msgpack 1.5.6 → 1.8.0

data/ext/msgpack/factory_class.c

@@ -37,15 +37,10 @@ struct msgpack_factory_t {
     int symbol_ext_type;
 };
 
-#define FACTORY(from, name) \
-    msgpack_factory_t *name = NULL; \
-    Data_Get_Struct(from, msgpack_factory_t, name); \
-    if(name == NULL) { \
-        rb_raise(rb_eArgError, "NULL found for " # name " when shouldn't be."); \
-    }
-
-static void Factory_free(msgpack_factory_t* fc)
+static void Factory_free(void *ptr)
 {
+    msgpack_factory_t *fc = ptr;
+
     if(fc == NULL) {
         return;
     }
@@ -54,25 +49,56 @@ static void Factory_free(msgpack_factory_t* fc)
     xfree(fc);
 }
 
-void Factory_mark(msgpack_factory_t* fc)
+void Factory_mark(void *ptr)
 {
+    msgpack_factory_t *fc = ptr;
     msgpack_packer_ext_registry_mark(&fc->pkrg);
     msgpack_unpacker_ext_registry_mark(fc->ukrg);
 }
 
-static VALUE Factory_alloc(VALUE klass)
+static size_t Factory_memsize(const void *ptr)
 {
-    msgpack_factory_t* fc = ZALLOC_N(msgpack_factory_t, 1);
+    const msgpack_factory_t *fc = ptr;
+    size_t total_size = sizeof(msgpack_factory_t);
 
-    VALUE self = Data_Wrap_Struct(klass, Factory_mark, Factory_free, fc);
-    return self;
+    if (fc->ukrg) {
+        total_size += sizeof(msgpack_unpacker_ext_registry_t) / (fc->ukrg->borrow_count + 1);
+    }
+
+    return total_size;
+}
+
+static const rb_data_type_t factory_data_type = {
+    .wrap_struct_name = "msgpack:factory",
+    .function = {
+        .dmark = Factory_mark,
+        .dfree = Factory_free,
+        .dsize = Factory_memsize,
+    },
+    .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED
+};
+
+static inline msgpack_factory_t *Factory_get(VALUE object)
+{
+    msgpack_factory_t *factory;
+    TypedData_Get_Struct(object, msgpack_factory_t, &factory_data_type, factory);
+    if (!factory) {
+        rb_raise(rb_eArgError, "Uninitialized Factory object");
+    }
+    return factory;
+}
+
+static VALUE Factory_alloc(VALUE klass)
+{
+    msgpack_factory_t *fc;
+    return TypedData_Make_Struct(klass, msgpack_factory_t, &factory_data_type, fc);
 }
 
 static VALUE Factory_initialize(int argc, VALUE* argv, VALUE self)
 {
-    FACTORY(self, fc);
+    msgpack_factory_t *fc = Factory_get(self);
 
-    msgpack_packer_ext_registry_init(&fc->pkrg);
+    msgpack_packer_ext_registry_init(self, &fc->pkrg);
     // fc->ukrg is lazily initialized
 
     fc->has_symbol_ext_type = false;
@@ -92,20 +118,20 @@ static VALUE Factory_dup(VALUE self)
 {
     VALUE clone = Factory_alloc(rb_obj_class(self));
 
-    FACTORY(self, fc);
-    FACTORY(clone, cloned_fc);
+    msgpack_factory_t *fc = Factory_get(self);
+    msgpack_factory_t *cloned_fc = Factory_get(clone);
 
     cloned_fc->has_symbol_ext_type = fc->has_symbol_ext_type;
     cloned_fc->pkrg = fc->pkrg;
     msgpack_unpacker_ext_registry_borrow(fc->ukrg, &cloned_fc->ukrg);
-    msgpack_packer_ext_registry_dup(&fc->pkrg, &cloned_fc->pkrg);
+    msgpack_packer_ext_registry_dup(clone, &fc->pkrg, &cloned_fc->pkrg);
 
     return clone;
 }
 
 static VALUE Factory_freeze(VALUE self) {
     if(!rb_obj_frozen_p(self)) {
-        FACTORY(self, fc);
+        msgpack_factory_t *fc = Factory_get(self);
 
         if (RTEST(fc->pkrg.hash)) {
             rb_hash_freeze(fc->pkrg.hash);
@@ -113,7 +139,7 @@ static VALUE Factory_freeze(VALUE self) {
                 // If the factory is frozen, we can safely share the packer cache between
                 // all packers. So we eagerly create it now so it's available when #packer
                 // is called.
-                fc->pkrg.cache = rb_hash_new();
+                RB_OBJ_WRITE(self, &fc->pkrg.cache, rb_hash_new());
             }
         }
 
@@ -125,16 +151,14 @@ static VALUE Factory_freeze(VALUE self) {
 
 VALUE MessagePack_Factory_packer(int argc, VALUE* argv, VALUE self)
 {
-    FACTORY(self, fc);
+    msgpack_factory_t *fc = Factory_get(self);
 
     VALUE packer = MessagePack_Packer_alloc(cMessagePack_Packer);
     MessagePack_Packer_initialize(argc, argv, packer);
 
-    msgpack_packer_t* pk;
-    Data_Get_Struct(packer, msgpack_packer_t, pk);
-
+    msgpack_packer_t* pk = MessagePack_Packer_get(packer);
     msgpack_packer_ext_registry_destroy(&pk->ext_registry);
-    msgpack_packer_ext_registry_dup(&fc->pkrg, &pk->ext_registry);
+    msgpack_packer_ext_registry_borrow(packer, &fc->pkrg, &pk->ext_registry);
     pk->has_bigint_ext_type = fc->has_bigint_ext_type;
     pk->has_symbol_ext_type = fc->has_symbol_ext_type;
 
@@ -143,13 +167,12 @@ VALUE MessagePack_Factory_packer(int argc, VALUE* argv, VALUE self)
 
 VALUE MessagePack_Factory_unpacker(int argc, VALUE* argv, VALUE self)
 {
-    FACTORY(self, fc);
+    msgpack_factory_t *fc = Factory_get(self);
 
     VALUE unpacker = MessagePack_Unpacker_alloc(cMessagePack_Unpacker);
     MessagePack_Unpacker_initialize(argc, argv, unpacker);
 
-    msgpack_unpacker_t* uk;
-    Data_Get_Struct(unpacker, msgpack_unpacker_t, uk);
+    msgpack_unpacker_t* uk = MessagePack_Unpacker_get(unpacker);
     msgpack_unpacker_ext_registry_borrow(fc->ukrg, &uk->ext_registry);
     uk->optimized_symbol_ext_type = fc->optimized_symbol_ext_type;
     uk->symbol_ext_type = fc->symbol_ext_type;
@@ -159,12 +182,12 @@ VALUE MessagePack_Factory_unpacker(int argc, VALUE* argv, VALUE self)
 
 static VALUE Factory_registered_types_internal(VALUE self)
 {
-    FACTORY(self, fc);
+    msgpack_factory_t *fc = Factory_get(self);
 
     VALUE uk_mapping = rb_hash_new();
     if (fc->ukrg) {
         for(int i=0; i < 256; i++) {
-            if(fc->ukrg->array[i] != Qnil) {
+            if(!NIL_P(fc->ukrg->array[i])) {
                 rb_hash_aset(uk_mapping, INT2FIX(i - 128), fc->ukrg->array[i]);
             }
         }
@@ -177,73 +200,39 @@ static VALUE Factory_registered_types_internal(VALUE self)
     );
 }
 
-static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
+static VALUE Factory_register_type_internal(VALUE self, VALUE rb_ext_type, VALUE ext_module, VALUE options)
 {
-    FACTORY(self, fc);
+    msgpack_factory_t *fc = Factory_get(self);
 
-    int ext_type;
-    int flags = 0;
-    VALUE ext_module;
-    VALUE options = Qnil;
-    VALUE packer_arg, unpacker_arg;
-    VALUE packer_proc, unpacker_proc;
+    Check_Type(rb_ext_type, T_FIXNUM);
 
-    if (OBJ_FROZEN(self)) {
-        rb_raise(rb_eRuntimeError, "can't modify frozen Factory");
+    if(rb_type(ext_module) != T_MODULE && rb_type(ext_module) != T_CLASS) {
+        rb_raise(rb_eArgError, "expected Module/Class but found %s.", rb_obj_classname(ext_module));
     }
 
-    switch (argc) {
-    case 2:
-        /* register_type(0x7f, Time) */
-        packer_arg = ID2SYM(rb_intern("to_msgpack_ext"));
-        unpacker_arg = ID2SYM(rb_intern("from_msgpack_ext"));
-        break;
-    case 3:
-        /* register_type(0x7f, Time, packer: proc-like, unapcker: proc-like) */
-        options = argv[2];
-        if(rb_type(options) != T_HASH) {
-            rb_raise(rb_eArgError, "expected Hash but found %s.", rb_obj_classname(options));
-        }
-        packer_arg = rb_hash_aref(options, ID2SYM(rb_intern("packer")));
-        unpacker_arg = rb_hash_aref(options, ID2SYM(rb_intern("unpacker")));
-        break;
-    default:
-        rb_raise(rb_eArgError, "wrong number of arguments (%d for 2..3)", argc);
-    }
+    int flags = 0;
 
-    if (options != Qnil) {
+    VALUE packer_proc = Qnil;
+    VALUE unpacker_proc = Qnil;
+    if(!NIL_P(options)) {
         Check_Type(options, T_HASH);
+        packer_proc = rb_hash_aref(options, ID2SYM(rb_intern("packer")));
+        unpacker_proc = rb_hash_aref(options, ID2SYM(rb_intern("unpacker")));
     }
 
-    ext_type = NUM2INT(argv[0]);
-    if(ext_type < -128 || ext_type > 127) {
-        rb_raise(rb_eRangeError, "integer %d too big to convert to `signed char'", ext_type);
-    }
-
-    ext_module = argv[1];
-    if(rb_type(ext_module) != T_MODULE && rb_type(ext_module) != T_CLASS) {
-        rb_raise(rb_eArgError, "expected Module/Class but found %s.", rb_obj_classname(ext_module));
-    }
-
-    packer_proc = Qnil;
-    unpacker_proc = Qnil;
-
-    if(packer_arg != Qnil) {
-        packer_proc = rb_funcall(packer_arg, rb_intern("to_proc"), 0);
+    if (OBJ_FROZEN(self)) {
+        rb_raise(rb_eFrozenError, "can't modify frozen MessagePack::Factory");
     }
 
-    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")));
-        }
+    int ext_type = NUM2INT(rb_ext_type);
+    if(ext_type < -128 || ext_type > 127) {
+        rb_raise(rb_eRangeError, "integer %d too big to convert to `signed char'", ext_type);
     }
 
     if(ext_module == rb_cSymbol) {
-        fc->has_symbol_ext_type = true;
+        if(NIL_P(options) || RTEST(rb_hash_aref(options, ID2SYM(rb_intern("packer"))))) {
+            fc->has_symbol_ext_type = true;
+        }
         if(RTEST(options) && RTEST(rb_hash_aref(options, ID2SYM(rb_intern("optimized_symbols_parsing"))))) {
             fc->optimized_symbol_ext_type = true;
         }
@@ -263,8 +252,8 @@ static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
         }
     }
 
-    msgpack_packer_ext_registry_put(&fc->pkrg, ext_module, ext_type, flags, packer_proc, packer_arg);
-    msgpack_unpacker_ext_registry_put(&fc->ukrg, ext_module, ext_type, flags, unpacker_proc, unpacker_arg);
+    msgpack_packer_ext_registry_put(self, &fc->pkrg, ext_module, ext_type, flags, packer_proc);
+    msgpack_unpacker_ext_registry_put(self, &fc->ukrg, ext_module, ext_type, flags, unpacker_proc);
 
     return Qnil;
 }
@@ -283,5 +272,5 @@ void MessagePack_Factory_module_init(VALUE mMessagePack)
     rb_define_method(cMessagePack_Factory, "unpacker", MessagePack_Factory_unpacker, -1);
 
     rb_define_private_method(cMessagePack_Factory, "registered_types_internal", Factory_registered_types_internal, 0);
-    rb_define_method(cMessagePack_Factory, "register_type", Factory_register_type, -1);
+    rb_define_private_method(cMessagePack_Factory, "register_type_internal", Factory_register_type_internal, 3);
 }

data/ext/msgpack/packer.c

@@ -17,21 +17,14 @@
  */
 
 #include "packer.h"
+#include "buffer_class.h"
 
-static ID s_call;
-
-void msgpack_packer_static_init()
-{
-    s_call = rb_intern("call");
-}
-
-void msgpack_packer_static_destroy()
-{ }
+#if !defined(HAVE_RB_PROC_CALL_WITH_BLOCK)
+#define rb_proc_call_with_block(recv, argc, argv, block) rb_funcallv(recv, rb_intern("call"), argc, argv)
+#endif
 
 void msgpack_packer_init(msgpack_packer_t* pk)
 {
-    memset(pk, 0, sizeof(msgpack_packer_t));
-
     msgpack_buffer_init(PACKER_BUFFER_(pk));
 }
 
@@ -45,6 +38,7 @@ void msgpack_packer_mark(msgpack_packer_t* pk)
     /* See MessagePack_Buffer_wrap */
     /* msgpack_buffer_mark(PACKER_BUFFER_(pk)); */
     rb_gc_mark(pk->buffer_ref);
+    rb_gc_mark(pk->to_msgpack_arg);
 }
 
 void msgpack_packer_reset(msgpack_packer_t* pk)
@@ -101,14 +95,13 @@ struct msgpack_call_proc_args_t;
 typedef struct msgpack_call_proc_args_t msgpack_call_proc_args_t;
 struct msgpack_call_proc_args_t {
     VALUE proc;
-    VALUE arg;
-    VALUE packer;
+    VALUE args[2];
 };
 
 VALUE msgpack_packer_try_calling_proc(VALUE value)
 {
     msgpack_call_proc_args_t *args = (msgpack_call_proc_args_t *)value;
-    return rb_funcall(args->proc, s_call, 2, args->arg, args->packer);
+    return rb_proc_call_with_block(args->proc, 2, args->args, Qnil);
 }
 
 bool msgpack_packer_try_write_with_ext_type_lookup(msgpack_packer_t* pk, VALUE v)
@@ -122,11 +115,13 @@ bool msgpack_packer_try_write_with_ext_type_lookup(msgpack_packer_t* pk, VALUE v
     }
 
     if(ext_flags & MSGPACK_EXT_RECURSIVE) {
+        VALUE held_buffer = MessagePack_Buffer_hold(&pk->buffer);
+
         msgpack_buffer_t parent_buffer = pk->buffer;
         msgpack_buffer_init(PACKER_BUFFER_(pk));
 
         int exception_occured = 0;
-        msgpack_call_proc_args_t args = { proc, v, pk->to_msgpack_arg };
+        msgpack_call_proc_args_t args = { proc, { v, pk->to_msgpack_arg } };
         rb_protect(msgpack_packer_try_calling_proc, (VALUE)&args, &exception_occured);
 
         if (exception_occured) {
@@ -140,8 +135,10 @@ bool msgpack_packer_try_write_with_ext_type_lookup(msgpack_packer_t* pk, VALUE v
             pk->buffer = parent_buffer;
             msgpack_packer_write_ext(pk, ext_type, payload);
         }
+
+        RB_GC_GUARD(held_buffer);
     } else {
-        VALUE payload = rb_funcall(proc, s_call, 1, v);
+        VALUE payload = rb_proc_call_with_block(proc, 1, &v, Qnil);
         StringValue(payload);
         msgpack_packer_write_ext(pk, ext_type, payload);
     }

data/ext/msgpack/packer.h

@@ -25,21 +25,26 @@
 #define MSGPACK_PACKER_IO_FLUSH_THRESHOLD_TO_WRITE_STRING_BODY (1024)
 #endif
 
+#ifndef UNREACHABLE_RETURN
+// Ruby 2.5
+#define UNREACHABLE_RETURN() return
+#endif
+
 struct msgpack_packer_t;
 typedef struct msgpack_packer_t msgpack_packer_t;
 
 struct msgpack_packer_t {
     msgpack_buffer_t buffer;
 
-    bool compatibility_mode;
-    bool has_bigint_ext_type;
-    bool has_symbol_ext_type;
-
     ID to_msgpack_method;
     VALUE to_msgpack_arg;
 
     VALUE buffer_ref;
 
+    bool compatibility_mode;
+    bool has_bigint_ext_type;
+    bool has_symbol_ext_type;
+
     /* options */
     bool comaptibility_mode;
     msgpack_packer_ext_registry_t ext_registry;
@@ -47,10 +52,6 @@ struct msgpack_packer_t {
 
 #define PACKER_BUFFER_(pk) (&(pk)->buffer)
 
-void msgpack_packer_static_init();
-
-void msgpack_packer_static_destroy();
-
 void msgpack_packer_init(msgpack_packer_t* pk);
 
 void msgpack_packer_destroy(msgpack_packer_t* pk);
@@ -408,27 +409,33 @@ static inline bool msgpack_packer_is_utf8_compat_string(VALUE v, int encindex)
 {
     return encindex == msgpack_rb_encindex_utf8
         || encindex == msgpack_rb_encindex_usascii
-        || (rb_enc_asciicompat(rb_enc_from_index(encindex)) && ENC_CODERANGE_ASCIIONLY(v));
+        || ENC_CODERANGE_ASCIIONLY(v);
 }
 
 static inline void msgpack_packer_write_string_value(msgpack_packer_t* pk, VALUE v)
 {
-    /* actual return type of RSTRING_LEN is long */
-    unsigned long len = RSTRING_LEN(v);
-    if(len > 0xffffffffUL) {
-        // TODO rb_eArgError?
-        rb_raise(rb_eArgError, "size of string is too long to pack: %lu bytes should be <= %lu", len, 0xffffffffUL);
+    long len = RSTRING_LEN(v);
+
+    if(RB_UNLIKELY(len > 0xffffffffL)) {
+        rb_raise(rb_eArgError, "size of string is too long to pack: %lu bytes should be <= %ld", len, 0xffffffffL);
+        UNREACHABLE_RETURN();
     }
 
-    int encindex = ENCODING_GET(v);
-    if(msgpack_packer_is_binary(v, encindex) && !pk->compatibility_mode) {
+    if (RB_UNLIKELY(pk->compatibility_mode)) {
+        msgpack_packer_write_raw_header(pk, (unsigned int)len);
+        msgpack_buffer_append_string(PACKER_BUFFER_(pk), v);
+        return;
+    }
+
+    int encindex = ENCODING_GET_INLINED(v);
+    if(msgpack_packer_is_binary(v, encindex)) {
         /* write ASCII-8BIT string using Binary type */
         msgpack_packer_write_bin_header(pk, (unsigned int)len);
         msgpack_buffer_append_string(PACKER_BUFFER_(pk), v);
     } else {
         /* write UTF-8, US-ASCII, or 7bit-safe ascii-compatible string using String type directly */
         /* in compatibility mode, packer packs String values as is */
-        if(!pk->compatibility_mode && !msgpack_packer_is_utf8_compat_string(v, encindex)) {
+        if(RB_UNLIKELY(!msgpack_packer_is_utf8_compat_string(v, encindex))) {
             /* transcode other strings to UTF-8 and write using String type */
             VALUE enc = rb_enc_from_encoding(rb_utf8_encoding()); /* rb_enc_from_encoding_index is not extern */
             v = rb_str_encode(v, enc, 0, Qnil);
@@ -457,11 +464,7 @@ static inline void msgpack_packer_write_symbol_value(msgpack_packer_t* pk, VALUE
 
 static inline void msgpack_packer_write_fixnum_value(msgpack_packer_t* pk, VALUE v)
 {
-#ifdef JRUBY
-    msgpack_packer_write_long(pk, FIXNUM_P(v) ? FIX2LONG(v) : rb_num2ll(v));
-#else
     msgpack_packer_write_long(pk, FIX2LONG(v));
-#endif
 }
 
 static inline void msgpack_packer_write_bignum_value(msgpack_packer_t* pk, VALUE v)