msgpack 1.2.9 → 1.4.4

data/ext/msgpack/buffer.c

@@ -23,11 +23,11 @@
 static ID s_replace;
 #endif
 
-#ifdef COMPAT_HAVE_ENCODING  /* see compat.h*/
 int msgpack_rb_encindex_utf8;
 int msgpack_rb_encindex_usascii;
 int msgpack_rb_encindex_ascii8bit;
-#endif
+
+ID s_uminus;
 
 #ifndef DISABLE_RMEM
 static msgpack_rmem_t s_rmem;
@@ -35,11 +35,11 @@ static msgpack_rmem_t s_rmem;
 
 void msgpack_buffer_static_init()
 {
-#ifdef COMPAT_HAVE_ENCODING
+    s_uminus = rb_intern("-@");
+
     msgpack_rb_encindex_utf8 = rb_utf8_encindex();
     msgpack_rb_encindex_usascii = rb_usascii_encindex();
     msgpack_rb_encindex_ascii8bit = rb_ascii8bit_encindex();
-#endif
 
 #ifndef DISABLE_RMEM
     msgpack_rmem_init(&s_rmem);
@@ -164,12 +164,7 @@ size_t msgpack_buffer_read_to_string_nonblock(msgpack_buffer_t* b, VALUE string,
             b->head->mapped_string != NO_MAPPED_STRING &&
             length >= b->read_reference_threshold) {
         VALUE s = _msgpack_buffer_refer_head_mapped_string(b, length);
-#ifndef HAVE_RB_STR_REPLACE
-        /* TODO MRI 1.8 */
-        rb_funcall(string, s_replace, 1, s);
-#else
         rb_str_replace(string, s);
-#endif
         /* here doesn't have to call ENCODING_SET because
          * encoding of s is always ASCII-8BIT */
         _msgpack_buffer_consumed(b, length);
@@ -308,9 +303,7 @@ static inline void _msgpack_buffer_add_new_chunk(msgpack_buffer_t* b)
 static inline void _msgpack_buffer_append_reference(msgpack_buffer_t* b, VALUE string)
 {
     VALUE mapped_string = rb_str_dup(string);
-#ifdef COMPAT_HAVE_ENCODING
     ENCODING_SET(mapped_string, msgpack_rb_encindex_ascii8bit);
-#endif
 
     _msgpack_buffer_add_new_chunk(b);
 
@@ -337,7 +330,6 @@ void _msgpack_buffer_append_long_string(msgpack_buffer_t* b, VALUE string)
 
     if(b->io != Qnil) {
         msgpack_buffer_flush(b);
-#ifdef COMPAT_HAVE_ENCODING
         if (ENCODING_GET(string) == msgpack_rb_encindex_ascii8bit) {
             rb_funcall(b->io, b->io_write_all_method, 1, string);
         } else if(!STR_DUP_LIKELY_DOES_COPY(string)) {
@@ -347,10 +339,6 @@ void _msgpack_buffer_append_long_string(msgpack_buffer_t* b, VALUE string)
         } else {
             msgpack_buffer_append(b, RSTRING_PTR(string), length);
         }
-#else
-        rb_funcall(b->io, b->io_write_all_method, 1, string);
-#endif
-
     } else if(!STR_DUP_LIKELY_DOES_COPY(string)) {
         _msgpack_buffer_append_reference(b, string);
 

data/ext/msgpack/buffer.h

@@ -49,11 +49,11 @@
 
 #define NO_MAPPED_STRING ((VALUE)0)
 
-#ifdef COMPAT_HAVE_ENCODING  /* see compat.h*/
 extern int msgpack_rb_encindex_utf8;
 extern int msgpack_rb_encindex_usascii;
 extern int msgpack_rb_encindex_ascii8bit;
-#endif
+
+extern ID s_uminus;
 
 struct msgpack_buffer_chunk_t;
 typedef struct msgpack_buffer_chunk_t msgpack_buffer_chunk_t;
@@ -262,6 +262,20 @@ static inline size_t msgpack_buffer_append_string(msgpack_buffer_t* b, VALUE str
     return length;
 }
 
+static inline size_t msgpack_buffer_append_string_reference(msgpack_buffer_t* b, VALUE string)
+{
+    size_t length = RSTRING_LEN(string);
+
+    if(length > MSGPACK_BUFFER_STRING_WRITE_REFERENCE_MINIMUM) {
+        _msgpack_buffer_append_long_string(b, string);
+
+    } else {
+        msgpack_buffer_append(b, RSTRING_PTR(string), length);
+    }
+
+    return length;
+}
+
 
 /*
  * IO functions
@@ -424,7 +438,7 @@ static inline VALUE _msgpack_buffer_refer_head_mapped_string(msgpack_buffer_t* b
     return rb_str_substr(b->head->mapped_string, offset, length);
 }
 
-static inline VALUE msgpack_buffer_read_top_as_string(msgpack_buffer_t* b, size_t length, bool will_be_frozen)
+static inline VALUE msgpack_buffer_read_top_as_string(msgpack_buffer_t* b, size_t length, bool will_be_frozen, bool utf8)
 {
 #ifndef DISABLE_BUFFER_READ_REFERENCE_OPTIMIZE
     /* optimize */
@@ -432,16 +446,57 @@ static inline VALUE msgpack_buffer_read_top_as_string(msgpack_buffer_t* b, size_
             b->head->mapped_string != NO_MAPPED_STRING &&
             length >= b->read_reference_threshold) {
         VALUE result = _msgpack_buffer_refer_head_mapped_string(b, length);
+        if (utf8) ENCODING_SET(result, msgpack_rb_encindex_utf8);
         _msgpack_buffer_consumed(b, length);
         return result;
     }
 #endif
 
-    VALUE result = rb_str_new(b->read_buffer, length);
+    VALUE result;
+
+#ifdef HAVE_RB_ENC_INTERNED_STR
+    if (will_be_frozen) {
+        result = rb_enc_interned_str(b->read_buffer, length, utf8 ? rb_utf8_encoding() : rb_ascii8bit_encoding());
+    } else {
+        if (utf8) {
+            result = rb_utf8_str_new(b->read_buffer, length);
+        } else {
+            result = rb_str_new(b->read_buffer, length);
+        }
+    }
     _msgpack_buffer_consumed(b, length);
     return result;
+
+#else
+
+    if (utf8) {
+        result = rb_utf8_str_new(b->read_buffer, length);
+    } else {
+        result = rb_str_new(b->read_buffer, length);
+    }
+
+#if STR_UMINUS_DEDUPE
+    if (will_be_frozen) {
+#if STR_UMINUS_DEDUPE_FROZEN
+        // Starting from MRI 2.8 it is preferable to freeze the string
+        // before deduplication so that it can be interned directly
+        // otherwise it would be duplicated first which is wasteful.
+        rb_str_freeze(result);
+#endif //STR_UMINUS_DEDUPE_FROZEN
+        // MRI 2.5 and older do not deduplicate strings that are already
+        // frozen.
+        result = rb_funcall(result, s_uminus, 0);
+    }
+#endif // STR_UMINUS_DEDUPE
+    _msgpack_buffer_consumed(b, length);
+    return result;
+
+#endif // HAVE_RB_ENC_INTERNED_STR
 }
 
+static inline VALUE msgpack_buffer_read_top_as_symbol(msgpack_buffer_t* b, size_t length)
+{
+    return rb_str_intern(msgpack_buffer_read_top_as_string(b, length, true, false));
+}
 
 #endif
-

data/ext/msgpack/compat.h

@@ -20,6 +20,7 @@
 
 #include <stdbool.h>
 #include "ruby.h"
+#include "ruby/encoding.h"
 
 #if defined(HAVE_RUBY_ST_H)
 #  include "ruby/st.h"  /* ruby hash on Ruby 1.9 */
@@ -38,18 +39,6 @@
 #  define ZALLOC_N(type,n) RB_ZALLOC_N(type,n)
 #endif
 
-/*
- * COMPAT_HAVE_ENCODING
- */
-#ifdef HAVE_RUBY_ENCODING_H
-#  include "ruby/encoding.h"
-#  define COMPAT_HAVE_ENCODING
-#endif
-
-#if defined(__MACRUBY__)  /* MacRuby */
-#  undef COMPAT_HAVE_ENCODING
-#endif
-
 
 /*
  * define STR_DUP_LIKELY_DOES_COPY

data/ext/msgpack/extconf.rb

@@ -2,13 +2,7 @@ require 'mkmf'
 
 have_header("ruby/st.h")
 have_header("st.h")
-have_func("rb_str_replace", ["ruby.h"])
-have_func("rb_intern_str", ["ruby.h"])
-have_func("rb_sym2str", ["ruby.h"])
-have_func("rb_str_intern", ["ruby.h"])
-have_func("rb_block_lambda", ["ruby.h"])
-have_func("rb_hash_dup", ["ruby.h"])
-have_func("rb_hash_clear", ["ruby.h"])
+have_func("rb_enc_interned_str", "ruby.h")
 
 unless RUBY_PLATFORM.include? 'mswin'
   $CFLAGS << %[ -I.. -Wall -O3 -g -std=gnu99]
@@ -25,6 +19,44 @@ if defined?(RUBY_ENGINE) && RUBY_ENGINE == 'rbx'
   $CFLAGS << %[ -DDISABLE_RMEM]
 end
 
+# checking if Hash#[]= (rb_hash_aset) dedupes string keys
+h = {}
+x = {}
+r = rand.to_s
+h[%W(#{r}).join('')] = :foo
+x[%W(#{r}).join('')] = :foo
+if x.keys[0].equal?(h.keys[0])
+  $CFLAGS << ' -DHASH_ASET_DEDUPE=1 '
+else
+  $CFLAGS << ' -DHASH_ASET_DEDUPE=0 '
+end
+
+
+# checking if String#-@ (str_uminus) dedupes... '
+begin
+  a = -(%w(t e s t).join)
+  b = -(%w(t e s t).join)
+  if a.equal?(b)
+    $CFLAGS << ' -DSTR_UMINUS_DEDUPE=1 '
+  else
+    $CFLAGS += ' -DSTR_UMINUS_DEDUPE=0 '
+  end
+rescue NoMethodError
+  $CFLAGS << ' -DSTR_UMINUS_DEDUPE=0 '
+end
+
+# checking if String#-@ (str_uminus) directly interns frozen strings... '
+begin
+  s = rand.to_s.freeze
+  if (-s).equal?(s) && (-s.dup).equal?(s)
+    $CFLAGS << ' -DSTR_UMINUS_DEDUPE_FROZEN=1 '
+  else
+    $CFLAGS << ' -DSTR_UMINUS_DEDUPE_FROZEN=0 '
+  end
+rescue NoMethodError
+  $CFLAGS << ' -DSTR_UMINUS_DEDUPE_FROZEN=0 '
+end
+
 if warnflags = CONFIG['warnflags']
   warnflags.slice!(/ -Wdeclaration-after-statement/)
 end

data/ext/msgpack/factory_class.c

@@ -32,6 +32,8 @@ struct msgpack_factory_t {
     msgpack_packer_ext_registry_t pkrg;
     msgpack_unpacker_ext_registry_t ukrg;
     bool has_symbol_ext_type;
+    bool optimized_symbol_ext_type;
+    int symbol_ext_type;
 };
 
 #define FACTORY(from, name) \
@@ -114,6 +116,8 @@ VALUE MessagePack_Factory_unpacker(int argc, VALUE* argv, VALUE self)
 
     msgpack_unpacker_ext_registry_destroy(&uk->ext_registry);
     msgpack_unpacker_ext_registry_dup(&fc->ukrg, &uk->ext_registry);
+    uk->optimized_symbol_ext_type = fc->optimized_symbol_ext_type;
+    uk->symbol_ext_type = fc->symbol_ext_type;
 
     return unpacker;
 }
@@ -128,11 +132,7 @@ static VALUE Factory_registered_types_internal(VALUE self)
             rb_hash_aset(uk_mapping, INT2FIX(i - 128), fc->ukrg.array[i]);
         }
     }
-#ifdef HAVE_RB_HASH_DUP
     return rb_ary_new3(2, rb_hash_dup(fc->pkrg.hash), uk_mapping);
-#else
-    return rb_ary_new3(2, rb_funcall(fc->pkrg.hash, rb_intern("dup"), 0), uk_mapping);
-#endif
 }
 
 static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
@@ -141,7 +141,7 @@ static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
 
     int ext_type;
     VALUE ext_module;
-    VALUE options;
+    VALUE options = Qnil;
     VALUE packer_arg, unpacker_arg;
     VALUE packer_proc, unpacker_proc;
 
@@ -188,6 +188,8 @@ static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
     if(unpacker_arg != Qnil) {
         if(rb_type(unpacker_arg) == T_SYMBOL || rb_type(unpacker_arg) == T_STRING) {
             unpacker_proc = rb_obj_method(ext_module, unpacker_arg);
+        } else if (rb_respond_to(unpacker_arg, rb_intern("call"))) {
+            unpacker_proc = unpacker_arg;
         } else {
             unpacker_proc = rb_funcall(unpacker_arg, rb_intern("method"), 1, ID2SYM(rb_intern("call")));
         }
@@ -197,6 +199,9 @@ static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
 
     if (ext_module == rb_cSymbol) {
         fc->has_symbol_ext_type = true;
+        if(RB_TEST(options) && RB_TEST(rb_hash_aref(options, ID2SYM(rb_intern("optimized_symbols_parsing"))))) {
+            fc->optimized_symbol_ext_type = true;
+        }
     }
 
     msgpack_unpacker_ext_registry_put(&fc->ukrg, ext_module, ext_type, unpacker_proc, unpacker_arg);

data/ext/msgpack/packer.c

@@ -121,7 +121,7 @@ void msgpack_packer_write_hash_value(msgpack_packer_t* pk, VALUE v)
 #endif
 }
 
-void msgpack_packer_write_other_value(msgpack_packer_t* pk, VALUE v)
+bool msgpack_packer_try_write_with_ext_type_lookup(msgpack_packer_t* pk, VALUE v)
 {
     int ext_type;
 
@@ -131,7 +131,14 @@ void msgpack_packer_write_other_value(msgpack_packer_t* pk, VALUE v)
         VALUE payload = rb_funcall(proc, s_call, 1, v);
         StringValue(payload);
         msgpack_packer_write_ext(pk, ext_type, payload);
-    } else {
+        return true;
+    }
+    return false;
+}
+
+void msgpack_packer_write_other_value(msgpack_packer_t* pk, VALUE v)
+{
+    if(!(msgpack_packer_try_write_with_ext_type_lookup(pk, v))) {
         rb_funcall(v, pk->to_msgpack_method, 1, pk->to_msgpack_arg);
     }
 }
@@ -155,13 +162,19 @@ void msgpack_packer_write_value(msgpack_packer_t* pk, VALUE v)
         msgpack_packer_write_symbol_value(pk, v);
         break;
     case T_STRING:
-        msgpack_packer_write_string_value(pk, v);
+        if(rb_class_of(v) == rb_cString || !msgpack_packer_try_write_with_ext_type_lookup(pk, v)) {
+            msgpack_packer_write_string_value(pk, v);
+        }
         break;
     case T_ARRAY:
-        msgpack_packer_write_array_value(pk, v);
+        if(rb_class_of(v) == rb_cArray || !msgpack_packer_try_write_with_ext_type_lookup(pk, v)) {
+            msgpack_packer_write_array_value(pk, v);
+        }
         break;
     case T_HASH:
-        msgpack_packer_write_hash_value(pk, v);
+        if(rb_class_of(v) == rb_cHash || !msgpack_packer_try_write_with_ext_type_lookup(pk, v)) {
+            msgpack_packer_write_hash_value(pk, v);
+        }
         break;
     case T_BIGNUM:
         msgpack_packer_write_bignum_value(pk, v);

data/ext/msgpack/packer.h

@@ -396,7 +396,6 @@ static inline void msgpack_packer_write_ext(msgpack_packer_t* pk, int ext_type,
     msgpack_buffer_append_string(PACKER_BUFFER_(pk), payload);
 }
 
-#ifdef COMPAT_HAVE_ENCODING
 static inline bool msgpack_packer_is_binary(VALUE v, int encindex)
 {
     return encindex == msgpack_rb_encindex_ascii8bit;
@@ -414,7 +413,6 @@ static inline bool msgpack_packer_is_utf8_compat_string(VALUE v, int encindex)
 #endif
         ;
 }
-#endif
 
 static inline void msgpack_packer_write_string_value(msgpack_packer_t* pk, VALUE v)
 {
@@ -425,7 +423,6 @@ static inline void msgpack_packer_write_string_value(msgpack_packer_t* pk, VALUE
         rb_raise(rb_eArgError, "size of string is too long to pack: %lu bytes should be <= %lu", len, 0xffffffffUL);
     }
 
-#ifdef COMPAT_HAVE_ENCODING
     int encindex = ENCODING_GET(v);
     if(msgpack_packer_is_binary(v, encindex) && !pk->compatibility_mode) {
         /* write ASCII-8BIT string using Binary type */
@@ -443,24 +440,11 @@ static inline void msgpack_packer_write_string_value(msgpack_packer_t* pk, VALUE
         msgpack_packer_write_raw_header(pk, (unsigned int)len);
         msgpack_buffer_append_string(PACKER_BUFFER_(pk), v);
     }
-#else
-    msgpack_packer_write_raw_header(pk, (unsigned int)len);
-    msgpack_buffer_append_string(PACKER_BUFFER_(pk), v);
-#endif
 }
 
 static inline void msgpack_packer_write_symbol_string_value(msgpack_packer_t* pk, VALUE v)
 {
-#ifdef HAVE_RB_SYM2STR
-    /* rb_sym2str is added since MRI 2.2.0 */
     msgpack_packer_write_string_value(pk, rb_sym2str(v));
-#else
-    VALUE str = rb_id2str(SYM2ID(v));
-    if (!str) {
-       rb_raise(rb_eRuntimeError, "could not convert a symbol to string");
-    }
-    msgpack_packer_write_string_value(pk, str);
-#endif
 }
 
 void msgpack_packer_write_other_value(msgpack_packer_t* pk, VALUE v);

data/ext/msgpack/packer_class.c

@@ -339,11 +339,7 @@ static VALUE Packer_write_to(VALUE self, VALUE io)
 static VALUE Packer_registered_types_internal(VALUE self)
 {
     PACKER(self, pk);
-#ifdef HAVE_RB_HASH_DUP
     return rb_hash_dup(pk->ext_registry.hash);
-#else
-    return rb_funcall(pk->ext_registry.hash, rb_intern("dup"), 0);
-#endif
 }
 
 static VALUE Packer_register_type(int argc, VALUE* argv, VALUE self)
@@ -359,12 +355,7 @@ static VALUE Packer_register_type(int argc, VALUE* argv, VALUE self)
     case 2:
         /* register_type(0x7f, Time) {|obj| block... } */
         rb_need_block();
-#ifdef HAVE_RB_BLOCK_LAMBDA
         proc = rb_block_lambda();
-#else
-        /* MRI 1.8 */
-        proc = rb_block_proc();
-#endif
         arg = proc;
         break;
     case 3:

data/ext/msgpack/packer_ext_registry.c

@@ -43,37 +43,15 @@ void msgpack_packer_ext_registry_mark(msgpack_packer_ext_registry_t* pkrg)
 void msgpack_packer_ext_registry_dup(msgpack_packer_ext_registry_t* src,
         msgpack_packer_ext_registry_t* dst)
 {
-#ifdef HAVE_RB_HASH_DUP
     dst->hash = rb_hash_dup(src->hash);
     dst->cache = rb_hash_dup(src->cache);
-#else
-    dst->hash = rb_funcall(src->hash, rb_intern("dup"), 0);
-    dst->cache = rb_funcall(src->cache, rb_intern("dup"), 0);
-#endif
 }
 
-#ifndef HAVE_RB_HASH_CLEAR
-
-static int
-__rb_hash_clear_clear_i(key, value, dummy)
-    VALUE key, value, dummy;
-{
-    return ST_DELETE;
-}
-
-#endif
-
 VALUE msgpack_packer_ext_registry_put(msgpack_packer_ext_registry_t* pkrg,
         VALUE ext_module, int ext_type, VALUE proc, VALUE arg)
 {
     VALUE e = rb_ary_new3(3, INT2FIX(ext_type), proc, arg);
     /* clear lookup cache not to miss added type */
-#ifdef HAVE_RB_HASH_CLEAR
     rb_hash_clear(pkrg->cache);
-#else
-    if(FIX2INT(rb_funcall(pkrg->cache, rb_intern("size"), 0)) > 0) {
-        rb_hash_foreach(pkrg->cache, __rb_hash_clear_clear_i, 0);
-    }
-#endif
     return rb_hash_aset(pkrg->hash, ext_module, e);
 }

data/ext/msgpack/unpacker.c

@@ -142,32 +142,27 @@ static inline void reset_head_byte(msgpack_unpacker_t* uk)
 
 static inline int object_complete(msgpack_unpacker_t* uk, VALUE object)
 {
+    if(uk->freeze) {
+        rb_obj_freeze(object);
+    }
+
     uk->last_object = object;
     reset_head_byte(uk);
     return PRIMITIVE_OBJECT_COMPLETE;
 }
 
-static inline int object_complete_string(msgpack_unpacker_t* uk, VALUE str)
-{
-#ifdef COMPAT_HAVE_ENCODING
-    ENCODING_SET(str, msgpack_rb_encindex_utf8);
-#endif
-    return object_complete(uk, str);
-}
-
-static inline int object_complete_binary(msgpack_unpacker_t* uk, VALUE str)
+static inline int object_complete_symbol(msgpack_unpacker_t* uk, VALUE object)
 {
-#ifdef COMPAT_HAVE_ENCODING
-    ENCODING_SET(str, msgpack_rb_encindex_ascii8bit);
-#endif
-    return object_complete(uk, str);
+    uk->last_object = object;
+    reset_head_byte(uk);
+    return PRIMITIVE_OBJECT_COMPLETE;
 }
 
 static inline int object_complete_ext(msgpack_unpacker_t* uk, int ext_type, VALUE str)
 {
-#ifdef COMPAT_HAVE_ENCODING
-    ENCODING_SET(str, msgpack_rb_encindex_ascii8bit);
-#endif
+    if (uk->optimized_symbol_ext_type && ext_type == uk->symbol_ext_type) {
+        return object_complete_symbol(uk, rb_str_intern(str));
+    }
 
     VALUE proc = msgpack_unpacker_ext_registry_lookup(&uk->ext_registry, ext_type);
     if(proc != Qnil) {
@@ -271,9 +266,10 @@ static int read_raw_body_cont(msgpack_unpacker_t* uk)
 
     int ret;
     if(uk->reading_raw_type == RAW_TYPE_STRING) {
-        ret = object_complete_string(uk, uk->reading_raw);
-    } else if(uk->reading_raw_type == RAW_TYPE_BINARY) {
-        ret = object_complete_binary(uk, uk->reading_raw);
+        ENCODING_SET(uk->reading_raw, msgpack_rb_encindex_utf8);
+        ret = object_complete(uk, uk->reading_raw);
+    } else if (uk->reading_raw_type == RAW_TYPE_BINARY) {
+        ret = object_complete(uk, uk->reading_raw);
     } else {
         ret = object_complete_ext(uk, uk->reading_raw_type, uk->reading_raw);
     }
@@ -288,20 +284,26 @@ static inline int read_raw_body_begin(msgpack_unpacker_t* uk, int raw_type)
     /* try optimized read */
     size_t length = uk->reading_raw_remaining;
     if(length <= msgpack_buffer_top_readable_size(UNPACKER_BUFFER_(uk))) {
-        /* don't use zerocopy for hash keys but get a frozen string directly
-         * because rb_hash_aset freezes keys and it causes copying */
-        bool will_freeze = is_reading_map_key(uk);
-        VALUE string = msgpack_buffer_read_top_as_string(UNPACKER_BUFFER_(uk), length, will_freeze);
         int ret;
-        if(raw_type == RAW_TYPE_STRING) {
-            ret = object_complete_string(uk, string);
-        } else if(raw_type == RAW_TYPE_BINARY) {
-            ret = object_complete_binary(uk, string);
+        if ((uk->optimized_symbol_ext_type && uk->symbol_ext_type == raw_type) || (uk->symbolize_keys && is_reading_map_key(uk))) {
+            VALUE symbol = msgpack_buffer_read_top_as_symbol(UNPACKER_BUFFER_(uk), length);
+            ret = object_complete_symbol(uk, symbol);
         } else {
-            ret = object_complete_ext(uk, raw_type, string);
-        }
-        if(will_freeze) {
-            rb_obj_freeze(string);
+            /* don't use zerocopy for hash keys but get a frozen string directly
+             * because rb_hash_aset freezes keys and it causes copying */
+            bool will_freeze = uk->freeze || is_reading_map_key(uk);
+            VALUE string = msgpack_buffer_read_top_as_string(UNPACKER_BUFFER_(uk), length, will_freeze, raw_type == RAW_TYPE_STRING);
+            if(raw_type == RAW_TYPE_STRING || raw_type == RAW_TYPE_BINARY) {
+                ret = object_complete(uk, string);
+            } else {
+                ret = object_complete_ext(uk, raw_type, string);
+            }
+
+# if !HASH_ASET_DEDUPE
+            if(will_freeze) {
+                rb_obj_freeze(string);
+            }
+# endif
         }
         uk->reading_raw_remaining = 0;
         return ret;
@@ -332,7 +334,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
     SWITCH_RANGE(b, 0xa0, 0xbf)  // FixRaw / fixstr
         int count = b & 0x1f;
         if(count == 0) {
-            return object_complete_string(uk, rb_str_buf_new(0));
+            return object_complete(uk, rb_utf8_str_new_static("", 0));
         }
         /* read_raw_body_begin sets uk->reading_raw */
         uk->reading_raw_remaining = count;
@@ -517,7 +519,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
                 uint8_t count = cb->u8;
                 if(count == 0) {
-                    return object_complete_string(uk, rb_str_buf_new(0));
+                    return object_complete(uk, rb_utf8_str_new_static("", 0));
                 }
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
@@ -529,7 +531,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
                 uint16_t count = _msgpack_be16(cb->u16);
                 if(count == 0) {
-                    return object_complete_string(uk, rb_str_buf_new(0));
+                    return object_complete(uk, rb_utf8_str_new_static("", 0));
                 }
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
@@ -541,7 +543,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
                 uint32_t count = _msgpack_be32(cb->u32);
                 if(count == 0) {
-                    return object_complete_string(uk, rb_str_buf_new(0));
+                    return object_complete(uk, rb_utf8_str_new_static("", 0));
                 }
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
@@ -553,7 +555,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
                 uint8_t count = cb->u8;
                 if(count == 0) {
-                    return object_complete_binary(uk, rb_str_buf_new(0));
+                    return object_complete(uk, rb_str_new_static("", 0));
                 }
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
@@ -565,7 +567,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
                 uint16_t count = _msgpack_be16(cb->u16);
                 if(count == 0) {
-                    return object_complete_binary(uk, rb_str_buf_new(0));
+                    return object_complete(uk, rb_str_new_static("", 0));
                 }
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
@@ -577,7 +579,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
                 uint32_t count = _msgpack_be32(cb->u32);
                 if(count == 0) {
-                    return object_complete_binary(uk, rb_str_buf_new(0));
+                    return object_complete(uk, rb_str_new_static("", 0));
                 }
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
@@ -719,18 +721,8 @@ int msgpack_unpacker_read(msgpack_unpacker_t* uk, size_t target_stack_depth)
                 break;
             case STACK_TYPE_MAP_VALUE:
                 if(uk->symbolize_keys && rb_type(top->key) == T_STRING) {
-                    /* here uses rb_intern_str instead of rb_intern so that Ruby VM can GC unused symbols */
-#ifdef HAVE_RB_STR_INTERN
-                    /* rb_str_intern is added since MRI 2.2.0 */
+                    /* here uses rb_str_intern instead of rb_intern so that Ruby VM can GC unused symbols */
                     rb_hash_aset(top->object, rb_str_intern(top->key), uk->last_object);
-#else
-#ifndef HAVE_RB_INTERN_STR
-                    /* MRI 1.8 doesn't have rb_intern_str or rb_intern2 */
-                    rb_hash_aset(top->object, ID2SYM(rb_intern(RSTRING_PTR(top->key))), uk->last_object);
-#else
-                    rb_hash_aset(top->object, ID2SYM(rb_intern_str(top->key)), uk->last_object);
-#endif
-#endif
                 } else {
                     rb_hash_aset(top->object, top->key, uk->last_object);
                 }

data/ext/msgpack/unpacker.h

@@ -64,7 +64,10 @@ struct msgpack_unpacker_t {
 
     /* options */
     bool symbolize_keys;
+    bool freeze;
     bool allow_unknown_ext;
+    bool optimized_symbol_ext_type;
+    int symbol_ext_type;
 };
 
 #define UNPACKER_BUFFER_(uk) (&(uk)->buffer)
@@ -96,6 +99,11 @@ static inline void msgpack_unpacker_set_symbolized_keys(msgpack_unpacker_t* uk,
     uk->symbolize_keys = enable;
 }
 
+static inline void msgpack_unpacker_set_freeze(msgpack_unpacker_t* uk, bool enable)
+{
+    uk->freeze = enable;
+}
+
 static inline void msgpack_unpacker_set_allow_unknown_ext(msgpack_unpacker_t* uk, bool enable)
 {
     uk->allow_unknown_ext = enable;