1 // Protocol Buffers - Google's data interchange format
2 // Copyright 2014 Google Inc. All rights reserved.
3 // https://developers.google.com/protocol-buffers/
4 //
5 // Redistribution and use in source and binary forms, with or without
6 // modification, are permitted provided that the following conditions are
7 // met:
8 //
9 // * Redistributions of source code must retain the above copyright
10 // notice, this list of conditions and the following disclaimer.
11 // * Redistributions in binary form must reproduce the above
12 // copyright notice, this list of conditions and the following disclaimer
13 // in the documentation and/or other materials provided with the
14 // distribution.
15 // * Neither the name of Google Inc. nor the names of its
16 // contributors may be used to endorse or promote products derived from
17 // this software without specific prior written permission.
18 //
19 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30
31 #include "protobuf.h"
32
33 // -----------------------------------------------------------------------------
34 // Class/module creation from msgdefs and enumdefs, respectively.
35 // -----------------------------------------------------------------------------
36
Message_data(void * msg)37 void* Message_data(void* msg) {
38 return ((uint8_t *)msg) + sizeof(MessageHeader);
39 }
40
Message_mark(void * _self)41 void Message_mark(void* _self) {
42 MessageHeader* self = (MessageHeader *)_self;
43 layout_mark(self->descriptor->layout, Message_data(self));
44 }
45
Message_free(void * self)46 void Message_free(void* self) {
47 xfree(self);
48 }
49
50 rb_data_type_t Message_type = {
51 "Message",
52 { Message_mark, Message_free, NULL },
53 };
54
Message_alloc(VALUE klass)55 VALUE Message_alloc(VALUE klass) {
56 VALUE descriptor = rb_ivar_get(klass, descriptor_instancevar_interned);
57 Descriptor* desc = ruby_to_Descriptor(descriptor);
58 MessageHeader* msg = (MessageHeader*)ALLOC_N(
59 uint8_t, sizeof(MessageHeader) + desc->layout->size);
60 VALUE ret;
61
62 memset(Message_data(msg), 0, desc->layout->size);
63
64 // We wrap first so that everything in the message object is GC-rooted in case
65 // a collection happens during object creation in layout_init().
66 ret = TypedData_Wrap_Struct(klass, &Message_type, msg);
67 msg->descriptor = desc;
68 rb_ivar_set(ret, descriptor_instancevar_interned, descriptor);
69
70 layout_init(desc->layout, Message_data(msg));
71
72 return ret;
73 }
74
which_oneof_field(MessageHeader * self,const upb_oneofdef * o)75 static VALUE which_oneof_field(MessageHeader* self, const upb_oneofdef* o) {
76 upb_oneof_iter it;
77 size_t case_ofs;
78 uint32_t oneof_case;
79 const upb_fielddef* first_field;
80 const upb_fielddef* f;
81
82 // If no fields in the oneof, always nil.
83 if (upb_oneofdef_numfields(o) == 0) {
84 return Qnil;
85 }
86 // Grab the first field in the oneof so we can get its layout info to find the
87 // oneof_case field.
88 upb_oneof_begin(&it, o);
89 assert(!upb_oneof_done(&it));
90 first_field = upb_oneof_iter_field(&it);
91 assert(upb_fielddef_containingoneof(first_field) != NULL);
92
93 case_ofs =
94 self->descriptor->layout->
95 fields[upb_fielddef_index(first_field)].case_offset;
96 oneof_case = *((uint32_t*)((char*)Message_data(self) + case_ofs));
97
98 if (oneof_case == ONEOF_CASE_NONE) {
99 return Qnil;
100 }
101
102 // oneof_case is a field index, so find that field.
103 f = upb_oneofdef_itof(o, oneof_case);
104 assert(f != NULL);
105
106 return ID2SYM(rb_intern(upb_fielddef_name(f)));
107 }
108
109 /*
110 * call-seq:
111 * Message.method_missing(*args)
112 *
113 * Provides accessors and setters for message fields according to their field
114 * names. For any field whose name does not conflict with a built-in method, an
115 * accessor is provided with the same name as the field, and a setter is
116 * provided with the name of the field plus the '=' suffix. Thus, given a
117 * message instance 'msg' with field 'foo', the following code is valid:
118 *
119 * msg.foo = 42
120 * puts msg.foo
121 *
122 * This method also provides read-only accessors for oneofs. If a oneof exists
123 * with name 'my_oneof', then msg.my_oneof will return a Ruby symbol equal to
124 * the name of the field in that oneof that is currently set, or nil if none.
125 */
Message_method_missing(int argc,VALUE * argv,VALUE _self)126 VALUE Message_method_missing(int argc, VALUE* argv, VALUE _self) {
127 MessageHeader* self;
128 VALUE method_name, method_str;
129 char* name;
130 size_t name_len;
131 bool setter;
132 const upb_oneofdef* o;
133 const upb_fielddef* f;
134
135 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
136 if (argc < 1) {
137 rb_raise(rb_eArgError, "Expected method name as first argument.");
138 }
139 method_name = argv[0];
140 if (!SYMBOL_P(method_name)) {
141 rb_raise(rb_eArgError, "Expected symbol as method name.");
142 }
143 method_str = rb_id2str(SYM2ID(method_name));
144 name = RSTRING_PTR(method_str);
145 name_len = RSTRING_LEN(method_str);
146 setter = false;
147
148 // Setters have names that end in '='.
149 if (name[name_len - 1] == '=') {
150 setter = true;
151 name_len--;
152 }
153
154 // See if this name corresponds to either a oneof or field in this message.
155 if (!upb_msgdef_lookupname(self->descriptor->msgdef, name, name_len, &f,
156 &o)) {
157 return rb_call_super(argc, argv);
158 }
159
160 if (o != NULL) {
161 // This is a oneof -- return which field inside the oneof is set.
162 if (setter) {
163 rb_raise(rb_eRuntimeError, "Oneof accessors are read-only.");
164 }
165 return which_oneof_field(self, o);
166 } else {
167 // This is a field -- get or set the field's value.
168 assert(f);
169 if (setter) {
170 if (argc < 2) {
171 rb_raise(rb_eArgError, "No value provided to setter.");
172 }
173 layout_set(self->descriptor->layout, Message_data(self), f, argv[1]);
174 return Qnil;
175 } else {
176 return layout_get(self->descriptor->layout, Message_data(self), f);
177 }
178 }
179 }
180
Message_initialize_kwarg(VALUE key,VALUE val,VALUE _self)181 int Message_initialize_kwarg(VALUE key, VALUE val, VALUE _self) {
182 MessageHeader* self;
183 VALUE method_str;
184 char* name;
185 const upb_fielddef* f;
186 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
187
188 if (!SYMBOL_P(key)) {
189 rb_raise(rb_eArgError,
190 "Expected symbols as hash keys in initialization map.");
191 }
192
193 method_str = rb_id2str(SYM2ID(key));
194 name = RSTRING_PTR(method_str);
195 f = upb_msgdef_ntofz(self->descriptor->msgdef, name);
196 if (f == NULL) {
197 rb_raise(rb_eArgError,
198 "Unknown field name '%s' in initialization map entry.", name);
199 }
200
201 if (is_map_field(f)) {
202 VALUE map;
203
204 if (TYPE(val) != T_HASH) {
205 rb_raise(rb_eArgError,
206 "Expected Hash object as initializer value for map field '%s'.", name);
207 }
208 map = layout_get(self->descriptor->layout, Message_data(self), f);
209 Map_merge_into_self(map, val);
210 } else if (upb_fielddef_label(f) == UPB_LABEL_REPEATED) {
211 VALUE ary;
212
213 if (TYPE(val) != T_ARRAY) {
214 rb_raise(rb_eArgError,
215 "Expected array as initializer value for repeated field '%s'.", name);
216 }
217 ary = layout_get(self->descriptor->layout, Message_data(self), f);
218 for (int i = 0; i < RARRAY_LEN(val); i++) {
219 RepeatedField_push(ary, rb_ary_entry(val, i));
220 }
221 } else {
222 layout_set(self->descriptor->layout, Message_data(self), f, val);
223 }
224 return 0;
225 }
226
227 /*
228 * call-seq:
229 * Message.new(kwargs) => new_message
230 *
231 * Creates a new instance of the given message class. Keyword arguments may be
232 * provided with keywords corresponding to field names.
233 *
234 * Note that no literal Message class exists. Only concrete classes per message
235 * type exist, as provided by the #msgclass method on Descriptors after they
236 * have been added to a pool. The method definitions described here on the
237 * Message class are provided on each concrete message class.
238 */
Message_initialize(int argc,VALUE * argv,VALUE _self)239 VALUE Message_initialize(int argc, VALUE* argv, VALUE _self) {
240 VALUE hash_args;
241
242 if (argc == 0) {
243 return Qnil;
244 }
245 if (argc != 1) {
246 rb_raise(rb_eArgError, "Expected 0 or 1 arguments.");
247 }
248 hash_args = argv[0];
249 if (TYPE(hash_args) != T_HASH) {
250 rb_raise(rb_eArgError, "Expected hash arguments.");
251 }
252
253 rb_hash_foreach(hash_args, Message_initialize_kwarg, _self);
254 return Qnil;
255 }
256
257 /*
258 * call-seq:
259 * Message.dup => new_message
260 *
261 * Performs a shallow copy of this message and returns the new copy.
262 */
Message_dup(VALUE _self)263 VALUE Message_dup(VALUE _self) {
264 MessageHeader* self;
265 VALUE new_msg;
266 MessageHeader* new_msg_self;
267 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
268
269 new_msg = rb_class_new_instance(0, NULL, CLASS_OF(_self));
270 TypedData_Get_Struct(new_msg, MessageHeader, &Message_type, new_msg_self);
271
272 layout_dup(self->descriptor->layout,
273 Message_data(new_msg_self),
274 Message_data(self));
275
276 return new_msg;
277 }
278
279 // Internal only; used by Google::Protobuf.deep_copy.
Message_deep_copy(VALUE _self)280 VALUE Message_deep_copy(VALUE _self) {
281 MessageHeader* self;
282 MessageHeader* new_msg_self;
283 VALUE new_msg;
284 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
285
286 new_msg = rb_class_new_instance(0, NULL, CLASS_OF(_self));
287 TypedData_Get_Struct(new_msg, MessageHeader, &Message_type, new_msg_self);
288
289 layout_deep_copy(self->descriptor->layout,
290 Message_data(new_msg_self),
291 Message_data(self));
292
293 return new_msg;
294 }
295
296 /*
297 * call-seq:
298 * Message.==(other) => boolean
299 *
300 * Performs a deep comparison of this message with another. Messages are equal
301 * if they have the same type and if each field is equal according to the :==
302 * method's semantics (a more efficient comparison may actually be done if the
303 * field is of a primitive type).
304 */
Message_eq(VALUE _self,VALUE _other)305 VALUE Message_eq(VALUE _self, VALUE _other) {
306 MessageHeader* self;
307 MessageHeader* other;
308 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
309 TypedData_Get_Struct(_other, MessageHeader, &Message_type, other);
310
311 if (self->descriptor != other->descriptor) {
312 return Qfalse;
313 }
314
315 return layout_eq(self->descriptor->layout,
316 Message_data(self),
317 Message_data(other));
318 }
319
320 /*
321 * call-seq:
322 * Message.hash => hash_value
323 *
324 * Returns a hash value that represents this message's field values.
325 */
Message_hash(VALUE _self)326 VALUE Message_hash(VALUE _self) {
327 MessageHeader* self;
328 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
329
330 return layout_hash(self->descriptor->layout, Message_data(self));
331 }
332
333 /*
334 * call-seq:
335 * Message.inspect => string
336 *
337 * Returns a human-readable string representing this message. It will be
338 * formatted as "<MessageType: field1: value1, field2: value2, ...>". Each
339 * field's value is represented according to its own #inspect method.
340 */
Message_inspect(VALUE _self)341 VALUE Message_inspect(VALUE _self) {
342 MessageHeader* self;
343 VALUE str;
344 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
345
346 str = rb_str_new2("<");
347 str = rb_str_append(str, rb_str_new2(rb_class2name(CLASS_OF(_self))));
348 str = rb_str_cat2(str, ": ");
349 str = rb_str_append(str, layout_inspect(
350 self->descriptor->layout, Message_data(self)));
351 str = rb_str_cat2(str, ">");
352 return str;
353 }
354
355
Message_to_h(VALUE _self)356 VALUE Message_to_h(VALUE _self) {
357 MessageHeader* self;
358 VALUE hash;
359 upb_msg_field_iter it;
360 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
361
362 hash = rb_hash_new();
363
364 for (upb_msg_field_begin(&it, self->descriptor->msgdef);
365 !upb_msg_field_done(&it);
366 upb_msg_field_next(&it)) {
367 const upb_fielddef* field = upb_msg_iter_field(&it);
368 VALUE msg_value = layout_get(self->descriptor->layout, Message_data(self),
369 field);
370 VALUE msg_key = ID2SYM(rb_intern(upb_fielddef_name(field)));
371 if (upb_fielddef_label(field) == UPB_LABEL_REPEATED) {
372 msg_value = RepeatedField_to_ary(msg_value);
373 }
374 rb_hash_aset(hash, msg_key, msg_value);
375 }
376 return hash;
377 }
378
379
380
381 /*
382 * call-seq:
383 * Message.[](index) => value
384 *
385 * Accesses a field's value by field name. The provided field name should be a
386 * string.
387 */
Message_index(VALUE _self,VALUE field_name)388 VALUE Message_index(VALUE _self, VALUE field_name) {
389 MessageHeader* self;
390 const upb_fielddef* field;
391 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
392 Check_Type(field_name, T_STRING);
393 field = upb_msgdef_ntofz(self->descriptor->msgdef, RSTRING_PTR(field_name));
394 if (field == NULL) {
395 return Qnil;
396 }
397 return layout_get(self->descriptor->layout, Message_data(self), field);
398 }
399
400 /*
401 * call-seq:
402 * Message.[]=(index, value)
403 *
404 * Sets a field's value by field name. The provided field name should be a
405 * string.
406 */
Message_index_set(VALUE _self,VALUE field_name,VALUE value)407 VALUE Message_index_set(VALUE _self, VALUE field_name, VALUE value) {
408 MessageHeader* self;
409 const upb_fielddef* field;
410 TypedData_Get_Struct(_self, MessageHeader, &Message_type, self);
411 Check_Type(field_name, T_STRING);
412 field = upb_msgdef_ntofz(self->descriptor->msgdef, RSTRING_PTR(field_name));
413 if (field == NULL) {
414 rb_raise(rb_eArgError, "Unknown field: %s", RSTRING_PTR(field_name));
415 }
416 layout_set(self->descriptor->layout, Message_data(self), field, value);
417 return Qnil;
418 }
419
420 /*
421 * call-seq:
422 * Message.descriptor => descriptor
423 *
424 * Class method that returns the Descriptor instance corresponding to this
425 * message class's type.
426 */
Message_descriptor(VALUE klass)427 VALUE Message_descriptor(VALUE klass) {
428 return rb_ivar_get(klass, descriptor_instancevar_interned);
429 }
430
build_class_from_descriptor(Descriptor * desc)431 VALUE build_class_from_descriptor(Descriptor* desc) {
432 const char *name;
433 VALUE klass;
434
435 if (desc->layout == NULL) {
436 desc->layout = create_layout(desc->msgdef);
437 }
438 if (desc->fill_method == NULL) {
439 desc->fill_method = new_fillmsg_decodermethod(desc, &desc->fill_method);
440 }
441
442 name = upb_msgdef_fullname(desc->msgdef);
443 if (name == NULL) {
444 rb_raise(rb_eRuntimeError, "Descriptor does not have assigned name.");
445 }
446
447 klass = rb_define_class_id(
448 // Docs say this parameter is ignored. User will assign return value to
449 // their own toplevel constant class name.
450 rb_intern("Message"),
451 rb_cObject);
452 rb_ivar_set(klass, descriptor_instancevar_interned,
453 get_def_obj(desc->msgdef));
454 rb_define_alloc_func(klass, Message_alloc);
455 rb_require("google/protobuf/message_exts");
456 rb_include_module(klass, rb_eval_string("Google::Protobuf::MessageExts"));
457 rb_extend_object(
458 klass, rb_eval_string("Google::Protobuf::MessageExts::ClassMethods"));
459
460 rb_define_method(klass, "method_missing",
461 Message_method_missing, -1);
462 rb_define_method(klass, "initialize", Message_initialize, -1);
463 rb_define_method(klass, "dup", Message_dup, 0);
464 // Also define #clone so that we don't inherit Object#clone.
465 rb_define_method(klass, "clone", Message_dup, 0);
466 rb_define_method(klass, "==", Message_eq, 1);
467 rb_define_method(klass, "hash", Message_hash, 0);
468 rb_define_method(klass, "to_h", Message_to_h, 0);
469 rb_define_method(klass, "to_hash", Message_to_h, 0);
470 rb_define_method(klass, "inspect", Message_inspect, 0);
471 rb_define_method(klass, "[]", Message_index, 1);
472 rb_define_method(klass, "[]=", Message_index_set, 2);
473 rb_define_singleton_method(klass, "decode", Message_decode, 1);
474 rb_define_singleton_method(klass, "encode", Message_encode, 1);
475 rb_define_singleton_method(klass, "decode_json", Message_decode_json, 1);
476 rb_define_singleton_method(klass, "encode_json", Message_encode_json, -1);
477 rb_define_singleton_method(klass, "descriptor", Message_descriptor, 0);
478
479 return klass;
480 }
481
482 /*
483 * call-seq:
484 * Enum.lookup(number) => name
485 *
486 * This module method, provided on each generated enum module, looks up an enum
487 * value by number and returns its name as a Ruby symbol, or nil if not found.
488 */
enum_lookup(VALUE self,VALUE number)489 VALUE enum_lookup(VALUE self, VALUE number) {
490 int32_t num = NUM2INT(number);
491 VALUE desc = rb_ivar_get(self, descriptor_instancevar_interned);
492 EnumDescriptor* enumdesc = ruby_to_EnumDescriptor(desc);
493
494 const char* name = upb_enumdef_iton(enumdesc->enumdef, num);
495 if (name == NULL) {
496 return Qnil;
497 } else {
498 return ID2SYM(rb_intern(name));
499 }
500 }
501
502 /*
503 * call-seq:
504 * Enum.resolve(name) => number
505 *
506 * This module method, provided on each generated enum module, looks up an enum
507 * value by name (as a Ruby symbol) and returns its name, or nil if not found.
508 */
enum_resolve(VALUE self,VALUE sym)509 VALUE enum_resolve(VALUE self, VALUE sym) {
510 const char* name = rb_id2name(SYM2ID(sym));
511 VALUE desc = rb_ivar_get(self, descriptor_instancevar_interned);
512 EnumDescriptor* enumdesc = ruby_to_EnumDescriptor(desc);
513
514 int32_t num = 0;
515 bool found = upb_enumdef_ntoiz(enumdesc->enumdef, name, &num);
516 if (!found) {
517 return Qnil;
518 } else {
519 return INT2NUM(num);
520 }
521 }
522
523 /*
524 * call-seq:
525 * Enum.descriptor
526 *
527 * This module method, provided on each generated enum module, returns the
528 * EnumDescriptor corresponding to this enum type.
529 */
enum_descriptor(VALUE self)530 VALUE enum_descriptor(VALUE self) {
531 return rb_ivar_get(self, descriptor_instancevar_interned);
532 }
533
build_module_from_enumdesc(EnumDescriptor * enumdesc)534 VALUE build_module_from_enumdesc(EnumDescriptor* enumdesc) {
535 VALUE mod = rb_define_module_id(
536 rb_intern(upb_enumdef_fullname(enumdesc->enumdef)));
537
538 upb_enum_iter it;
539 for (upb_enum_begin(&it, enumdesc->enumdef);
540 !upb_enum_done(&it);
541 upb_enum_next(&it)) {
542 const char* name = upb_enum_iter_name(&it);
543 int32_t value = upb_enum_iter_number(&it);
544 if (name[0] < 'A' || name[0] > 'Z') {
545 rb_raise(rb_eTypeError,
546 "Enum value '%s' does not start with an uppercase letter "
547 "as is required for Ruby constants.",
548 name);
549 }
550 rb_define_const(mod, name, INT2NUM(value));
551 }
552
553 rb_define_singleton_method(mod, "lookup", enum_lookup, 1);
554 rb_define_singleton_method(mod, "resolve", enum_resolve, 1);
555 rb_define_singleton_method(mod, "descriptor", enum_descriptor, 0);
556 rb_ivar_set(mod, descriptor_instancevar_interned,
557 get_def_obj(enumdesc->enumdef));
558
559 return mod;
560 }
561
562 /*
563 * call-seq:
564 * Google::Protobuf.deep_copy(obj) => copy_of_obj
565 *
566 * Performs a deep copy of a RepeatedField instance, a Map instance, or a
567 * message object, recursively copying its members.
568 */
Google_Protobuf_deep_copy(VALUE self,VALUE obj)569 VALUE Google_Protobuf_deep_copy(VALUE self, VALUE obj) {
570 VALUE klass = CLASS_OF(obj);
571 if (klass == cRepeatedField) {
572 return RepeatedField_deep_copy(obj);
573 } else if (klass == cMap) {
574 return Map_deep_copy(obj);
575 } else {
576 return Message_deep_copy(obj);
577 }
578 }
579