| 1 | #ifndef SIMDJSON_INLINE_PADDED_STRING_H |
| 2 | #define SIMDJSON_INLINE_PADDED_STRING_H |
| 3 | |
| 4 | #include "simdjson/portability.h" |
| 5 | #include "simdjson/common_defs.h" // for SIMDJSON_PADDING |
| 6 | |
| 7 | #include <climits> |
| 8 | #include <cstring> |
| 9 | #include <memory> |
| 10 | #include <string> |
| 11 | |
| 12 | namespace simdjson { |
| 13 | namespace internal { |
| 14 | |
| 15 | // The allocate_padded_buffer function is a low-level function to allocate memory |
| 16 | // with padding so we can read past the "length" bytes safely. It is used by |
| 17 | // the padded_string class automatically. It returns nullptr in case |
| 18 | // of error: the caller should check for a null pointer. |
| 19 | // The length parameter is the maximum size in bytes of the string. |
| 20 | // The caller is responsible to free the memory (e.g., delete[] (...)). |
| 21 | inline char *allocate_padded_buffer(size_t length) noexcept { |
| 22 | const size_t totalpaddedlength = length + SIMDJSON_PADDING; |
| 23 | if(totalpaddedlength<length) { |
| 24 | // overflow |
| 25 | return nullptr; |
| 26 | } |
| 27 | #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION |
| 28 | // avoid getting out of memory |
| 29 | if (totalpaddedlength>(1UL<<20)) { |
| 30 | return nullptr; |
| 31 | } |
| 32 | #endif |
| 33 | |
| 34 | char *padded_buffer = new (std::nothrow) char[totalpaddedlength]; |
| 35 | if (padded_buffer == nullptr) { |
| 36 | return nullptr; |
| 37 | } |
| 38 | // We write zeroes in the padded region to avoid having uninitized |
| 39 | // garbage. If nothing else, garbage getting read might trigger a |
| 40 | // warning in a memory checking. |
| 41 | std::memset(s: padded_buffer + length, c: 0, n: totalpaddedlength - length); |
| 42 | return padded_buffer; |
| 43 | } // allocate_padded_buffer() |
| 44 | |
| 45 | } // namespace internal |
| 46 | |
| 47 | |
| 48 | inline padded_string::padded_string() noexcept = default; |
| 49 | inline padded_string::padded_string(size_t length) noexcept |
| 50 | : viable_size(length), data_ptr(internal::allocate_padded_buffer(length)) { |
| 51 | } |
| 52 | inline padded_string::padded_string(const char *data, size_t length) noexcept |
| 53 | : viable_size(length), data_ptr(internal::allocate_padded_buffer(length)) { |
| 54 | if ((data != nullptr) && (data_ptr != nullptr)) { |
| 55 | std::memcpy(dest: data_ptr, src: data, n: length); |
| 56 | } |
| 57 | } |
| 58 | // note: do not pass std::string arguments by value |
| 59 | inline padded_string::padded_string(const std::string & str_ ) noexcept |
| 60 | : viable_size(str_.size()), data_ptr(internal::allocate_padded_buffer(length: str_.size())) { |
| 61 | if (data_ptr != nullptr) { |
| 62 | std::memcpy(dest: data_ptr, src: str_.data(), n: str_.size()); |
| 63 | } |
| 64 | } |
| 65 | // note: do pass std::string_view arguments by value |
| 66 | inline padded_string::padded_string(std::string_view sv_) noexcept |
| 67 | : viable_size(sv_.size()), data_ptr(internal::allocate_padded_buffer(length: sv_.size())) { |
| 68 | if(simdjson_unlikely(!data_ptr)) { |
| 69 | //allocation failed or zero size |
| 70 | viable_size=0; |
| 71 | return; |
| 72 | } |
| 73 | if (sv_.size()) { |
| 74 | std::memcpy(dest: data_ptr, src: sv_.data(), n: sv_.size()); |
| 75 | } |
| 76 | } |
| 77 | inline padded_string::padded_string(padded_string &&o) noexcept |
| 78 | : viable_size(o.viable_size), data_ptr(o.data_ptr) { |
| 79 | o.data_ptr = nullptr; // we take ownership |
| 80 | } |
| 81 | |
| 82 | inline padded_string &padded_string::operator=(padded_string &&o) noexcept { |
| 83 | delete[] data_ptr; |
| 84 | data_ptr = o.data_ptr; |
| 85 | viable_size = o.viable_size; |
| 86 | o.data_ptr = nullptr; // we take ownership |
| 87 | o.viable_size = 0; |
| 88 | return *this; |
| 89 | } |
| 90 | |
| 91 | inline void padded_string::swap(padded_string &o) noexcept { |
| 92 | size_t tmp_viable_size = viable_size; |
| 93 | char *tmp_data_ptr = data_ptr; |
| 94 | viable_size = o.viable_size; |
| 95 | data_ptr = o.data_ptr; |
| 96 | o.data_ptr = tmp_data_ptr; |
| 97 | o.viable_size = tmp_viable_size; |
| 98 | } |
| 99 | |
| 100 | inline padded_string::~padded_string() noexcept { |
| 101 | delete[] data_ptr; |
| 102 | } |
| 103 | |
| 104 | inline size_t padded_string::size() const noexcept { return viable_size; } |
| 105 | |
| 106 | inline size_t padded_string::length() const noexcept { return viable_size; } |
| 107 | |
| 108 | inline const char *padded_string::data() const noexcept { return data_ptr; } |
| 109 | |
| 110 | inline char *padded_string::data() noexcept { return data_ptr; } |
| 111 | |
| 112 | inline padded_string::operator std::string_view() const { return std::string_view(data(), length()); } |
| 113 | |
| 114 | inline padded_string::operator padded_string_view() const noexcept { |
| 115 | return padded_string_view(data(), length(), length() + SIMDJSON_PADDING); |
| 116 | } |
| 117 | |
| 118 | inline simdjson_result<padded_string> padded_string::load(std::string_view filename) noexcept { |
| 119 | // Open the file |
| 120 | SIMDJSON_PUSH_DISABLE_WARNINGS |
| 121 | SIMDJSON_DISABLE_DEPRECATED_WARNING // Disable CRT_SECURE warning on MSVC: manually verified this is safe |
| 122 | std::FILE *fp = std::fopen(filename: filename.data(), modes: "rb" ); |
| 123 | SIMDJSON_POP_DISABLE_WARNINGS |
| 124 | |
| 125 | if (fp == nullptr) { |
| 126 | return IO_ERROR; |
| 127 | } |
| 128 | |
| 129 | // Get the file size |
| 130 | int ret; |
| 131 | #if SIMDJSON_VISUAL_STUDIO && !SIMDJSON_IS_32BITS |
| 132 | ret = _fseeki64(fp, 0, SEEK_END); |
| 133 | #else |
| 134 | ret = std::fseek(stream: fp, off: 0, SEEK_END); |
| 135 | #endif // _WIN64 |
| 136 | if(ret < 0) { |
| 137 | std::fclose(stream: fp); |
| 138 | return IO_ERROR; |
| 139 | } |
| 140 | #if SIMDJSON_VISUAL_STUDIO && !SIMDJSON_IS_32BITS |
| 141 | __int64 llen = _ftelli64(fp); |
| 142 | if(llen == -1L) { |
| 143 | std::fclose(fp); |
| 144 | return IO_ERROR; |
| 145 | } |
| 146 | #else |
| 147 | long llen = std::ftell(stream: fp); |
| 148 | if((llen < 0) || (llen == LONG_MAX)) { |
| 149 | std::fclose(stream: fp); |
| 150 | return IO_ERROR; |
| 151 | } |
| 152 | #endif |
| 153 | |
| 154 | // Allocate the padded_string |
| 155 | size_t len = static_cast<size_t>(llen); |
| 156 | padded_string s(len); |
| 157 | if (s.data() == nullptr) { |
| 158 | std::fclose(stream: fp); |
| 159 | return MEMALLOC; |
| 160 | } |
| 161 | |
| 162 | // Read the padded_string |
| 163 | std::rewind(stream: fp); |
| 164 | size_t bytes_read = std::fread(ptr: s.data(), size: 1, n: len, stream: fp); |
| 165 | if (std::fclose(stream: fp) != 0 || bytes_read != len) { |
| 166 | return IO_ERROR; |
| 167 | } |
| 168 | |
| 169 | return s; |
| 170 | } |
| 171 | |
| 172 | } // namespace simdjson |
| 173 | |
| 174 | #endif // SIMDJSON_INLINE_PADDED_STRING_H |
| 175 | |