[Feature] Add new function to parse mime addresses

diff --git a/src/libmime/email_addr.c b/src/libmime/email_addr.c
index 2c7964f873638ae210741db3307582b18ed56397..75fdd9fc6433efcc1823f9a40ee2a3b1c9b32d7a 100644 (file)
--- a/src/libmime/email_addr.c
+++ b/src/libmime/email_addr.c
@@ -19,6 +19,7 @@
 #include "message.h"
 #include "printf.h"
 #include "smtp_parsers.h"
+#include "mime_headers.h"
 
 static void
 rspamd_email_addr_dtor (struct rspamd_email_address *addr)
@@ -31,6 +32,10 @@ rspamd_email_addr_dtor (struct rspamd_email_address *addr)
                g_free ((void *)addr->user);
        }
 
+       if (addr->name) {
+               g_free ((gpointer)addr->name);
+       }
+
        g_slice_free1 (sizeof (*addr), addr);
 }
 
@@ -113,3 +118,195 @@ rspamd_email_address_unref (struct rspamd_email_address *addr)
 {
        REF_RELEASE (addr);
 }
+
+static inline void
+rspamd_email_address_add (rspamd_mempool_t *pool,
+               GPtrArray *ar,
+               struct rspamd_email_address *addr,
+               GString *name)
+{
+       struct rspamd_email_address *elt;
+       guint nlen;
+
+       elt = g_slice_alloc (sizeof (*elt));
+       memcpy (elt, addr, sizeof (*addr));
+
+       if ((elt->flags & RSPAMD_EMAIL_ADDR_QUOTED) && elt->addr[0] == '"') {
+               if (elt->flags & RSPAMD_EMAIL_ADDR_HAS_BACKSLASH) {
+                       /* We also need to unquote user */
+                       rspamd_email_address_unescape (elt);
+               }
+
+               /* We need to unquote addr */
+               nlen = elt->domain_len + elt->user_len + 2;
+               elt->addr = g_malloc (nlen + 1);
+               elt->addr_len = rspamd_snprintf ((char *)elt->addr, nlen, "%*s@%*s",
+                               (gint)elt->user_len, elt->user,
+                               (gint)elt->domain_len, elt->domain);
+               elt->flags |= RSPAMD_EMAIL_ADDR_ADDR_ALLOCATED;
+       }
+
+       REF_INIT_RETAIN (elt, rspamd_email_addr_dtor);
+
+       if (name->len > 0) {
+               elt->name = rspamd_mime_header_decode (pool, name->str, name->len);
+       }
+
+       g_ptr_array_add (ar, elt);
+}
+
+GPtrArray *
+rspamd_email_address_from_mime (rspamd_mempool_t *pool,
+               const gchar *hdr, guint len,
+               GPtrArray *src)
+{
+       GPtrArray *res = src;
+       struct rspamd_email_address addr;
+       const gchar *p = hdr, *end = hdr + len, *c = hdr, *t;
+       GString *ns;
+       enum {
+               parse_name = 0,
+               parse_quoted,
+               parse_addr,
+               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;
+                                       while (t > c && g_ascii_isspace (*t)) {
+                                               t --;
+                                       }
+
+                                       g_string_append_len (ns, c, t - c);
+                               }
+
+                               state = parse_quoted;
+                               c = p + 1;
+                       }
+                       else if (*p == '<') {
+                               if (p > c) {
+                                       t = p;
+                                       while (t > c && g_ascii_isspace (*t)) {
+                                               t --;
+                                       }
+
+                                       g_string_append_len (ns, c, t - c);
+                               }
+
+                               c = p;
+                               state = parse_addr;
+                       }
+                       else if (*p == ',') {
+                               if (p > c) {
+                                       /*
+                                        * Last token must be the address:
+                                        * e.g. Some name name@domain.com
+                                        */
+                                       t = p;
+                                       while (t > c && g_ascii_isspace (*t)) {
+                                               t --;
+                                       }
+
+                                       rspamd_smtp_addr_parse (c, t - c, &addr);
+
+                                       if (addr.flags & RSPAMD_EMAIL_ADDR_VALID) {
+                                               rspamd_email_address_add (pool, res, &addr, ns);
+                                       }
+
+                                       /* Cleanup for the next use */
+                                       g_string_set_size (ns, 0);
+                               }
+
+                               state = skip_spaces;
+                               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);
+                               }
+
+                               /* Cleanup for the next use */
+                               g_string_set_size (ns, 0);
+
+                               state = skip_spaces;
+                               next_state = parse_name;
+                       }
+                       p ++;
+                       break;
+               case skip_spaces:
+                       if (!g_ascii_isspace (*p)) {
+                               c = p;
+                               state = next_state;
+                       }
+                       else {
+                               p ++;
+                       }
+                       break;
+               }
+       }
+
+       /* Handle leftover */
+       switch (state) {
+       case parse_name:
+       case parse_addr:
+               if (p > c) {
+                       rspamd_smtp_addr_parse (c, p - c + 1, &addr);
+
+                       if (addr.flags & RSPAMD_EMAIL_ADDR_VALID) {
+                               rspamd_email_address_add (pool, res, &addr, ns);
+                       }
+               }
+               break;
+       case parse_quoted:
+               /* Unfinished quoted string */
+               break;
+       default:
+               /* Do nothing */
+               break;
+       }
+
+       return res;
+}
+
+void
+rspamd_email_address_list_destroy (gpointer ptr)
+{
+       GPtrArray *ar = ptr;
+       guint i;
+       struct rspamd_email_address *addr;
+
+       PTR_ARRAY_FOREACH (ar, i, addr) {
+               REF_RELEASE (addr);
+       }
+
+       g_ptr_array_free (ar, TRUE);
+}

diff --git a/src/libmime/email_addr.h b/src/libmime/email_addr.h
index de91237db8ae6da2614fd7dc7ba1f96ae15ffef7..f6a16d6823ccc94a33bebcb1abe6566643624576 100644 (file)
--- a/src/libmime/email_addr.h
+++ b/src/libmime/email_addr.h
@@ -17,7 +17,8 @@
 #define SRC_LIBMIME_EMAIL_ADDR_H_
 
 #include "config.h"
-#include "ref.h"
+#include "libutil/mem_pool.h"
+#include "libutil/ref.h"
 
 struct rspamd_mime_header;
 
@@ -62,6 +63,26 @@ struct rspamd_email_address {
 struct rspamd_email_address * rspamd_email_address_from_smtp (
                const gchar *str, guint len);
 
+/**
+ * Parses email address from the mime header, decodes names and return the array
+ * of `rspamd_email_address`. If `src` is NULL, then this function creates a new
+ * array and adds a destructor to remove elements when `pool` is destroyed.
+ * Otherwise, addresses are appended to `src`.
+ * @param hdr
+ * @param len
+ * @return
+ */
+GPtrArray *rspamd_email_address_from_mime (rspamd_mempool_t *pool,
+               const gchar *hdr,
+               guint len,
+               GPtrArray *src);
+
+/**
+ * Destroys list of email addresses
+ * @param ptr
+ */
+void rspamd_email_address_list_destroy (gpointer ptr);
+
 struct rspamd_email_address * rspamd_email_address_ref (
                struct rspamd_email_address *addr);