better duplicate detection

src/config.h

@@ -14,7 +14,7 @@
 
 #define VERSION "0.1.25-master-branch"
 
-#define BUILD 867
+#define BUILD 868
 
 #define HOSTID "mailarchiver"
 

src/defs.h

@@ -148,6 +148,7 @@ struct _state {
    int abufpos;
    char attachedfile[RND_STR_LEN+SMALLBUFSIZE];
    char message_id[SMALLBUFSIZE];
+   char message_id_hash[2*DIGEST_LENGTH+1];
    char miscbuf[MAX_TOKEN_LEN];
    char qpbuf[MAX_TOKEN_LEN];
    unsigned long n_token;

src/import.c

@@ -98,6 +98,7 @@ int import_message(char *filename, struct session_data *sdata, struct __data *da
       }
 
       rc = process_message(sdata, &state, data, cfg);
+      unlink(state.message_id_hash);
    }
 
    unlink(sdata->tmpframe);

src/message.c

@@ -240,7 +240,7 @@ void remove_stripped_attachments(struct _state *state){
 
 
 int process_message(struct session_data *sdata, struct _state *state, struct __data *data, struct __config *cfg){
-   int rc;
+   int rc, fd;
 
    /* discard if existing message_id */
 
@@ -257,6 +257,12 @@ int process_message(struct session_data *sdata, struct _state *state, struct __d
       return ERR_EXISTS;
    }
 
+   fd = open(state->message_id_hash, O_CREAT|O_EXCL|O_RDWR|O_TRUNC, S_IRUSR|S_IWUSR);
+   if(fd == -1){
+      remove_stripped_attachments(state);
+      return ERR_EXISTS;
+   }
+
 
    /* store base64 encoded file attachments */
 

src/parser.c

@@ -145,6 +145,9 @@ void post_parse(struct session_data *sdata, struct _state *state, struct __confi
       else snprintf(state->message_id, SMALLBUFSIZE-1, "null");
    }
 
+
+   digest_string(state->message_id, &(state->message_id_hash[0]));
+
 }
 
 

src/parser_utils.c

@@ -48,6 +48,7 @@ void init_state(struct _state *state){
    state->content_type_is_set = 0;
 
    memset(state->message_id, 0, SMALLBUFSIZE);
+   memset(state->message_id_hash, 0, 2*DIGEST_LENGTH+1);
    memset(state->miscbuf, 0, MAX_TOKEN_LEN);
    memset(state->qpbuf, 0, MAX_TOKEN_LEN);
 

src/session.c

@@ -231,6 +231,7 @@ int handle_smtp_session(int new_sd, struct __data *data, struct __config *cfg){
                         }
                         else {
                            inj = process_message(&sdata, &sstate, data, cfg);
+                           unlink(sstate.message_id_hash);
                            counters.c_size += sdata.tot_len;
                         }
 
@@ -258,7 +259,7 @@ int handle_smtp_session(int new_sd, struct __data *data, struct __config *cfg){
                   counters.c_rcvd++;
 
                   if(inj == ERR_EXISTS){
-                     syslog(LOG_PRIORITY, "%s: discarding: duplicate message, id: %llu", sdata.ttmpfile, sdata.duplicate_id);
+                     syslog(LOG_PRIORITY, "%s: discarding: duplicate message, id: %llu, message-id: %s", sdata.ttmpfile, sdata.duplicate_id, sstate.message_id);
                      counters.c_duplicate++;
                      status = S_STATUS_DUPLICATE;
                   }

src/test.c

@@ -78,7 +78,7 @@ int main(int argc, char **argv){
    printf("post parsing...\n");
    post_parse(&sdata, &state, &cfg);
 
-   printf("message-id: %s\n", state.message_id);
+   printf("message-id: %s / %s\n", state.message_id, state.message_id_hash);
    printf("from: *%s (%s)*\n", state.b_from, state.b_from_domain);
    printf("to: *%s (%s)*\n", state.b_to, state.b_to_domain);
    printf("reference: *%s*\n", state.reference);