msgpack 1.4.5 → 1.5.6

data/ext/msgpack/factory_class.c

@@ -31,6 +31,7 @@ typedef struct msgpack_factory_t msgpack_factory_t;
 struct msgpack_factory_t {
     msgpack_packer_ext_registry_t pkrg;
     msgpack_unpacker_ext_registry_t *ukrg;
+    bool has_bigint_ext_type;
     bool has_symbol_ext_type;
     bool optimized_symbol_ext_type;
     int symbol_ext_type;
@@ -87,6 +88,41 @@ static VALUE Factory_initialize(int argc, VALUE* argv, VALUE self)
     return Qnil;
 }
 
+static VALUE Factory_dup(VALUE self)
+{
+    VALUE clone = Factory_alloc(rb_obj_class(self));
+
+    FACTORY(self, fc);
+    FACTORY(clone, cloned_fc);
+
+    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);
+
+    return clone;
+}
+
+static VALUE Factory_freeze(VALUE self) {
+    if(!rb_obj_frozen_p(self)) {
+        FACTORY(self, fc);
+
+        if (RTEST(fc->pkrg.hash)) {
+            rb_hash_freeze(fc->pkrg.hash);
+            if (!RTEST(fc->pkrg.cache)) {
+                // 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_freeze(self);
+    }
+
+    return self;
+}
+
 VALUE MessagePack_Factory_packer(int argc, VALUE* argv, VALUE self)
 {
     FACTORY(self, fc);
@@ -99,6 +135,7 @@ VALUE MessagePack_Factory_packer(int argc, VALUE* argv, VALUE self)
 
     msgpack_packer_ext_registry_destroy(&pk->ext_registry);
     msgpack_packer_ext_registry_dup(&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;
 
     return packer;
@@ -145,6 +182,7 @@ static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
     FACTORY(self, fc);
 
     int ext_type;
+    int flags = 0;
     VALUE ext_module;
     VALUE options = Qnil;
     VALUE packer_arg, unpacker_arg;
@@ -173,6 +211,10 @@ static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
         rb_raise(rb_eArgError, "wrong number of arguments (%d for 2..3)", argc);
     }
 
+    if (options != Qnil) {
+        Check_Type(options, T_HASH);
+    }
+
     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);
@@ -200,16 +242,29 @@ static VALUE Factory_register_type(int argc, VALUE* argv, VALUE self)
         }
     }
 
-    msgpack_packer_ext_registry_put(&fc->pkrg, ext_module, ext_type, packer_proc, packer_arg);
-
-    if (ext_module == rb_cSymbol) {
+    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"))))) {
+        if(RTEST(options) && RTEST(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);
+    if(RTEST(options)) {
+        if(RTEST(rb_hash_aref(options, ID2SYM(rb_intern("oversized_integer_extension"))))) {
+            if(ext_module == rb_cInteger) {
+                fc->has_bigint_ext_type = true;
+            } else {
+                rb_raise(rb_eArgError, "oversized_integer_extension: true is only for Integer class");
+            }
+        }
+
+        if(RTEST(rb_hash_aref(options, ID2SYM(rb_intern("recursive"))))) {
+            flags |= MSGPACK_EXT_RECURSIVE;
+        }
+    }
+
+    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);
 
     return Qnil;
 }
@@ -221,6 +276,8 @@ void MessagePack_Factory_module_init(VALUE mMessagePack)
     rb_define_alloc_func(cMessagePack_Factory, Factory_alloc);
 
     rb_define_method(cMessagePack_Factory, "initialize", Factory_initialize, -1);
+    rb_define_method(cMessagePack_Factory, "dup", Factory_dup, 0);
+    rb_define_method(cMessagePack_Factory, "freeze", Factory_freeze, 0);
 
     rb_define_method(cMessagePack_Factory, "packer", MessagePack_Factory_packer, -1);
     rb_define_method(cMessagePack_Factory, "unpacker", MessagePack_Factory_unpacker, -1);

data/ext/msgpack/packer.c

@@ -18,24 +18,10 @@
 
 #include "packer.h"
 
-#ifdef RUBINIUS
-static ID s_to_iter;
-static ID s_next;
-static ID s_key;
-static ID s_value;
-#endif
-
 static ID s_call;
 
 void msgpack_packer_static_init()
 {
-#ifdef RUBINIUS
-    s_to_iter = rb_intern("to_iter");
-    s_next = rb_intern("next");
-    s_key = rb_intern("key");
-    s_value = rb_intern("value");
-#endif
-
     s_call = rb_intern("call");
 }
 
@@ -108,32 +94,59 @@ void msgpack_packer_write_hash_value(msgpack_packer_t* pk, VALUE v)
     unsigned int len32 = (unsigned int)len;
     msgpack_packer_write_map_header(pk, len32);
 
-#ifdef RUBINIUS
-    VALUE iter = rb_funcall(v, s_to_iter, 0);
-    VALUE entry = Qnil;
-    while(RTEST(entry = rb_funcall(iter, s_next, 1, entry))) {
-        VALUE key = rb_funcall(entry, s_key, 0);
-        VALUE val = rb_funcall(entry, s_value, 0);
-        write_hash_foreach(key, val, (VALUE) pk);
-    }
-#else
     rb_hash_foreach(v, write_hash_foreach, (VALUE) pk);
-#endif
+}
+
+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 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);
 }
 
 bool msgpack_packer_try_write_with_ext_type_lookup(msgpack_packer_t* pk, VALUE v)
 {
-    int ext_type;
+    int ext_type, ext_flags;
+
+    VALUE proc = msgpack_packer_ext_registry_lookup(&pk->ext_registry, v, &ext_type, &ext_flags);
 
-    VALUE proc = msgpack_packer_ext_registry_lookup(&pk->ext_registry, v, &ext_type);
+    if(proc == Qnil) {
+        return false;
+    }
 
-    if(proc != Qnil) {
+    if(ext_flags & MSGPACK_EXT_RECURSIVE) {
+        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 };
+        rb_protect(msgpack_packer_try_calling_proc, (VALUE)&args, &exception_occured);
+
+        if (exception_occured) {
+            msgpack_buffer_destroy(PACKER_BUFFER_(pk));
+            pk->buffer = parent_buffer;
+            rb_jump_tag(exception_occured); // re-raise the exception
+        } else {
+            VALUE payload = msgpack_buffer_all_as_string(PACKER_BUFFER_(pk));
+            StringValue(payload);
+            msgpack_buffer_destroy(PACKER_BUFFER_(pk));
+            pk->buffer = parent_buffer;
+            msgpack_packer_write_ext(pk, ext_type, payload);
+        }
+    } else {
         VALUE payload = rb_funcall(proc, s_call, 1, v);
         StringValue(payload);
         msgpack_packer_write_ext(pk, ext_type, payload);
-        return true;
     }
-    return false;
+
+    return true;
 }
 
 void msgpack_packer_write_other_value(msgpack_packer_t* pk, VALUE v)

data/ext/msgpack/packer.h

@@ -32,6 +32,7 @@ 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;
@@ -56,6 +57,8 @@ void msgpack_packer_destroy(msgpack_packer_t* pk);
 
 void msgpack_packer_mark(msgpack_packer_t* pk);
 
+bool msgpack_packer_try_write_with_ext_type_lookup(msgpack_packer_t* pk, VALUE v);
+
 static inline void msgpack_packer_set_to_msgpack_method(msgpack_packer_t* pk,
         ID to_msgpack_method, VALUE to_msgpack_arg)
 {
@@ -405,13 +408,7 @@ static inline bool msgpack_packer_is_utf8_compat_string(VALUE v, int encindex)
 {
     return encindex == msgpack_rb_encindex_utf8
         || encindex == msgpack_rb_encindex_usascii
-#ifdef ENC_CODERANGE_ASCIIONLY
-        /* Because ENC_CODERANGE_ASCIIONLY does not scan string, it may return ENC_CODERANGE_UNKNOWN unlike */
-        /* rb_enc_str_asciionly_p. It is always faster than rb_str_encode if it is available. */
-        /* Very old Rubinius (< v1.3.1) doesn't have ENC_CODERANGE_ASCIIONLY. */
-        || (rb_enc_asciicompat(rb_enc_from_index(encindex)) && ENC_CODERANGE_ASCIIONLY(v))
-#endif
-        ;
+        || (rb_enc_asciicompat(rb_enc_from_index(encindex)) && ENC_CODERANGE_ASCIIONLY(v));
 }
 
 static inline void msgpack_packer_write_string_value(msgpack_packer_t* pk, VALUE v)
@@ -469,9 +466,30 @@ static inline void msgpack_packer_write_fixnum_value(msgpack_packer_t* pk, VALUE
 
 static inline void msgpack_packer_write_bignum_value(msgpack_packer_t* pk, VALUE v)
 {
+    int leading_zero_bits;
+    size_t required_size = rb_absint_size(v, &leading_zero_bits);
+
     if(RBIGNUM_POSITIVE_P(v)) {
+        if(required_size > 8 && pk->has_bigint_ext_type) {
+            if(msgpack_packer_try_write_with_ext_type_lookup(pk, v)) {
+                return;
+            }
+            // if we didn't return here `msgpack_packer_write_u64` will raise a RangeError
+        }
+
         msgpack_packer_write_u64(pk, rb_big2ull(v));
     } else {
+        if(leading_zero_bits == 0) {
+            required_size += 1;
+        }
+
+        if(required_size > 8 && pk->has_bigint_ext_type) {
+            if(msgpack_packer_try_write_with_ext_type_lookup(pk, v)) {
+                return;
+            }
+            // if we didn't return here `msgpack_packer_write_u64` will raise a RangeError
+        }
+
         msgpack_packer_write_long_long(pk, rb_big2ll(v));
     }
 }

data/ext/msgpack/packer_class.c

@@ -28,6 +28,8 @@ VALUE cMessagePack_Packer;
 static ID s_to_msgpack;
 static ID s_write;
 
+static VALUE sym_compatibility_mode;
+
 //static VALUE s_packer_value;
 //static msgpack_packer_t* s_packer;
 
@@ -68,29 +70,28 @@ VALUE MessagePack_Packer_alloc(VALUE klass)
 
 VALUE MessagePack_Packer_initialize(int argc, VALUE* argv, VALUE self)
 {
+    if(argc > 2) {
+        rb_raise(rb_eArgError, "wrong number of arguments (%d for 0..2)", argc);
+    }
+
     VALUE io = Qnil;
     VALUE options = Qnil;
 
-    if(argc == 0 || (argc == 1 && argv[0] == Qnil)) {
-        /* Qnil */
-
-    } else if(argc == 1) {
-        VALUE v = argv[0];
-        if(rb_type(v) == T_HASH) {
-            options = v;
-        } else {
-            io = v;
-        }
-
-    } else if(argc == 2) {
+    if(argc >= 1) {
         io = argv[0];
+    }
+
+    if(argc == 2) {
         options = argv[1];
-        if(rb_type(options) != T_HASH) {
-            rb_raise(rb_eArgError, "expected Hash but found %s.", rb_obj_classname(options));
-        }
+    }
 
-    } else {
-        rb_raise(rb_eArgError, "wrong number of arguments (%d for 0..2)", argc);
+    if (options == Qnil && rb_type(io) == T_HASH) {
+        options = io;
+        io = Qnil;
+    }
+
+    if(options != Qnil) {
+        Check_Type(options, T_HASH);
     }
 
     PACKER(self, pk);
@@ -103,7 +104,7 @@ VALUE MessagePack_Packer_initialize(int argc, VALUE* argv, VALUE self)
     if(options != Qnil) {
         VALUE v;
 
-        v = rb_hash_aref(options, ID2SYM(rb_intern("compatibility_mode")));
+        v = rb_hash_aref(options, sym_compatibility_mode);
         msgpack_packer_set_compat(pk, RTEST(v));
     }
 
@@ -321,7 +322,7 @@ static VALUE Packer_write_to(VALUE self, VALUE io)
 {
     PACKER(self, pk);
     size_t sz = msgpack_buffer_flush_to_io(PACKER_BUFFER_(pk), io, s_write, true);
-    return ULONG2NUM(sz);
+    return SIZET2NUM(sz);
 }
 
 //static VALUE Packer_append(VALUE self, VALUE string_or_buffer)
@@ -380,7 +381,7 @@ static VALUE Packer_register_type(int argc, VALUE* argv, VALUE self)
         rb_raise(rb_eArgError, "expected Module/Class but found %s.", rb_obj_classname(ext_module));
     }
 
-    msgpack_packer_ext_registry_put(&pk->ext_registry, ext_module, ext_type, proc, arg);
+    msgpack_packer_ext_registry_put(&pk->ext_registry, ext_module, ext_type, 0, proc, arg);
 
     if (ext_module == rb_cSymbol) {
         pk->has_symbol_ext_type = true;
@@ -412,6 +413,8 @@ void MessagePack_Packer_module_init(VALUE mMessagePack)
     s_to_msgpack = rb_intern("to_msgpack");
     s_write = rb_intern("write");
 
+    sym_compatibility_mode = ID2SYM(rb_intern("compatibility_mode"));
+
     msgpack_packer_static_init();
     msgpack_packer_ext_registry_static_init();
 

data/ext/msgpack/packer_ext_registry.c

@@ -43,12 +43,18 @@ 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)
 {
-    dst->hash = RTEST(src->hash) ? rb_hash_dup(src->hash) : Qnil;
-    dst->cache = RTEST(src->cache) ? rb_hash_dup(src->cache): Qnil;
+    if(RTEST(src->hash) && !rb_obj_frozen_p(src->hash)) {
+        dst->hash = rb_hash_dup(src->hash);
+        dst->cache = RTEST(src->cache) ? rb_hash_dup(src->cache) : Qnil;
+    } else {
+        // If the type registry is frozen we can safely share it, and share the cache as well.
+        dst->hash = src->hash;
+        dst->cache = src->cache;
+    }
 }
 
 VALUE msgpack_packer_ext_registry_put(msgpack_packer_ext_registry_t* pkrg,
-        VALUE ext_module, int ext_type, VALUE proc, VALUE arg)
+        VALUE ext_module, int ext_type, int flags, VALUE proc, VALUE arg)
 {
     if (!RTEST(pkrg->hash)) {
         pkrg->hash = rb_hash_new();
@@ -58,5 +64,8 @@ VALUE msgpack_packer_ext_registry_put(msgpack_packer_ext_registry_t* pkrg,
         /* clear lookup cache not to miss added type */
         rb_hash_clear(pkrg->cache);
     }
-    return rb_hash_aset(pkrg->hash, ext_module, rb_ary_new3(3, INT2FIX(ext_type), proc, arg));
+
+    // TODO: Ruby embeded array limit is 3, merging `proc` and `arg` would be good.
+    VALUE entry = rb_ary_new3(4, INT2FIX(ext_type), proc, arg, INT2FIX(flags));
+    return rb_hash_aset(pkrg->hash, ext_module, entry);
 }

data/ext/msgpack/packer_ext_registry.h

@@ -21,6 +21,8 @@
 #include "compat.h"
 #include "ruby.h"
 
+#define MSGPACK_EXT_RECURSIVE 0b0001
+
 struct msgpack_packer_ext_registry_t;
 typedef struct msgpack_packer_ext_registry_t msgpack_packer_ext_registry_t;
 
@@ -44,7 +46,7 @@ void msgpack_packer_ext_registry_dup(msgpack_packer_ext_registry_t* src,
         msgpack_packer_ext_registry_t* dst);
 
 VALUE msgpack_packer_ext_registry_put(msgpack_packer_ext_registry_t* pkrg,
-        VALUE ext_module, int ext_type, VALUE proc, VALUE arg);
+        VALUE ext_module, int ext_type, int flags, VALUE proc, VALUE arg);
 
 static int msgpack_packer_ext_find_superclass(VALUE key, VALUE value, VALUE arg)
 {
@@ -60,12 +62,13 @@ static int msgpack_packer_ext_find_superclass(VALUE key, VALUE value, VALUE arg)
 }
 
 static inline VALUE msgpack_packer_ext_registry_fetch(msgpack_packer_ext_registry_t* pkrg,
-        VALUE lookup_class, int* ext_type_result)
+        VALUE lookup_class, int* ext_type_result, int* ext_flags_result)
 {
     // fetch lookup_class from hash, which is a hash to register classes
     VALUE type = rb_hash_lookup(pkrg->hash, lookup_class);
     if(type != Qnil) {
         *ext_type_result = FIX2INT(rb_ary_entry(type, 0));
+        *ext_flags_result = FIX2INT(rb_ary_entry(type, 3));
         return rb_ary_entry(type, 1);
     }
 
@@ -74,6 +77,7 @@ static inline VALUE msgpack_packer_ext_registry_fetch(msgpack_packer_ext_registr
         VALUE type_inht = rb_hash_lookup(pkrg->cache, lookup_class);
         if(type_inht != Qnil) {
             *ext_type_result = FIX2INT(rb_ary_entry(type_inht, 0));
+            *ext_flags_result = FIX2INT(rb_ary_entry(type_inht, 3));
             return rb_ary_entry(type_inht, 1);
         }
     }
@@ -82,7 +86,7 @@ static inline VALUE msgpack_packer_ext_registry_fetch(msgpack_packer_ext_registr
 }
 
 static inline VALUE msgpack_packer_ext_registry_lookup(msgpack_packer_ext_registry_t* pkrg,
-        VALUE instance, int* ext_type_result)
+        VALUE instance, int* ext_type_result, int* ext_flags_result)
 {
     VALUE type;
 
@@ -97,7 +101,7 @@ static inline VALUE msgpack_packer_ext_registry_lookup(msgpack_packer_ext_regist
     * `rb_class_of` returns the singleton_class if the object has one, or the "real class" otherwise.
     */
     VALUE lookup_class = rb_class_of(instance);
-    type = msgpack_packer_ext_registry_fetch(pkrg, lookup_class, ext_type_result);
+    type = msgpack_packer_ext_registry_fetch(pkrg, lookup_class, ext_type_result, ext_flags_result);
     if(type != Qnil) {
         return type;
     }
@@ -108,7 +112,7 @@ static inline VALUE msgpack_packer_ext_registry_lookup(msgpack_packer_ext_regist
      */
     VALUE real_class = rb_obj_class(instance);
     if(lookup_class != real_class) {
-        type = msgpack_packer_ext_registry_fetch(pkrg, real_class, ext_type_result);
+        type = msgpack_packer_ext_registry_fetch(pkrg, real_class, ext_type_result, ext_flags_result);
         if(type != Qnil) {
             return type;
         }
@@ -130,6 +134,7 @@ static inline VALUE msgpack_packer_ext_registry_lookup(msgpack_packer_ext_regist
         }
         rb_hash_aset(pkrg->cache, lookup_class, superclass_type);
         *ext_type_result = FIX2INT(rb_ary_entry(superclass_type, 0));
+        *ext_flags_result = FIX2INT(rb_ary_entry(superclass_type, 3));
         return rb_ary_entry(superclass_type, 1);
     }