struct rspamd_email_address *
rspamd_email_address_from_smtp (const gchar *str, guint len)
{
struct rspamd_email_address addr, *ret;
gsize nlen;
if (str == NULL || len == 0) {
return NULL;
}
rspamd_smtp_addr_parse (str, len, &addr);
if (addr.flags & RSPAMD_EMAIL_ADDR_VALID) {
ret = g_slice_alloc (sizeof (*ret));
memcpy (ret, &addr, sizeof (addr));
if ((ret->flags & RSPAMD_EMAIL_ADDR_QUOTED) && ret->addr[0] == '"') {
if (ret->flags & RSPAMD_EMAIL_ADDR_HAS_BACKSLASH) {
/* We also need to unquote user */
rspamd_email_address_unescape (ret);
}
/* We need to unquote addr */
nlen = ret->domain_len + ret->user_len + 2;
ret->addr = g_malloc (nlen + 1);
ret->addr_len = rspamd_snprintf ((char *)ret->addr, nlen, "%*s@%*s",
(gint)ret->user_len, ret->user,
(gint)ret->domain_len, ret->domain);
ret->flags |= RSPAMD_EMAIL_ADDR_ADDR_ALLOCATED;
}
REF_INIT_RETAIN (ret, rspamd_email_addr_dtor);
return ret;
}
return NULL;
}
GPtrArray *
rspamd_email_address_from_mime (rspamd_mempool_t *pool,
const gchar *hdr, guint len,
GPtrArray *src)
{
GPtrArray *res = src;
gboolean seen_at = FALSE;
struct rspamd_email_address addr;
const gchar *p = hdr, *end = hdr + len, *c = hdr, *t;
GString *ns;
gint obraces, ebraces;
enum {
parse_name = 0,
parse_quoted,
parse_addr,
skip_comment,
skip_spaces
} state = parse_name, next_state = parse_name;
if (res == NULL) {
res = g_ptr_array_sized_new (2);
rspamd_mempool_add_destructor (pool, rspamd_email_address_list_destroy,
res);
}
ns = g_string_sized_new (127);
while (p < end) {
switch (state) {
case parse_name:
if (*p == '"') {
/* We need to strip last spaces and update `ns` */
if (p > c) {
t = p - 1;
while (t > c && g_ascii_isspace (*t)) {
t --;
}
g_string_append_len (ns, c, t - c + 1);
}
state = parse_quoted;
c = p + 1;
}
else if (*p == '<') {
if (p > c) {
t = p - 1;
while (t > c && g_ascii_isspace (*t)) {
t --;
}
g_string_append_len (ns, c, t - c + 1);
}
c = p;
state = parse_addr;
}
else if (*p == ',') {
if (p > c) {
/*
* Last token must be the address:
* e.g. Some name [email protected]
*/
t = p - 1;
while (t > c && g_ascii_isspace (*t)) {
t --;
}
rspamd_smtp_addr_parse (c, t - c + 1, &addr);
if (addr.flags & RSPAMD_EMAIL_ADDR_VALID) {
rspamd_email_address_add (pool, res, &addr, ns);
}
else {
/* Try heuristic */
if (seen_at &&
rspamd_email_address_parse_heuristic (c,
t - c + 1, &addr)) {
rspamd_email_address_add (pool, res, &addr, ns);
}
else {
rspamd_email_address_add (pool, res, NULL, ns);
}
}
/* Cleanup for the next use */
g_string_set_size (ns, 0);
seen_at = FALSE;
}
state = skip_spaces;
next_state = parse_name;
}
else if (*p == '@') {
seen_at = TRUE;
}
else if (*p == '(') {
if (p > c) {
t = p - 1;
while (t > c && g_ascii_isspace (*t)) {
t --;
}
g_string_append_len (ns, c, t - c + 1);
}
c = p;
obraces = 1;
ebraces = 0;
state = skip_comment;
next_state = parse_name;
}
p ++;
break;
case parse_quoted:
if (*p == '"') {
if (p > c) {
g_string_append_len (ns, c, p - c);
}
state = skip_spaces;
next_state = parse_name;
}
p ++;
break;
case parse_addr:
if (*p == '>') {
rspamd_smtp_addr_parse (c, p - c + 1, &addr);
if (addr.flags & RSPAMD_EMAIL_ADDR_VALID) {
rspamd_email_address_add (pool, res, &addr, ns);
}
else {
/* Try heuristic */
if (seen_at &&
rspamd_email_address_parse_heuristic (c,
p - c + 1, &addr)) {
rspamd_email_address_add (pool, res, &addr, ns);
}
else {
rspamd_email_address_add (pool, res, NULL, ns);
}
}
/* Cleanup for the next use */
g_string_set_size (ns, 0);
seen_at = FALSE;
state = skip_spaces;
next_state = parse_name;
}
else if (*p == '@') {
seen_at = TRUE;
}
else if (*p == '(') {
obraces = 1;
ebraces = 0;
state = skip_comment;
next_state = parse_addr;
}
p ++;
break;
case skip_spaces:
if (!g_ascii_isspace (*p)) {
c = p;
state = next_state;
}
else {
p ++;
}
break;
case skip_comment:
if (*p == '(') {
obraces ++;
}
else if (*p == ')') {
ebraces ++;
}
if (obraces == ebraces) {
if (next_state == parse_name) {
/* Include comment in name */
if (p > c) {
t = p - 1;
while (t > c && g_ascii_isspace (*t)) {
t --;
}
g_string_append_len (ns, c, t - c + 1);
}
c = p;
}
state = next_state;
}
p ++;
break;
}
}
/* Handle leftover */
switch (state) {
case parse_name:
/* Assume the whole header as name (bad thing) */
if (p > c) {
while (p > c && g_ascii_isspace (*p)) {
p --;
}
if (seen_at) {
/* The whole email is likely address */
rspamd_smtp_addr_parse (c, p - c, &addr);
if (addr.flags & RSPAMD_EMAIL_ADDR_VALID) {
rspamd_email_address_add (pool, res, &addr, ns);
}
else {
/* Try heuristic */
if (rspamd_email_address_parse_heuristic (c,
p - c, &addr)) {
rspamd_email_address_add (pool, res, &addr, ns);
}
else {
rspamd_email_address_add (pool, res, NULL, ns);
}
}
}
else {
/* No @ seen */
g_string_append_len (ns, c, p - c);
rspamd_email_address_add (pool, res, NULL, ns);
}
}
break;
case parse_addr:
if (p > c) {
rspamd_smtp_addr_parse (c, p - c, &addr);
if (addr.flags & RSPAMD_EMAIL_ADDR_VALID) {
rspamd_email_address_add (pool, res, &addr, ns);
}
else {
/* Try heuristic */
if (rspamd_email_address_parse_heuristic (c, p - c,
&addr)) {
rspamd_email_address_add (pool, res, &addr, ns);
}
}
}
break;
case parse_quoted:
case skip_comment:
/* Unfinished quoted string or a comment */
break;
default:
/* Do nothing */
break;
}
g_string_free (ns, TRUE);
return res;
}