1 // Protocol Buffers - Google's data interchange format
2 // Copyright 2023 Google LLC. All rights reserved.
3 //
4 // 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
7
8 #include "upb/reflection/internal/message_def.h"
9
10 #include <stddef.h>
11 #include <stdint.h>
12 #include <string.h>
13
14 #include "upb/base/descriptor_constants.h"
15 #include "upb/base/string_view.h"
16 #include "upb/hash/common.h"
17 #include "upb/hash/int_table.h"
18 #include "upb/hash/str_table.h"
19 #include "upb/mem/arena.h"
20 #include "upb/mini_descriptor/decode.h"
21 #include "upb/mini_descriptor/internal/encode.h"
22 #include "upb/mini_descriptor/internal/modifiers.h"
23 #include "upb/mini_table/file.h"
24 #include "upb/mini_table/message.h"
25 #include "upb/reflection/def.h"
26 #include "upb/reflection/internal/def_builder.h"
27 #include "upb/reflection/internal/desc_state.h"
28 #include "upb/reflection/internal/enum_def.h"
29 #include "upb/reflection/internal/extension_range.h"
30 #include "upb/reflection/internal/field_def.h"
31 #include "upb/reflection/internal/file_def.h"
32 #include "upb/reflection/internal/message_reserved_range.h"
33 #include "upb/reflection/internal/oneof_def.h"
34 #include "upb/reflection/internal/strdup2.h"
35
36 // Must be last.
37 #include "upb/port/def.inc"
38
39 struct upb_MessageDef {
40 const UPB_DESC(MessageOptions*) opts;
41 const UPB_DESC(FeatureSet*) resolved_features;
42 const upb_MiniTable* layout;
43 const upb_FileDef* file;
44 const upb_MessageDef* containing_type;
45 const char* full_name;
46
47 // Tables for looking up fields by number and name.
48 upb_inttable itof;
49 upb_strtable ntof;
50
51 // Looking up fields by json name.
52 upb_strtable jtof;
53
54 /* All nested defs.
55 * MEM: We could save some space here by putting nested defs in a contiguous
56 * region and calculating counts from offsets or vice-versa. */
57 const upb_FieldDef* fields;
58 const upb_OneofDef* oneofs;
59 const upb_ExtensionRange* ext_ranges;
60 const upb_StringView* res_names;
61 const upb_MessageDef* nested_msgs;
62 const upb_MessageReservedRange* res_ranges;
63 const upb_EnumDef* nested_enums;
64 const upb_FieldDef* nested_exts;
65
66 // TODO: These counters don't need anywhere near 32 bits.
67 int field_count;
68 int real_oneof_count;
69 int oneof_count;
70 int ext_range_count;
71 int res_range_count;
72 int res_name_count;
73 int nested_msg_count;
74 int nested_enum_count;
75 int nested_ext_count;
76 bool in_message_set;
77 bool is_sorted;
78 upb_WellKnown well_known_type;
79 #if UINTPTR_MAX == 0xffffffff
80 uint32_t padding; // Increase size to a multiple of 8.
81 #endif
82 };
83
assign_msg_wellknowntype(upb_MessageDef * m)84 static void assign_msg_wellknowntype(upb_MessageDef* m) {
85 const char* name = m->full_name;
86 if (name == NULL) {
87 m->well_known_type = kUpb_WellKnown_Unspecified;
88 return;
89 }
90 if (!strcmp(name, "google.protobuf.Any")) {
91 m->well_known_type = kUpb_WellKnown_Any;
92 } else if (!strcmp(name, "google.protobuf.FieldMask")) {
93 m->well_known_type = kUpb_WellKnown_FieldMask;
94 } else if (!strcmp(name, "google.protobuf.Duration")) {
95 m->well_known_type = kUpb_WellKnown_Duration;
96 } else if (!strcmp(name, "google.protobuf.Timestamp")) {
97 m->well_known_type = kUpb_WellKnown_Timestamp;
98 } else if (!strcmp(name, "google.protobuf.DoubleValue")) {
99 m->well_known_type = kUpb_WellKnown_DoubleValue;
100 } else if (!strcmp(name, "google.protobuf.FloatValue")) {
101 m->well_known_type = kUpb_WellKnown_FloatValue;
102 } else if (!strcmp(name, "google.protobuf.Int64Value")) {
103 m->well_known_type = kUpb_WellKnown_Int64Value;
104 } else if (!strcmp(name, "google.protobuf.UInt64Value")) {
105 m->well_known_type = kUpb_WellKnown_UInt64Value;
106 } else if (!strcmp(name, "google.protobuf.Int32Value")) {
107 m->well_known_type = kUpb_WellKnown_Int32Value;
108 } else if (!strcmp(name, "google.protobuf.UInt32Value")) {
109 m->well_known_type = kUpb_WellKnown_UInt32Value;
110 } else if (!strcmp(name, "google.protobuf.BoolValue")) {
111 m->well_known_type = kUpb_WellKnown_BoolValue;
112 } else if (!strcmp(name, "google.protobuf.StringValue")) {
113 m->well_known_type = kUpb_WellKnown_StringValue;
114 } else if (!strcmp(name, "google.protobuf.BytesValue")) {
115 m->well_known_type = kUpb_WellKnown_BytesValue;
116 } else if (!strcmp(name, "google.protobuf.Value")) {
117 m->well_known_type = kUpb_WellKnown_Value;
118 } else if (!strcmp(name, "google.protobuf.ListValue")) {
119 m->well_known_type = kUpb_WellKnown_ListValue;
120 } else if (!strcmp(name, "google.protobuf.Struct")) {
121 m->well_known_type = kUpb_WellKnown_Struct;
122 } else {
123 m->well_known_type = kUpb_WellKnown_Unspecified;
124 }
125 }
126
_upb_MessageDef_At(const upb_MessageDef * m,int i)127 upb_MessageDef* _upb_MessageDef_At(const upb_MessageDef* m, int i) {
128 return (upb_MessageDef*)&m[i];
129 }
130
_upb_MessageDef_IsValidExtensionNumber(const upb_MessageDef * m,int n)131 bool _upb_MessageDef_IsValidExtensionNumber(const upb_MessageDef* m, int n) {
132 for (int i = 0; i < m->ext_range_count; i++) {
133 const upb_ExtensionRange* r = upb_MessageDef_ExtensionRange(m, i);
134 if (upb_ExtensionRange_Start(r) <= n && n < upb_ExtensionRange_End(r)) {
135 return true;
136 }
137 }
138 return false;
139 }
140
UPB_DESC(MessageOptions)141 const UPB_DESC(MessageOptions) *
142 upb_MessageDef_Options(const upb_MessageDef* m) {
143 return m->opts;
144 }
145
upb_MessageDef_HasOptions(const upb_MessageDef * m)146 bool upb_MessageDef_HasOptions(const upb_MessageDef* m) {
147 return m->opts != (void*)kUpbDefOptDefault;
148 }
149
UPB_DESC(FeatureSet)150 const UPB_DESC(FeatureSet) *
151 upb_MessageDef_ResolvedFeatures(const upb_MessageDef* m) {
152 return m->resolved_features;
153 }
154
upb_MessageDef_FullName(const upb_MessageDef * m)155 const char* upb_MessageDef_FullName(const upb_MessageDef* m) {
156 return m->full_name;
157 }
158
upb_MessageDef_File(const upb_MessageDef * m)159 const upb_FileDef* upb_MessageDef_File(const upb_MessageDef* m) {
160 return m->file;
161 }
162
upb_MessageDef_ContainingType(const upb_MessageDef * m)163 const upb_MessageDef* upb_MessageDef_ContainingType(const upb_MessageDef* m) {
164 return m->containing_type;
165 }
166
upb_MessageDef_Name(const upb_MessageDef * m)167 const char* upb_MessageDef_Name(const upb_MessageDef* m) {
168 return _upb_DefBuilder_FullToShort(m->full_name);
169 }
170
upb_MessageDef_Syntax(const upb_MessageDef * m)171 upb_Syntax upb_MessageDef_Syntax(const upb_MessageDef* m) {
172 return upb_FileDef_Syntax(m->file);
173 }
174
upb_MessageDef_FindFieldByNumber(const upb_MessageDef * m,uint32_t i)175 const upb_FieldDef* upb_MessageDef_FindFieldByNumber(const upb_MessageDef* m,
176 uint32_t i) {
177 upb_value val;
178 return upb_inttable_lookup(&m->itof, i, &val) ? upb_value_getconstptr(val)
179 : NULL;
180 }
181
upb_MessageDef_FindFieldByNameWithSize(const upb_MessageDef * m,const char * name,size_t size)182 const upb_FieldDef* upb_MessageDef_FindFieldByNameWithSize(
183 const upb_MessageDef* m, const char* name, size_t size) {
184 upb_value val;
185
186 if (!upb_strtable_lookup2(&m->ntof, name, size, &val)) {
187 return NULL;
188 }
189
190 return _upb_DefType_Unpack(val, UPB_DEFTYPE_FIELD);
191 }
192
upb_MessageDef_FindOneofByNameWithSize(const upb_MessageDef * m,const char * name,size_t size)193 const upb_OneofDef* upb_MessageDef_FindOneofByNameWithSize(
194 const upb_MessageDef* m, const char* name, size_t size) {
195 upb_value val;
196
197 if (!upb_strtable_lookup2(&m->ntof, name, size, &val)) {
198 return NULL;
199 }
200
201 return _upb_DefType_Unpack(val, UPB_DEFTYPE_ONEOF);
202 }
203
_upb_MessageDef_Insert(upb_MessageDef * m,const char * name,size_t len,upb_value v,upb_Arena * a)204 bool _upb_MessageDef_Insert(upb_MessageDef* m, const char* name, size_t len,
205 upb_value v, upb_Arena* a) {
206 return upb_strtable_insert(&m->ntof, name, len, v, a);
207 }
208
upb_MessageDef_FindByNameWithSize(const upb_MessageDef * m,const char * name,size_t len,const upb_FieldDef ** out_f,const upb_OneofDef ** out_o)209 bool upb_MessageDef_FindByNameWithSize(const upb_MessageDef* m,
210 const char* name, size_t len,
211 const upb_FieldDef** out_f,
212 const upb_OneofDef** out_o) {
213 upb_value val;
214
215 if (!upb_strtable_lookup2(&m->ntof, name, len, &val)) {
216 return false;
217 }
218
219 const upb_FieldDef* f = _upb_DefType_Unpack(val, UPB_DEFTYPE_FIELD);
220 const upb_OneofDef* o = _upb_DefType_Unpack(val, UPB_DEFTYPE_ONEOF);
221 if (out_f) *out_f = f;
222 if (out_o) *out_o = o;
223 return f || o; /* False if this was a JSON name. */
224 }
225
upb_MessageDef_FindByJsonNameWithSize(const upb_MessageDef * m,const char * name,size_t size)226 const upb_FieldDef* upb_MessageDef_FindByJsonNameWithSize(
227 const upb_MessageDef* m, const char* name, size_t size) {
228 upb_value val;
229
230 if (upb_strtable_lookup2(&m->jtof, name, size, &val)) {
231 return upb_value_getconstptr(val);
232 }
233
234 if (!upb_strtable_lookup2(&m->ntof, name, size, &val)) {
235 return NULL;
236 }
237
238 return _upb_DefType_Unpack(val, UPB_DEFTYPE_FIELD);
239 }
240
upb_MessageDef_ExtensionRangeCount(const upb_MessageDef * m)241 int upb_MessageDef_ExtensionRangeCount(const upb_MessageDef* m) {
242 return m->ext_range_count;
243 }
244
upb_MessageDef_ReservedRangeCount(const upb_MessageDef * m)245 int upb_MessageDef_ReservedRangeCount(const upb_MessageDef* m) {
246 return m->res_range_count;
247 }
248
upb_MessageDef_ReservedNameCount(const upb_MessageDef * m)249 int upb_MessageDef_ReservedNameCount(const upb_MessageDef* m) {
250 return m->res_name_count;
251 }
252
upb_MessageDef_FieldCount(const upb_MessageDef * m)253 int upb_MessageDef_FieldCount(const upb_MessageDef* m) {
254 return m->field_count;
255 }
256
upb_MessageDef_OneofCount(const upb_MessageDef * m)257 int upb_MessageDef_OneofCount(const upb_MessageDef* m) {
258 return m->oneof_count;
259 }
260
upb_MessageDef_RealOneofCount(const upb_MessageDef * m)261 int upb_MessageDef_RealOneofCount(const upb_MessageDef* m) {
262 return m->real_oneof_count;
263 }
264
upb_MessageDef_NestedMessageCount(const upb_MessageDef * m)265 int upb_MessageDef_NestedMessageCount(const upb_MessageDef* m) {
266 return m->nested_msg_count;
267 }
268
upb_MessageDef_NestedEnumCount(const upb_MessageDef * m)269 int upb_MessageDef_NestedEnumCount(const upb_MessageDef* m) {
270 return m->nested_enum_count;
271 }
272
upb_MessageDef_NestedExtensionCount(const upb_MessageDef * m)273 int upb_MessageDef_NestedExtensionCount(const upb_MessageDef* m) {
274 return m->nested_ext_count;
275 }
276
upb_MessageDef_MiniTable(const upb_MessageDef * m)277 const upb_MiniTable* upb_MessageDef_MiniTable(const upb_MessageDef* m) {
278 return m->layout;
279 }
280
upb_MessageDef_ExtensionRange(const upb_MessageDef * m,int i)281 const upb_ExtensionRange* upb_MessageDef_ExtensionRange(const upb_MessageDef* m,
282 int i) {
283 UPB_ASSERT(0 <= i && i < m->ext_range_count);
284 return _upb_ExtensionRange_At(m->ext_ranges, i);
285 }
286
upb_MessageDef_ReservedRange(const upb_MessageDef * m,int i)287 const upb_MessageReservedRange* upb_MessageDef_ReservedRange(
288 const upb_MessageDef* m, int i) {
289 UPB_ASSERT(0 <= i && i < m->res_range_count);
290 return _upb_MessageReservedRange_At(m->res_ranges, i);
291 }
292
upb_MessageDef_ReservedName(const upb_MessageDef * m,int i)293 upb_StringView upb_MessageDef_ReservedName(const upb_MessageDef* m, int i) {
294 UPB_ASSERT(0 <= i && i < m->res_name_count);
295 return m->res_names[i];
296 }
297
upb_MessageDef_Field(const upb_MessageDef * m,int i)298 const upb_FieldDef* upb_MessageDef_Field(const upb_MessageDef* m, int i) {
299 UPB_ASSERT(0 <= i && i < m->field_count);
300 return _upb_FieldDef_At(m->fields, i);
301 }
302
upb_MessageDef_Oneof(const upb_MessageDef * m,int i)303 const upb_OneofDef* upb_MessageDef_Oneof(const upb_MessageDef* m, int i) {
304 UPB_ASSERT(0 <= i && i < m->oneof_count);
305 return _upb_OneofDef_At(m->oneofs, i);
306 }
307
upb_MessageDef_NestedMessage(const upb_MessageDef * m,int i)308 const upb_MessageDef* upb_MessageDef_NestedMessage(const upb_MessageDef* m,
309 int i) {
310 UPB_ASSERT(0 <= i && i < m->nested_msg_count);
311 return &m->nested_msgs[i];
312 }
313
upb_MessageDef_NestedEnum(const upb_MessageDef * m,int i)314 const upb_EnumDef* upb_MessageDef_NestedEnum(const upb_MessageDef* m, int i) {
315 UPB_ASSERT(0 <= i && i < m->nested_enum_count);
316 return _upb_EnumDef_At(m->nested_enums, i);
317 }
318
upb_MessageDef_NestedExtension(const upb_MessageDef * m,int i)319 const upb_FieldDef* upb_MessageDef_NestedExtension(const upb_MessageDef* m,
320 int i) {
321 UPB_ASSERT(0 <= i && i < m->nested_ext_count);
322 return _upb_FieldDef_At(m->nested_exts, i);
323 }
324
upb_MessageDef_WellKnownType(const upb_MessageDef * m)325 upb_WellKnown upb_MessageDef_WellKnownType(const upb_MessageDef* m) {
326 return m->well_known_type;
327 }
328
_upb_MessageDef_InMessageSet(const upb_MessageDef * m)329 bool _upb_MessageDef_InMessageSet(const upb_MessageDef* m) {
330 return m->in_message_set;
331 }
332
upb_MessageDef_FindFieldByName(const upb_MessageDef * m,const char * name)333 const upb_FieldDef* upb_MessageDef_FindFieldByName(const upb_MessageDef* m,
334 const char* name) {
335 return upb_MessageDef_FindFieldByNameWithSize(m, name, strlen(name));
336 }
337
upb_MessageDef_FindOneofByName(const upb_MessageDef * m,const char * name)338 const upb_OneofDef* upb_MessageDef_FindOneofByName(const upb_MessageDef* m,
339 const char* name) {
340 return upb_MessageDef_FindOneofByNameWithSize(m, name, strlen(name));
341 }
342
upb_MessageDef_IsMapEntry(const upb_MessageDef * m)343 bool upb_MessageDef_IsMapEntry(const upb_MessageDef* m) {
344 return UPB_DESC(MessageOptions_map_entry)(m->opts);
345 }
346
upb_MessageDef_IsMessageSet(const upb_MessageDef * m)347 bool upb_MessageDef_IsMessageSet(const upb_MessageDef* m) {
348 return UPB_DESC(MessageOptions_message_set_wire_format)(m->opts);
349 }
350
_upb_MessageDef_MakeMiniTable(upb_DefBuilder * ctx,const upb_MessageDef * m)351 static upb_MiniTable* _upb_MessageDef_MakeMiniTable(upb_DefBuilder* ctx,
352 const upb_MessageDef* m) {
353 upb_StringView desc;
354 // Note: this will assign layout_index for fields, so upb_FieldDef_MiniTable()
355 // is safe to call only after this call.
356 bool ok = upb_MessageDef_MiniDescriptorEncode(m, ctx->tmp_arena, &desc);
357 if (!ok) _upb_DefBuilder_OomErr(ctx);
358
359 void** scratch_data = _upb_DefPool_ScratchData(ctx->symtab);
360 size_t* scratch_size = _upb_DefPool_ScratchSize(ctx->symtab);
361 upb_MiniTable* ret = upb_MiniTable_BuildWithBuf(
362 desc.data, desc.size, ctx->platform, ctx->arena, scratch_data,
363 scratch_size, ctx->status);
364 if (!ret) _upb_DefBuilder_FailJmp(ctx);
365
366 return ret;
367 }
368
_upb_MessageDef_Resolve(upb_DefBuilder * ctx,upb_MessageDef * m)369 void _upb_MessageDef_Resolve(upb_DefBuilder* ctx, upb_MessageDef* m) {
370 for (int i = 0; i < m->field_count; i++) {
371 upb_FieldDef* f = (upb_FieldDef*)upb_MessageDef_Field(m, i);
372 _upb_FieldDef_Resolve(ctx, m->full_name, f);
373 }
374
375 m->in_message_set = false;
376 for (int i = 0; i < upb_MessageDef_NestedExtensionCount(m); i++) {
377 upb_FieldDef* ext = (upb_FieldDef*)upb_MessageDef_NestedExtension(m, i);
378 _upb_FieldDef_Resolve(ctx, m->full_name, ext);
379 if (upb_FieldDef_Type(ext) == kUpb_FieldType_Message &&
380 upb_FieldDef_Label(ext) == kUpb_Label_Optional &&
381 upb_FieldDef_MessageSubDef(ext) == m &&
382 UPB_DESC(MessageOptions_message_set_wire_format)(
383 upb_MessageDef_Options(upb_FieldDef_ContainingType(ext)))) {
384 m->in_message_set = true;
385 }
386 }
387
388 for (int i = 0; i < upb_MessageDef_NestedMessageCount(m); i++) {
389 upb_MessageDef* n = (upb_MessageDef*)upb_MessageDef_NestedMessage(m, i);
390 _upb_MessageDef_Resolve(ctx, n);
391 }
392 }
393
_upb_MessageDef_InsertField(upb_DefBuilder * ctx,upb_MessageDef * m,const upb_FieldDef * f)394 void _upb_MessageDef_InsertField(upb_DefBuilder* ctx, upb_MessageDef* m,
395 const upb_FieldDef* f) {
396 const int32_t field_number = upb_FieldDef_Number(f);
397
398 if (field_number <= 0 || field_number > kUpb_MaxFieldNumber) {
399 _upb_DefBuilder_Errf(ctx, "invalid field number (%u)", field_number);
400 }
401
402 const char* json_name = upb_FieldDef_JsonName(f);
403 const char* shortname = upb_FieldDef_Name(f);
404 const size_t shortnamelen = strlen(shortname);
405
406 upb_value v = upb_value_constptr(f);
407
408 upb_value existing_v;
409 if (upb_strtable_lookup(&m->ntof, shortname, &existing_v)) {
410 _upb_DefBuilder_Errf(ctx, "duplicate field name (%s)", shortname);
411 }
412
413 const upb_value field_v = _upb_DefType_Pack(f, UPB_DEFTYPE_FIELD);
414 bool ok =
415 _upb_MessageDef_Insert(m, shortname, shortnamelen, field_v, ctx->arena);
416 if (!ok) _upb_DefBuilder_OomErr(ctx);
417
418 bool skip_json_conflicts =
419 UPB_DESC(MessageOptions_deprecated_legacy_json_field_conflicts)(
420 upb_MessageDef_Options(m));
421 if (!skip_json_conflicts && strcmp(shortname, json_name) != 0 &&
422 UPB_DESC(FeatureSet_json_format)(m->resolved_features) ==
423 UPB_DESC(FeatureSet_ALLOW) &&
424 upb_strtable_lookup(&m->ntof, json_name, &v)) {
425 _upb_DefBuilder_Errf(
426 ctx, "duplicate json_name for (%s) with original field name (%s)",
427 shortname, json_name);
428 }
429
430 if (upb_strtable_lookup(&m->jtof, json_name, &v)) {
431 if (!skip_json_conflicts) {
432 _upb_DefBuilder_Errf(ctx, "duplicate json_name (%s)", json_name);
433 }
434 } else {
435 const size_t json_size = strlen(json_name);
436 ok = upb_strtable_insert(&m->jtof, json_name, json_size,
437 upb_value_constptr(f), ctx->arena);
438 if (!ok) _upb_DefBuilder_OomErr(ctx);
439 }
440
441 if (upb_inttable_lookup(&m->itof, field_number, NULL)) {
442 _upb_DefBuilder_Errf(ctx, "duplicate field number (%u)", field_number);
443 }
444
445 ok = upb_inttable_insert(&m->itof, field_number, v, ctx->arena);
446 if (!ok) _upb_DefBuilder_OomErr(ctx);
447 }
448
_upb_MessageDef_CreateMiniTable(upb_DefBuilder * ctx,upb_MessageDef * m)449 void _upb_MessageDef_CreateMiniTable(upb_DefBuilder* ctx, upb_MessageDef* m) {
450 if (ctx->layout == NULL) {
451 m->layout = _upb_MessageDef_MakeMiniTable(ctx, m);
452 } else {
453 m->layout = upb_MiniTableFile_Message(ctx->layout, ctx->msg_count++);
454 UPB_ASSERT(m->field_count == upb_MiniTable_FieldCount(m->layout));
455
456 // We don't need the result of this call, but it will assign layout_index
457 // for all the fields in O(n lg n) time.
458 _upb_FieldDefs_Sorted(m->fields, m->field_count, ctx->tmp_arena);
459 }
460
461 for (int i = 0; i < m->nested_msg_count; i++) {
462 upb_MessageDef* nested =
463 (upb_MessageDef*)upb_MessageDef_NestedMessage(m, i);
464 _upb_MessageDef_CreateMiniTable(ctx, nested);
465 }
466 }
467
_upb_MessageDef_LinkMiniTable(upb_DefBuilder * ctx,const upb_MessageDef * m)468 void _upb_MessageDef_LinkMiniTable(upb_DefBuilder* ctx,
469 const upb_MessageDef* m) {
470 for (int i = 0; i < upb_MessageDef_NestedExtensionCount(m); i++) {
471 const upb_FieldDef* ext = upb_MessageDef_NestedExtension(m, i);
472 _upb_FieldDef_BuildMiniTableExtension(ctx, ext);
473 }
474
475 for (int i = 0; i < m->nested_msg_count; i++) {
476 _upb_MessageDef_LinkMiniTable(ctx, upb_MessageDef_NestedMessage(m, i));
477 }
478
479 if (ctx->layout) return;
480
481 for (int i = 0; i < m->field_count; i++) {
482 const upb_FieldDef* f = upb_MessageDef_Field(m, i);
483 const upb_MessageDef* sub_m = upb_FieldDef_MessageSubDef(f);
484 const upb_EnumDef* sub_e = upb_FieldDef_EnumSubDef(f);
485 const int layout_index = _upb_FieldDef_LayoutIndex(f);
486 upb_MiniTable* mt = (upb_MiniTable*)upb_MessageDef_MiniTable(m);
487
488 UPB_ASSERT(layout_index < m->field_count);
489 upb_MiniTableField* mt_f =
490 (upb_MiniTableField*)&m->layout->UPB_PRIVATE(fields)[layout_index];
491 if (sub_m) {
492 if (!mt->UPB_PRIVATE(subs)) {
493 _upb_DefBuilder_Errf(ctx, "unexpected submsg for (%s)", m->full_name);
494 }
495 UPB_ASSERT(mt_f);
496 UPB_ASSERT(sub_m->layout);
497 if (UPB_UNLIKELY(!upb_MiniTable_SetSubMessage(mt, mt_f, sub_m->layout))) {
498 _upb_DefBuilder_Errf(ctx, "invalid submsg for (%s)", m->full_name);
499 }
500 } else if (_upb_FieldDef_IsClosedEnum(f)) {
501 const upb_MiniTableEnum* mt_e = _upb_EnumDef_MiniTable(sub_e);
502 if (UPB_UNLIKELY(!upb_MiniTable_SetSubEnum(mt, mt_f, mt_e))) {
503 _upb_DefBuilder_Errf(ctx, "invalid subenum for (%s)", m->full_name);
504 }
505 }
506 }
507
508 #ifndef NDEBUG
509 for (int i = 0; i < m->field_count; i++) {
510 const upb_FieldDef* f = upb_MessageDef_Field(m, i);
511 const int layout_index = _upb_FieldDef_LayoutIndex(f);
512 UPB_ASSERT(layout_index < upb_MiniTable_FieldCount(m->layout));
513 const upb_MiniTableField* mt_f =
514 &m->layout->UPB_PRIVATE(fields)[layout_index];
515 UPB_ASSERT(upb_FieldDef_Type(f) == upb_MiniTableField_Type(mt_f));
516 UPB_ASSERT(upb_FieldDef_CType(f) == upb_MiniTableField_CType(mt_f));
517 UPB_ASSERT(upb_FieldDef_HasPresence(f) ==
518 upb_MiniTableField_HasPresence(mt_f));
519 }
520 #endif
521 }
522
_upb_MessageDef_ValidateUtf8(const upb_MessageDef * m)523 static bool _upb_MessageDef_ValidateUtf8(const upb_MessageDef* m) {
524 bool has_string = false;
525 for (int i = 0; i < m->field_count; i++) {
526 const upb_FieldDef* f = upb_MessageDef_Field(m, i);
527 // Old binaries do not recognize the field-level "FlipValidateUtf8" wire
528 // modifier, so we do not actually have field-level control for old
529 // binaries. Given this, we judge that the better failure mode is to be
530 // more lax than intended, rather than more strict. To achieve this, we
531 // only mark the message with the ValidateUtf8 modifier if *all* fields
532 // validate UTF-8.
533 if (!_upb_FieldDef_ValidateUtf8(f)) return false;
534 if (upb_FieldDef_Type(f) == kUpb_FieldType_String) has_string = true;
535 }
536 return has_string;
537 }
538
_upb_MessageDef_Modifiers(const upb_MessageDef * m)539 static uint64_t _upb_MessageDef_Modifiers(const upb_MessageDef* m) {
540 uint64_t out = 0;
541
542 if (UPB_DESC(FeatureSet_repeated_field_encoding(m->resolved_features)) ==
543 UPB_DESC(FeatureSet_PACKED)) {
544 out |= kUpb_MessageModifier_DefaultIsPacked;
545 }
546
547 if (_upb_MessageDef_ValidateUtf8(m)) {
548 out |= kUpb_MessageModifier_ValidateUtf8;
549 }
550
551 if (m->ext_range_count) {
552 out |= kUpb_MessageModifier_IsExtendable;
553 }
554
555 return out;
556 }
557
_upb_MessageDef_EncodeMap(upb_DescState * s,const upb_MessageDef * m,upb_Arena * a)558 static bool _upb_MessageDef_EncodeMap(upb_DescState* s, const upb_MessageDef* m,
559 upb_Arena* a) {
560 if (m->field_count != 2) return false;
561
562 const upb_FieldDef* key_field = upb_MessageDef_Field(m, 0);
563 const upb_FieldDef* val_field = upb_MessageDef_Field(m, 1);
564 if (key_field == NULL || val_field == NULL) return false;
565
566 UPB_ASSERT(_upb_FieldDef_LayoutIndex(key_field) == 0);
567 UPB_ASSERT(_upb_FieldDef_LayoutIndex(val_field) == 1);
568
569 s->ptr = upb_MtDataEncoder_EncodeMap(
570 &s->e, s->ptr, upb_FieldDef_Type(key_field), upb_FieldDef_Type(val_field),
571 _upb_FieldDef_Modifiers(key_field), _upb_FieldDef_Modifiers(val_field));
572 return true;
573 }
574
_upb_MessageDef_EncodeMessage(upb_DescState * s,const upb_MessageDef * m,upb_Arena * a)575 static bool _upb_MessageDef_EncodeMessage(upb_DescState* s,
576 const upb_MessageDef* m,
577 upb_Arena* a) {
578 const upb_FieldDef** sorted = NULL;
579 if (!m->is_sorted) {
580 sorted = _upb_FieldDefs_Sorted(m->fields, m->field_count, a);
581 if (!sorted) return false;
582 }
583
584 s->ptr = upb_MtDataEncoder_StartMessage(&s->e, s->ptr,
585 _upb_MessageDef_Modifiers(m));
586
587 for (int i = 0; i < m->field_count; i++) {
588 const upb_FieldDef* f = sorted ? sorted[i] : upb_MessageDef_Field(m, i);
589 const upb_FieldType type = upb_FieldDef_Type(f);
590 const int number = upb_FieldDef_Number(f);
591 const uint64_t modifiers = _upb_FieldDef_Modifiers(f);
592
593 if (!_upb_DescState_Grow(s, a)) return false;
594 s->ptr = upb_MtDataEncoder_PutField(&s->e, s->ptr, type, number, modifiers);
595 }
596
597 for (int i = 0; i < m->real_oneof_count; i++) {
598 if (!_upb_DescState_Grow(s, a)) return false;
599 s->ptr = upb_MtDataEncoder_StartOneof(&s->e, s->ptr);
600
601 const upb_OneofDef* o = upb_MessageDef_Oneof(m, i);
602 const int field_count = upb_OneofDef_FieldCount(o);
603 for (int j = 0; j < field_count; j++) {
604 const int number = upb_FieldDef_Number(upb_OneofDef_Field(o, j));
605
606 if (!_upb_DescState_Grow(s, a)) return false;
607 s->ptr = upb_MtDataEncoder_PutOneofField(&s->e, s->ptr, number);
608 }
609 }
610
611 return true;
612 }
613
_upb_MessageDef_EncodeMessageSet(upb_DescState * s,const upb_MessageDef * m,upb_Arena * a)614 static bool _upb_MessageDef_EncodeMessageSet(upb_DescState* s,
615 const upb_MessageDef* m,
616 upb_Arena* a) {
617 s->ptr = upb_MtDataEncoder_EncodeMessageSet(&s->e, s->ptr);
618
619 return true;
620 }
621
upb_MessageDef_MiniDescriptorEncode(const upb_MessageDef * m,upb_Arena * a,upb_StringView * out)622 bool upb_MessageDef_MiniDescriptorEncode(const upb_MessageDef* m, upb_Arena* a,
623 upb_StringView* out) {
624 upb_DescState s;
625 _upb_DescState_Init(&s);
626
627 if (!_upb_DescState_Grow(&s, a)) return false;
628
629 if (upb_MessageDef_IsMapEntry(m)) {
630 if (!_upb_MessageDef_EncodeMap(&s, m, a)) return false;
631 } else if (UPB_DESC(MessageOptions_message_set_wire_format)(m->opts)) {
632 if (!_upb_MessageDef_EncodeMessageSet(&s, m, a)) return false;
633 } else {
634 if (!_upb_MessageDef_EncodeMessage(&s, m, a)) return false;
635 }
636
637 if (!_upb_DescState_Grow(&s, a)) return false;
638 *s.ptr = '\0';
639
640 out->data = s.buf;
641 out->size = s.ptr - s.buf;
642 return true;
643 }
644
_upb_ReservedNames_New(upb_DefBuilder * ctx,int n,const upb_StringView * protos)645 static upb_StringView* _upb_ReservedNames_New(upb_DefBuilder* ctx, int n,
646 const upb_StringView* protos) {
647 upb_StringView* sv = _upb_DefBuilder_Alloc(ctx, sizeof(upb_StringView) * n);
648 for (int i = 0; i < n; i++) {
649 sv[i].data =
650 upb_strdup2(protos[i].data, protos[i].size, _upb_DefBuilder_Arena(ctx));
651 sv[i].size = protos[i].size;
652 }
653 return sv;
654 }
655
create_msgdef(upb_DefBuilder * ctx,const char * prefix,const UPB_DESC (DescriptorProto *)msg_proto,const UPB_DESC (FeatureSet *)parent_features,const upb_MessageDef * containing_type,upb_MessageDef * m)656 static void create_msgdef(upb_DefBuilder* ctx, const char* prefix,
657 const UPB_DESC(DescriptorProto*) msg_proto,
658 const UPB_DESC(FeatureSet*) parent_features,
659 const upb_MessageDef* containing_type,
660 upb_MessageDef* m) {
661 const UPB_DESC(OneofDescriptorProto)* const* oneofs;
662 const UPB_DESC(FieldDescriptorProto)* const* fields;
663 const UPB_DESC(DescriptorProto_ExtensionRange)* const* ext_ranges;
664 const UPB_DESC(DescriptorProto_ReservedRange)* const* res_ranges;
665 const upb_StringView* res_names;
666 size_t n_oneof, n_field, n_enum, n_ext, n_msg;
667 size_t n_ext_range, n_res_range, n_res_name;
668 upb_StringView name;
669
670 UPB_DEF_SET_OPTIONS(m->opts, DescriptorProto, MessageOptions, msg_proto);
671 m->resolved_features = _upb_DefBuilder_ResolveFeatures(
672 ctx, parent_features, UPB_DESC(MessageOptions_features)(m->opts));
673
674 // Must happen before _upb_DefBuilder_Add()
675 m->file = _upb_DefBuilder_File(ctx);
676
677 m->containing_type = containing_type;
678 m->is_sorted = true;
679
680 name = UPB_DESC(DescriptorProto_name)(msg_proto);
681
682 m->full_name = _upb_DefBuilder_MakeFullName(ctx, prefix, name);
683 _upb_DefBuilder_Add(ctx, m->full_name, _upb_DefType_Pack(m, UPB_DEFTYPE_MSG));
684
685 oneofs = UPB_DESC(DescriptorProto_oneof_decl)(msg_proto, &n_oneof);
686 fields = UPB_DESC(DescriptorProto_field)(msg_proto, &n_field);
687 ext_ranges =
688 UPB_DESC(DescriptorProto_extension_range)(msg_proto, &n_ext_range);
689 res_ranges =
690 UPB_DESC(DescriptorProto_reserved_range)(msg_proto, &n_res_range);
691 res_names = UPB_DESC(DescriptorProto_reserved_name)(msg_proto, &n_res_name);
692
693 bool ok = upb_inttable_init(&m->itof, ctx->arena);
694 if (!ok) _upb_DefBuilder_OomErr(ctx);
695
696 ok = upb_strtable_init(&m->ntof, n_oneof + n_field, ctx->arena);
697 if (!ok) _upb_DefBuilder_OomErr(ctx);
698
699 ok = upb_strtable_init(&m->jtof, n_field, ctx->arena);
700 if (!ok) _upb_DefBuilder_OomErr(ctx);
701
702 m->oneof_count = n_oneof;
703 m->oneofs = _upb_OneofDefs_New(ctx, n_oneof, oneofs, m->resolved_features, m);
704
705 m->field_count = n_field;
706 m->fields = _upb_FieldDefs_New(ctx, n_field, fields, m->resolved_features,
707 m->full_name, m, &m->is_sorted);
708
709 // Message Sets may not contain fields.
710 if (UPB_UNLIKELY(UPB_DESC(MessageOptions_message_set_wire_format)(m->opts))) {
711 if (UPB_UNLIKELY(n_field > 0)) {
712 _upb_DefBuilder_Errf(ctx, "invalid message set (%s)", m->full_name);
713 }
714 }
715
716 m->ext_range_count = n_ext_range;
717 m->ext_ranges = _upb_ExtensionRanges_New(ctx, n_ext_range, ext_ranges,
718 m->resolved_features, m);
719
720 m->res_range_count = n_res_range;
721 m->res_ranges =
722 _upb_MessageReservedRanges_New(ctx, n_res_range, res_ranges, m);
723
724 m->res_name_count = n_res_name;
725 m->res_names = _upb_ReservedNames_New(ctx, n_res_name, res_names);
726
727 const size_t synthetic_count = _upb_OneofDefs_Finalize(ctx, m);
728 m->real_oneof_count = m->oneof_count - synthetic_count;
729
730 assign_msg_wellknowntype(m);
731 upb_inttable_compact(&m->itof, ctx->arena);
732
733 const UPB_DESC(EnumDescriptorProto)* const* enums =
734 UPB_DESC(DescriptorProto_enum_type)(msg_proto, &n_enum);
735 m->nested_enum_count = n_enum;
736 m->nested_enums =
737 _upb_EnumDefs_New(ctx, n_enum, enums, m->resolved_features, m);
738
739 const UPB_DESC(FieldDescriptorProto)* const* exts =
740 UPB_DESC(DescriptorProto_extension)(msg_proto, &n_ext);
741 m->nested_ext_count = n_ext;
742 m->nested_exts = _upb_Extensions_New(ctx, n_ext, exts, m->resolved_features,
743 m->full_name, m);
744
745 const UPB_DESC(DescriptorProto)* const* msgs =
746 UPB_DESC(DescriptorProto_nested_type)(msg_proto, &n_msg);
747 m->nested_msg_count = n_msg;
748 m->nested_msgs =
749 _upb_MessageDefs_New(ctx, n_msg, msgs, m->resolved_features, m);
750 }
751
752 // Allocate and initialize an array of |n| message defs.
_upb_MessageDefs_New(upb_DefBuilder * ctx,int n,const UPB_DESC (DescriptorProto *)const * protos,const UPB_DESC (FeatureSet *)parent_features,const upb_MessageDef * containing_type)753 upb_MessageDef* _upb_MessageDefs_New(upb_DefBuilder* ctx, int n,
754 const UPB_DESC(DescriptorProto*)
755 const* protos,
756 const UPB_DESC(FeatureSet*)
757 parent_features,
758 const upb_MessageDef* containing_type) {
759 _upb_DefType_CheckPadding(sizeof(upb_MessageDef));
760
761 const char* name = containing_type ? containing_type->full_name
762 : _upb_FileDef_RawPackage(ctx->file);
763
764 upb_MessageDef* m = _upb_DefBuilder_Alloc(ctx, sizeof(upb_MessageDef) * n);
765 for (int i = 0; i < n; i++) {
766 create_msgdef(ctx, name, protos[i], parent_features, containing_type,
767 &m[i]);
768 }
769 return m;
770 }
771