smtp input

7 files changed, 76 insertions, 58 deletions

diff --git a/src/src/dns.c b/src/src/dns.c
index a2f430993..f1619f4a4 100644
--- a/src/src/dns.c
+++ b/src/src/dns.c
@@ -576,6 +576,23 @@ if (previous != NULL)
   return previous->data.val;
   }
 
+#ifdef EXPERIMENTAL_INTERNATIONAL
+/* Convert all names to a-label form before doing lookup */
+  {
+  uschar * alabel;
+  uschar * errstr = NULL;
+  if ((alabel = string_domain_utf8_to_alabel(name, &errstr)), errstr)
+    {
+    DEBUG(D_dns)
+      debug_printf("DNS name '%s' utf8 conversion to alabel failed: %s", name,
+        errstr);
+    host_find_failed_syntax = TRUE;
+    return DNS_NOMATCH;
+    }
+  name = alabel;
+  }
+#endif
+
 /* If configured, check the hygene of the name passed to lookup. Otherwise,
 although DNS lookups may give REFUSED at the lower level, some resolvers
 turn this into TRY_AGAIN, which is silly. Give a NOMATCH return, since such
diff --git a/src/src/exim.c b/src/src/exim.c
index f6d9be6ef..121c6c2e3 100644
--- a/src/src/exim.c
+++ b/src/src/exim.c
@@ -3705,6 +3705,10 @@ is equivalent to the ability to modify a setuid binary!
 This needs to happen before we read the main configuration. */
 init_lookup_list();
 
+#ifdef EXPERIMENTAL_INTERNATIONAL
+if (running_in_test_harness) smtputf8_advertise_hosts = NULL;
+#endif
+
 /* Read the main runtime configuration data; this gives up if there
 is a failure. It leaves the configuration file open so that the subsequent
 configuration data for delivery can be read if needed. */
diff --git a/src/src/functions.h b/src/src/functions.h
index ac93c1635..fdd629228 100644
--- a/src/src/functions.h
+++ b/src/src/functions.h
@@ -422,6 +422,7 @@ extern const uschar *string_printing2(const uschar *, BOOL);
 extern uschar *string_split_message(uschar *);
 extern uschar *string_unprinting(uschar *);
 #ifdef EXPERIMENTAL_INTERNATIONAL
+extern uschar *string_address_utf8_to_alabel(uschar *, uschar **, int *);
 extern uschar *string_domain_alabel_to_utf8(const uschar *, uschar **);
 extern uschar *string_domain_utf8_to_alabel(const uschar *, uschar **);
 extern uschar *string_localpart_alabel_to_utf8(const uschar *, uschar **);
diff --git a/src/src/globals.c b/src/src/globals.c
index cb93a0192..2cbafcdce 100644
--- a/src/src/globals.c
+++ b/src/src/globals.c
@@ -1274,7 +1274,7 @@ int     smtp_rlr_threshold     = INT_MAX;
 BOOL    smtp_use_pipelining    = FALSE;
 BOOL    smtp_use_size          = FALSE;
 #ifdef EXPERIMENTAL_INTERNATIONAL
-uschar *smtputf8_advertise_hosts = US"*";
+uschar *smtputf8_advertise_hosts = US"*";	/* overridden under test-harness */
 #endif
 
 #ifdef WITH_CONTENT_SCAN
diff --git a/src/src/parse.c b/src/src/parse.c
index ff814e738..a648f755a 100644
--- a/src/src/parse.c
+++ b/src/src/parse.c
@@ -550,9 +550,7 @@ read_addr_spec(uschar *s, uschar *t, int term, uschar **errorptr,
 {
 s = read_local_part(s, t, errorptr, FALSE);
 if (*errorptr == NULL)
-  {
   if (*s != term)
-    {
     if (*s != '@')
       *errorptr = string_sprintf("\"@\" or \".\" expected after \"%s\"", t);
     else
@@ -562,8 +560,6 @@ if (*errorptr == NULL)
       *domainptr = t;
       s = read_domain(s, t, errorptr);
       }
-    }
-  }
 return s;
 }
 
@@ -744,7 +740,7 @@ if (*s == '<')
   while (bracket_count-- > 0) if (*s++ != '>')
     {
     *errorptr = (s[-1] == 0)? US"'>' missing at end of address" :
-      string_sprintf("malformed address: %.32s may not follow %.*s",
+      string_sprintf("malformed address A: %.32s may not follow %.*s",
         s-1, s - (uschar *)mailbox - 1, mailbox);
     goto PARSE_FAILED;
     }
@@ -797,7 +793,7 @@ if (*s != 0)
     }
   else
     {
-    *errorptr = string_sprintf("malformed address: %.32s may not follow %.*s",
+    *errorptr = string_sprintf("malformed address B: %.32s may not follow %.*s",
       s, s - (uschar *)mailbox, mailbox);
     goto PARSE_FAILED;
     }
@@ -817,7 +813,7 @@ if (*end - *start > ADDRESS_MAXLENGTH)
   return NULL;
   }
 
-return (uschar *)yield;
+return yield;
 
 /* Use goto (via the macro FAILED) to get to here from a variety of places.
 We might have an empty address in a group - the caller can choose to ignore
diff --git a/src/src/smtp_in.c b/src/src/smtp_in.c
index 8086e9456..a0e44d89a 100644
--- a/src/src/smtp_in.c
+++ b/src/src/smtp_in.c
@@ -3828,10 +3828,8 @@ while (done <= 0)
            (char *)mail_args < (char *)env_mail_type_list + sizeof(env_mail_type_list);
            mail_args++
           )
-        {
         if (strcmpic(name, mail_args->name) == 0)
           break;
-        }
       if (mail_args->need_value && strcmpic(value, US"") == 0)
         break;
 
@@ -3859,16 +3857,17 @@ while (done <= 0)
         and "7BIT" as body types, but take no action. */
         case ENV_MAIL_OPT_BODY:
           if (accept_8bitmime) {
-            if (strcmpic(value, US"8BITMIME") == 0) {
+            if (strcmpic(value, US"8BITMIME") == 0)
               body_8bitmime = 8;
-            } else if (strcmpic(value, US"7BIT") == 0) {
+            else if (strcmpic(value, US"7BIT") == 0)
               body_8bitmime = 7;
-            } else {
+            else
+	      {
               body_8bitmime = 0;
               done = synprot_error(L_smtp_syntax_error, 501, NULL,
                 US"invalid data for BODY");
               goto COMMAND_LOOP;
-            }
+              }
             DEBUG(D_receive) debug_printf("8BITMIME: %d\n", body_8bitmime);
 	    break;
           }
@@ -3880,35 +3879,43 @@ while (done <= 0)
         is included only if configured in at build time. */
 
         case ENV_MAIL_OPT_RET:
-          if (dsn_advertised) {
+          if (dsn_advertised)
+	    {
             /* Check if RET has already been set */
-            if (dsn_ret > 0) {
+            if (dsn_ret > 0)
+	      {
               synprot_error(L_smtp_syntax_error, 501, NULL,
                 US"RET can be specified once only");
               goto COMMAND_LOOP;
-            }
-            dsn_ret = (strcmpic(value, US"HDRS") == 0)? dsn_ret_hdrs :
-                    (strcmpic(value, US"FULL") == 0)? dsn_ret_full : 0;
+	      }
+            dsn_ret = strcmpic(value, US"HDRS") == 0
+	      ? dsn_ret_hdrs
+	      : strcmpic(value, US"FULL") == 0
+	      ? dsn_ret_full
+	      : 0;
             DEBUG(D_receive) debug_printf("DSN_RET: %d\n", dsn_ret);
             /* Check for invalid invalid value, and exit with error */
-            if (dsn_ret == 0) {
+            if (dsn_ret == 0)
+	      {
               synprot_error(L_smtp_syntax_error, 501, NULL,
                 US"Value for RET is invalid");
               goto COMMAND_LOOP;
-            }
-          }
+	      }
+	    }
           break;
         case ENV_MAIL_OPT_ENVID:
-          if (dsn_advertised) {
+          if (dsn_advertised)
+	    {
             /* Check if the dsn envid has been already set */
-            if (dsn_envid != NULL) {
+            if (dsn_envid != NULL)
+	      {
               synprot_error(L_smtp_syntax_error, 501, NULL,
                 US"ENVID can be specified once only");
               goto COMMAND_LOOP;
-            }
+	      }
             dsn_envid = string_copy(value);
             DEBUG(D_receive) debug_printf("DSN_ENVID: %s\n", dsn_envid);
-          }
+	    }
           break;
 
         /* Handle the AUTH extension. If the value given is not "<>" and either
@@ -3948,34 +3955,34 @@ while (done <= 0)
             switch (rc)
               {
               case OK:
-              if (authenticated_by == NULL ||
-                  authenticated_by->mail_auth_condition == NULL ||
-                  expand_check_condition(authenticated_by->mail_auth_condition,
-                      authenticated_by->name, US"authenticator"))
-                break;     /* Accept the AUTH */
-  
-              ignore_msg = US"server_mail_auth_condition failed";
-              if (authenticated_id != NULL)
-                ignore_msg = string_sprintf("%s: authenticated ID=\"%s\"",
-                  ignore_msg, authenticated_id);
+		if (authenticated_by == NULL ||
+		    authenticated_by->mail_auth_condition == NULL ||
+		    expand_check_condition(authenticated_by->mail_auth_condition,
+			authenticated_by->name, US"authenticator"))
+		  break;     /* Accept the AUTH */
+
+		ignore_msg = US"server_mail_auth_condition failed";
+		if (authenticated_id != NULL)
+		  ignore_msg = string_sprintf("%s: authenticated ID=\"%s\"",
+		    ignore_msg, authenticated_id);
   
               /* Fall through */
   
               case FAIL:
-              authenticated_sender = NULL;
-              log_write(0, LOG_MAIN, "ignoring AUTH=%s from %s (%s)",
-                value, host_and_ident(TRUE), ignore_msg);
-              break;
+		authenticated_sender = NULL;
+		log_write(0, LOG_MAIN, "ignoring AUTH=%s from %s (%s)",
+		  value, host_and_ident(TRUE), ignore_msg);
+		break;
   
               /* Should only get DEFER or ERROR here. Put back terminator
               overrides for error message */
   
               default:
-              value[-1] = '=';
-              name[-1] = ' ';
-              (void)smtp_handle_acl_fail(ACL_WHERE_MAILAUTH, rc, user_msg,
-                log_msg);
-              goto COMMAND_LOOP;
+		value[-1] = '=';
+		name[-1] = ' ';
+		(void)smtp_handle_acl_fail(ACL_WHERE_MAILAUTH, rc, user_msg,
+		  log_msg);
+		goto COMMAND_LOOP;
               }
             }
             break;
@@ -3990,7 +3997,7 @@ while (done <= 0)
 #ifdef EXPERIMENTAL_INTERNATIONAL
         case ENV_MAIL_OPT_UTF8:
 	  if (smtputf8_advertised)
-	    message_smtputf8 = TRUE;
+	    message_smtputf8 = allow_utf8_domains = TRUE;
 	  break;
 #endif
         /* Unknown option. Stick back the terminator characters and break
@@ -4025,9 +4032,10 @@ while (done <= 0)
     /* Now extract the address, first applying any SMTP-time rewriting. The
     TRUE flag allows "<>" as a sender address. */
 
-    raw_sender = ((rewrite_existflags & rewrite_smtp) != 0)?
-      rewrite_one(smtp_cmd_data, rewrite_smtp, NULL, FALSE, US"",
-        global_rewrite_rules) : smtp_cmd_data;
+    raw_sender = rewrite_existflags & rewrite_smtp
+      ? rewrite_one(smtp_cmd_data, rewrite_smtp, NULL, FALSE, US"",
+		    global_rewrite_rules)
+      : smtp_cmd_data;
 
     /* rfc821_domains = TRUE; << no longer needed */
     raw_sender =
diff --git a/src/src/utf8.c b/src/src/utf8.c
index 9a2b8656e..32d2eaed0 100644
--- a/src/src/utf8.c
+++ b/src/src/utf8.c
@@ -78,9 +78,6 @@ size_t p_len = ucs4_len*4;	/* this multiplier is pure guesswork */
 uschar * res = store_get(p_len+5);
 int rc;
 
-DEBUG(D_expand) debug_printf("l_u2a: ulen %d  plen %d\n", ucs4_len, p_len);
-DEBUG(D_expand) for (rc = 0; rc < ucs4_len; rc++) debug_printf("%08x ", p[rc]);
-
 res[0] = 'x'; res[1] = 'n'; res[2] = res[3] = '-';
 
 if ((rc = punycode_encode(ucs4_len, p, NULL, &p_len, res+4)) != PUNYCODE_SUCCESS)
@@ -90,10 +87,7 @@ if ((rc = punycode_encode(ucs4_len, p, NULL, &p_len, res+4)) != PUNYCODE_SUCCESS
   if (err) *err = US punycode_strerror(rc);
   return NULL;
   }
-DEBUG(D_expand) debug_printf("l_u2a: plen %d\n", p_len);
 p_len += 4;
-DEBUG(D_expand) for (rc = 0; rc < p_len; rc++) debug_printf("%02x ", res[rc]);
-DEBUG(D_expand) debug_printf("\n");
 free(p);
 res[p_len] = '\0';
 return res;
@@ -114,9 +108,8 @@ if (alabel[0] != 'x' || alabel[1] != 'n' || alabel[2] != '-' || alabel[3] != '-'
   if (err) *err = US"bad alabel prefix";
   return NULL;
   }
-p_len -= 4;
-DEBUG(D_expand) debug_printf("l_a2u: plen %d\n", p_len);
 
+p_len -= 4;
 p = (punycode_uint *) store_get((p_len+1) * sizeof(*p));
 
 if ((rc = punycode_decode(p_len, CCS alabel+4, &p_len, p, NULL)) != PUNYCODE_SUCCESS)
@@ -124,7 +117,6 @@ if ((rc = punycode_decode(p_len, CCS alabel+4, &p_len, p, NULL)) != PUNYCODE_SUC
   if (err) *err = US punycode_strerror(rc);
   return NULL;
   }
-DEBUG(D_expand) debug_printf("l_a2u: dlen %d\n", p_len);
 
 s = stringprep_ucs4_to_utf8(p, p_len, NULL, &p_len);
 res = string_copyn(s, p_len);