msgpack 1.4.2 → 1.7.2

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (95) hide show
  1. checksums.yaml +4 -4
  2. data/ChangeLog +85 -0
  3. data/README.md +52 -1
  4. data/ext/java/org/msgpack/jruby/Buffer.java +26 -19
  5. data/ext/java/org/msgpack/jruby/Decoder.java +29 -21
  6. data/ext/java/org/msgpack/jruby/Encoder.java +68 -30
  7. data/ext/java/org/msgpack/jruby/ExtensionRegistry.java +43 -64
  8. data/ext/java/org/msgpack/jruby/ExtensionValue.java +6 -9
  9. data/ext/java/org/msgpack/jruby/Factory.java +43 -42
  10. data/ext/java/org/msgpack/jruby/Packer.java +37 -40
  11. data/ext/java/org/msgpack/jruby/Unpacker.java +80 -73
  12. data/ext/msgpack/buffer.c +54 -74
  13. data/ext/msgpack/buffer.h +21 -18
  14. data/ext/msgpack/buffer_class.c +161 -52
  15. data/ext/msgpack/buffer_class.h +1 -0
  16. data/ext/msgpack/compat.h +0 -99
  17. data/ext/msgpack/extconf.rb +25 -46
  18. data/ext/msgpack/factory_class.c +143 -87
  19. data/ext/msgpack/packer.c +66 -43
  20. data/ext/msgpack/packer.h +25 -20
  21. data/ext/msgpack/packer_class.c +102 -130
  22. data/ext/msgpack/packer_class.h +11 -0
  23. data/ext/msgpack/packer_ext_registry.c +35 -40
  24. data/ext/msgpack/packer_ext_registry.h +41 -38
  25. data/ext/msgpack/rbinit.c +1 -1
  26. data/ext/msgpack/rmem.c +3 -4
  27. data/ext/msgpack/sysdep.h +5 -2
  28. data/ext/msgpack/unpacker.c +126 -108
  29. data/ext/msgpack/unpacker.h +16 -13
  30. data/ext/msgpack/unpacker_class.c +86 -126
  31. data/ext/msgpack/unpacker_class.h +11 -0
  32. data/ext/msgpack/unpacker_ext_registry.c +40 -28
  33. data/ext/msgpack/unpacker_ext_registry.h +21 -18
  34. data/lib/msgpack/bigint.rb +69 -0
  35. data/lib/msgpack/buffer.rb +9 -0
  36. data/lib/msgpack/factory.rb +140 -10
  37. data/lib/msgpack/packer.rb +10 -1
  38. data/lib/msgpack/symbol.rb +21 -4
  39. data/lib/msgpack/time.rb +1 -1
  40. data/lib/msgpack/unpacker.rb +14 -1
  41. data/lib/msgpack/version.rb +1 -1
  42. data/lib/msgpack.rb +6 -7
  43. data/msgpack.gemspec +8 -5
  44. metadata +37 -82
  45. data/.gitignore +0 -23
  46. data/.rubocop.yml +0 -36
  47. data/.travis.yml +0 -39
  48. data/Gemfile +0 -9
  49. data/Rakefile +0 -71
  50. data/appveyor.yml +0 -18
  51. data/bench/pack.rb +0 -23
  52. data/bench/pack_log.rb +0 -33
  53. data/bench/pack_log_long.rb +0 -65
  54. data/bench/pack_symbols.rb +0 -28
  55. data/bench/run.sh +0 -14
  56. data/bench/run_long.sh +0 -35
  57. data/bench/run_symbols.sh +0 -26
  58. data/bench/unpack.rb +0 -21
  59. data/bench/unpack_log.rb +0 -34
  60. data/bench/unpack_log_long.rb +0 -67
  61. data/doclib/msgpack/buffer.rb +0 -193
  62. data/doclib/msgpack/core_ext.rb +0 -101
  63. data/doclib/msgpack/error.rb +0 -19
  64. data/doclib/msgpack/extension_value.rb +0 -9
  65. data/doclib/msgpack/factory.rb +0 -101
  66. data/doclib/msgpack/packer.rb +0 -208
  67. data/doclib/msgpack/time.rb +0 -22
  68. data/doclib/msgpack/timestamp.rb +0 -44
  69. data/doclib/msgpack/unpacker.rb +0 -183
  70. data/doclib/msgpack.rb +0 -87
  71. data/msgpack.org.md +0 -46
  72. data/spec/cases.json +0 -1
  73. data/spec/cases.msg +0 -0
  74. data/spec/cases_compact.msg +0 -0
  75. data/spec/cases_spec.rb +0 -39
  76. data/spec/cruby/buffer_io_spec.rb +0 -255
  77. data/spec/cruby/buffer_packer.rb +0 -29
  78. data/spec/cruby/buffer_spec.rb +0 -575
  79. data/spec/cruby/buffer_unpacker.rb +0 -19
  80. data/spec/cruby/unpacker_spec.rb +0 -70
  81. data/spec/ext_value_spec.rb +0 -99
  82. data/spec/exttypes.rb +0 -51
  83. data/spec/factory_spec.rb +0 -367
  84. data/spec/format_spec.rb +0 -301
  85. data/spec/jruby/benchmarks/shootout_bm.rb +0 -73
  86. data/spec/jruby/benchmarks/symbolize_keys_bm.rb +0 -25
  87. data/spec/jruby/unpacker_spec.rb +0 -186
  88. data/spec/msgpack_spec.rb +0 -214
  89. data/spec/pack_spec.rb +0 -61
  90. data/spec/packer_spec.rb +0 -557
  91. data/spec/random_compat.rb +0 -24
  92. data/spec/spec_helper.rb +0 -55
  93. data/spec/timestamp_spec.rb +0 -121
  94. data/spec/unpack_spec.rb +0 -57
  95. data/spec/unpacker_spec.rb +0 -819
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: 451ed743fe37bd3726f897f4daae9e7fd9b2cb065a34cf438ca786350060e0ec
4
- data.tar.gz: 87335c4b1d35f9ac88af648963b1fe60caa72059ec7e6c1b6b507027da9118d8
3
+ metadata.gz: 2f5b1af6b3a51f5ccc6bcf67c94c1fc6193b02fe01b123e2cfb06a6df9607116
4
+ data.tar.gz: cc057f24e1ffa4cdc3e331499eb04de4c2383b0657dcf0baeba08300fd20862e
5
5
  SHA512:
6
- metadata.gz: eb519049e9d8e4b36682e4b2c2625f10172745ce74ad1cfc8e6d93468b219071ed579bcfd36b7f0e1210eeb177197b435da2a028f46d2c71c73a3bcf2b53dcf8
7
- data.tar.gz: e8c4765ac836226f5a2ce68a62194f112fc77bcdd9744d39040ff2db5a2c27d876348fb6002110c00b122bccdff2d539223d82ff73ea0bcb4cf1a3d9fb62defe
6
+ metadata.gz: 3eb06321a534ca9b16e321cc4a71458532578dafe7967314a662223b1fbf4aa93449c98177fa982aa532ce3732ddda4a6d497704df0e9c874da07f378c73595c
7
+ data.tar.gz: 8e540755e3db9e21d7dfa4354854e8b0486f5a1bbf82c3994c6095022205f7873153d364df9310d8072c481de38ca2b4c3e088e4221c3451ceb9438312489419
data/ChangeLog CHANGED
@@ -1,3 +1,88 @@
1
+ 2023-07-18 1.7.2:
2
+
3
+ * Fix a potential GC bug when packing data using recursive extensions and buffers containing over 512KkiB of data (See #341).
4
+ * Fix a regression where feeding an empty string to an Unpacker would be considered like the end of the buffer.
5
+
6
+ 2023-05-19 1.7.1:
7
+
8
+ * Fix JRuby 9.4 compatibility.
9
+ * Fix compilation on older compilers (gcc 4.x).
10
+ * Fix an infinite recursion issue when registering a Symbol type with a `nil` packer.
11
+
12
+ 2023-03-29 1.7.0:
13
+
14
+ * Fix a possible double-free issue when GC triggers inside `_msgpack_rmem_alloc2`.
15
+ * `Unpacker#feed` now always directly read in provided strings instead of copying content in its buffer.
16
+ * `Unpacker#feed` is now an alias of `Unpacker#feed_reference`.
17
+ * Implement `Factory::Pool#unpacker` and `Factory::Pool#packer` to allow for more precise serialization.
18
+ * Require Ruby 2.5+.
19
+
20
+ 2023-03-03 1.6.1:
21
+
22
+ * Undefine `#clone` and `#dup` on `MessagePack::Buffer`, `MessagePack::Packer` and `MessagePack::Unpacker`.
23
+ These methods were never intended, and using them could cause leaks or crashes or worse.
24
+ * Fix a possible GC crash when GC trigger inside `MessagePack::Buffer.new` (#314).
25
+
26
+ 2022-09-30 1.6.0:
27
+
28
+ * Fix a potential use-after-free bug in Buffer_free when accessing a packer or unpacker buffer.
29
+ * `old-style-definition` compilation warnings.
30
+ * Restore zero-copy buffer feed when provided a Ruby string. This was accidentally broken in 1.5.4.
31
+ * Provide implementations for `ObjectSpace.memsize`. Message pack objects now properly report their size to Ruby.
32
+ * Fix an endianess bug on Windows platform.
33
+
34
+ 2022-08-23 1.5.6:
35
+
36
+ * No actual code change, just re-release the `java` version properly.
37
+
38
+ 2022-08-22 1.5.5:
39
+
40
+ * Fix a segfault when GC triggers inside a recursive extension.
41
+
42
+ 2022-07-25 1.5.4:
43
+
44
+ * Fix a segfault when deserializing empty symbol (`:""`).
45
+ * Improve compilation flags to not strip debug symbols.
46
+
47
+ 2022-05-30 version 1.5.3:
48
+
49
+ * Fix deduplication of empty strings when using the `freeze: true` option.
50
+ * Use `rb_hash_new_capa` when available (Ruby 3.2) for improved performance when parsing large hashes.
51
+
52
+ 2022-05-27 version 1.5.2:
53
+
54
+ * Fix bug about unpacking ext type objects with the recursive option
55
+
56
+ 2022-04-07 version 1.5.1:
57
+
58
+ * Fix bug about packing/unpacking ext type objects with the recursive option
59
+
60
+ 2022-04-06 version 1.5.0:
61
+
62
+ * Add recursive option on Factory#register_type to operate Packer/Unpacker manually
63
+ * Add oversized_integer_extension option on Factory#register_type to pack/unpack bigint using ext types
64
+ * Add Factory#pool method and Factory::Pool class to provide pooled Packer and Unpacker instances
65
+
66
+ 2022-02-15 version 1.4.5:
67
+
68
+ * Fix to create UTF-8 Symbol keys when symbolize_keys: true
69
+ * Fix to assume Symbols as US-ASCII or UTF-8
70
+ * Optimize Packer/Unpacker initialization
71
+ * Optimize extension class lookup
72
+ * Rename Packer#clear as Packer#reset (#clear is still available as an alias)
73
+
74
+ 2022-01-22 version 1.4.4:
75
+
76
+ * Specify the build option --platform=8 for older Java platforms
77
+
78
+ 2022-01-20 version 1.4.3:
79
+
80
+ * Optimize serialization/deserialization of Symbols
81
+ * Support registering ext types for objects of subclasses of primitive types (like Hash)
82
+ * Add optimized_symbols_parsing option to Factory#register_type on MRI implementation
83
+ * Optimize to deduplicate Hash keys on JRuby
84
+ * Support JRuby 9.3 (and drop 9.1)
85
+
1
86
  2021-02-01 version 1.4.2:
2
87
 
3
88
  * Add the required Ruby version (>= 2.4) to avoid compilation errors on older Ruby runtimes
data/README.md CHANGED
@@ -40,7 +40,7 @@ or build msgpack-ruby and install:
40
40
  MessagePack for Ruby should run on x86, ARM, PowerPC, SPARC and other CPU architectures.
41
41
 
42
42
  And it works with MRI (CRuby) and Rubinius.
43
- Patches to improve portability is highly welcomed.
43
+ Patches to improve portability are highly welcomed.
44
44
 
45
45
 
46
46
  ## Serializing objects
@@ -51,6 +51,7 @@ Use `MessagePack.pack` or `to_msgpack`:
51
51
  require 'msgpack'
52
52
  msg = MessagePack.pack(obj) # or
53
53
  msg = obj.to_msgpack
54
+ File.binwrite('mydata.msgpack', msg)
54
55
  ```
55
56
 
56
57
  ### Streaming serialization
@@ -71,6 +72,7 @@ Use `MessagePack.unpack`:
71
72
 
72
73
  ```ruby
73
74
  require 'msgpack'
75
+ msg = File.binread('mydata.msgpack')
74
76
  obj = MessagePack.unpack(msg)
75
77
  ```
76
78
 
@@ -187,6 +189,55 @@ MessagePack::DefaultFactory.register_type(0x03, MyClass3)
187
189
  MessagePack.unpack(data_with_ext_typeid_03) #=> MyClass3 instance
188
190
  ```
189
191
 
192
+ Alternatively, extension types can call the packer or unpacker recursively to generate the extension data:
193
+
194
+ ```ruby
195
+ Point = Struct.new(:x, :y)
196
+ factory = MessagePack::Factory.new
197
+ factory.register_type(
198
+ 0x01,
199
+ Point,
200
+ packer: ->(point, packer) {
201
+ packer.write(point.x)
202
+ packer.write(point.y)
203
+ },
204
+ unpacker: ->(unpacker) {
205
+ x = unpacker.read
206
+ y = unpacker.read
207
+ Point.new(x, y)
208
+ },
209
+ recursive: true,
210
+ )
211
+ factory.load(factory.dump(Point.new(12, 34))) # => #<struct Point x=12, y=34>
212
+ ```
213
+
214
+ ## Pooling
215
+
216
+ Creating `Packer` and `Unpacker` objects is expensive. For best performance it is preferable to re-use these objects.
217
+
218
+ `MessagePack::Factory#pool` makes that easier:
219
+
220
+ ```ruby
221
+ factory = MessagePack::Factory.new
222
+ factory.register_type(
223
+ 0x01,
224
+ Point,
225
+ packer: ->(point, packer) {
226
+ packer.write(point.x)
227
+ packer.write(point.y)
228
+ },
229
+ unpacker: ->(unpacker) {
230
+ x = unpacker.read
231
+ y = unpacker.read
232
+ Point.new(x, y)
233
+ },
234
+ recursive: true,
235
+ )
236
+ pool = factory.pool(5) # The pool size should match the number of threads expected to use the factory concurrently.
237
+
238
+ pool.load(pool.dump(Point.new(12, 34))) # => #<struct Point x=12, y=34>
239
+ ```
240
+
190
241
  ## Buffer API
191
242
 
192
243
  MessagePack for Ruby provides a buffer API so that you can read or write data by hand, not via Packer or Unpacker API.
@@ -21,10 +21,11 @@ import org.jcodings.Encoding;
21
21
 
22
22
  @JRubyClass(name="MessagePack::Buffer")
23
23
  public class Buffer extends RubyObject {
24
- private IRubyObject io;
25
- private ByteBuffer buffer;
24
+ private static final long serialVersionUID = 8441244627425629412L;
25
+ private transient IRubyObject io;
26
+ private transient ByteBuffer buffer;
26
27
  private boolean writeMode;
27
- private Encoding binaryEncoding;
28
+ private transient Encoding binaryEncoding;
28
29
 
29
30
  private static final int CACHE_LINE_SIZE = 64;
30
31
  private static final int ARRAY_HEADER_SIZE = 24;
@@ -49,7 +50,7 @@ public class Buffer extends RubyObject {
49
50
  }
50
51
  this.buffer = ByteBuffer.allocate(CACHE_LINE_SIZE - ARRAY_HEADER_SIZE);
51
52
  this.writeMode = true;
52
- this.binaryEncoding = ctx.getRuntime().getEncodingService().getAscii8bitEncoding();
53
+ this.binaryEncoding = ctx.runtime.getEncodingService().getAscii8bitEncoding();
53
54
  return this;
54
55
  }
55
56
 
@@ -87,17 +88,17 @@ public class Buffer extends RubyObject {
87
88
  writeMode = true;
88
89
  }
89
90
  buffer.clear();
90
- return ctx.getRuntime().getNil();
91
+ return ctx.runtime.getNil();
91
92
  }
92
93
 
93
94
  @JRubyMethod(name = "size")
94
95
  public IRubyObject size(ThreadContext ctx) {
95
- return ctx.getRuntime().newFixnum(rawSize());
96
+ return ctx.runtime.newFixnum(rawSize());
96
97
  }
97
98
 
98
99
  @JRubyMethod(name = "empty?")
99
100
  public IRubyObject isEmpty(ThreadContext ctx) {
100
- return rawSize() == 0 ? ctx.getRuntime().getTrue() : ctx.getRuntime().getFalse();
101
+ return rawSize() == 0 ? ctx.runtime.getTrue() : ctx.runtime.getFalse();
101
102
  }
102
103
 
103
104
  private IRubyObject bufferWrite(ThreadContext ctx, IRubyObject str) {
@@ -105,7 +106,7 @@ public class Buffer extends RubyObject {
105
106
  int length = bytes.length();
106
107
  ensureRemainingCapacity(length);
107
108
  buffer.put(bytes.unsafeBytes(), bytes.begin(), length);
108
- return ctx.getRuntime().newFixnum(length);
109
+ return ctx.runtime.newFixnum(length);
109
110
 
110
111
  }
111
112
 
@@ -131,19 +132,19 @@ public class Buffer extends RubyObject {
131
132
  length = (int) args[0].convertToInteger().getLongValue();
132
133
  }
133
134
  if (raiseOnUnderflow && rawSize() < length) {
134
- throw ctx.getRuntime().newEOFError();
135
+ throw ctx.runtime.newEOFError();
135
136
  }
136
137
  int readLength = Math.min(length, rawSize());
137
138
  if (readLength == 0 && length > 0) {
138
- return ctx.getRuntime().getNil();
139
+ return ctx.runtime.getNil();
139
140
  } else if (readLength == 0) {
140
- return ctx.getRuntime().newString();
141
+ return ctx.runtime.newString();
141
142
  } else {
142
143
  ensureReadMode();
143
144
  byte[] bytes = new byte[readLength];
144
145
  buffer.get(bytes);
145
146
  ByteList byteList = new ByteList(bytes, binaryEncoding);
146
- return ctx.getRuntime().newString(byteList);
147
+ return ctx.runtime.newString(byteList);
147
148
  }
148
149
  }
149
150
 
@@ -161,12 +162,12 @@ public class Buffer extends RubyObject {
161
162
  feed(ctx);
162
163
  int length = (int) _length.convertToInteger().getLongValue();
163
164
  if (raiseOnUnderflow && rawSize() < length) {
164
- throw ctx.getRuntime().newEOFError();
165
+ throw ctx.runtime.newEOFError();
165
166
  }
166
167
  ensureReadMode();
167
168
  int skipLength = Math.min(length, rawSize());
168
169
  buffer.position(buffer.position() + skipLength);
169
- return ctx.getRuntime().newFixnum(skipLength);
170
+ return ctx.runtime.newFixnum(skipLength);
170
171
  }
171
172
 
172
173
  @JRubyMethod(name = "skip")
@@ -188,23 +189,23 @@ public class Buffer extends RubyObject {
188
189
  ensureReadMode();
189
190
  int length = buffer.limit() - buffer.position();
190
191
  ByteList str = new ByteList(buffer.array(), buffer.position(), length, binaryEncoding, true);
191
- return ctx.getRuntime().newString(str);
192
+ return ctx.runtime.newString(str);
192
193
  }
193
194
 
194
195
  @JRubyMethod(name = "to_a")
195
196
  public IRubyObject toA(ThreadContext ctx) {
196
- return ctx.getRuntime().newArray(toS(ctx));
197
+ return ctx.runtime.newArray(toS(ctx));
197
198
  }
198
199
 
199
200
  @JRubyMethod(name = "io")
200
201
  public IRubyObject getIo(ThreadContext ctx) {
201
- return io == null ? ctx.getRuntime().getNil() : io;
202
+ return io == null ? ctx.runtime.getNil() : io;
202
203
  }
203
204
 
204
205
  @JRubyMethod(name = "flush")
205
206
  public IRubyObject flush(ThreadContext ctx) {
206
207
  if (io == null) {
207
- return ctx.getRuntime().getNil();
208
+ return ctx.runtime.getNil();
208
209
  } else {
209
210
  return io.callMethod(ctx, "flush");
210
211
  }
@@ -213,7 +214,7 @@ public class Buffer extends RubyObject {
213
214
  @JRubyMethod(name = "close")
214
215
  public IRubyObject close(ThreadContext ctx) {
215
216
  if (io == null) {
216
- return ctx.getRuntime().getNil();
217
+ return ctx.runtime.getNil();
217
218
  } else {
218
219
  return io.callMethod(ctx, "close");
219
220
  }
@@ -223,4 +224,10 @@ public class Buffer extends RubyObject {
223
224
  public IRubyObject writeTo(ThreadContext ctx, IRubyObject io) {
224
225
  return io.callMethod(ctx, "write", readCommon(ctx, null, false));
225
226
  }
227
+
228
+ public ByteList getBytes() {
229
+ byte[] bytes = new byte[rawSize()];
230
+ buffer.get(bytes);
231
+ return new ByteList(bytes, binaryEncoding);
232
+ }
226
233
  }
@@ -14,6 +14,7 @@ import org.jruby.RubyBignum;
14
14
  import org.jruby.RubyString;
15
15
  import org.jruby.RubyArray;
16
16
  import org.jruby.RubyHash;
17
+ import org.jruby.RubyInteger;
17
18
  import org.jruby.exceptions.RaiseException;
18
19
  import org.jruby.runtime.builtin.IRubyObject;
19
20
  import org.jruby.util.ByteList;
@@ -35,7 +36,7 @@ public class Decoder implements Iterator<IRubyObject> {
35
36
  private final RubyClass unexpectedTypeErrorClass;
36
37
  private final RubyClass unknownExtTypeErrorClass;
37
38
 
38
- private ExtensionRegistry registry;
39
+ private Unpacker unpacker;
39
40
  private ByteBuffer buffer;
40
41
  private boolean symbolizeKeys;
41
42
  private boolean freeze;
@@ -45,29 +46,29 @@ public class Decoder implements Iterator<IRubyObject> {
45
46
  this(runtime, null, new byte[] {}, 0, 0, false, false, false);
46
47
  }
47
48
 
48
- public Decoder(Ruby runtime, ExtensionRegistry registry) {
49
- this(runtime, registry, new byte[] {}, 0, 0, false, false, false);
49
+ public Decoder(Ruby runtime, Unpacker unpacker) {
50
+ this(runtime, unpacker, new byte[] {}, 0, 0, false, false, false);
50
51
  }
51
52
 
52
53
  public Decoder(Ruby runtime, byte[] bytes) {
53
54
  this(runtime, null, bytes, 0, bytes.length, false, false, false);
54
55
  }
55
56
 
56
- public Decoder(Ruby runtime, ExtensionRegistry registry, byte[] bytes) {
57
- this(runtime, registry, bytes, 0, bytes.length, false, false, false);
57
+ public Decoder(Ruby runtime, Unpacker unpacker, byte[] bytes) {
58
+ this(runtime, unpacker, bytes, 0, bytes.length, false, false, false);
58
59
  }
59
60
 
60
- public Decoder(Ruby runtime, ExtensionRegistry registry, byte[] bytes, boolean symbolizeKeys, boolean freeze, boolean allowUnknownExt) {
61
- this(runtime, registry, bytes, 0, bytes.length, symbolizeKeys, freeze, allowUnknownExt);
61
+ public Decoder(Ruby runtime, Unpacker unpacker, byte[] bytes, boolean symbolizeKeys, boolean freeze, boolean allowUnknownExt) {
62
+ this(runtime, unpacker, bytes, 0, bytes.length, symbolizeKeys, freeze, allowUnknownExt);
62
63
  }
63
64
 
64
- public Decoder(Ruby runtime, ExtensionRegistry registry, byte[] bytes, int offset, int length) {
65
- this(runtime, registry, bytes, offset, length, false, false, false);
65
+ public Decoder(Ruby runtime, Unpacker unpacker, byte[] bytes, int offset, int length) {
66
+ this(runtime, unpacker, bytes, offset, length, false, false, false);
66
67
  }
67
68
 
68
- public Decoder(Ruby runtime, ExtensionRegistry registry, byte[] bytes, int offset, int length, boolean symbolizeKeys, boolean freeze, boolean allowUnknownExt) {
69
+ public Decoder(Ruby runtime, Unpacker unpacker, byte[] bytes, int offset, int length, boolean symbolizeKeys, boolean freeze, boolean allowUnknownExt) {
69
70
  this.runtime = runtime;
70
- this.registry = registry;
71
+ this.unpacker = unpacker;
71
72
  this.symbolizeKeys = symbolizeKeys;
72
73
  this.freeze = freeze;
73
74
  this.allowUnknownExt = allowUnknownExt;
@@ -122,7 +123,6 @@ public class Decoder implements Iterator<IRubyObject> {
122
123
  ByteList byteList = new ByteList(bytes, encoding);
123
124
  RubyString string = runtime.newString(byteList);
124
125
  if (this.freeze) {
125
- string.setFrozen(true);
126
126
  string = runtime.freezeAndDedupString(string);
127
127
  }
128
128
  return string;
@@ -140,9 +140,14 @@ public class Decoder implements Iterator<IRubyObject> {
140
140
  RubyHash hash = RubyHash.newHash(runtime);
141
141
  for (int i = 0; i < size; i++) {
142
142
  IRubyObject key = next();
143
- if (this.symbolizeKeys && key instanceof RubyString) {
143
+ if (key instanceof RubyString) {
144
+ if (this.symbolizeKeys) {
144
145
  key = ((RubyString) key).intern();
146
+ } else {
147
+ key = runtime.freezeAndDedupString((RubyString) key);
148
+ }
145
149
  }
150
+
146
151
  hash.fastASet(key, next());
147
152
  }
148
153
  return hash;
@@ -150,18 +155,21 @@ public class Decoder implements Iterator<IRubyObject> {
150
155
 
151
156
  private IRubyObject consumeExtension(int size) {
152
157
  int type = buffer.get();
153
- byte[] payload = readBytes(size);
154
-
155
- if (registry != null) {
156
- IRubyObject proc = registry.lookupUnpackerByTypeId(type);
157
- if (proc != null) {
158
- ByteList byteList = new ByteList(payload, runtime.getEncodingService().getAscii8bitEncoding());
159
- return proc.callMethod(runtime.getCurrentContext(), "call", runtime.newString(byteList));
158
+ if (unpacker != null) {
159
+ ExtensionRegistry.ExtensionEntry entry = unpacker.lookupExtensionByTypeId(type);
160
+ if (entry != null) {
161
+ IRubyObject proc = entry.getUnpackerProc();
162
+ if (entry.isRecursive()) {
163
+ return proc.callMethod(runtime.getCurrentContext(), "call", unpacker);
164
+ } else {
165
+ ByteList byteList = new ByteList(readBytes(size), runtime.getEncodingService().getAscii8bitEncoding());
166
+ return proc.callMethod(runtime.getCurrentContext(), "call", runtime.newString(byteList));
167
+ }
160
168
  }
161
169
  }
162
170
 
163
171
  if (this.allowUnknownExt) {
164
- return ExtensionValue.newExtensionValue(runtime, type, payload);
172
+ return ExtensionValue.newExtensionValue(runtime, type, readBytes(size));
165
173
  }
166
174
 
167
175
  throw runtime.newRaiseException(unknownExtTypeErrorClass, "unexpected extension type");
@@ -38,12 +38,16 @@ public class Encoder {
38
38
  private final Encoding utf8Encoding;
39
39
  private final boolean compatibilityMode;
40
40
  private final ExtensionRegistry registry;
41
+ private final Packer packer;
41
42
 
42
43
  public boolean hasSymbolExtType;
44
+ private boolean hasBigintExtType;
45
+ private boolean recursiveExtension;
43
46
 
44
47
  private ByteBuffer buffer;
45
48
 
46
- public Encoder(Ruby runtime, boolean compatibilityMode, ExtensionRegistry registry, boolean hasSymbolExtType) {
49
+ public Encoder(Ruby runtime, Packer packer, boolean compatibilityMode, ExtensionRegistry registry, boolean hasSymbolExtType, boolean hasBigintExtType) {
50
+ this.packer = packer;
47
51
  this.runtime = runtime;
48
52
  this.buffer = ByteBuffer.allocate(CACHE_LINE_SIZE - ARRAY_HEADER_SIZE);
49
53
  this.binaryEncoding = runtime.getEncodingService().getAscii8bitEncoding();
@@ -51,6 +55,7 @@ public class Encoder {
51
55
  this.compatibilityMode = compatibilityMode;
52
56
  this.registry = registry;
53
57
  this.hasSymbolExtType = hasSymbolExtType;
58
+ this.hasBigintExtType = hasBigintExtType;
54
59
  }
55
60
 
56
61
  public boolean isCompatibilityMode() {
@@ -66,9 +71,17 @@ public class Encoder {
66
71
  }
67
72
 
68
73
  private IRubyObject readRubyString() {
69
- IRubyObject str = runtime.newString(new ByteList(buffer.array(), 0, buffer.position(), binaryEncoding, false));
70
- buffer.clear();
71
- return str;
74
+ if (recursiveExtension) {
75
+ // If recursiveExtension is true, it means we re-entered encode, so we MUST NOT flush the buffer.
76
+ // Instead we return an empty string to act as a null object for the caller. The buffer will actually
77
+ // be flushed once we're done serializing the recursive extension.
78
+ // All other method that consume the buffer should do so through readRubyString or implement the same logic.
79
+ return runtime.newString();
80
+ } else {
81
+ IRubyObject str = runtime.newString(new ByteList(buffer.array(), 0, buffer.position(), binaryEncoding, false));
82
+ buffer.clear();
83
+ return str;
84
+ }
72
85
  }
73
86
 
74
87
  public IRubyObject encode(IRubyObject object) {
@@ -119,7 +132,9 @@ public class Encoder {
119
132
  } else if (object instanceof RubyFloat) {
120
133
  appendFloat((RubyFloat) object);
121
134
  } else if (object instanceof RubyString) {
122
- appendString((RubyString) object);
135
+ if (object.getType() == runtime.getString() || !tryAppendWithExtTypeLookup(object)) {
136
+ appendString((RubyString) object);
137
+ }
123
138
  } else if (object instanceof RubySymbol) {
124
139
  if (hasSymbolExtType) {
125
140
  appendOther(object, destination);
@@ -127,9 +142,13 @@ public class Encoder {
127
142
  appendString(((RubySymbol) object).asString());
128
143
  }
129
144
  } else if (object instanceof RubyArray) {
130
- appendArray((RubyArray) object);
145
+ if (object.getType() == runtime.getArray() || !tryAppendWithExtTypeLookup(object)) {
146
+ appendArray((RubyArray) object);
147
+ }
131
148
  } else if (object instanceof RubyHash) {
132
- appendHash((RubyHash) object);
149
+ if (object.getType() == runtime.getHash() || !tryAppendWithExtTypeLookup(object)) {
150
+ appendHash((RubyHash) object);
151
+ }
133
152
  } else if (object instanceof ExtensionValue) {
134
153
  appendExtensionValue((ExtensionValue) object);
135
154
  } else {
@@ -141,7 +160,10 @@ public class Encoder {
141
160
  BigInteger value = object.getBigIntegerValue();
142
161
  if (value.compareTo(RubyBignum.LONG_MIN) < 0 || value.compareTo(RubyBignum.LONG_MAX) > 0) {
143
162
  if (value.bitLength() > 64 || (value.bitLength() > 63 && value.signum() < 0)) {
144
- throw runtime.newArgumentError(String.format("Cannot pack big integer: %s", value));
163
+ if (hasBigintExtType && tryAppendWithExtTypeLookup(object)) {
164
+ return;
165
+ }
166
+ throw runtime.newRangeError(String.format("Cannot pack big integer: %s", value));
145
167
  }
146
168
  ensureRemainingCapacity(9);
147
169
  buffer.put(value.signum() < 0 ? INT64 : UINT64);
@@ -153,7 +175,7 @@ public class Encoder {
153
175
  }
154
176
 
155
177
  private void appendInteger(RubyInteger object) {
156
- long value = ((RubyInteger) object).getLongValue();
178
+ long value = object.getLongValue();
157
179
  if (value < 0) {
158
180
  if (value < Short.MIN_VALUE) {
159
181
  if (value < Integer.MIN_VALUE) {
@@ -241,7 +263,7 @@ public class Encoder {
241
263
  } else {
242
264
  ensureRemainingCapacity(5 + length);
243
265
  buffer.put(binary ? BIN32 : STR32);
244
- buffer.putInt((int) length);
266
+ buffer.putInt(length);
245
267
  }
246
268
  }
247
269
 
@@ -249,7 +271,7 @@ public class Encoder {
249
271
  Encoding encoding = object.getEncoding();
250
272
  boolean binary = !compatibilityMode && encoding == binaryEncoding;
251
273
  if (encoding != utf8Encoding && encoding != binaryEncoding) {
252
- object = (RubyString) ((RubyString) object).encode(runtime.getCurrentContext(), runtime.getEncodingService().getEncoding(utf8Encoding));
274
+ object = (RubyString)(object).encode(runtime.getCurrentContext(), runtime.getEncodingService().getEncoding(utf8Encoding));
253
275
  }
254
276
  ByteList bytes = object.getByteList();
255
277
  int length = bytes.length();
@@ -257,12 +279,12 @@ public class Encoder {
257
279
  buffer.put(bytes.unsafeBytes(), bytes.begin(), length);
258
280
  }
259
281
 
260
- private void appendArray(RubyArray object) {
282
+ private void appendArray(RubyArray<?> object) {
261
283
  appendArrayHeader(object);
262
284
  appendArrayElements(object);
263
285
  }
264
286
 
265
- private void appendArrayHeader(RubyArray object) {
287
+ private void appendArrayHeader(RubyArray<?> object) {
266
288
  appendArrayHeader(object.size());
267
289
  }
268
290
 
@@ -281,7 +303,7 @@ public class Encoder {
281
303
  }
282
304
  }
283
305
 
284
- private void appendArrayElements(RubyArray object) {
306
+ private void appendArrayElements(RubyArray<?> object) {
285
307
  int size = object.size();
286
308
  for (int i = 0; i < size; i++) {
287
309
  appendObject(object.eltOk(i));
@@ -383,25 +405,41 @@ public class Encoder {
383
405
  appendExt((int) type, payloadBytes);
384
406
  }
385
407
 
386
- private void appendOther(IRubyObject object, IRubyObject destination) {
408
+ private boolean tryAppendWithExtTypeLookup(IRubyObject object) {
387
409
  if (registry != null) {
388
- RubyModule lookupClass;
389
-
390
- if (object.getType() == runtime.getSymbol()) {
391
- lookupClass = object.getType();
392
- } else {
393
- lookupClass = object.getSingletonClass();
394
- }
395
-
396
- IRubyObject[] pair = registry.lookupPackerForObject(object);
397
- if (pair != null) {
398
- RubyString bytes = pair[0].callMethod(runtime.getCurrentContext(), "call", object).asString();
399
- int type = (int) ((RubyFixnum) pair[1]).getLongValue();
400
- appendExt(type, bytes.getByteList());
401
- return;
410
+ ExtensionRegistry.ExtensionEntry entry = registry.lookupExtensionForObject(object);
411
+ if (entry != null) {
412
+ IRubyObject proc = entry.getPackerProc();
413
+ int type = entry.getTypeId();
414
+
415
+ if (entry.isRecursive()) {
416
+ ByteBuffer oldBuffer = buffer;
417
+ buffer = ByteBuffer.allocate(CACHE_LINE_SIZE - ARRAY_HEADER_SIZE);
418
+ boolean previousRecursiveExtension = recursiveExtension;
419
+ recursiveExtension = true;
420
+
421
+ ByteList payload;
422
+ try {
423
+ IRubyObject args[] = { object, packer };
424
+ proc.callMethod(runtime.getCurrentContext(), "call", args);
425
+ payload = new ByteList(buffer.array(), 0, buffer.position(), binaryEncoding, false);
426
+ } finally {
427
+ recursiveExtension = previousRecursiveExtension;
428
+ buffer = oldBuffer;
429
+ }
430
+ appendExt(type, payload);
431
+ } else {
432
+ RubyString bytes = proc.callMethod(runtime.getCurrentContext(), "call", object).asString();
433
+ appendExt(type, bytes.getByteList());
434
+ }
435
+ return true;
402
436
  }
403
437
  }
404
- appendCustom(object, destination);
438
+ return false;
439
+ }
440
+
441
+ private void appendOther(IRubyObject object, IRubyObject destination) {
442
+ if (!tryAppendWithExtTypeLookup(object)) { appendCustom(object, destination); }
405
443
  }
406
444
 
407
445
  private void appendCustom(IRubyObject object, IRubyObject destination) {