--- CBOR-XS/XS.xs	2013/11/20 02:03:09	1.19
+++ CBOR-XS/XS.xs	2013/11/20 11:06:42	1.20
@@ -137,7 +137,8 @@
   SV *sv;     // result scalar
   CBOR cbor;
   U32 depth;  // recursion level
-  HV *stringref; // string => index, or 0
+  HV *stringref[2]; // string => index, or 0 ([0] = bytes, [1] = utf-8)
+  UV stringref_idx;
   HV *shareable; // ptr => index, or 0
   UV shareable_idx;
 } enc_t;
@@ -239,6 +240,15 @@
   --enc->depth;
 }
 
+ecb_inline void
+encode_he (enc_t *enc, HE *he)
+{
+  if (HeKLEN (he) == HEf_SVKEY)
+    encode_sv (enc, HeSVKEY (he));
+  else
+    encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
+}
+
 static void
 encode_hv (enc_t *enc, HV *hv)
 {
@@ -259,10 +269,29 @@
 
   while ((he = hv_iternext (hv)))
     {
-      if (HeKLEN (he) == HEf_SVKEY)
-        encode_sv (enc, HeSVKEY (he));
+      if (ecb_expect_false (enc->cbor.flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS)))
+        {
+          SV **svp;
+
+          if (HeKLEN (he) == HEf_SVKEY)
+            svp = hv_fetch_ent (enc->stringref[!! SvUTF8 (HeSVKEY (he))], HeSVKEY (he)           , 1, 0);//TODO return HE :/
+          else
+            svp = hv_fetch     (enc->stringref[!! HeKUTF8 (he)         ], HeKEY (he), HeKLEN (he), 1);
+
+          if (SvOK (*svp))
+            {
+              encode_tag (enc, CBOR_TAG_STRINGREF);
+              encode_uint (enc, 0x00, SvUV (*svp));
+            }
+          else
+            {
+              sv_setuv (*svp, enc->stringref_idx);
+              ++enc->stringref_idx;
+              encode_he (enc, he);
+            }
+        }
       else
-        encode_str (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he));
+        encode_he (enc, he);
 
       encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
     }
@@ -477,6 +506,14 @@
   enc.end       = SvEND (enc.sv);
 
   SvPOK_only (enc.sv);
+
+  if (cbor->flags & (F_DEDUP_STRINGS | F_DEDUP_KEYS))
+    {
+      encode_tag (&enc, CBOR_TAG_STRINGREF_NAMESPACE);
+      enc.stringref[0]= (HV *)sv_2mortal ((SV *)newHV ());
+      enc.stringref[1]= (HV *)sv_2mortal ((SV *)newHV ());
+    }
+
   encode_sv (&enc, scalar);
 
   SvCUR_set (enc.sv, enc.cur - SvPVX (enc.sv));
@@ -501,6 +538,7 @@
   U32 depth; // recursion depth
   U32 maxdepth; // recursion depth limit
   AV *shareable;
+  AV *stringref;
 } dec_t;
 
 #define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE
@@ -617,6 +655,9 @@
 
       dec->cur += len;
 
+      if (ecb_expect_false (dec->stringref))
+        av_push (dec->stringref, newSVpvn (key, len));
+
       hv_store (hv, key,  len, decode_sv (dec), 0);
     }
   else if (*dec->cur >= 0x60 && *dec->cur <= 0x60 + 27)
@@ -626,6 +667,9 @@
 
       dec->cur += len;
 
+      if (ecb_expect_false (dec->stringref))
+        av_push (dec->stringref, newSVpvn_utf8 (key, len, 1));
+
       hv_store (hv, key, -len, decode_sv (dec), 0);
     }
   else
@@ -716,6 +760,9 @@
   if (utf8)
     SvUTF8_on (sv);
 
+  if (ecb_expect_false (dec->stringref))
+    av_push (dec->stringref, SvREFCNT_inc_NN (sv));
+
   return sv;
 
 fail:
@@ -734,10 +781,39 @@
   switch (tag)
     {
       case CBOR_TAG_MAGIC:
-        return decode_sv (dec);
+        sv = decode_sv (dec);
+        break;
 
       case CBOR_TAG_INDIRECTION:
-        return newRV_noinc (decode_sv (dec));
+        sv = newRV_noinc (decode_sv (dec));
+        break;
+
+      case CBOR_TAG_STRINGREF_NAMESPACE:
+        {
+          ENTER; SAVETMPS;
+
+          SAVESPTR (dec->stringref);
+          dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
+
+          sv = decode_sv (dec);
+
+          FREETMPS; LEAVE;
+        }
+        break;
+
+      case CBOR_TAG_STRINGREF:
+        {
+          if ((*dec->cur >> 5) != 0)
+            ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
+
+          UV idx = decode_uint (dec);
+
+          if (!dec->stringref || (int)idx > AvFILLp (dec->stringref))
+            ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
+
+          sv = newSVsv (AvARRAY (dec->stringref)[idx]);
+        }
+        break;
 
       case CBOR_TAG_VALUE_SHAREABLE:
         {
@@ -751,8 +827,7 @@
           sv_setsv (sv, osv);
           SvREFCNT_dec_NN (osv);
         }
-
-        return sv;
+        break;
 
       case CBOR_TAG_VALUE_SHAREDREF:
         {
@@ -761,11 +836,12 @@
 
           UV idx = decode_uint (dec);
 
-          if (!dec->shareable || idx > AvFILLp (dec->shareable))
+          if (!dec->shareable || (int)idx > AvFILLp (dec->shareable))
             ERR ("corrupted CBOR data (sharedref index out of bounds)");
 
-          return SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
+          sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
         }
+        break;
 
       case CBOR_TAG_PERL_OBJECT:
         {
@@ -815,9 +891,8 @@
           PUTBACK;
 
           FREETMPS; LEAVE;
-
-          return sv;
         }
+        break;
 
       default:
         {
@@ -831,10 +906,13 @@
                               ? cbor_tagged_stash
                               : gv_stashpv ("CBOR::XS::Tagged" , 1);
 
-          return sv_bless (newRV_noinc ((SV *)av), tagged_stash);
+          sv = sv_bless (newRV_noinc ((SV *)av), tagged_stash);
         }
+        break;
     }
 
+  return sv;
+
 fail:
   SvREFCNT_dec (sv);
   return &PL_sv_undef;