• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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/message/promote.h"
9 
10 #include <stdbool.h>
11 #include <stdint.h>
12 #include <string.h>
13 
14 #include "upb/base/descriptor_constants.h"
15 #include "upb/mem/arena.h"
16 #include "upb/message/accessors.h"
17 #include "upb/message/array.h"
18 #include "upb/message/internal/array.h"
19 #include "upb/message/internal/extension.h"
20 #include "upb/message/internal/message.h"
21 #include "upb/message/internal/tagged_ptr.h"
22 #include "upb/message/map.h"
23 #include "upb/message/message.h"
24 #include "upb/message/tagged_ptr.h"
25 #include "upb/mini_table/extension.h"
26 #include "upb/mini_table/field.h"
27 #include "upb/mini_table/message.h"
28 #include "upb/mini_table/sub.h"
29 #include "upb/wire/decode.h"
30 #include "upb/wire/eps_copy_input_stream.h"
31 #include "upb/wire/reader.h"
32 
33 // Must be last.
34 #include "upb/port/def.inc"
35 
36 // Parses unknown data by merging into existing base_message or creating a
37 // new message usingg mini_table.
upb_MiniTable_ParseUnknownMessage(const char * unknown_data,size_t unknown_size,const upb_MiniTable * mini_table,upb_Message * base_message,int decode_options,upb_Arena * arena)38 static upb_UnknownToMessageRet upb_MiniTable_ParseUnknownMessage(
39     const char* unknown_data, size_t unknown_size,
40     const upb_MiniTable* mini_table, upb_Message* base_message,
41     int decode_options, upb_Arena* arena) {
42   upb_UnknownToMessageRet ret;
43   ret.message =
44       base_message ? base_message : _upb_Message_New(mini_table, arena);
45   if (!ret.message) {
46     ret.status = kUpb_UnknownToMessage_OutOfMemory;
47     return ret;
48   }
49   // Decode sub message using unknown field contents.
50   const char* data = unknown_data;
51   uint32_t tag;
52   uint64_t message_len = 0;
53   data = upb_WireReader_ReadTag(data, &tag);
54   data = upb_WireReader_ReadVarint(data, &message_len);
55   upb_DecodeStatus status = upb_Decode(data, message_len, ret.message,
56                                        mini_table, NULL, decode_options, arena);
57   if (status == kUpb_DecodeStatus_OutOfMemory) {
58     ret.status = kUpb_UnknownToMessage_OutOfMemory;
59   } else if (status == kUpb_DecodeStatus_Ok) {
60     ret.status = kUpb_UnknownToMessage_Ok;
61   } else {
62     ret.status = kUpb_UnknownToMessage_ParseError;
63   }
64   return ret;
65 }
66 
upb_Message_GetOrPromoteExtension(upb_Message * msg,const upb_MiniTableExtension * ext_table,int decode_options,upb_Arena * arena,upb_MessageValue * value)67 upb_GetExtension_Status upb_Message_GetOrPromoteExtension(
68     upb_Message* msg, const upb_MiniTableExtension* ext_table,
69     int decode_options, upb_Arena* arena, upb_MessageValue* value) {
70   UPB_ASSERT(!upb_Message_IsFrozen(msg));
71   UPB_ASSERT(upb_MiniTableExtension_CType(ext_table) == kUpb_CType_Message);
72   const upb_Extension* extension =
73       UPB_PRIVATE(_upb_Message_Getext)(msg, ext_table);
74   if (extension) {
75     memcpy(value, &extension->data, sizeof(upb_MessageValue));
76     return kUpb_GetExtension_Ok;
77   }
78 
79   // Check unknown fields, if available promote.
80   int field_number = upb_MiniTableExtension_Number(ext_table);
81   upb_FindUnknownRet result = upb_Message_FindUnknown(msg, field_number, 0);
82   if (result.status != kUpb_FindUnknown_Ok) {
83     return kUpb_GetExtension_NotPresent;
84   }
85   size_t len;
86   size_t ofs = result.ptr - upb_Message_GetUnknown(msg, &len);
87   // Decode and promote from unknown.
88   const upb_MiniTable* extension_table =
89       upb_MiniTableExtension_GetSubMessage(ext_table);
90   upb_UnknownToMessageRet parse_result = upb_MiniTable_ParseUnknownMessage(
91       result.ptr, result.len, extension_table,
92       /* base_message= */ NULL, decode_options, arena);
93   switch (parse_result.status) {
94     case kUpb_UnknownToMessage_OutOfMemory:
95       return kUpb_GetExtension_OutOfMemory;
96     case kUpb_UnknownToMessage_ParseError:
97       return kUpb_GetExtension_ParseError;
98     case kUpb_UnknownToMessage_NotFound:
99       return kUpb_GetExtension_NotPresent;
100     case kUpb_UnknownToMessage_Ok:
101       break;
102   }
103   upb_Message* extension_msg = parse_result.message;
104   // Add to extensions.
105   upb_Extension* ext =
106       UPB_PRIVATE(_upb_Message_GetOrCreateExtension)(msg, ext_table, arena);
107   if (!ext) {
108     return kUpb_GetExtension_OutOfMemory;
109   }
110   ext->data.msg_val = extension_msg;
111   value->msg_val = extension_msg;
112   const char* delete_ptr = upb_Message_GetUnknown(msg, &len) + ofs;
113   upb_Message_DeleteUnknown(msg, delete_ptr, result.len);
114   return kUpb_GetExtension_Ok;
115 }
116 
upb_FindUnknownRet_ParseError(void)117 static upb_FindUnknownRet upb_FindUnknownRet_ParseError(void) {
118   return (upb_FindUnknownRet){.status = kUpb_FindUnknown_ParseError};
119 }
120 
upb_Message_FindUnknown(const upb_Message * msg,uint32_t field_number,int depth_limit)121 upb_FindUnknownRet upb_Message_FindUnknown(const upb_Message* msg,
122                                            uint32_t field_number,
123                                            int depth_limit) {
124   depth_limit = depth_limit ? depth_limit : 100;
125 
126   size_t size;
127   upb_FindUnknownRet ret;
128 
129   const char* ptr = upb_Message_GetUnknown(msg, &size);
130   upb_EpsCopyInputStream stream;
131   upb_EpsCopyInputStream_Init(&stream, &ptr, size, true);
132 
133   while (!upb_EpsCopyInputStream_IsDone(&stream, &ptr)) {
134     uint32_t tag;
135     const char* unknown_begin = ptr;
136     ptr = upb_WireReader_ReadTag(ptr, &tag);
137     if (!ptr) return upb_FindUnknownRet_ParseError();
138     if (field_number == upb_WireReader_GetFieldNumber(tag)) {
139       ret.status = kUpb_FindUnknown_Ok;
140       ret.ptr = upb_EpsCopyInputStream_GetAliasedPtr(&stream, unknown_begin);
141       ptr = _upb_WireReader_SkipValue(ptr, tag, depth_limit, &stream);
142       // Because we know that the input is a flat buffer, it is safe to perform
143       // pointer arithmetic on aliased pointers.
144       ret.len = upb_EpsCopyInputStream_GetAliasedPtr(&stream, ptr) - ret.ptr;
145       return ret;
146     }
147 
148     ptr = _upb_WireReader_SkipValue(ptr, tag, depth_limit, &stream);
149     if (!ptr) return upb_FindUnknownRet_ParseError();
150   }
151   ret.status = kUpb_FindUnknown_NotPresent;
152   ret.ptr = NULL;
153   ret.len = 0;
154   return ret;
155 }
156 
upb_Message_PromoteOne(upb_TaggedMessagePtr * tagged,const upb_MiniTable * mini_table,int decode_options,upb_Arena * arena)157 static upb_DecodeStatus upb_Message_PromoteOne(upb_TaggedMessagePtr* tagged,
158                                                const upb_MiniTable* mini_table,
159                                                int decode_options,
160                                                upb_Arena* arena) {
161   upb_Message* empty =
162       UPB_PRIVATE(_upb_TaggedMessagePtr_GetEmptyMessage)(*tagged);
163   size_t unknown_size;
164   const char* unknown_data = upb_Message_GetUnknown(empty, &unknown_size);
165   upb_Message* promoted = upb_Message_New(mini_table, arena);
166   if (!promoted) return kUpb_DecodeStatus_OutOfMemory;
167   upb_DecodeStatus status = upb_Decode(unknown_data, unknown_size, promoted,
168                                        mini_table, NULL, decode_options, arena);
169   if (status == kUpb_DecodeStatus_Ok) {
170     *tagged = UPB_PRIVATE(_upb_TaggedMessagePtr_Pack)(promoted, false);
171   }
172   return status;
173 }
174 
upb_Message_PromoteMessage(upb_Message * parent,const upb_MiniTable * mini_table,const upb_MiniTableField * field,int decode_options,upb_Arena * arena,upb_Message ** promoted)175 upb_DecodeStatus upb_Message_PromoteMessage(upb_Message* parent,
176                                             const upb_MiniTable* mini_table,
177                                             const upb_MiniTableField* field,
178                                             int decode_options,
179                                             upb_Arena* arena,
180                                             upb_Message** promoted) {
181   UPB_ASSERT(!upb_Message_IsFrozen(parent));
182   const upb_MiniTable* sub_table =
183       upb_MiniTable_GetSubMessageTable(mini_table, field);
184   UPB_ASSERT(sub_table);
185   upb_TaggedMessagePtr tagged =
186       upb_Message_GetTaggedMessagePtr(parent, field, NULL);
187   upb_DecodeStatus ret =
188       upb_Message_PromoteOne(&tagged, sub_table, decode_options, arena);
189   if (ret == kUpb_DecodeStatus_Ok) {
190     *promoted = upb_TaggedMessagePtr_GetNonEmptyMessage(tagged);
191     upb_Message_SetMessage(parent, field, *promoted);
192   }
193   return ret;
194 }
195 
upb_Array_PromoteMessages(upb_Array * arr,const upb_MiniTable * mini_table,int decode_options,upb_Arena * arena)196 upb_DecodeStatus upb_Array_PromoteMessages(upb_Array* arr,
197                                            const upb_MiniTable* mini_table,
198                                            int decode_options,
199                                            upb_Arena* arena) {
200   void** data = upb_Array_MutableDataPtr(arr);
201   size_t size = upb_Array_Size(arr);
202   for (size_t i = 0; i < size; i++) {
203     upb_TaggedMessagePtr tagged;
204     memcpy(&tagged, &data[i], sizeof(tagged));
205     if (!upb_TaggedMessagePtr_IsEmpty(tagged)) continue;
206     upb_DecodeStatus status =
207         upb_Message_PromoteOne(&tagged, mini_table, decode_options, arena);
208     if (status != kUpb_DecodeStatus_Ok) return status;
209     memcpy(&data[i], &tagged, sizeof(tagged));
210   }
211   return kUpb_DecodeStatus_Ok;
212 }
213 
upb_Map_PromoteMessages(upb_Map * map,const upb_MiniTable * mini_table,int decode_options,upb_Arena * arena)214 upb_DecodeStatus upb_Map_PromoteMessages(upb_Map* map,
215                                          const upb_MiniTable* mini_table,
216                                          int decode_options, upb_Arena* arena) {
217   size_t iter = kUpb_Map_Begin;
218   upb_MessageValue key, val;
219   while (upb_Map_Next(map, &key, &val, &iter)) {
220     if (!upb_TaggedMessagePtr_IsEmpty(val.tagged_msg_val)) continue;
221     upb_DecodeStatus status = upb_Message_PromoteOne(
222         &val.tagged_msg_val, mini_table, decode_options, arena);
223     if (status != kUpb_DecodeStatus_Ok) return status;
224     upb_Map_SetEntryValue(map, iter, val);
225   }
226   return kUpb_DecodeStatus_Ok;
227 }
228 
229 ////////////////////////////////////////////////////////////////////////////////
230 // OLD promotion functions, will be removed!
231 ////////////////////////////////////////////////////////////////////////////////
232 
233 // Warning: See TODO
upb_MiniTable_PromoteUnknownToMessage(upb_Message * msg,const upb_MiniTable * mini_table,const upb_MiniTableField * field,const upb_MiniTable * sub_mini_table,int decode_options,upb_Arena * arena)234 upb_UnknownToMessageRet upb_MiniTable_PromoteUnknownToMessage(
235     upb_Message* msg, const upb_MiniTable* mini_table,
236     const upb_MiniTableField* field, const upb_MiniTable* sub_mini_table,
237     int decode_options, upb_Arena* arena) {
238   UPB_ASSERT(!upb_Message_IsFrozen(msg));
239   upb_FindUnknownRet unknown;
240   // We need to loop and merge unknowns that have matching tag field->number.
241   upb_Message* message = NULL;
242   // Callers should check that message is not set first before calling
243   // PromotoUnknownToMessage.
244   UPB_ASSERT(upb_MiniTable_GetSubMessageTable(mini_table, field) ==
245              sub_mini_table);
246   bool is_oneof = upb_MiniTableField_IsInOneof(field);
247   if (!is_oneof || UPB_PRIVATE(_upb_Message_GetOneofCase)(msg, field) ==
248                        upb_MiniTableField_Number(field)) {
249     UPB_ASSERT(upb_Message_GetMessage(msg, field) == NULL);
250   }
251   upb_UnknownToMessageRet ret;
252   ret.status = kUpb_UnknownToMessage_Ok;
253   do {
254     unknown =
255         upb_Message_FindUnknown(msg, upb_MiniTableField_Number(field),
256                                 upb_DecodeOptions_GetMaxDepth(decode_options));
257     switch (unknown.status) {
258       case kUpb_FindUnknown_Ok: {
259         const char* unknown_data = unknown.ptr;
260         size_t unknown_size = unknown.len;
261         ret = upb_MiniTable_ParseUnknownMessage(unknown_data, unknown_size,
262                                                 sub_mini_table, message,
263                                                 decode_options, arena);
264         if (ret.status == kUpb_UnknownToMessage_Ok) {
265           message = ret.message;
266           upb_Message_DeleteUnknown(msg, unknown_data, unknown_size);
267         }
268       } break;
269       case kUpb_FindUnknown_ParseError:
270         ret.status = kUpb_UnknownToMessage_ParseError;
271         break;
272       case kUpb_FindUnknown_NotPresent:
273         // If we parsed at least one unknown, we are done.
274         ret.status =
275             message ? kUpb_UnknownToMessage_Ok : kUpb_UnknownToMessage_NotFound;
276         break;
277     }
278   } while (unknown.status == kUpb_FindUnknown_Ok);
279   if (message) {
280     if (is_oneof) {
281       UPB_PRIVATE(_upb_Message_SetOneofCase)(msg, field);
282     }
283     upb_Message_SetMessage(msg, field, message);
284     ret.message = message;
285   }
286   return ret;
287 }
288 
289 // Moves repeated messages in unknowns to a upb_Array.
290 //
291 // Since the repeated field is not a scalar type we don't check for
292 // kUpb_LabelFlags_IsPacked.
293 // TODO: Optimize. Instead of converting messages one at a time,
294 // scan all unknown data once and compact.
upb_MiniTable_PromoteUnknownToMessageArray(upb_Message * msg,const upb_MiniTableField * field,const upb_MiniTable * mini_table,int decode_options,upb_Arena * arena)295 upb_UnknownToMessage_Status upb_MiniTable_PromoteUnknownToMessageArray(
296     upb_Message* msg, const upb_MiniTableField* field,
297     const upb_MiniTable* mini_table, int decode_options, upb_Arena* arena) {
298   UPB_ASSERT(!upb_Message_IsFrozen(msg));
299 
300   upb_Array* repeated_messages = upb_Message_GetMutableArray(msg, field);
301   // Find all unknowns with given field number and parse.
302   upb_FindUnknownRet unknown;
303   do {
304     unknown =
305         upb_Message_FindUnknown(msg, upb_MiniTableField_Number(field),
306                                 upb_DecodeOptions_GetMaxDepth(decode_options));
307     if (unknown.status == kUpb_FindUnknown_Ok) {
308       upb_UnknownToMessageRet ret = upb_MiniTable_ParseUnknownMessage(
309           unknown.ptr, unknown.len, mini_table,
310           /* base_message= */ NULL, decode_options, arena);
311       if (ret.status == kUpb_UnknownToMessage_Ok) {
312         upb_MessageValue value;
313         value.msg_val = ret.message;
314         // Allocate array on demand before append.
315         if (!repeated_messages) {
316           upb_Message_ResizeArrayUninitialized(msg, field, 0, arena);
317           repeated_messages = upb_Message_GetMutableArray(msg, field);
318         }
319         if (!upb_Array_Append(repeated_messages, value, arena)) {
320           return kUpb_UnknownToMessage_OutOfMemory;
321         }
322         upb_Message_DeleteUnknown(msg, unknown.ptr, unknown.len);
323       } else {
324         return ret.status;
325       }
326     }
327   } while (unknown.status == kUpb_FindUnknown_Ok);
328   return kUpb_UnknownToMessage_Ok;
329 }
330 
331 // Moves repeated messages in unknowns to a upb_Map.
upb_MiniTable_PromoteUnknownToMap(upb_Message * msg,const upb_MiniTable * mini_table,const upb_MiniTableField * field,int decode_options,upb_Arena * arena)332 upb_UnknownToMessage_Status upb_MiniTable_PromoteUnknownToMap(
333     upb_Message* msg, const upb_MiniTable* mini_table,
334     const upb_MiniTableField* field, int decode_options, upb_Arena* arena) {
335   UPB_ASSERT(!upb_Message_IsFrozen(msg));
336 
337   const upb_MiniTable* map_entry_mini_table =
338       upb_MiniTable_MapEntrySubMessage(mini_table, field);
339   UPB_ASSERT(upb_MiniTable_FieldCount(map_entry_mini_table) == 2);
340   // Find all unknowns with given field number and parse.
341   upb_FindUnknownRet unknown;
342   while (1) {
343     unknown =
344         upb_Message_FindUnknown(msg, upb_MiniTableField_Number(field),
345                                 upb_DecodeOptions_GetMaxDepth(decode_options));
346     if (unknown.status != kUpb_FindUnknown_Ok) break;
347     upb_UnknownToMessageRet ret = upb_MiniTable_ParseUnknownMessage(
348         unknown.ptr, unknown.len, map_entry_mini_table,
349         /* base_message= */ NULL, decode_options, arena);
350     if (ret.status != kUpb_UnknownToMessage_Ok) return ret.status;
351     // Allocate map on demand before append.
352     upb_Map* map = upb_Message_GetOrCreateMutableMap(msg, map_entry_mini_table,
353                                                      field, arena);
354     upb_Message* map_entry_message = ret.message;
355     bool insert_success = upb_Message_SetMapEntry(map, mini_table, field,
356                                                   map_entry_message, arena);
357     if (!insert_success) return kUpb_UnknownToMessage_OutOfMemory;
358     upb_Message_DeleteUnknown(msg, unknown.ptr, unknown.len);
359   }
360   return kUpb_UnknownToMessage_Ok;
361 }
362