cli_dbgmsg("Found the start of another bounce candidate (%s)\n", s); |
be32043e |
lookahead_definately_is_bounce = TRUE; |
ebe57840 |
break;
}
} |
a585329e |
} while(!fileblobInfected(fb));
if(fileblobScanAndDestroy(fb) == CL_VIRUS)
rc = VIRUS; |
001ad879 |
mctx->files++; |
ebe57840 |
if(topofbounce)
t = topofbounce; |
6fd711b2 |
} |
f12d2498 |
textDestroy(aText);
aText = NULL;
}
|
b0b860f1 |
/*
* No attachments - scan the text portions, often files
* are hidden in HTML code
*/ |
ecc3d638 |
if(mainMessage && (rc != VIRUS)) { |
47d9cc65 |
text *t_line;
|
e3aaff8e |
/* |
b0b860f1 |
* Look for uu-encoded main file |
e3aaff8e |
*/ |
f75c04b0 |
if(mainMessage->body_first != NULL &&
(encodingLine(mainMessage) != NULL) &&
((t_line = bounceBegin(mainMessage)) != NULL)) |
001ad879 |
rc = (exportBounceMessage(mctx, t_line) == CL_VIRUS) ? VIRUS : OK; |
7dde984e |
else { |
b0b860f1 |
bool saveIt; |
d4d14218 |
|
b0b860f1 |
if(messageGetMimeType(mainMessage) == MESSAGE) |
f01bbfe8 |
/* |
b0b860f1 |
* Quick peek, if the encapsulated
* message has no
* content encoding statement don't
* bother saving to scan, it's safe |
f01bbfe8 |
*/ |
47d9cc65 |
saveIt = (bool)(encodingLine(mainMessage) != NULL); |
7cf2a701 |
else if(mainMessage->body_last != NULL && (t_line = encodingLine(mainMessage)) != NULL) { |
a7527b1f |
/* |
b0b860f1 |
* Some bounces include the message
* body without the headers. |
ebe57840 |
* FIXME: Unfortunately this generates a |
b0b860f1 |
* lot of false positives that a bounce
* has been found when it hasn't. |
a7527b1f |
*/ |
b0b860f1 |
if((fb = fileblobCreate()) != NULL) { |
ae5c693a |
cli_dbgmsg("Found a bounce message with no header at '%s'\n",
lineGetData(t_line->t_line)); |
c1fce7f7 |
fileblobSetFilename(fb, mctx->dir, "bounce"); |
f24bf390 |
fileblobAddData(fb,
(const unsigned char *)"Received: by clamd (bounce)\n",
28); |
cca4efe4 |
|
7dde984e |
fileblobSetCTX(fb, mctx->ctx);
if(fileblobScanAndDestroy(textToFileblob(t_line, fb, 1)) == CL_VIRUS)
rc = VIRUS; |
001ad879 |
mctx->files++; |
5c1150ac |
} |
b0b860f1 |
saveIt = FALSE; |
2673dc74 |
} else |
b0b860f1 |
/*
* Save the entire text portion,
* since it it may be an HTML file with |
a05e6d45 |
* a JavaScript virus or a phish |
b0b860f1 |
*/
saveIt = TRUE; |
e3aaff8e |
|
b0b860f1 |
if(saveIt) { |
7dde984e |
cli_dbgmsg("Saving text part to scan, rc = %d\n", |
8affc406 |
(int)rc); |
001ad879 |
if(saveTextPart(mctx, mainMessage, 1) == CL_VIRUS) |
7dde984e |
rc = VIRUS;
|
59da5a4f |
if(mainMessage != messageIn) {
messageDestroy(mainMessage);
mainMessage = NULL;
} else
messageReset(mainMessage); |
049a18b9 |
} |
e3aaff8e |
} |
69c62847 |
} /*else |
96435bdc |
rc = OK_ATTACHMENTS_NOT_SAVED; */ /* nothing saved */ |
e3aaff8e |
|
ad9c6836 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage);
|
ecc3d638 |
if((rc != FAIL) && infected)
rc = VIRUS; |
a603478f |
|
ede9939c |
#if HAVE_JSON
mctx->wrkobj = saveobj;
#endif
|
8affc406 |
cli_dbgmsg("parseEmailBody() returning %d\n", (int)rc); |
e3aaff8e |
|
e06d34dc |
return rc; |
e3aaff8e |
}
/*
* Is the current line the start of a new section?
*
* New sections start with --boundary
*/
static int
boundaryStart(const char *line, const char *boundary)
{ |
95e11e5a |
const char *ptr;
char *out; |
df8806fd |
int rc; |
85bb253e |
char buf[RFC2821LENGTH + 1]; |
23889d27 |
char *newline; |
2ad0c86e |
|
6b579c8d |
if(line == NULL || *line == '\0') |
0e4e16d4 |
return 0; /* empty line */ |
300a8ae9 |
if(boundary == NULL)
return 0; |
0e4e16d4 |
|
23889d27 |
newline = strdup(line);
if (!(newline)) |
cd94be7a |
newline = (char *)line; |
23889d27 |
|
6b579c8d |
if (newline != line && strlen(line)) { |
23889d27 |
char *p;
/* Trim trailing spaces */ |
77326673 |
p = newline + strlen(line) - 1; |
26f42733 |
while (p >= newline && *p == ' ') |
23889d27 |
*(p--) = '\0';
}
if (newline != line)
cli_chomp(newline); |
2ad0c86e |
|
d00281e1 |
/* cli_dbgmsg("boundaryStart: line = '%s' boundary = '%s'\n", line, boundary); */ |
23889d27 |
if((*newline != '-') && (*newline != '(')) {
if (newline != line)
free(newline); |
ebe57840 |
return 0; |
23889d27 |
} |
ebe57840 |
|
23889d27 |
if(strchr(newline, '-') == NULL) {
if (newline != line)
free(newline); |
ebe57840 |
return 0; |
23889d27 |
} |
ebe57840 |
|
23889d27 |
if(strlen(newline) <= sizeof(buf)) { |
d72749e0 |
out = NULL; |
23889d27 |
ptr = rfc822comments(newline, buf); |
d72749e0 |
} else |
23889d27 |
ptr = out = rfc822comments(newline, NULL); |
d72749e0 |
|
2ad0c86e |
if(ptr == NULL) |
23889d27 |
ptr = newline; |
2ad0c86e |
|
a9d251e0 |
if((*ptr++ != '-') || (*ptr == '\0')) { |
d72749e0 |
if(out)
free(out); |
23889d27 |
if (newline != line)
free(newline);
|
0e4e16d4 |
return 0; |
2ad0c86e |
} |
0e4e16d4 |
|
e3aaff8e |
/* |
0e4e16d4 |
* Gibe.B3 is broken, it has: |
e3aaff8e |
* boundary="---- =_NextPart_000_01C31177.9DC7C000"
* but it's boundaries look like
* ------ =_NextPart_000_01C31177.9DC7C000 |
0e4e16d4 |
* 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. |
7c56033f |
*
* Look with and without RFC822 comments stripped, I've seen some
* samples where () are taken as comments in boundaries and some where |
69c62847 |
* they're not. Irrespective of whatever RFC2822 says, we need to find
* viruses in both types of mails. |
e3aaff8e |
*/ |
23889d27 |
if((strstr(&ptr[1], boundary) != NULL) || (strstr(newline, boundary) != NULL)) { |
69c62847 |
const char *k = ptr;
/*
* We need to ensure that we don't match --11=-=-=11 when
* looking for --1=-=-=1 in well behaved headers, that's a
* false positive problem mentioned above
*/
rc = 0;
do
if(strcmp(++k, boundary) == 0) {
rc = 1;
break;
}
while(*k == '-');
if(rc == 0) {
k = &line[1];
do
if(strcmp(++k, boundary) == 0) {
rc = 1;
break;
}
while(*k == '-');
}
} else if(*ptr++ != '-')
rc = 0; |
df8806fd |
else |
79179da5 |
rc = (strcasecmp(ptr, boundary) == 0); |
df8806fd |
|
d72749e0 |
if(out)
free(out); |
df8806fd |
|
69c62847 |
if(rc == 1) |
df8806fd |
cli_dbgmsg("boundaryStart: found %s in %s\n", boundary, line);
|
23889d27 |
if (newline != line)
free(newline);
|
df8806fd |
return rc; |
e3aaff8e |
}
/*
* Is the current line the end?
*
* The message ends with with --boundary--
*/
static int |
69c62847 |
boundaryEnd(const char *line, const char *boundary) |
e3aaff8e |
{
size_t len; |
23889d27 |
char *newline, *p, *p2; |
e3aaff8e |
|
6b579c8d |
if(line == NULL || *line == '\0') |
02927896 |
return 0; |
69c62847 |
|
23889d27 |
p = newline = strdup(line); |
d00281e1 |
if (!(newline)) { |
cd94be7a |
p = (char *)line;
newline = (char *)line; |
d00281e1 |
} |
23889d27 |
|
6b579c8d |
if (newline != line && strlen(line)) { |
23889d27 |
/* Trim trailing spaces */ |
77326673 |
p2 = newline + strlen(line) - 1; |
4a5d422d |
while (p2 >= newline && *p2 == ' ') |
23889d27 |
*(p2--) = '\0';
}
|
d00281e1 |
/* cli_dbgmsg("boundaryEnd: line = '%s' boundary = '%s'\n", newline, boundary); */ |
69c62847 |
|
95890079 |
if(*p++ != '-') {
if (newline != line)
free(newline); |
e3aaff8e |
return 0; |
95890079 |
}
if(*p++ != '-') {
if (newline != line)
free(newline);
|
e3aaff8e |
return 0; |
95890079 |
}
|
e3aaff8e |
len = strlen(boundary); |
23889d27 |
if(strncasecmp(p, boundary, len) != 0) {
if (newline != line)
free(newline);
|
049a18b9 |
return 0; |
23889d27 |
} |
0e523db2 |
/*
* Use < rather than == because some broken mails have white
* space after the boundary
*/ |
d00281e1 |
if(strlen(p) < (len + 2)) {
if (newline != line)
free(newline);
|
e3aaff8e |
return 0; |
d00281e1 |
} |
95890079 |
|
23889d27 |
p = &p[len];
if(*p++ != '-') {
if (newline != line)
free(newline);
|
e3aaff8e |
return 0; |
23889d27 |
} |
95890079 |
|
23889d27 |
if(*p == '-') { |
d00281e1 |
/* cli_dbgmsg("boundaryEnd: found %s in %s\n", boundary, p); */ |
23889d27 |
if (newline != line)
free(newline); |
d00281e1 |
|
69c62847 |
return 1;
} |
745cda08 |
if (newline != line)
free(newline);
|
69c62847 |
return 0; |
e3aaff8e |
}
/*
* 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++) |
51fc2aa8 |
if(tableInsert(*rfc821Table, tableinit->key, tableinit->value) < 0) {
tableDestroy(*rfc821Table); |
767f16ab |
*rfc821Table = NULL; |
e3aaff8e |
return -1; |
51fc2aa8 |
} |
e3aaff8e |
*subtypeTable = tableCreate();
assert(*subtypeTable != NULL);
for(tableinit = mimeSubtypes; tableinit->key; tableinit++)
if(tableInsert(*subtypeTable, tableinit->key, tableinit->value) < 0) {
tableDestroy(*rfc821Table); |
51fc2aa8 |
tableDestroy(*subtypeTable); |
767f16ab |
*rfc821Table = NULL;
*subtypeTable = NULL; |
e3aaff8e |
return -1;
}
return 0;
}
/* |
d4d14218 |
* If there's a HTML text version use that, otherwise |
e3aaff8e |
* use the first text part, otherwise just use the |
d4d14218 |
* first one around. HTML text is most likely to include
* a scripting worm |
e3aaff8e |
*
* If we can't find one, return -1
*/
static int
getTextPart(message *const messages[], size_t size)
{
size_t i; |
28010d29 |
int textpart = -1; |
e3aaff8e |
|
73ddf91f |
for(i = 0; i < size; i++)
if(messages[i] && (messageGetMimeType(messages[i]) == TEXT)) { |
28010d29 |
if(strcasecmp(messageGetMimeSubtype(messages[i]), "html") == 0)
return (int)i;
textpart = (int)i;
} |
73ddf91f |
|
28010d29 |
return textpart; |
e3aaff8e |
}
/*
* strip - |
767f16ab |
* 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
* |
e3aaff8e |
* 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)) |
767f16ab |
return 0; |
e3aaff8e |
i = strlen(buf);
if(len > (int)(i + 1)) |
767f16ab |
return i; |
e3aaff8e |
ptr = &buf[--len];
|
ecc3d638 |
#if defined(UNIX) || defined(C_LINUX) || defined(C_DARWIN) /* watch - it may be in shared text area */ |
e3aaff8e |
do
if(*ptr)
*ptr = '\0'; |
87c9313e |
while((--len >= 0) && (!isgraph(*--ptr)) && (*ptr != '\n') && (*ptr != '\r')); |
e3aaff8e |
#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'; |
ced371fe |
while((--len >= 0) && ((*--ptr == '\0') || isspace((int)(*ptr & 0xFF)))); |
e3aaff8e |
#endif
return((size_t)(len + 1));
}
/*
* strstrip:
* Strip a given string
*/ |
f0627588 |
size_t |
e3aaff8e |
strstrip(char *s)
{
if(s == (char *)NULL)
return(0); |
02927896 |
|
5cd3f734 |
return(strip(s, strlen(s) + 1)); |
e3aaff8e |
}
|
30e18caf |
/*
* Returns 0 for OK, -1 for error
*/ |
e3aaff8e |
static int
parseMimeHeader(message *m, const char *cmd, const table_t *rfc821Table, const char *arg)
{ |
30e18caf |
char *copy, *p, *buf; |
38d07186 |
const char *ptr; |
8b3563f2 |
int commandNumber; |
4f1d0bfc |
|
e3aaff8e |
cli_dbgmsg("parseMimeHeader: cmd='%s', arg='%s'\n", cmd, arg); |
8b3563f2 |
|
d72749e0 |
copy = rfc822comments(cmd, NULL); |
38d07186 |
if(copy) {
commandNumber = tableFind(rfc821Table, copy);
free(copy); |
f017fbdd |
} else
commandNumber = tableFind(rfc821Table, cmd); |
8b3563f2 |
|
d72749e0 |
copy = rfc822comments(arg, NULL); |
e3aaff8e |
|
38d07186 |
if(copy)
ptr = copy;
else
ptr = arg; |
e75e1ad1 |
|
30e18caf |
buf = NULL;
|
8b3563f2 |
switch(commandNumber) { |
e3aaff8e |
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:"
*/ |
5e394e73 |
if(arg == NULL) |
69543a9d |
/*
* According to section 4 of RFC1521:
* "Note also that a subtype specification is
* MANDATORY. There are no default subtypes"
* |
1eec55a6 |
* We have to break this and make an assumption |
69543a9d |
* for the subtype because virus writers and
* email client writers don't get it right
*/ |
6351aa86 |
cli_dbgmsg("Empty content-type received, no subtype specified, assuming text/plain; charset=us-ascii\n"); |
38d07186 |
else if(strchr(ptr, '/') == NULL) |
69543a9d |
/*
* Empty field, such as
* Content-Type:
* which I believe is illegal according to
* RFC1521
*/ |
38d07186 |
cli_dbgmsg("Invalid content-type '%s' received, no subtype specified, assuming text/plain; charset=us-ascii\n", ptr); |
e3aaff8e |
else { |
8037334b |
int i; |
6d312569 |
|
30e18caf |
buf = cli_malloc(strlen(ptr) + 1);
if(buf == NULL) { |
059ca614 |
cli_errmsg("parseMimeHeader: Unable to allocate memory for buf %llu\n", (long long unsigned)(strlen(ptr) + 1)); |
30e18caf |
if(copy)
free(copy);
return -1;
} |
e3aaff8e |
/*
* Some clients are broken and
* put white space after the ;
*/ |
2625d6a0 |
if(*arg == '/') { |
6351aa86 |
cli_dbgmsg("Content-type '/' received, assuming application/octet-stream\n"); |
2625d6a0 |
messageSetMimeType(m, "application");
messageSetMimeSubtype(m, "octet-stream");
} else {
/* |
0960ff5e |
* 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 |
2625d6a0 |
*/ |
38d07186 |
while(isspace(*ptr))
ptr++;
if(ptr[0] == '\"')
ptr++; |
0960ff5e |
|
38d07186 |
if(ptr[0] != '/') { |
f017fbdd |
char *s; |
38d07186 |
#ifdef CL_THREAD_SAFE |
1adb6fa8 |
char *strptr = NULL; |
38d07186 |
#endif |
f017fbdd |
|
30e18caf |
s = cli_strtokbuf(ptr, 0, ";", buf); |
0960ff5e |
/* |
6d312569 |
* Handle
* Content-Type: foo/bar multipart/mixed
* and
* Content-Type: multipart/mixed foo/bar |
0960ff5e |
*/ |
584aadd5 |
if(s && *s) {
char *buf2 = cli_strdup(buf);
if(buf2 == NULL) {
if(copy)
free(copy);
free(buf);
return -1;
}
for(;;) { |
548a5f96 |
#ifdef CL_THREAD_SAFE |
584aadd5 |
int set = messageSetMimeType(m, strtok_r(s, "/", &strptr)); |
548a5f96 |
#else |
584aadd5 |
int set = messageSetMimeType(m, strtok(s, "/")); |
548a5f96 |
#endif |
6d312569 |
|
548a5f96 |
#ifdef CL_THREAD_SAFE |
584aadd5 |
s = strtok_r(NULL, ";", &strptr); |
548a5f96 |
#else |
584aadd5 |
s = strtok(NULL, ";");
#endif
if(s == NULL)
break;
if(set) {
size_t len = strstrip(s) - 1;
if(s[len] == '\"') {
s[len] = '\0';
len = strstrip(s);
}
if(len) {
if(strchr(s, ' '))
messageSetMimeSubtype(m,
cli_strtokbuf(s, 0, " ", buf2));
else
messageSetMimeSubtype(m, s);
} |
6d312569 |
}
|
584aadd5 |
while(*s && !isspace(*s))
s++;
if(*s++ == '\0')
break;
if(*s == '\0')
break;
}
free(buf2); |
0960ff5e |
} |
f017fbdd |
} |
2625d6a0 |
} |
e3aaff8e |
/* |
20d3dde9 |
* 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 |
e3aaff8e |
*/ |
8037334b |
i = 1; |
94f051b0 |
while(cli_strtokbuf(ptr, i++, ";", buf) != NULL) {
cli_dbgmsg("mimeArgs = '%s'\n", buf); |
8037334b |
|
94f051b0 |
messageAddArguments(m, buf); |
6d312569 |
} |
e3aaff8e |
}
break;
case CONTENT_TRANSFER_ENCODING: |
38d07186 |
messageSetEncoding(m, ptr); |
e3aaff8e |
break;
case CONTENT_DISPOSITION: |
30e18caf |
buf = cli_malloc(strlen(ptr) + 1);
if(buf == NULL) { |
059ca614 |
cli_errmsg("parseMimeHeader: Unable to allocate memory for buf %llu\n", (long long unsigned)(strlen(ptr) + 1)); |
30e18caf |
if(copy)
free(copy);
return -1;
}
p = cli_strtokbuf(ptr, 0, ";", buf); |
faa0d267 |
if(p && *p) {
messageSetDispositionType(m, p);
messageAddArgument(m, cli_strtokbuf(ptr, 1, ";", buf)); |
548a5f96 |
} |
be32043e |
if(!messageHasFilename(m)) |
ab84808e |
/*
* 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"); |
e3aaff8e |
} |
38d07186 |
if(copy)
free(copy); |
30e18caf |
if(buf)
free(buf); |
e3aaff8e |
|
4f1d0bfc |
return 0; |
e3aaff8e |
}
|
e06d34dc |
/* |
cca4efe4 |
* Save the text portion of the message
*/ |
7dde984e |
static int |
001ad879 |
saveTextPart(mbox_ctx *mctx, message *m, int destroy_text) |
cca4efe4 |
{ |
0e5a0129 |
fileblob *fb; |
cca4efe4 |
messageAddArgument(m, "filename=textportion"); |
001ad879 |
if((fb = messageToFileblob(m, mctx->dir, destroy_text)) != NULL) { |
cca4efe4 |
/*
* Save main part to scan that
*/ |
c77c8809 |
cli_dbgmsg("Saving main message\n"); |
cca4efe4 |
|
001ad879 |
mctx->files++; |
7dde984e |
return fileblobScanAndDestroy(fb); |
cca4efe4 |
} |
7dde984e |
return CL_ETMPFILE; |
cca4efe4 |
}
|
a9714c49 |
/* |
f017fbdd |
* Handle RFC822 comments in headers. |
d72749e0 |
* 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. |
7cd9337a |
* See section 3.4.3 of RFC822 |
a9714c49 |
* TODO: handle comments that go on to more than one line
*/
static char * |
d72749e0 |
rfc822comments(const char *in, char *out) |
a9714c49 |
{
const char *iptr; |
d72749e0 |
char *optr; |
a9714c49 |
int backslash, inquote, commentlevel;
if(in == NULL) |
f017fbdd |
return NULL; |
a9714c49 |
if(strchr(in, '(') == NULL) |
f017fbdd |
return NULL; |
a9714c49 |
|
d72749e0 |
assert(out != in);
|
faa0d267 |
while(isspace(*in))
in++;
|
d72749e0 |
if(out == NULL) {
out = cli_malloc(strlen(in) + 1); |
241e7eb1 |
if(out == NULL) { |
059ca614 |
cli_errmsg("rfc822comments: Unable to allocate memory for out %llu\n", (long long unsigned)(strlen(in) + 1)); |
d72749e0 |
return NULL; |
241e7eb1 |
} |
d72749e0 |
} |
a9714c49 |
backslash = commentlevel = inquote = 0;
optr = out;
cli_dbgmsg("rfc822comments: contains a comment\n");
for(iptr = in; *iptr; iptr++)
if(backslash) { |
b814e02a |
if(commentlevel == 0)
*optr++ = *iptr; |
a9714c49 |
backslash = 0;
} else switch(*iptr) {
case '\\':
backslash = 1;
break;
case '\"': |
7c56033f |
*optr++ = '\"'; |
a9714c49 |
inquote = !inquote;
break;
case '(': |
7c56033f |
if(inquote)
*optr++ = '(';
else
commentlevel++; |
a9714c49 |
break;
case ')': |
7c56033f |
if(inquote)
*optr++ = ')';
else if(commentlevel > 0) |
a9714c49 |
commentlevel--;
break;
default:
if(commentlevel == 0)
*optr++ = *iptr;
}
if(backslash) /* last character was a single backslash */
*optr++ = '\\';
*optr = '\0';
|
d72749e0 |
/*strstrip(out);*/ |
a9714c49 |
cli_dbgmsg("rfc822comments '%s'=>'%s'\n", in, out);
return out;
} |
50df4118 |
/*
* 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;
|
4de5fffd |
if((strstr(in, "=?") == NULL) || (strstr(in, "?=") == NULL)) |
0cf4cea7 |
return cli_strdup(in); |
50df4118 |
cli_dbgmsg("rfc2047 '%s'\n", in);
out = cli_malloc(strlen(in) + 1);
|
241e7eb1 |
if(out == NULL) { |
059ca614 |
cli_errmsg("rfc2047: Unable to allocate memory for out %llu\n", (long long unsigned)(strlen(in) + 1)); |
50df4118 |
return NULL; |
241e7eb1 |
} |
50df4118 |
pout = out;
/* For each RFC2047 string */
while(*in) { |
cf569541 |
char encoding, *ptr, *enctext; |
50df4118 |
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; |
3471518d |
encoding = (char)tolower(encoding); |
50df4118 |
if((encoding != 'q') && (encoding != 'b')) { |
6fcf5624 |
cli_warnmsg("Unsupported RFC2047 encoding type '%c' - if you believe this file contains a virus, submit it to www.clamav.net\n", encoding); |
1b00d9a4 |
free(out);
out = NULL; |
50df4118 |
break;
}
/* Skip to encoded text */
if(*++in != '?')
break;
if(*++in == '\0')
break;
|
0cf4cea7 |
enctext = cli_strdup(in); |
cf569541 |
if(enctext == NULL) {
free(out);
out = NULL;
break;
} |
50df4118 |
in = strstr(in, "?="); |
cf569541 |
if(in == NULL) {
free(enctext); |
50df4118 |
break; |
cf569541 |
} |
50df4118 |
in += 2;
ptr = strstr(enctext, "?=");
assert(ptr != NULL);
*ptr = '\0';
/*cli_dbgmsg("Need to decode '%s' with method '%c'\n", enctext, encoding);*/
m = messageCreate(); |
c77c8809 |
if(m == NULL) |
50df4118 |
break; |
564b3e07 |
messageAddStr(m, enctext); |
cf569541 |
free(enctext); |
767f16ab |
switch(encoding) { |
50df4118 |
case 'q':
messageSetEncoding(m, "quoted-printable");
break;
case 'b':
messageSetEncoding(m, "base64");
break;
} |
2673dc74 |
b = messageToBlob(m, 1); |
39c89d14 |
if (b == NULL) {
messageDestroy(m);
break;
} |
50df4118 |
len = blobGetDataSize(b); |
95e11e5a |
cli_dbgmsg("Decoded as '%*.*s'\n", (int)len, (int)len, |
8affc406 |
(const char *)blobGetData(b)); |
50df4118 |
memcpy(pout, blobGetData(b), len);
blobDestroy(b);
messageDestroy(m); |
d1100be3 |
if(len > 0 && pout[len - 1] == '\n') |
50df4118 |
pout += len - 1;
else
pout += len;
} |
ec8e31fa |
if(out == NULL)
return NULL;
*pout = '\0'; |
50df4118 |
|
ec8e31fa |
cli_dbgmsg("rfc2047 returns '%s'\n", out); |
50df4118 |
return out;
}
|
f10460ed |
/*
* Handle partial messages
*/
static int
rfc1341(message *m, const char *dir)
{ |
13462674 |
char *arg, *id, *number, *total, *oldfilename; |
fb79b576 |
const char *tmpdir; |
4270f93b |
int n; |
16ea58a2 |
char pdir[NAME_MAX + 1]; |
4270f93b |
unsigned char md5_val[16];
char *md5_hex; |
fb79b576 |
|
f003b79e |
id = (char *)messageFindArgument(m, "id");
if(id == NULL)
return -1;
|
081f6473 |
tmpdir = cli_gettmpdir(); |
f10460ed |
|
58481352 |
snprintf(pdir, sizeof(pdir) - 1, "%s"PATHSEP"clamav-partial", tmpdir); |
fb79b576 |
|
ba74b333 |
if((mkdir(pdir, S_IRWXU) < 0) && (errno != EEXIST)) { |
fb79b576 |
cli_errmsg("Can't create the directory '%s'\n", pdir); |
ba74b333 |
free(id); |
f10460ed |
return -1; |
7dde984e |
} else if(errno == EEXIST) { |
a2a004df |
STATBUF statb; |
13462674 |
|
d9b6b8c7 |
if(CLAMSTAT(pdir, &statb) < 0) { |
e68d70e7 |
char err[128]; |
ba74b333 |
cli_errmsg("Partial directory %s: %s\n", pdir, |
e68d70e7 |
cli_strerror(errno, err, sizeof(err))); |
ba74b333 |
free(id); |
13462674 |
return -1;
} |
1cb424e6 |
if(statb.st_mode & 077) |
13462674 |
cli_warnmsg("Insecure partial directory %s (mode 0%o)\n", |
ba74b333 |
pdir,
#ifdef ACCESSPERMS
(int)(statb.st_mode&ACCESSPERMS)
#else
(int)(statb.st_mode & 0777)
#endif
); |
f10460ed |
}
number = (char *)messageFindArgument(m, "number");
if(number == NULL) {
free(id);
return -1;
}
|
ba74b333 |
oldfilename = messageGetFilename(m); |
f10460ed |
arg = cli_malloc(10 + strlen(id) + strlen(number)); |
a43dd3cd |
if(arg) {
sprintf(arg, "filename=%s%s", id, number);
messageAddArgument(m, arg);
free(arg);
} |
f10460ed |
if(oldfilename) { |
6351aa86 |
cli_dbgmsg("Must reset to %s\n", oldfilename); |
f10460ed |
free(oldfilename);
}
|
4270f93b |
n = atoi(number); |
7fb5036f |
cl_hash_data("md5", id, strlen(id), md5_val, NULL); |
4270f93b |
md5_hex = cli_str2hex((const char*)md5_val, 16);
if(!md5_hex) { |
f10460ed |
free(id);
free(number); |
4270f93b |
return CL_EMEM; |
f10460ed |
}
|
4270f93b |
if(messageSavePartial(m, pdir, md5_hex, n) < 0) {
free(md5_hex);
free(id);
free(number);
return -1;
} |
f10460ed |
total = (char *)messageFindArgument(m, "total");
cli_dbgmsg("rfc1341: %s, %s of %s\n", id, number, (total) ? total : "?");
if(total) {
int t = atoi(total);
DIR *dd = NULL;
|
f003b79e |
free(total); |
f10460ed |
/*
* If it's the last one - reassemble it |
3a0f75c6 |
* FIXME: this assumes that we receive the parts in order |
f10460ed |
*/ |
fb79b576 |
if((n == t) && ((dd = opendir(pdir)) != NULL)) { |
f10460ed |
FILE *fout;
char outname[NAME_MAX + 1]; |
e6be10f7 |
time_t now; |
f10460ed |
|
11b50569 |
sanitiseName(id); |
95e11e5a |
|
58481352 |
snprintf(outname, sizeof(outname) - 1, "%s"PATHSEP"%s", dir, id); |
f10460ed |
cli_dbgmsg("outname: %s\n", outname);
fout = fopen(outname, "wb");
if(fout == NULL) { |
3a0f75c6 |
cli_errmsg("Can't open '%s' for writing", outname); |
f10460ed |
free(id);
free(number); |
4270f93b |
free(md5_hex); |
f10460ed |
closedir(dd);
return -1;
}
|
e6be10f7 |
time(&now); |
f10460ed |
for(n = 1; n <= t; n++) {
char filename[NAME_MAX + 1]; |
43f1a140 |
struct dirent *dent; |
13462674 |
#if defined(HAVE_READDIR_R_3) || defined(HAVE_READDIR_R_2) |
242bfde8 |
union {
struct dirent d;
char b[offsetof(struct dirent, d_name) + NAME_MAX + 1];
} result; |
13462674 |
#endif |
f10460ed |
|
4270f93b |
snprintf(filename, sizeof(filename), "_%s-%u", md5_hex, n); |
fdc6adbe |
|
242bfde8 |
#ifdef HAVE_READDIR_R_3
while((readdir_r(dd, &result.d, &dent) == 0) && dent) { |
f10460ed |
#elif defined(HAVE_READDIR_R_2) |
242bfde8 |
while((dent = (struct dirent *)readdir_r(dd, &result.d))) { |
fdc6adbe |
#else /*!HAVE_READDIR_R*/ |
f10460ed |
while((dent = readdir(dd))) {
#endif
FILE *fin; |
e6be10f7 |
char buffer[BUFSIZ], fullname[NAME_MAX + 1]; |
f10460ed |
int nblanks; |
a2a004df |
STATBUF statb; |
4270f93b |
const char *dentry_idpart; |
b2c04b6c |
int test_fd; |
be4bf7f4 |
|
970ab7d0 |
if(dent->d_ino == 0)
continue; |
f10460ed |
|
1cb424e6 |
if(!strcmp(".",dent->d_name) ||
!strcmp("..", dent->d_name)) |
f10460ed |
continue; |
a43dd3cd |
snprintf(fullname, sizeof(fullname) - 1, |
58481352 |
"%s"PATHSEP"%s", pdir, dent->d_name); |
4270f93b |
dentry_idpart = strchr(dent->d_name, '_'); |
a43dd3cd |
|
4270f93b |
if(!dentry_idpart ||
strcmp(filename, dentry_idpart) != 0) { |
33068e09 |
if(!m->ctx->engine->keeptmp) |
e6be10f7 |
continue; |
b2c04b6c |
if ((test_fd = open(fullname, O_RDONLY)) < 0)
continue;
|
a2a004df |
if(FSTAT(test_fd, &statb) < 0) { |
b2c04b6c |
close(test_fd); |
e6be10f7 |
continue; |
b2c04b6c |
}
|
997a0e0b |
if(now - statb.st_mtime > (time_t)(7 * 24 * 3600)) {
if (cli_unlink(fullname)) {
cli_unlink(outname);
fclose(fout); |
4270f93b |
free(md5_hex); |
997a0e0b |
free(id);
free(number);
closedir(dd); |
b2c04b6c |
close(test_fd); |
997a0e0b |
return -1;
}
} |
b2c04b6c |
close(test_fd); |
f10460ed |
continue; |
e6be10f7 |
} |
f10460ed |
|
e6be10f7 |
fin = fopen(fullname, "rb"); |
f10460ed |
if(fin == NULL) { |
e6be10f7 |
cli_errmsg("Can't open '%s' for reading", fullname); |
f10460ed |
fclose(fout); |
c0a95e0c |
cli_unlink(outname); |
bebd86a6 |
free(md5_hex); |
f10460ed |
free(id);
free(number);
closedir(dd);
return -1;
}
nblanks = 0; |
9f2024cc |
while(fgets(buffer, sizeof(buffer) - 1, fin) != NULL) |
f10460ed |
/*
* Ensure that trailing newlines
* aren't copied
*/ |
f003b79e |
if(buffer[0] == '\n') |
f10460ed |
nblanks++; |
f003b79e |
else { |
f10460ed |
if(nblanks) |
6c06c7b3 |
do {
if (putc('\n', fout)==EOF) break;
} while(--nblanks > 0);
if (nblanks || fputs(buffer, fout)==EOF) {
fclose(fin);
fclose(fout);
cli_unlink(outname); |
4270f93b |
free(md5_hex); |
6c06c7b3 |
free(id);
free(number);
closedir(dd);
return -1;
} |
f10460ed |
}
fclose(fin); |
fb79b576 |
|
997a0e0b |
/* don't unlink if leave temps */ |
33068e09 |
if(!m->ctx->engine->keeptmp) { |
997a0e0b |
if(cli_unlink(fullname)) {
fclose(fout);
cli_unlink(outname); |
4270f93b |
free(md5_hex); |
997a0e0b |
free(id);
free(number);
closedir(dd);
return -1;
}
} |
f10460ed |
break;
}
rewinddir(dd);
}
closedir(dd);
fclose(fout);
}
} |
f003b79e |
free(number); |
f10460ed |
free(id); |
4270f93b |
free(md5_hex); |
f10460ed |
return 0;
}
|
c52d991e |
static void
hrefs_done(blob *b, tag_arguments_t *hrefs)
{
if(b)
blobDestroy(b);
html_tag_arg_free(hrefs);
}
|
f2b71eb9 |
/* extract URLs from static text */
static void extract_text_urls(const unsigned char *mem, size_t len, tag_arguments_t *hrefs)
{
char url[1024];
size_t off;
for (off=0;off + 10 < len;off++) {
/* check whether this is the start of a URL */
int32_t proto = cli_readint32(mem + off);
/* convert to lowercase */
proto |= 0x20202020;
/* 'http:', 'https:', or 'ftp:' in little-endian */
if ((proto == 0x70747468 &&
(mem[off+4] == ':' || (mem[off+5] == 's' && mem[off+6] == ':')))
|| proto == 0x3a707466) {
size_t url_len;
for (url_len=4; off + url_len < len && url_len < (sizeof(url)-1); url_len++) {
unsigned char c = mem[off + url_len];
/* smart compilers will compile this if into
* a single bt + jb instruction */
if (c == ' ' || c == '\n' || c == '\t')
break;
}
memcpy(url, mem + off, url_len);
url[url_len] = '\0';
html_tag_arg_add(hrefs, "href", url);
off += url_len;
}
}
}
|
c52d991e |
/*
* This used to be part of checkURLs, split out, because phishingScan needs it
* too, and phishingScan might be used in situations where checkURLs is
* disabled (see ifdef)
*/
static blob *
getHrefs(message *m, tag_arguments_t *hrefs)
{ |
f2b71eb9 |
unsigned char *mem; |
8386c723 |
blob *b = messageToBlob(m, 0); |
c52d991e |
size_t len;
if(b == NULL)
return NULL;
len = blobGetDataSize(b);
if(len == 0) {
blobDestroy(b);
return NULL;
}
/* TODO: make this size customisable */
if(len > 100*1024) { |
e25960d2 |
cli_dbgmsg("Viruses pointed to by URLs not scanned in large message\n"); |
c52d991e |
blobDestroy(b);
return NULL;
}
hrefs->count = 0;
hrefs->tag = hrefs->value = NULL;
hrefs->contents = NULL;
|
8386c723 |
cli_dbgmsg("getHrefs: calling html_normalise_mem\n"); |
f2b71eb9 |
mem = blobGetData(b);
if(!html_normalise_mem(mem, (off_t)len, NULL, hrefs,m->ctx->dconf)) { |
c52d991e |
blobDestroy(b);
return NULL;
} |
8386c723 |
cli_dbgmsg("getHrefs: html_normalise_mem returned\n"); |
f2b71eb9 |
if (!hrefs->count && hrefs->scanContents) {
extract_text_urls(mem, len, hrefs);
} |
c52d991e |
/* TODO: Do we need to call remove_html_comments? */
return b;
}
|
ad422cc9 |
/* |
7dde984e |
* validate URLs for phishes |
ad422cc9 |
* followurls: see if URLs point to malware
*/ |
c52d991e |
static void |
ecc3d638 |
checkURLs(message *mainMessage, mbox_ctx *mctx, mbox_status *rc, int is_html) |
c52d991e |
{ |
94aea271 |
blob *b; |
ad422cc9 |
tag_arguments_t hrefs; |
c52d991e |
|
cd94be7a |
UNUSEDPARAM(is_html);
|
7dde984e |
if(*rc == VIRUS)
return;
|
d77ac7de |
hrefs.scanContents = mctx->ctx->engine->dboptions&CL_DB_PHISHING_URLS && (DCONF_PHISHING & PHISHING_CONF_ENGINE); |
c52d991e |
|
94aea271 |
if(!hrefs.scanContents) |
093e013c |
/*
* Don't waste time extracting hrefs (parsing html), nobody
* will need it
*/ |
c52d991e |
return;
|
94aea271 |
hrefs.count = 0;
hrefs.tag = hrefs.value = NULL;
hrefs.contents = NULL;
b = getHrefs(mainMessage, &hrefs);
if(b) { |
d77ac7de |
if(hrefs.scanContents) { |
08402afa |
if(phishingScan(mctx->ctx, &hrefs) == CL_VIRUS) { |
8affc406 |
/*
* FIXME: message objects' contents are
* encapsulated so we should not access
* the members directly
*/ |
94aea271 |
mainMessage->isInfected = TRUE; |
ecc3d638 |
*rc = VIRUS; |
94aea271 |
cli_dbgmsg("PH:Phishing found\n");
}
}
}
hrefs_done(b,&hrefs); |
c52d991e |
}
|
f2f25418 |
#ifdef HAVE_BACKTRACE |
4f1d0bfc |
static void |
02927896 |
sigsegv(int sig)
{
signal(SIGSEGV, SIG_DFL); |
f2f25418 |
print_trace(1); |
02927896 |
exit(SIGSEGV);
}
|
4f1d0bfc |
static void |
02927896 |
print_trace(int use_syslog)
{
void *array[10];
size_t size;
char **strings;
size_t i;
pid_t pid = getpid();
|
d563818f |
cli_errmsg("Segmentation fault, attempting to print backtrace\n");
|
02927896 |
size = backtrace(array, 10);
strings = backtrace_symbols(array, size);
|
d563818f |
cli_errmsg("Backtrace of pid %d:\n", pid);
if(use_syslog) |
02927896 |
syslog(LOG_ERR, "Backtrace of pid %d:", pid);
|
d563818f |
for(i = 0; i < size; i++) {
cli_errmsg("%s\n", strings[i]); |
02927896 |
if(use_syslog) |
ce2dcb53 |
syslog(LOG_ERR, "bt[%llu]: %s", (unsigned long long)i, strings[i]); |
d563818f |
} |
02927896 |
|
d563818f |
#ifdef SAVE_TMP
cli_errmsg("The errant mail file has been saved\n");
#endif
/* #else TODO: dump the current email */ |
ef822cfc |
|
02927896 |
free(strings);
}
#endif |
f24bf390 |
|
6cad6a15 |
/* See also clamav-milter */ |
ddea752e |
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; |
6cad6a15 |
if(strcasecmp(cmd, "Received") == 0) |
ddea752e |
return TRUE; |
6cad6a15 |
if(strcasecmp(cmd, "De") == 0) |
ddea752e |
return TRUE;
}
return FALSE;
}
|
ae5c693a |
/* |
72cf1461 |
* Like fgets but cope with end of line by "\n", "\r\n", "\n\r", "\r"
*/
static char * |
6bcd0c86 |
getline_from_mbox(char *buffer, size_t buffer_len, fmap_t *map, size_t *at) |
72cf1461 |
{ |
f304dc68 |
const char *src, *cursrc;
char *curbuf; |
ee1b2a6c |
size_t i; |
6bcd0c86 |
size_t input_len = MIN(map->len - *at, buffer_len + 1);
src = cursrc = fmap_need_off_once(map, *at, input_len); |
72cf1461 |
|
f87a92ca |
/* we check for eof from the result of GETC()
* if(feof(fin))
return NULL;*/ |
ee1b2a6c |
if(!src) {
cli_dbgmsg("getline_from_mbox: fmap need failed\n");
return NULL;
} |
6bcd0c86 |
if((buffer_len == 0) || (buffer == NULL)) { |
964a1e73 |
cli_errmsg("Invalid call to getline_from_mbox(). Refer to https://www.clamav.net/documents/installing-clamav\n"); |
ee1b2a6c |
return NULL;
} |
72cf1461 |
|
ee1b2a6c |
curbuf = buffer;
|
6bcd0c86 |
for(i=0; i<buffer_len-1; i++) {
char c; |
72cf1461 |
|
6bcd0c86 |
if(!input_len--) {
if(curbuf == buffer) {
/* EOF on first char */ |
72cf1461 |
return NULL; |
6bcd0c86 |
}
break; |
72cf1461 |
}
|
6bcd0c86 |
switch((c = *cursrc++)) { |
ee1b2a6c |
case '\0':
continue;
case '\n':
*curbuf++ = '\n'; |
6bcd0c86 |
if(input_len && *cursrc == '\r') { |
ee1b2a6c |
i++;
cursrc++;
}
break;
case '\r':
*curbuf++ = '\r'; |
6bcd0c86 |
if(input_len && *cursrc == '\n') { |
ee1b2a6c |
i++;
cursrc++;
}
break;
default:
*curbuf++ = c;
continue; |
72cf1461 |
} |
ee1b2a6c |
break;
}
*at += cursrc - src;
*curbuf = '\0';
return buffer; |
72cf1461 |
} |
b2ba24f5 |
|
a603478f |
/*
* Is this line a candidate for the start of a bounce message?
*/ |
b2ba24f5 |
static bool |
7021b545 |
isBounceStart(mbox_ctx *mctx, const char *line) |
b2ba24f5 |
{ |
be32043e |
size_t len;
|
b2ba24f5 |
if(line == NULL)
return FALSE;
if(*line == '\0')
return FALSE; |
25071deb |
/*if((strncmp(line, "From ", 5) == 0) && !isalnum(line[5])) |
b2ba24f5 |
return FALSE;
if((strncmp(line, ">From ", 6) == 0) && !isalnum(line[6])) |
25071deb |
return FALSE;*/ |
b2ba24f5 |
|
be32043e |
len = strlen(line);
if((len < 6) || (len >= 72))
return FALSE;
if((memcmp(line, "From ", 5) == 0) ||
(memcmp(line, ">From ", 6) == 0)) { |
b2ba24f5 |
int numSpaces = 0, numDigits = 0;
|
d563818f |
line += 4;
|
b2ba24f5 |
do
if(*line == ' ')
numSpaces++; |
8affc406 |
else if(isdigit((*line) & 0xFF)) |
b2ba24f5 |
numDigits++;
while(*++line != '\0');
if(numSpaces < 6)
return FALSE;
if(numDigits < 11)
return FALSE; |
164ba3b8 |
return TRUE; |
b2ba24f5 |
} |
1f6e52cb |
return (bool)(cli_filetype((const unsigned char *)line, len, mctx->ctx->engine) == CL_TYPE_MAIL); |
b2ba24f5 |
} |
a05e6d45 |
/*
* Extract a binhexEncoded message, return if it's found to be infected as we
* extract it
*/
static bool |
001ad879 |
exportBinhexMessage(mbox_ctx *mctx, message *m) |
a05e6d45 |
{
bool infected = FALSE;
fileblob *fb;
if(messageGetEncoding(m) == NOENCODING)
messageSetEncoding(m, "x-binhex");
|
001ad879 |
fb = messageToFileblob(m, mctx->dir, 0); |
a05e6d45 |
if(fb) {
cli_dbgmsg("Binhex file decoded to %s\n",
fileblobGetFilename(fb)); |
a585329e |
if(fileblobScanAndDestroy(fb) == CL_VIRUS)
infected = TRUE; |
001ad879 |
mctx->files++; |
a05e6d45 |
} else |
001ad879 |
cli_errmsg("Couldn't decode binhex file to %s\n", mctx->dir); |
a05e6d45 |
return infected;
} |
c1fce7f7 |
/* |
7dde984e |
* Locate any bounce message and extract it. Return cl_status |
47d9cc65 |
*/
static int |
001ad879 |
exportBounceMessage(mbox_ctx *mctx, text *start) |
47d9cc65 |
{ |
7dde984e |
int rc = CL_CLEAN; |
47d9cc65 |
text *t;
fileblob *fb;
/*
* Attempt to save the original (unbounced)
* message - clamscan will find that in the
* directory and call us again (with any luck)
* having found an e-mail message to handle.
*
* This finds a lot of false positives, the
* search that a content type is in the
* bounce (i.e. it's after the bounce header)
* helps a bit.
*
* messageAddLine |
7cd9337a |
* optimization could help here, but needs |
47d9cc65 |
* 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
*/
for(t = start; t; t = t->t_next) {
const char *txt = lineGetData(t->t_line); |
ecc3d638 |
char cmd[RFC2821LENGTH + 1]; |
47d9cc65 |
if(txt == NULL)
continue;
if(cli_strtokbuf(txt, 0, ":", cmd) == NULL)
continue;
switch(tableFind(mctx->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;
else if(strcasecmp(cmd, "Received") == 0)
start = t;
continue;
}
break;
}
if(t && ((fb = fileblobCreate()) != NULL)) {
cli_dbgmsg("Found a bounce message\n");
fileblobSetFilename(fb, mctx->dir, "bounce"); |
7dde984e |
fileblobSetCTX(fb, mctx->ctx);
if(textToFileblob(start, fb, 1) == NULL) { |
47d9cc65 |
cli_dbgmsg("Nothing new to save in the bounce message\n"); |
7dde984e |
fileblobDestroy(fb);
} else
rc = fileblobScanAndDestroy(fb); |
001ad879 |
mctx->files++; |
47d9cc65 |
} else
cli_dbgmsg("Not found a bounce message\n");
return rc;
}
/* |
ede9939c |
* Get string representation of mimetype
*/
static const char *getMimeTypeStr(mime_type mimetype)
{
const struct tableinit *entry = mimeTypeStr;
while (entry->key) {
if (mimetype == entry->value)
return entry->key;
entry++;
}
return "UNKNOWN";
}
/*
* Get string representation of encoding type
*/
static const char *getEncTypeStr(encoding_type enctype)
{
const struct tableinit *entry = encTypeStr;
while (entry->key) {
if (enctype == entry->value)
return entry->key;
entry++;
}
return "UNKNOWN";
}
/* |
c1fce7f7 |
* Handle the ith element of a number of multiparts, e.g. multipart/alternative
*/
static message * |
ecc3d638 |
do_multipart(message *mainMessage, message **messages, int i, mbox_status *rc, mbox_ctx *mctx, message *messageIn, text **tptr, unsigned int recursion_level) |
c1fce7f7 |
{
bool addToText = FALSE;
const char *dtype;
#ifndef SAVE_TO_DISC
message *body;
#endif
message *aMessage = messages[i]; |
d77ac7de |
const int doPhishingScan = mctx->ctx->engine->dboptions&CL_DB_PHISHING_URLS && (DCONF_PHISHING&PHISHING_CONF_ENGINE); |
ede9939c |
#if HAVE_JSON
const char *mtype = NULL; |
c44d7e7a |
json_object *thisobj = NULL, *saveobj = mctx->wrkobj; |
ede9939c |
if (mctx->wrkobj != NULL) { |
c44d7e7a |
json_object *multiobj = cli_jsonarray(mctx->wrkobj, "Multipart"); |
ede9939c |
if (multiobj == NULL) {
cli_errmsg("Cannot get multipart preclass array\n");
*rc = -1;
return mainMessage;
} |
d77ac7de |
|
ede9939c |
thisobj = messageGetJObj(aMessage);
if (thisobj == NULL) {
cli_errmsg("Cannot get message preclass object\n");
*rc = -1;
return mainMessage;
}
if (cli_json_addowner(multiobj, thisobj, NULL, -1) != CL_SUCCESS) {
cli_errmsg("Cannot assign message preclass object to multipart preclass array\n");
*rc = -1;
return mainMessage;
}
}
#endif
if(aMessage == NULL) {
#if HAVE_JSON
if (thisobj != NULL)
cli_jsonstr(thisobj, "MimeType", "NULL");
#endif |
c1fce7f7 |
return mainMessage; |
ede9939c |
} |
c1fce7f7 |
|
69c62847 |
if(*rc != OK)
return mainMessage;
|
c1fce7f7 |
cli_dbgmsg("Mixed message part %d is of type %d\n",
i, messageGetMimeType(aMessage));
|
ede9939c |
#if HAVE_JSON
if (thisobj != NULL) {
cli_jsonstr(thisobj, "MimeType", getMimeTypeStr(messageGetMimeType(aMessage)));
cli_jsonstr(thisobj, "MimeSubtype", messageGetMimeSubtype(aMessage));
cli_jsonstr(thisobj, "EncodingType", getEncTypeStr(messageGetEncoding(aMessage)));
cli_jsonstr(thisobj, "Disposition", messageGetDispositionType(aMessage));
cli_jsonstr(thisobj, "Filename", messageHasFilename(aMessage) ?
messageGetFilename(aMessage): "(inline)");
}
#endif
|
c1fce7f7 |
switch(messageGetMimeType(aMessage)) {
case APPLICATION:
case AUDIO:
case IMAGE:
case VIDEO:
break;
case NOMIME:
cli_dbgmsg("No mime headers found in multipart part %d\n", i);
if(mainMessage) {
if(binhexBegin(aMessage)) {
cli_dbgmsg("Found binhex message in multipart/mixed mainMessage\n");
|
001ad879 |
if(exportBinhexMessage(mctx, mainMessage)) |
5684fccf |
*rc = VIRUS; |
c1fce7f7 |
}
if(mainMessage != messageIn)
messageDestroy(mainMessage);
mainMessage = NULL;
} else if(aMessage) {
if(binhexBegin(aMessage)) {
cli_dbgmsg("Found binhex message in multipart/mixed non mime part\n"); |
001ad879 |
if(exportBinhexMessage(mctx, aMessage)) |
5684fccf |
*rc = VIRUS; |
c1fce7f7 |
assert(aMessage == messages[i]);
messageReset(messages[i]);
}
}
addToText = TRUE;
if(messageGetBody(aMessage) == NULL)
/*
* No plain text version
*/
cli_dbgmsg("No plain text alternative\n");
break;
case TEXT:
dtype = messageGetDispositionType(aMessage);
cli_dbgmsg("Mixed message text part disposition \"%s\"\n",
dtype);
if(strcasecmp(dtype, "attachment") == 0)
break;
if((*dtype == '\0') || (strcasecmp(dtype, "inline") == 0)) {
const char *cptr;
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage);
mainMessage = NULL;
cptr = messageGetMimeSubtype(aMessage);
cli_dbgmsg("Mime subtype \"%s\"\n", cptr);
if((tableFind(mctx->subtypeTable, cptr) == PLAIN) && |
be32043e |
(messageGetEncoding(aMessage) == NOENCODING)) { |
c1fce7f7 |
/* |
be32043e |
* Strictly speaking, a text/plain part
* is not an attachment. We pretend it
* is so that we can decode and scan it |
c1fce7f7 |
*/ |
be32043e |
if(!messageHasFilename(aMessage)) { |
c1fce7f7 |
cli_dbgmsg("Adding part to main message\n");
addToText = TRUE; |
be32043e |
} else
cli_dbgmsg("Treating inline as attachment\n"); |
c1fce7f7 |
} else { |
56aea026 |
const int is_html = (tableFind(mctx->subtypeTable, cptr) == HTML); |
6a4dd9dc |
if(doPhishingScan) |
56aea026 |
checkURLs(aMessage, mctx, rc, is_html); |
c1fce7f7 |
messageAddArgument(aMessage,
"filename=mixedtextportion");
}
break;
}
cli_dbgmsg("Text type %s is not supported\n", dtype);
return mainMessage;
case MESSAGE:
/* Content-Type: message/rfc822 */
cli_dbgmsg("Found message inside multipart (encoding type %d)\n",
messageGetEncoding(aMessage));
#ifndef SCAN_UNENCODED_BOUNCES
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("Unencoded multipart/message will not be scanned\n");
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL;
return mainMessage;
}
/* FALLTHROUGH */
default:
cli_dbgmsg("Encoded multipart/message will be scanned\n");
}
#endif
#if 0
messageAddStrAtTop(aMessage,
"Received: by clamd (message/rfc822)");
#endif
#ifdef SAVE_TO_DISC
/*
* Save this embedded message
* to a temporary file
*/ |
001ad879 |
if(saveTextPart(mctx, aMessage, 1) == CL_VIRUS) |
7dde984e |
*rc = VIRUS; |
c1fce7f7 |
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL;
#else
/* |
b912eaf2 |
* Scan in memory, faster but is open to DoS attacks
* when many nested levels are involved. |
c1fce7f7 |
*/ |
0072fa21 |
body = parseEmailHeaders(aMessage, mctx->rfc821Table);
|
c1fce7f7 |
/* |
7cd9337a |
* We've finished with the |
c1fce7f7 |
* 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; |
ede9939c |
#if HAVE_JSON
mctx->wrkobj = thisobj;
#endif |
c1fce7f7 |
if(body) { |
0072fa21 |
messageSetCTX(body, mctx->ctx);
*rc = parseEmailBody(body, NULL, mctx, recursion_level + 1); |
69c62847 |
if((*rc == OK) && messageContainsVirus(body)) |
5684fccf |
*rc = VIRUS; |
c1fce7f7 |
messageDestroy(body);
} |
ede9939c |
#if HAVE_JSON
mctx->wrkobj = saveobj;
#endif |
c1fce7f7 |
#endif
return mainMessage;
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"); |
ede9939c |
#if HAVE_JSON
mctx->wrkobj = thisobj;
#endif |
c1fce7f7 |
if(aMessage) {
/*
* The headers were parsed when reading in the
* whole multipart section
*/ |
242ffd7a |
*rc = parseEmailBody(aMessage, *tptr, mctx, recursion_level + 1); |
8affc406 |
cli_dbgmsg("Finished recursion, rc = %d\n", (int)*rc); |
c1fce7f7 |
assert(aMessage == messages[i]);
messageDestroy(messages[i]);
messages[i] = NULL;
} else { |
242ffd7a |
*rc = parseEmailBody(NULL, NULL, mctx, recursion_level + 1); |
c1fce7f7 |
if(mainMessage && (mainMessage != messageIn))
messageDestroy(mainMessage);
mainMessage = NULL;
} |
ede9939c |
#if HAVE_JSON
mctx->wrkobj = saveobj;
#endif |
c1fce7f7 |
return mainMessage;
default: |
6351aa86 |
cli_dbgmsg("Only text and application attachments are fully supported, type = %d\n", |
c1fce7f7 |
messageGetMimeType(aMessage)); |
5b2bcb65 |
/* fall through - we may be able to salvage something */ |
c1fce7f7 |
}
|
a585329e |
if(*rc != VIRUS) { |
cfeeb327 |
fileblob *fb = messageToFileblob(aMessage, mctx->dir, 1); |
ede9939c |
#if HAVE_JSON
json_object *arrobj;
int arrlen = 0;
if (thisobj != NULL) {
/* attempt to determine container size - prevents incorrect type reporting */
if (json_object_object_get_ex(mctx->ctx->wrkproperty, "ContainedObjects", &arrobj))
arrlen = json_object_array_length(arrobj);
} |
c1fce7f7 |
|
ede9939c |
#endif |
cfeeb327 |
if(fb) {
/* aMessage doesn't always have a ctx set */
fileblobSetCTX(fb, mctx->ctx);
if(fileblobScanAndDestroy(fb) == CL_VIRUS)
*rc = VIRUS;
if (!addToText) |
001ad879 |
mctx->files++; |
c1fce7f7 |
} |
ede9939c |
#if HAVE_JSON
if (thisobj != NULL) {
json_object *entry = NULL;
const char *dtype = NULL;
/* attempt to acquire container type */
if (json_object_object_get_ex(mctx->ctx->wrkproperty, "ContainedObjects", &arrobj))
if (json_object_array_length(arrobj) > arrlen)
entry = json_object_array_get_idx(arrobj, arrlen);
if (entry) {
json_object_object_get_ex(entry, "FileType", &entry);
if (entry)
dtype = json_object_get_string(entry);
}
cli_jsonint(thisobj, "ContainedObjectsIndex", arrlen);
cli_jsonstr(thisobj, "ClamAVFileType", dtype ? dtype : "UNKNOWN");
}
#endif |
a585329e |
if(messageContainsVirus(aMessage))
*rc = VIRUS; |
c1fce7f7 |
}
messageDestroy(aMessage);
messages[i] = NULL;
return mainMessage;
} |
4f4a8f4a |
/*
* Returns the number of quote characters in the given string
*/
static int
count_quotes(const char *buf)
{
int quotes = 0;
while(*buf)
if(*buf++ == '\"')
quotes++;
return quotes;
} |
842c7d49 |
/*
* Will the next line be a folded header? See RFC2822 section 2.2.3
*/
static bool
next_is_folded_header(const text *t)
{
const text *next = t->t_next;
const char *data, *ptr;
if(next == NULL)
return FALSE;
if(next->t_line == NULL)
return FALSE;
data = lineGetData(next->t_line);
/* |
b912eaf2 |
* Section B.2 of RFC822 says TAB or SPACE means a continuation of the |
842c7d49 |
* previous entry.
*/
if(isblank(data[0]))
return TRUE;
if(strchr(data, '=') == NULL)
/*
* Avoid false positives with
* Content-Type: text/html;
* Content-Transfer-Encoding: quoted-printable
*/
return FALSE; |
bc6bbeff |
|
842c7d49 |
/*
* Some are broken and don't fold headers lines
* correctly as per section 2.2.3 of RFC2822.
* Generally they miss the white space at
* the start of the fold line:
* Content-Type: multipart/related;
* type="multipart/alternative";
* boundary="----=_NextPart_000_006A_01C6AC47.348CB550"
* should read:
* Content-Type: multipart/related;
* type="multipart/alternative";
* boundary="----=_NextPart_000_006A_01C6AC47.348CB550"
* Since we're a virus checker not an RFC
* verifier we need to handle these
*/
data = lineGetData(t->t_line);
ptr = strchr(data, '\0');
while(--ptr > data)
switch(*ptr) {
case ';':
return TRUE;
case '\n':
case ' ':
case '\r':
case '\t':
continue; /* white space at end of line */
default:
return FALSE;
}
return FALSE;
} |
0cf4cea7 |
/*
* This routine is called on the first line of the body of
* an email to handle broken messages that have newlines
* in the middle of its headers
*/
static bool
newline_in_header(const char *line)
{
cli_dbgmsg("newline_in_header, check \"%s\"\n", line);
if(strncmp(line, "Message-Id: ", 12) == 0)
return TRUE;
if(strncmp(line, "Date: ", 6) == 0)
return TRUE;
return FALSE;
} |