msgpack-ruby-fstring-hash-key-dedup-hack.diff

diff --git a/ext/msgpack/buffer.h b/ext/msgpack/buffer.h
index 9387254..bcbe99b 100644
--- a/ext/msgpack/buffer.h
+++ b/ext/msgpack/buffer.h
@@ -424,22 +424,61 @@ 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)
+#ifdef COMPAT_HAVE_ENCODING
+#ifdef HAVE_RB_FSTRING_ENC_NEW
+#ifndef DISABLE_HASH_KEY_FSTRING
+extern VALUE rb_fstring_enc_new(const char *ptr, long len, rb_encoding *enc);
+#endif
+#endif
+#endif
+
+static inline VALUE msgpack_buffer_read_top_as_string(msgpack_buffer_t* b, size_t length, bool frozen, bool string_enc)
 {
 #ifndef DISABLE_BUFFER_READ_REFERENCE_OPTIMIZE
     /* optimize */
-    if(!will_be_frozen &&
+    /* don't use _msgpack_buffer_refer_head_mapped_string optimization for
+     * frozen strings because freezing strings cause copying */
+    if(!frozen &&
             b->head->mapped_string != NO_MAPPED_STRING &&
             length >= b->read_reference_threshold) {
         VALUE result = _msgpack_buffer_refer_head_mapped_string(b, length);
+        ENCODING_SET(result,
+                string_enc ? msgpack_rb_encindex_utf8 : msgpack_rb_encindex_ascii8bit);
         _msgpack_buffer_consumed(b, length);
         return result;
     }
 #endif
 
+#ifdef COMPAT_HAVE_ENCODING
+#ifdef HAVE_RB_FSTRING_ENC_NEW
+#ifndef DISABLE_HASH_KEY_FSTRING
+    /* optimize */
+    if(frozen) {
+        VALUE fstring = rb_fstring_enc_new(b->read_buffer, length,
+                rb_enc_from_index(string_enc ? msgpack_rb_encindex_utf8 : msgpack_rb_encindex_ascii8bit));
+        _msgpack_buffer_consumed(b, length);
+        return fstring;
+    }
+#endif
+#endif
+
+    VALUE result = rb_str_new(b->read_buffer, length);
+    ENCODING_SET(result,
+            string_enc ? msgpack_rb_encindex_utf8 : msgpack_rb_encindex_ascii8bit);
+    if(frozen) {
+        rb_obj_freeze(result);
+    }
+    _msgpack_buffer_consumed(b, length);
+    return result;
+
+#else
     VALUE result = rb_str_new(b->read_buffer, length);
     _msgpack_buffer_consumed(b, length);
+    if(frozen) {
+        rb_obj_freeze(result);
+    }
     return result;
+#endif
 }
 
 
diff --git a/ext/msgpack/extconf.rb b/ext/msgpack/extconf.rb
index 4ca8cda..fb72137 100644
--- a/ext/msgpack/extconf.rb
+++ b/ext/msgpack/extconf.rb
@@ -6,6 +6,7 @@ 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_fstring_enc_new", ["ruby.h"])
 have_func("rb_block_lambda", ["ruby.h"])
 have_func("rb_hash_dup", ["ruby.h"])
 have_func("rb_hash_clear", ["ruby.h"])
@@ -17,6 +18,7 @@ end
 #$CFLAGS << %[ -DDISABLE_RMEM_REUSE_INTERNAL_FRAGMENT]
 #$CFLAGS << %[ -DDISABLE_BUFFER_READ_REFERENCE_OPTIMIZE]
 #$CFLAGS << %[ -DDISABLE_BUFFER_READ_TO_S_OPTIMIZE]
+#$CFLAGS << %[ -DDISABLE_HASH_KEY_FSTRING]
 
 if defined?(RUBY_ENGINE) && RUBY_ENGINE == 'rbx'
   # msgpack-ruby doesn't modify data came from RSTRING_PTR(str)
diff --git a/ext/msgpack/unpacker.c b/ext/msgpack/unpacker.c
index 8b1f699..54faaa4 100644
--- a/ext/msgpack/unpacker.c
+++ b/ext/msgpack/unpacker.c
@@ -163,12 +163,8 @@ static inline int object_complete_binary(msgpack_unpacker_t* uk, VALUE str)
     return object_complete(uk, str);
 }
 
-static inline int object_complete_ext(msgpack_unpacker_t* uk, int ext_type, VALUE str)
+static inline int object_complete_ext2(msgpack_unpacker_t* uk, int ext_type, VALUE str)
 {
-#ifdef COMPAT_HAVE_ENCODING
-    ENCODING_SET(str, msgpack_rb_encindex_ascii8bit);
-#endif
-
     VALUE proc = msgpack_unpacker_ext_registry_lookup(&uk->ext_registry, ext_type);
     if(proc != Qnil) {
         VALUE obj = rb_funcall(proc, s_call, 1, str);
@@ -183,6 +179,14 @@ static inline int object_complete_ext(msgpack_unpacker_t* uk, int ext_type, VALU
     return PRIMITIVE_UNEXPECTED_EXT_TYPE;
 }
 
+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
+    return object_complete_ext2(uk, ext_type, str);
+}
+
 /* stack funcs */
 static inline msgpack_unpacker_stack_t* _msgpack_unpacker_stack_top(msgpack_unpacker_t* uk)
 {
@@ -288,20 +292,16 @@ 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);
+        bool frozen = is_reading_map_key(uk);
+        VALUE string = msgpack_buffer_read_top_as_string(UNPACKER_BUFFER_(uk), length,
+                frozen, raw_type == RAW_TYPE_STRING);
         int ret;
         if(raw_type == RAW_TYPE_STRING) {
-            ret = object_complete_string(uk, string);
+            ret = object_complete(uk, string);
         } else if(raw_type == RAW_TYPE_BINARY) {
-            ret = object_complete_binary(uk, string);
+            ret = object_complete(uk, string);
         } else {
-            ret = object_complete_ext(uk, raw_type, string);
-        }
-        if(will_freeze) {
-            rb_obj_freeze(string);
+            ret = object_complete_ext2(uk, raw_type, string);
         }
         uk->reading_raw_remaining = 0;
         return ret;