blob: c15a1caa6cfa121dfa8373fd9058e104d86301aa [file] [log] [blame]
Chris Fallin973f4252014-11-18 14:19:58 -08001// Protocol Buffers - Google's data interchange format
2// Copyright 2014 Google Inc. All rights reserved.
Chris Fallin973f4252014-11-18 14:19:58 -08003//
Joshua Habermanba9e2af2023-09-08 17:13:26 -07004// Use of this source code is governed by a BSD-style
5// license that can be found in the LICENSE file or at
6// https://developers.google.com/open-source/licenses/bsd
Chris Fallin973f4252014-11-18 14:19:58 -08007
8#include "protobuf.h"
9
Joshua Haberman9abf6e22021-01-13 12:16:25 -080010#include <ruby/version.h>
11
12#include "defs.h"
13#include "map.h"
14#include "message.h"
15#include "repeated_field.h"
16
Aaron Pattersonaddd0612021-03-10 13:54:04 -080017VALUE cParseError;
Erik Benoist74f8e242018-06-26 22:24:24 -050018VALUE cTypeError;
Chris Fallin973f4252014-11-18 14:19:58 -080019
Joshua Habermanc153dd92022-01-21 15:46:56 -080020const upb_FieldDef *map_field_key(const upb_FieldDef *field) {
21 const upb_MessageDef *entry = upb_FieldDef_MessageSubDef(field);
Joshua Haberman7ecf43f2022-03-14 13:11:29 -070022 return upb_MessageDef_FindFieldByNumber(entry, 1);
Joshua Haberman1e37a942019-08-13 04:54:11 -070023}
24
Joshua Habermanc153dd92022-01-21 15:46:56 -080025const upb_FieldDef *map_field_value(const upb_FieldDef *field) {
26 const upb_MessageDef *entry = upb_FieldDef_MessageSubDef(field);
Joshua Haberman7ecf43f2022-03-14 13:11:29 -070027 return upb_MessageDef_FindFieldByNumber(entry, 2);
Joshua Haberman9abf6e22021-01-13 12:16:25 -080028}
29
30// -----------------------------------------------------------------------------
31// StringBuilder, for inspect
32// -----------------------------------------------------------------------------
33
34struct StringBuilder {
35 size_t size;
36 size_t cap;
37 char *data;
38};
39
40typedef struct StringBuilder StringBuilder;
41
42static size_t StringBuilder_SizeOf(size_t cap) {
43 return sizeof(StringBuilder) + cap;
44}
45
Joshua Habermanc153dd92022-01-21 15:46:56 -080046StringBuilder *StringBuilder_New() {
Joshua Haberman9abf6e22021-01-13 12:16:25 -080047 const size_t cap = 128;
Joshua Habermanc153dd92022-01-21 15:46:56 -080048 StringBuilder *builder = malloc(sizeof(*builder));
Joshua Haberman9abf6e22021-01-13 12:16:25 -080049 builder->size = 0;
50 builder->cap = cap;
51 builder->data = malloc(builder->cap);
52 return builder;
53}
54
Joshua Habermanc153dd92022-01-21 15:46:56 -080055void StringBuilder_Free(StringBuilder *b) {
Joshua Haberman9abf6e22021-01-13 12:16:25 -080056 free(b->data);
57 free(b);
58}
59
Joshua Habermanc153dd92022-01-21 15:46:56 -080060void StringBuilder_Printf(StringBuilder *b, const char *fmt, ...) {
Joshua Haberman9abf6e22021-01-13 12:16:25 -080061 size_t have = b->cap - b->size;
62 size_t n;
63 va_list args;
64
65 va_start(args, fmt);
66 n = vsnprintf(&b->data[b->size], have, fmt, args);
67 va_end(args);
68
69 if (have <= n) {
70 while (have <= n) {
71 b->cap *= 2;
72 have = b->cap - b->size;
73 }
74 b->data = realloc(b->data, StringBuilder_SizeOf(b->cap));
75 va_start(args, fmt);
76 n = vsnprintf(&b->data[b->size], have, fmt, args);
77 va_end(args);
78 PBRUBY_ASSERT(n < have);
79 }
80
81 b->size += n;
82}
83
Joshua Habermanc153dd92022-01-21 15:46:56 -080084VALUE StringBuilder_ToRubyString(StringBuilder *b) {
Joshua Haberman9abf6e22021-01-13 12:16:25 -080085 VALUE ret = rb_str_new(b->data, b->size);
86 rb_enc_associate(ret, rb_utf8_encoding());
87 return ret;
88}
89
Joshua Habermanc153dd92022-01-21 15:46:56 -080090static void StringBuilder_PrintEnum(StringBuilder *b, int32_t val,
91 const upb_EnumDef *e) {
92 const upb_EnumValueDef *ev = upb_EnumDef_FindValueByNumber(e, val);
93 if (ev) {
94 StringBuilder_Printf(b, ":%s", upb_EnumValueDef_Name(ev));
Joshua Haberman1e37a942019-08-13 04:54:11 -070095 } else {
Joshua Haberman9abf6e22021-01-13 12:16:25 -080096 StringBuilder_Printf(b, "%" PRId32, val);
97 }
98}
99
Joshua Habermanc153dd92022-01-21 15:46:56 -0800100void StringBuilder_PrintMsgval(StringBuilder *b, upb_MessageValue val,
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800101 TypeInfo info) {
102 switch (info.type) {
Joshua Habermanc153dd92022-01-21 15:46:56 -0800103 case kUpb_CType_Bool:
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800104 StringBuilder_Printf(b, "%s", val.bool_val ? "true" : "false");
105 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800106 case kUpb_CType_Float: {
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800107 VALUE str = rb_inspect(DBL2NUM(val.float_val));
108 StringBuilder_Printf(b, "%s", RSTRING_PTR(str));
109 break;
110 }
Joshua Habermanc153dd92022-01-21 15:46:56 -0800111 case kUpb_CType_Double: {
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800112 VALUE str = rb_inspect(DBL2NUM(val.double_val));
113 StringBuilder_Printf(b, "%s", RSTRING_PTR(str));
114 break;
115 }
Joshua Habermanc153dd92022-01-21 15:46:56 -0800116 case kUpb_CType_Int32:
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800117 StringBuilder_Printf(b, "%" PRId32, val.int32_val);
118 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800119 case kUpb_CType_UInt32:
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800120 StringBuilder_Printf(b, "%" PRIu32, val.uint32_val);
121 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800122 case kUpb_CType_Int64:
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800123 StringBuilder_Printf(b, "%" PRId64, val.int64_val);
124 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800125 case kUpb_CType_UInt64:
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800126 StringBuilder_Printf(b, "%" PRIu64, val.uint64_val);
127 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800128 case kUpb_CType_String:
129 StringBuilder_Printf(b, "\"%.*s\"", (int)val.str_val.size,
130 val.str_val.data);
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800131 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800132 case kUpb_CType_Bytes:
133 StringBuilder_Printf(b, "\"%.*s\"", (int)val.str_val.size,
134 val.str_val.data);
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800135 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800136 case kUpb_CType_Enum:
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800137 StringBuilder_PrintEnum(b, val.int32_val, info.def.enumdef);
138 break;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800139 case kUpb_CType_Message:
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800140 Message_PrintMessage(b, val.msg_val, info.def.msgdef);
141 break;
Joshua Haberman1e37a942019-08-13 04:54:11 -0700142 }
143}
144
Chris Fallin973f4252014-11-18 14:19:58 -0800145// -----------------------------------------------------------------------------
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800146// Arena
Chris Fallin973f4252014-11-18 14:19:58 -0800147// -----------------------------------------------------------------------------
148
Joshua Haberman9879f422021-02-24 16:41:35 -0800149typedef struct {
Joshua Habermanc153dd92022-01-21 15:46:56 -0800150 upb_Arena *arena;
Jean byroot Boussierd82d8a42023-02-23 13:44:39 -0800151 // IMPORTANT: WB_PROTECTED objects must only use the RB_OBJ_WRITE()
152 // macro to update VALUE references, as to trigger write barriers.
Joshua Haberman9879f422021-02-24 16:41:35 -0800153 VALUE pinned_objs;
154} Arena;
155
156static void Arena_mark(void *data) {
157 Arena *arena = data;
158 rb_gc_mark(arena->pinned_objs);
159}
160
161static void Arena_free(void *data) {
162 Arena *arena = data;
Joshua Habermanc153dd92022-01-21 15:46:56 -0800163 upb_Arena_Free(arena->arena);
Joshua Haberman67fee912021-04-07 10:30:51 -0700164 xfree(arena);
Joshua Haberman9879f422021-02-24 16:41:35 -0800165}
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800166
167static VALUE cArena;
168
169const rb_data_type_t Arena_type = {
Joshua Habermanc153dd92022-01-21 15:46:56 -0800170 "Google::Protobuf::Internal::Arena",
171 {Arena_mark, Arena_free, NULL},
Jean byroot Boussierd82d8a42023-02-23 13:44:39 -0800172 .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800173};
174
Matt Fowles Kulukundisa01d0472023-07-08 16:17:11 -0700175static void *ruby_upb_allocfunc(upb_alloc *alloc, void *ptr, size_t oldsize,
176 size_t size) {
zhangskz276add02022-03-08 12:05:34 -0500177 if (size == 0) {
178 xfree(ptr);
179 return NULL;
180 } else {
181 return xrealloc(ptr, size);
182 }
183}
184
185upb_alloc ruby_upb_alloc = {&ruby_upb_allocfunc};
186
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800187static VALUE Arena_alloc(VALUE klass) {
Joshua Haberman9879f422021-02-24 16:41:35 -0800188 Arena *arena = ALLOC(Arena);
zhangskz276add02022-03-08 12:05:34 -0500189 arena->arena = upb_Arena_Init(NULL, 0, &ruby_upb_alloc);
Joshua Haberman9879f422021-02-24 16:41:35 -0800190 arena->pinned_objs = Qnil;
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800191 return TypedData_Wrap_Struct(klass, &Arena_type, arena);
Chris Fallin973f4252014-11-18 14:19:58 -0800192}
193
Joshua Habermanc153dd92022-01-21 15:46:56 -0800194upb_Arena *Arena_get(VALUE _arena) {
Joshua Haberman9879f422021-02-24 16:41:35 -0800195 Arena *arena;
196 TypedData_Get_Struct(_arena, Arena, &Arena_type, arena);
197 return arena->arena;
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800198}
Chris Fallin973f4252014-11-18 14:19:58 -0800199
Joshua Habermanc153dd92022-01-21 15:46:56 -0800200void Arena_fuse(VALUE _arena, upb_Arena *other) {
Joshua Haberman367e4692021-05-21 23:04:09 -0700201 Arena *arena;
202 TypedData_Get_Struct(_arena, Arena, &Arena_type, arena);
Joshua Habermanc153dd92022-01-21 15:46:56 -0800203 if (!upb_Arena_Fuse(arena->arena, other)) {
Joshua Haberman367e4692021-05-21 23:04:09 -0700204 rb_raise(rb_eRuntimeError,
205 "Unable to fuse arenas. This should never happen since Ruby does "
206 "not use initial blocks");
207 }
208}
209
Joshua Habermanc153dd92022-01-21 15:46:56 -0800210VALUE Arena_new() { return Arena_alloc(cArena); }
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800211
Joshua Haberman9879f422021-02-24 16:41:35 -0800212void Arena_Pin(VALUE _arena, VALUE obj) {
213 Arena *arena;
214 TypedData_Get_Struct(_arena, Arena, &Arena_type, arena);
215 if (arena->pinned_objs == Qnil) {
Jean byroot Boussierd82d8a42023-02-23 13:44:39 -0800216 RB_OBJ_WRITE(_arena, &arena->pinned_objs, rb_ary_new());
Joshua Haberman9879f422021-02-24 16:41:35 -0800217 }
218 rb_ary_push(arena->pinned_objs, obj);
219}
220
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800221void Arena_register(VALUE module) {
222 VALUE internal = rb_define_module_under(module, "Internal");
223 VALUE klass = rb_define_class_under(internal, "Arena", rb_cObject);
224 rb_define_alloc_func(klass, Arena_alloc);
225 rb_gc_register_address(&cArena);
226 cArena = klass;
227}
228
229// -----------------------------------------------------------------------------
230// Object Cache
231// -----------------------------------------------------------------------------
232
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800233// Public ObjectCache API.
234
Joshua Haberman9879f422021-02-24 16:41:35 -0800235VALUE weak_obj_cache = Qnil;
236ID item_get;
Matt Fowles Kulukundisa01d0472023-07-08 16:17:11 -0700237ID item_try_add;
Joshua Haberman9879f422021-02-24 16:41:35 -0800238
Matt Fowles Kulukundisa01d0472023-07-08 16:17:11 -0700239static void ObjectCache_Init(VALUE protobuf) {
240 item_get = rb_intern("get");
241 item_try_add = rb_intern("try_add");
242
Joshua Haberman9879f422021-02-24 16:41:35 -0800243 rb_gc_register_address(&weak_obj_cache);
Joshua Haberman3f98af22023-07-11 11:21:22 -0700244#if SIZEOF_LONG >= SIZEOF_VALUE
Matt Fowles Kulukundisa01d0472023-07-08 16:17:11 -0700245 VALUE cache_class = rb_const_get(protobuf, rb_intern("ObjectCache"));
246#else
247 VALUE cache_class = rb_const_get(protobuf, rb_intern("LegacyObjectCache"));
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800248#endif
Matt Fowles Kulukundisa01d0472023-07-08 16:17:11 -0700249
250 weak_obj_cache = rb_class_new_instance(0, NULL, cache_class);
251 rb_const_set(protobuf, rb_intern("OBJECT_CACHE"), weak_obj_cache);
252 rb_const_set(protobuf, rb_intern("SIZEOF_LONG"), INT2NUM(SIZEOF_LONG));
253 rb_const_set(protobuf, rb_intern("SIZEOF_VALUE"), INT2NUM(SIZEOF_VALUE));
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800254}
255
Stan Hu055c3782023-08-15 12:25:13 -0700256static VALUE ObjectCache_GetKey(const void *key) {
Matt Fowles Kulukundisa01d0472023-07-08 16:17:11 -0700257 VALUE key_val = (VALUE)key;
258 PBRUBY_ASSERT((key_val & 3) == 0);
Stan Hu055c3782023-08-15 12:25:13 -0700259 // Ensure the key can be stored as a Fixnum since 1 bit is needed for
260 // FIXNUM_FLAG and 1 bit is needed for the sign bit.
261 VALUE new_key = LL2NUM(key_val >> 2);
262 PBRUBY_ASSERT(FIXNUM_P(new_key));
263 return new_key;
264}
265
266VALUE ObjectCache_TryAdd(const void *key, VALUE val) {
267 VALUE key_val = ObjectCache_GetKey(key);
268 return rb_funcall(weak_obj_cache, item_try_add, 2, key_val, val);
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800269}
270
271// Returns the cached object for this key, if any. Otherwise returns Qnil.
Joshua Habermanc153dd92022-01-21 15:46:56 -0800272VALUE ObjectCache_Get(const void *key) {
Stan Hu055c3782023-08-15 12:25:13 -0700273 VALUE key_val = ObjectCache_GetKey(key);
274 return rb_funcall(weak_obj_cache, item_get, 1, key_val);
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800275}
276
277/*
278 * call-seq:
279 * Google::Protobuf.discard_unknown(msg)
280 *
281 * Discard unknown fields in the given message object and recursively discard
282 * unknown fields in submessages.
283 */
284static VALUE Google_Protobuf_discard_unknown(VALUE self, VALUE msg_rb) {
Joshua Habermanc153dd92022-01-21 15:46:56 -0800285 const upb_MessageDef *m;
286 upb_Message *msg = Message_GetMutable(msg_rb, &m);
287 if (!upb_Message_DiscardUnknown(msg, m, 128)) {
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800288 rb_raise(rb_eRuntimeError, "Messages nested too deeply.");
289 }
290
291 return Qnil;
292}
293
294/*
295 * call-seq:
296 * Google::Protobuf.deep_copy(obj) => copy_of_obj
297 *
298 * Performs a deep copy of a RepeatedField instance, a Map instance, or a
299 * message object, recursively copying its members.
300 */
301VALUE Google_Protobuf_deep_copy(VALUE self, VALUE obj) {
302 VALUE klass = CLASS_OF(obj);
303 if (klass == cRepeatedField) {
304 return RepeatedField_deep_copy(obj);
305 } else if (klass == cMap) {
306 return Map_deep_copy(obj);
307 } else {
308 VALUE new_arena_rb = Arena_new();
Joshua Habermanc153dd92022-01-21 15:46:56 -0800309 upb_Arena *new_arena = Arena_get(new_arena_rb);
310 const upb_MessageDef *m;
311 const upb_Message *msg = Message_Get(obj, &m);
312 upb_Message *new_msg = Message_deep_copy(msg, m, new_arena);
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800313 return Message_GetRubyWrapper(new_msg, m, new_arena_rb);
314 }
315}
Chris Fallin231886f2015-05-19 15:33:48 -0700316
Chris Fallin973f4252014-11-18 14:19:58 -0800317// -----------------------------------------------------------------------------
318// Initialization/entry point.
319// -----------------------------------------------------------------------------
320
321// This must be named "Init_protobuf_c" because the Ruby module is named
322// "protobuf_c" -- the VM looks for this symbol in our .so.
Joshua Habermanc153dd92022-01-21 15:46:56 -0800323__attribute__((visibility("default"))) void Init_protobuf_c() {
Chris Fallin973f4252014-11-18 14:19:58 -0800324 VALUE google = rb_define_module("Google");
325 VALUE protobuf = rb_define_module_under(google, "Protobuf");
Josh Habermana1daeab2015-07-10 11:56:06 -0700326
Matt Fowles Kulukundisa01d0472023-07-08 16:17:11 -0700327 ObjectCache_Init(protobuf);
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800328 Arena_register(protobuf);
329 Defs_register(protobuf);
Chris Fallin973f4252014-11-18 14:19:58 -0800330 RepeatedField_register(protobuf);
Chris Fallinfd1a3ff2015-01-06 15:44:09 -0800331 Map_register(protobuf);
Joshua Haberman9abf6e22021-01-13 12:16:25 -0800332 Message_register(protobuf);
Chris Fallin973f4252014-11-18 14:19:58 -0800333
Aaron Pattersonaddd0612021-03-10 13:54:04 -0800334 cParseError = rb_const_get(protobuf, rb_intern("ParseError"));
335 rb_gc_register_mark_object(cParseError);
Erik Benoist74f8e242018-06-26 22:24:24 -0500336 cTypeError = rb_const_get(protobuf, rb_intern("TypeError"));
Aaron Pattersonaddd0612021-03-10 13:54:04 -0800337 rb_gc_register_mark_object(cTypeError);
Josh Haberman181c7f22015-07-15 11:05:10 -0700338
Paul Yang0e7b5892017-12-07 14:18:38 -0800339 rb_define_singleton_method(protobuf, "discard_unknown",
340 Google_Protobuf_discard_unknown, 1);
Joshua Habermanc153dd92022-01-21 15:46:56 -0800341 rb_define_singleton_method(protobuf, "deep_copy", Google_Protobuf_deep_copy,
342 1);
Chris Fallin973f4252014-11-18 14:19:58 -0800343}