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/check.h"
12 #include "base/i18n/icu_util.h"
13 #include "third_party/icu/source/common/unicode/locid.h"
14 #include "third_party/icu/source/common/unicode/uchar.h"
15 #include "third_party/icu/source/common/unicode/unistr.h"
16
17 struct IcuEnvironment {
IcuEnvironmentIcuEnvironment18 IcuEnvironment() {
19 CHECK(base::i18n::InitializeICU());
20 }
21 };
22
23 // Create RNG and seed it from data.
CreateRng(const uint8_t * data,size_t size)24 std::mt19937_64 CreateRng(const uint8_t* data, size_t size) {
25 std::mt19937_64 rng;
26 std::string str = std::string(reinterpret_cast<const char*>(data), size);
27 std::size_t data_hash = std::hash<std::string>()(str);
28 rng.seed(data_hash);
29 return rng;
30 }
31
GetRandomLocale(std::mt19937_64 * rng)32 const icu::Locale& GetRandomLocale(std::mt19937_64* rng) {
33 int32_t num_locales = 0;
34 const icu::Locale* locales = icu::Locale::getAvailableLocales(num_locales);
35 assert(num_locales > 0);
36 return locales[(*rng)() % num_locales];
37 }
38
UnicodeStringFromUtf8(const uint8_t * data,size_t size)39 icu::UnicodeString UnicodeStringFromUtf8(const uint8_t* data, size_t size) {
40 return icu::UnicodeString::fromUTF8(
41 icu::StringPiece(reinterpret_cast<const char*>(data), size));
42 }
43
UnicodeStringFromUtf32(const uint8_t * data,size_t size)44 icu::UnicodeString UnicodeStringFromUtf32(const uint8_t* data, size_t size) {
45 std::vector<UChar32> uchars;
46 uchars.resize(size * sizeof(uint8_t) / (sizeof(UChar32)));
47 memcpy(uchars.data(), data, uchars.size() * sizeof(UChar32));
48 for (size_t i = 0; i < uchars.size(); ++i) {
49 // The valid range for UTF32 is [0, UCHAR_MAX_VALUE]
50 // By % with (UCHAR_MAX_VALUE + 2) we make the output mostly valid with
51 // a small percentage of (1 / UCHAR_MAX_VALUE) invalid data in UTF8.
52 uchars[i] = uchars[i] % (UCHAR_MAX_VALUE + 2);
53 }
54
55 return icu::UnicodeString::fromUTF32(uchars.data(), uchars.size());
56 }
57
RandomChar16Array(size_t random_value,const uint8_t * data,size_t size)58 std::vector<char16_t> RandomChar16Array(size_t random_value,
59 const uint8_t* data,
60 size_t size) {
61 std::vector<char16_t> arr;
62 arr.resize(random_value % size * sizeof(uint8_t) / sizeof(char16_t));
63 memcpy(arr.data(), data, arr.size() * sizeof(char16_t) / sizeof(uint8_t));
64 return arr;
65 }
66
67 #endif // THIRD_PARTY_ICU_FUZZERS_FUZZER_UTILS_H_
68