A-label expansion operators

2 files changed, 97 insertions, 10 deletions

diff --git a/src/src/expand.c b/src/src/expand.c
index b613ef2b4..ad97f6fef 100644
--- a/src/src/expand.c
+++ b/src/src/expand.c
@@ -168,7 +168,14 @@ static uschar *op_table_underscore[] = {
   US"quote_local_part",
   US"reverse_ip",
   US"time_eval",
-  US"time_interval"};
+  US"time_interval"
+#ifdef EXPERIMENTAL_INTERNATIONAL
+ ,US"utf8_domain_from_alabel",
+  US"utf8_domain_to_alabel",
+  US"utf8_localpart_from_alabel",
+  US"utf8_localpart_to_alabel"
+#endif
+  };
 
 enum {
   EOP_FROM_UTF8,
@@ -176,7 +183,14 @@ enum {
   EOP_QUOTE_LOCAL_PART,
   EOP_REVERSE_IP,
   EOP_TIME_EVAL,
-  EOP_TIME_INTERVAL };
+  EOP_TIME_INTERVAL
+#ifdef EXPERIMENTAL_INTERNATIONAL
+ ,EOP_UTF8_DOMAIN_FROM_ALABEL,
+  EOP_UTF8_DOMAIN_TO_ALABEL,
+  EOP_UTF8_LOCALPART_FROM_ALABEL,
+  EOP_UTF8_LOCALPART_TO_ALABEL
+#endif
+  };
 
 static uschar *op_table_main[] = {
   US"address",
@@ -6555,16 +6569,13 @@ while (*s != 0)
 	  if (bytes_left)
 	    {
 	    if ((c & 0xc0) != 0x80)
-	      {
 		    /* wrong continuation byte; invalidate all bytes */
 	      complete = 1; /* error */
-	      }
 	    else
 	      {
 	      codepoint = (codepoint << 6) | (c & 0x3f);
 	      seq_buff[index++] = c;
 	      if (--bytes_left == 0)		/* codepoint complete */
-		{
 		if(codepoint > 0x10FFFF)	/* is it too large? */
 		  complete = -1;	/* error (RFC3629 limit) */
 		else
@@ -6572,7 +6583,6 @@ while (*s != 0)
 		  yield = string_cat(yield, &size, &ptr, seq_buff, seq_len);
 		  index = 0;
 		  }
-		}
 	      }
 	    }
 	  else	/* no bytes left: new sequence */
@@ -6615,13 +6625,75 @@ while (*s != 0)
 	    yield = string_cat(yield, &size, &ptr, UTF8_REPLACEMENT_CHAR, 1);
 	    }
 	  if ((complete == 1) && ((c & 0x80) == 0))
-	    { /* ASCII character follows incomplete sequence */
+			/* ASCII character follows incomplete sequence */
 	      yield = string_cat(yield, &size, &ptr, &c, 1);
-	    }
 	  }
         continue;
         }
 
+#ifdef EXPERIMENTAL_INTERNATIONAL
+      case EOP_UTF8_DOMAIN_TO_ALABEL:
+	{
+        uschar * error = NULL;
+	uschar * s = string_domain_utf8_to_alabel(sub, &error);
+	if (error)
+	  {
+	  expand_string_message = string_sprintf(
+	    "error converting utf8 (%s) to alabel: %s",
+	    string_printing(sub), error);
+	  goto EXPAND_FAILED;
+	  }
+	yield = string_cat(yield, &size, &ptr, s, Ustrlen(s));
+        continue;
+	}
+
+      case EOP_UTF8_DOMAIN_FROM_ALABEL:
+	{
+        uschar * error = NULL;
+	uschar * s = string_domain_alabel_to_utf8(sub, &error);
+	if (error)
+	  {
+	  expand_string_message = string_sprintf(
+	    "error converting alabel (%s) to utf8: %s",
+	    string_printing(sub), error);
+	  goto EXPAND_FAILED;
+	  }
+	yield = string_cat(yield, &size, &ptr, s, Ustrlen(s));
+        continue;
+	}
+
+      case EOP_UTF8_LOCALPART_TO_ALABEL:
+	{
+        uschar * error = NULL;
+	uschar * s = string_localpart_utf8_to_alabel(sub, &error);
+	if (error)
+	  {
+	  expand_string_message = string_sprintf(
+	    "error converting utf8 (%s) to alabel: %s",
+	    string_printing(sub), error);
+	  goto EXPAND_FAILED;
+	  }
+	yield = string_cat(yield, &size, &ptr, s, Ustrlen(s));
+	DEBUG(D_expand) debug_printf("yield: '%s'\n", yield);
+        continue;
+	}
+
+      case EOP_UTF8_LOCALPART_FROM_ALABEL:
+	{
+        uschar * error = NULL;
+	uschar * s = string_localpart_alabel_to_utf8(sub, &error);
+	if (error)
+	  {
+	  expand_string_message = string_sprintf(
+	    "error converting alabel (%s) to utf8: %s",
+	    string_printing(sub), error);
+	  goto EXPAND_FAILED;
+	  }
+	yield = string_cat(yield, &size, &ptr, s, Ustrlen(s));
+        continue;
+	}
+#endif	/* EXPERIMENTAL_INTERNATIONAL */
+
       /* escape turns all non-printing characters into escape sequences. */
 
       case EOP_ESCAPE:
diff --git a/src/src/utf8.c b/src/src/utf8.c
index 2f8173dc1..9a2b8656e 100644
--- a/src/src/utf8.c
+++ b/src/src/utf8.c
@@ -78,14 +78,22 @@ 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)
   {
+  DEBUG(D_expand) debug_printf("l_u2a: bad '%s'\n", punycode_strerror(rc));
   free(p);
   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;
@@ -97,6 +105,8 @@ string_localpart_alabel_to_utf8(const uschar * alabel, uschar ** err)
 {
 size_t p_len = strlen(alabel);
 punycode_uint * p;
+uschar * s;
+uschar * res;
 int rc;
 
 if (alabel[0] != 'x' || alabel[1] != 'n' || alabel[2] != '-' || alabel[3] != '-')
@@ -105,6 +115,7 @@ if (alabel[0] != 'x' || alabel[1] != 'n' || alabel[2] != '-' || alabel[3] != '-'
   return NULL;
   }
 p_len -= 4;
+DEBUG(D_expand) debug_printf("l_a2u: plen %d\n", p_len);
 
 p = (punycode_uint *) store_get((p_len+1) * sizeof(*p));
 
@@ -113,8 +124,12 @@ if ((rc = punycode_decode(p_len, CCS alabel+4, &p_len, p, NULL)) != PUNYCODE_SUC
   if (err) *err = US punycode_strerror(rc);
   return NULL;
   }
-p[p_len] = 0;
-return US p;
+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);
+free(s);
+return res;
 }