• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // Copyright 2016 The Chromium Authors. All rights reserved.
2 
3 #ifndef THIRD_PARTY_ICU_FUZZERS_FUZZER_UTILS_H_
4 #define THIRD_PARTY_ICU_FUZZERS_FUZZER_UTILS_H_
5 
6 #include <assert.h>
7 #include <algorithm>
8 #include <random>
9 
10 #include "base/at_exit.h"
11 #include "base/i18n/icu_util.h"
12 #include "third_party/icu/source/common/unicode/locid.h"
13 #include "third_party/icu/source/common/unicode/uchar.h"
14 #include "third_party/icu/source/common/unicode/unistr.h"
15 
16 struct IcuEnvironment {
IcuEnvironmentIcuEnvironment17   IcuEnvironment() {
18     base::i18n::InitializeICU();
19   }
20 };
21 
22 // Create RNG and seed it from data.
CreateRng(const uint8_t * data,size_t size)23 std::mt19937_64 CreateRng(const uint8_t* data, size_t size) {
24   std::mt19937_64 rng;
25   std::string str = std::string(reinterpret_cast<const char*>(data), size);
26   std::size_t data_hash = std::hash<std::string>()(str);
27   rng.seed(data_hash);
28   return rng;
29 }
30 
GetRandomLocale(std::mt19937_64 * rng)31 const icu::Locale& GetRandomLocale(std::mt19937_64* rng) {
32   int32_t num_locales = 0;
33   const icu::Locale* locales = icu::Locale::getAvailableLocales(num_locales);
34   assert(num_locales > 0);
35   return locales[(*rng)() % num_locales];
36 }
37 
UnicodeStringFromUtf8(const uint8_t * data,size_t size)38 icu::UnicodeString UnicodeStringFromUtf8(const uint8_t* data, size_t size) {
39   return icu::UnicodeString::fromUTF8(
40       icu::StringPiece(reinterpret_cast<const char*>(data), size));
41 }
42 
UnicodeStringFromUtf32(const uint8_t * data,size_t size)43 icu::UnicodeString UnicodeStringFromUtf32(const uint8_t* data, size_t size) {
44   std::vector<UChar32> uchars;
45   uchars.resize(size * sizeof(uint8_t) / (sizeof(UChar32)));
46   memcpy(uchars.data(), data, uchars.size() * sizeof(UChar32));
47   for (size_t i = 0; i < uchars.size(); ++i) {
48     // The valid range for UTF32 is [0, UCHAR_MAX_VALUE]
49     // By  % with (UCHAR_MAX_VALUE + 2) we make the output mostly valid  with
50     // a small percentage of (1 / UCHAR_MAX_VALUE) invalid data in UTF8.
51     uchars[i] = uchars[i] % (UCHAR_MAX_VALUE + 2);
52   }
53 
54   return icu::UnicodeString::fromUTF32(uchars.data(), uchars.size());
55 }
56 
RandomChar16Array(size_t random_value,const uint8_t * data,size_t size)57 std::vector<char16_t> RandomChar16Array(size_t random_value,
58                                         const uint8_t* data,
59                                         size_t size) {
60   std::vector<char16_t> arr;
61   arr.resize(random_value % size * sizeof(uint8_t) / sizeof(char16_t));
62   memcpy(arr.data(), data, arr.size() * sizeof(char16_t) / sizeof(uint8_t));
63   return arr;
64 }
65 
66 #endif  // THIRD_PARTY_ICU_FUZZERS_FUZZER_UTILS_H_
67