void
rspamd_content_type_add_param (rspamd_mempool_t *pool,
struct rspamd_content_type *ct,
const gchar *name_start, const gchar *name_end,
const gchar *value_start, const gchar *value_end)
{
rspamd_ftok_t srch;
struct rspamd_content_type_param *found = NULL, *nparam;
g_assert (ct != NULL);
srch.begin = name_start;
srch.len = name_end - name_start;
if (ct->attrs) {
found = g_hash_table_lookup (ct->attrs, &srch);
}
else {
ct->attrs = g_hash_table_new (rspamd_ftok_icase_hash,
rspamd_ftok_icase_equal);
}
nparam = rspamd_mempool_alloc (pool, sizeof (*nparam));
nparam->name.begin = name_start;
nparam->name.len = name_end - name_start;
nparam->value.begin = value_start;
nparam->value.len = value_end - value_start;
if (!found) {
DL_APPEND (found, nparam);
g_hash_table_insert (ct->attrs, &nparam->name, nparam);
}
else {
DL_APPEND (found, nparam);
}
RSPAMD_FTOK_ASSIGN (&srch, "charset");
if (rspamd_ftok_cmp (&nparam->name, &srch) == 0) {
/* Adjust charset */
ct->charset.begin = nparam->value.begin;
ct->charset.len = nparam->value.len;
}
RSPAMD_FTOK_ASSIGN (&srch, "boundary");
if (rspamd_ftok_cmp (&nparam->name, &srch) == 0) {
/* Adjust boundary */
ct->boundary.begin = nparam->value.begin;
ct->boundary.len = nparam->value.len;
}
}
void
rspamd_content_disposition_add_param (rspamd_mempool_t *pool,
struct rspamd_content_disposition *cd,
const gchar *name_start, const gchar *name_end,
const gchar *value_start, const gchar *value_end)
{
rspamd_ftok_t srch;
gchar *decoded;
struct rspamd_content_type_param *found = NULL, *nparam;
g_assert (cd != NULL);
srch.begin = name_start;
srch.len = name_end - name_start;
if (cd->attrs) {
found = g_hash_table_lookup (cd->attrs, &srch);
}
else {
cd->attrs = g_hash_table_new (rspamd_ftok_icase_hash,
rspamd_ftok_icase_equal);
}
nparam = rspamd_mempool_alloc (pool, sizeof (*nparam));
nparam->name.begin = name_start;
nparam->name.len = name_end - name_start;
decoded = rspamd_mime_header_decode (pool, value_start, value_end - value_start);
RSPAMD_FTOK_FROM_STR (&nparam->value, decoded);
if (!found) {
g_hash_table_insert (cd->attrs, &nparam->name, nparam);
}
DL_APPEND (found, nparam);
srch.begin = "filename";
srch.len = 8;
if (rspamd_ftok_cmp (&nparam->name, &srch) == 0) {
/* Adjust filename */
cd->filename.begin = nparam->value.begin;
cd->filename.len = nparam->value.len;
}
}
static gboolean
rspamd_archive_cheat_detect (struct rspamd_mime_part *part, const gchar *str,
const guchar *magic_start, gsize magic_len)
{
struct rspamd_content_type *ct;
const gchar *p;
rspamd_ftok_t srch, *fname;
ct = part->ct;
RSPAMD_FTOK_ASSIGN (&srch, "application");
if (ct && ct->type.len && ct->subtype.len > 0 && rspamd_ftok_cmp (&ct->type,
&srch) == 0) {
if (rspamd_substring_search_caseless (ct->subtype.begin, ct->subtype.len,
str, strlen (str)) != -1) {
return TRUE;
}
}
if (part->cd) {
fname = &part->cd->filename;
if (fname && fname->len > strlen (str)) {
p = fname->begin + fname->len - strlen (str);
if (rspamd_lc_cmp (p, str, strlen (str)) == 0) {
if (*(p - 1) == '.') {
return TRUE;
}
}
}
if (magic_start != NULL) {
if (part->parsed_data.len > magic_len && memcmp (part->parsed_data.begin,
magic_start, magic_len) == 0) {
return TRUE;
}
}
}
return FALSE;
}
struct rspamd_content_type *
rspamd_content_type_parse (const gchar *in,
gsize len, rspamd_mempool_t *pool)
{
struct rspamd_content_type *res = NULL, val;
rspamd_ftok_t srch;
val.lc_data = rspamd_mempool_alloc (pool, len);
memcpy (val.lc_data, in, len);
rspamd_str_lc (val.lc_data, len);
if (rspamd_content_type_parser (val.lc_data, len, &val, pool)) {
res = rspamd_mempool_alloc (pool, sizeof (val));
memcpy (res, &val, sizeof (val));
if (res->attrs) {
rspamd_mempool_add_destructor (pool,
(rspamd_mempool_destruct_t)g_hash_table_unref, res->attrs);
}
/* Now do some hacks to work with broken content types */
if (res->subtype.len == 0) {
res->flags |= RSPAMD_CONTENT_TYPE_BROKEN;
RSPAMD_FTOK_ASSIGN (&srch, "text");
if (rspamd_ftok_cmp (&res->type, &srch) == 0) {
/* Workaround for Content-Type: text */
/* Assume text/plain */
RSPAMD_FTOK_ASSIGN (&srch, "plain");
}
else {
RSPAMD_FTOK_ASSIGN (&srch, "html");
if (rspamd_ftok_cmp (&res->type, &srch) == 0) {
/* Workaround for Content-Type: html */
RSPAMD_FTOK_ASSIGN (&res->type, "text");
RSPAMD_FTOK_ASSIGN (&res->subtype, "html");
}
else {
RSPAMD_FTOK_ASSIGN (&srch, "application");
if (rspamd_ftok_cmp (&res->type, &srch) == 0) {
RSPAMD_FTOK_ASSIGN (&res->subtype, "octet-stream");
}
}
}
}
else {
/* Common mistake done by retards */
RSPAMD_FTOK_ASSIGN (&srch, "alternate");
if (rspamd_ftok_cmp (&res->subtype, &srch) == 0) {
res->flags |= RSPAMD_CONTENT_TYPE_BROKEN;
RSPAMD_FTOK_ASSIGN (&res->subtype, "alternative");
}
}
RSPAMD_FTOK_ASSIGN (&srch, "multipart");
if (rspamd_ftok_cmp (&res->type, &srch) == 0) {
res->flags |= RSPAMD_CONTENT_TYPE_MULTIPART;
}
else {
RSPAMD_FTOK_ASSIGN (&srch, "text");
if (rspamd_ftok_cmp (&res->type, &srch) == 0) {
res->flags |= RSPAMD_CONTENT_TYPE_TEXT;
}
else {
RSPAMD_FTOK_ASSIGN (&srch, "message");
if (rspamd_ftok_cmp (&res->type, &srch) == 0) {
res->flags |= RSPAMD_CONTENT_TYPE_MESSAGE;
}
}
}
}
else {
msg_warn_pool ("cannot parse content type: %*s", (gint)len, val.lc_data);
}
return res;
}