• 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/internal/compare_unknown.h"
9 
10 #include <stdlib.h>
11 
12 #include "upb/base/string_view.h"
13 #include "upb/mem/alloc.h"
14 #include "upb/wire/eps_copy_input_stream.h"
15 #include "upb/wire/reader.h"
16 #include "upb/wire/types.h"
17 
18 // Must be last.
19 #include "upb/port/def.inc"
20 
21 typedef struct upb_UnknownFields upb_UnknownFields;
22 
23 typedef struct {
24   uint32_t tag;
25   union {
26     uint64_t varint;
27     uint64_t uint64;
28     uint32_t uint32;
29     upb_StringView delimited;
30     upb_UnknownFields* group;
31   } data;
32 } upb_UnknownField;
33 
34 struct upb_UnknownFields {
35   size_t size;
36   size_t capacity;
37   upb_UnknownField* fields;
38 };
39 
40 typedef struct {
41   upb_EpsCopyInputStream stream;
42   upb_Arena* arena;
43   upb_UnknownField* tmp;
44   size_t tmp_size;
45   int depth;
46   upb_UnknownCompareResult status;
47   jmp_buf err;
48 } upb_UnknownField_Context;
49 
upb_UnknownFields_OutOfMemory(upb_UnknownField_Context * ctx)50 UPB_NORETURN static void upb_UnknownFields_OutOfMemory(
51     upb_UnknownField_Context* ctx) {
52   ctx->status = kUpb_UnknownCompareResult_OutOfMemory;
53   UPB_LONGJMP(ctx->err, 1);
54 }
55 
upb_UnknownFields_Grow(upb_UnknownField_Context * ctx,upb_UnknownField ** base,upb_UnknownField ** ptr,upb_UnknownField ** end)56 static void upb_UnknownFields_Grow(upb_UnknownField_Context* ctx,
57                                    upb_UnknownField** base,
58                                    upb_UnknownField** ptr,
59                                    upb_UnknownField** end) {
60   size_t old = (*ptr - *base);
61   size_t new = UPB_MAX(4, old * 2);
62 
63   *base = upb_Arena_Realloc(ctx->arena, *base, old * sizeof(**base),
64                             new * sizeof(**base));
65   if (!*base) upb_UnknownFields_OutOfMemory(ctx);
66 
67   *ptr = *base + old;
68   *end = *base + new;
69 }
70 
71 // We have to implement our own sort here, since qsort() is not an in-order
72 // sort. Here we use merge sort, the simplest in-order sort.
upb_UnknownFields_Merge(upb_UnknownField * arr,size_t start,size_t mid,size_t end,upb_UnknownField * tmp)73 static void upb_UnknownFields_Merge(upb_UnknownField* arr, size_t start,
74                                     size_t mid, size_t end,
75                                     upb_UnknownField* tmp) {
76   memcpy(tmp, &arr[start], (end - start) * sizeof(*tmp));
77 
78   upb_UnknownField* ptr1 = tmp;
79   upb_UnknownField* end1 = &tmp[mid - start];
80   upb_UnknownField* ptr2 = &tmp[mid - start];
81   upb_UnknownField* end2 = &tmp[end - start];
82   upb_UnknownField* out = &arr[start];
83 
84   while (ptr1 < end1 && ptr2 < end2) {
85     if (ptr1->tag <= ptr2->tag) {
86       *out++ = *ptr1++;
87     } else {
88       *out++ = *ptr2++;
89     }
90   }
91 
92   if (ptr1 < end1) {
93     memcpy(out, ptr1, (end1 - ptr1) * sizeof(*out));
94   } else if (ptr2 < end2) {
95     memcpy(out, ptr1, (end2 - ptr2) * sizeof(*out));
96   }
97 }
98 
upb_UnknownFields_SortRecursive(upb_UnknownField * arr,size_t start,size_t end,upb_UnknownField * tmp)99 static void upb_UnknownFields_SortRecursive(upb_UnknownField* arr, size_t start,
100                                             size_t end, upb_UnknownField* tmp) {
101   if (end - start > 1) {
102     size_t mid = start + ((end - start) / 2);
103     upb_UnknownFields_SortRecursive(arr, start, mid, tmp);
104     upb_UnknownFields_SortRecursive(arr, mid, end, tmp);
105     upb_UnknownFields_Merge(arr, start, mid, end, tmp);
106   }
107 }
108 
upb_UnknownFields_Sort(upb_UnknownField_Context * ctx,upb_UnknownFields * fields)109 static void upb_UnknownFields_Sort(upb_UnknownField_Context* ctx,
110                                    upb_UnknownFields* fields) {
111   if (ctx->tmp_size < fields->size) {
112     const int oldsize = ctx->tmp_size * sizeof(*ctx->tmp);
113     ctx->tmp_size = UPB_MAX(8, ctx->tmp_size);
114     while (ctx->tmp_size < fields->size) ctx->tmp_size *= 2;
115     const int newsize = ctx->tmp_size * sizeof(*ctx->tmp);
116     ctx->tmp = upb_grealloc(ctx->tmp, oldsize, newsize);
117   }
118   upb_UnknownFields_SortRecursive(fields->fields, 0, fields->size, ctx->tmp);
119 }
120 
upb_UnknownFields_DoBuild(upb_UnknownField_Context * ctx,const char ** buf)121 static upb_UnknownFields* upb_UnknownFields_DoBuild(
122     upb_UnknownField_Context* ctx, const char** buf) {
123   upb_UnknownField* arr_base = NULL;
124   upb_UnknownField* arr_ptr = NULL;
125   upb_UnknownField* arr_end = NULL;
126   const char* ptr = *buf;
127   uint32_t last_tag = 0;
128   bool sorted = true;
129   while (!upb_EpsCopyInputStream_IsDone(&ctx->stream, &ptr)) {
130     uint32_t tag;
131     ptr = upb_WireReader_ReadTag(ptr, &tag);
132     UPB_ASSERT(tag <= UINT32_MAX);
133     int wire_type = upb_WireReader_GetWireType(tag);
134     if (wire_type == kUpb_WireType_EndGroup) break;
135     if (tag < last_tag) sorted = false;
136     last_tag = tag;
137 
138     if (arr_ptr == arr_end) {
139       upb_UnknownFields_Grow(ctx, &arr_base, &arr_ptr, &arr_end);
140     }
141     upb_UnknownField* field = arr_ptr;
142     field->tag = tag;
143     arr_ptr++;
144 
145     switch (wire_type) {
146       case kUpb_WireType_Varint:
147         ptr = upb_WireReader_ReadVarint(ptr, &field->data.varint);
148         break;
149       case kUpb_WireType_64Bit:
150         ptr = upb_WireReader_ReadFixed64(ptr, &field->data.uint64);
151         break;
152       case kUpb_WireType_32Bit:
153         ptr = upb_WireReader_ReadFixed32(ptr, &field->data.uint32);
154         break;
155       case kUpb_WireType_Delimited: {
156         int size;
157         ptr = upb_WireReader_ReadSize(ptr, &size);
158         const char* s_ptr = ptr;
159         ptr = upb_EpsCopyInputStream_ReadStringAliased(&ctx->stream, &s_ptr,
160                                                        size);
161         field->data.delimited.data = s_ptr;
162         field->data.delimited.size = size;
163         break;
164       }
165       case kUpb_WireType_StartGroup:
166         if (--ctx->depth == 0) {
167           ctx->status = kUpb_UnknownCompareResult_MaxDepthExceeded;
168           UPB_LONGJMP(ctx->err, 1);
169         }
170         field->data.group = upb_UnknownFields_DoBuild(ctx, &ptr);
171         ctx->depth++;
172         break;
173       default:
174         UPB_UNREACHABLE();
175     }
176   }
177 
178   *buf = ptr;
179   upb_UnknownFields* ret = upb_Arena_Malloc(ctx->arena, sizeof(*ret));
180   if (!ret) upb_UnknownFields_OutOfMemory(ctx);
181   ret->fields = arr_base;
182   ret->size = arr_ptr - arr_base;
183   ret->capacity = arr_end - arr_base;
184   if (!sorted) {
185     upb_UnknownFields_Sort(ctx, ret);
186   }
187   return ret;
188 }
189 
190 // Builds a upb_UnknownFields data structure from the binary data in buf.
upb_UnknownFields_Build(upb_UnknownField_Context * ctx,const char * ptr,size_t size)191 static upb_UnknownFields* upb_UnknownFields_Build(upb_UnknownField_Context* ctx,
192                                                   const char* ptr,
193                                                   size_t size) {
194   upb_EpsCopyInputStream_Init(&ctx->stream, &ptr, size, true);
195   upb_UnknownFields* fields = upb_UnknownFields_DoBuild(ctx, &ptr);
196   UPB_ASSERT(upb_EpsCopyInputStream_IsDone(&ctx->stream, &ptr) &&
197              !upb_EpsCopyInputStream_IsError(&ctx->stream));
198   return fields;
199 }
200 
201 // Compares two sorted upb_UnknownFields structures for equality.
upb_UnknownFields_IsEqual(const upb_UnknownFields * uf1,const upb_UnknownFields * uf2)202 static bool upb_UnknownFields_IsEqual(const upb_UnknownFields* uf1,
203                                       const upb_UnknownFields* uf2) {
204   if (uf1->size != uf2->size) return false;
205   for (size_t i = 0, n = uf1->size; i < n; i++) {
206     upb_UnknownField* f1 = &uf1->fields[i];
207     upb_UnknownField* f2 = &uf2->fields[i];
208     if (f1->tag != f2->tag) return false;
209     int wire_type = f1->tag & 7;
210     switch (wire_type) {
211       case kUpb_WireType_Varint:
212         if (f1->data.varint != f2->data.varint) return false;
213         break;
214       case kUpb_WireType_64Bit:
215         if (f1->data.uint64 != f2->data.uint64) return false;
216         break;
217       case kUpb_WireType_32Bit:
218         if (f1->data.uint32 != f2->data.uint32) return false;
219         break;
220       case kUpb_WireType_Delimited:
221         if (!upb_StringView_IsEqual(f1->data.delimited, f2->data.delimited)) {
222           return false;
223         }
224         break;
225       case kUpb_WireType_StartGroup:
226         if (!upb_UnknownFields_IsEqual(f1->data.group, f2->data.group)) {
227           return false;
228         }
229         break;
230       default:
231         UPB_UNREACHABLE();
232     }
233   }
234   return true;
235 }
236 
upb_UnknownField_DoCompare(upb_UnknownField_Context * ctx,const char * buf1,size_t size1,const char * buf2,size_t size2)237 static upb_UnknownCompareResult upb_UnknownField_DoCompare(
238     upb_UnknownField_Context* ctx, const char* buf1, size_t size1,
239     const char* buf2, size_t size2) {
240   upb_UnknownCompareResult ret;
241   // First build both unknown fields into a sorted data structure (similar
242   // to the UnknownFieldSet in C++).
243   upb_UnknownFields* uf1 = upb_UnknownFields_Build(ctx, buf1, size1);
244   upb_UnknownFields* uf2 = upb_UnknownFields_Build(ctx, buf2, size2);
245 
246   // Now perform the equality check on the sorted structures.
247   if (upb_UnknownFields_IsEqual(uf1, uf2)) {
248     ret = kUpb_UnknownCompareResult_Equal;
249   } else {
250     ret = kUpb_UnknownCompareResult_NotEqual;
251   }
252   return ret;
253 }
254 
upb_UnknownField_Compare(upb_UnknownField_Context * const ctx,const char * const buf1,const size_t size1,const char * const buf2,const size_t size2)255 static upb_UnknownCompareResult upb_UnknownField_Compare(
256     upb_UnknownField_Context* const ctx, const char* const buf1,
257     const size_t size1, const char* const buf2, const size_t size2) {
258   upb_UnknownCompareResult ret;
259   if (UPB_SETJMP(ctx->err) == 0) {
260     ret = upb_UnknownField_DoCompare(ctx, buf1, size1, buf2, size2);
261   } else {
262     ret = ctx->status;
263     UPB_ASSERT(ret != kUpb_UnknownCompareResult_Equal);
264   }
265 
266   upb_Arena_Free(ctx->arena);
267   upb_gfree(ctx->tmp);
268   return ret;
269 }
270 
UPB_PRIVATE(_upb_Message_UnknownFieldsAreEqual)271 upb_UnknownCompareResult UPB_PRIVATE(_upb_Message_UnknownFieldsAreEqual)(
272     const char* buf1, size_t size1, const char* buf2, size_t size2,
273     int max_depth) {
274   if (size1 == 0 && size2 == 0) return kUpb_UnknownCompareResult_Equal;
275   if (size1 == 0 || size2 == 0) return kUpb_UnknownCompareResult_NotEqual;
276   if (memcmp(buf1, buf2, size1) == 0) return kUpb_UnknownCompareResult_Equal;
277 
278   upb_UnknownField_Context ctx = {
279       .arena = upb_Arena_New(),
280       .depth = max_depth,
281       .tmp = NULL,
282       .tmp_size = 0,
283       .status = kUpb_UnknownCompareResult_Equal,
284   };
285 
286   if (!ctx.arena) return kUpb_UnknownCompareResult_OutOfMemory;
287 
288   return upb_UnknownField_Compare(&ctx, buf1, size1, buf2, size2);
289 }
290