Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 1 | //===- BinaryStreamReader.h - Reads objects from a binary stream *- C++ -*-===// |
| 2 | // |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame] | 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | |
| 9 | #ifndef LLVM_SUPPORT_BINARYSTREAMREADER_H |
| 10 | #define LLVM_SUPPORT_BINARYSTREAMREADER_H |
| 11 | |
| 12 | #include "llvm/ADT/ArrayRef.h" |
| 13 | #include "llvm/ADT/STLExtras.h" |
| 14 | #include "llvm/Support/BinaryStreamArray.h" |
| 15 | #include "llvm/Support/BinaryStreamRef.h" |
| 16 | #include "llvm/Support/ConvertUTF.h" |
| 17 | #include "llvm/Support/Endian.h" |
| 18 | #include "llvm/Support/Error.h" |
| 19 | #include "llvm/Support/type_traits.h" |
| 20 | |
| 21 | #include <string> |
| 22 | #include <type_traits> |
| 23 | |
| 24 | namespace llvm { |
| 25 | |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame] | 26 | /// Provides read only access to a subclass of `BinaryStream`. Provides |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 27 | /// bounds checking and helpers for writing certain common data types such as |
| 28 | /// null-terminated strings, integers in various flavors of endianness, etc. |
| 29 | /// Can be subclassed to provide reading of custom datatypes, although no |
| 30 | /// are overridable. |
| 31 | class BinaryStreamReader { |
| 32 | public: |
| 33 | BinaryStreamReader() = default; |
| 34 | explicit BinaryStreamReader(BinaryStreamRef Ref); |
| 35 | explicit BinaryStreamReader(BinaryStream &Stream); |
| 36 | explicit BinaryStreamReader(ArrayRef<uint8_t> Data, |
| 37 | llvm::support::endianness Endian); |
| 38 | explicit BinaryStreamReader(StringRef Data, llvm::support::endianness Endian); |
| 39 | |
| 40 | BinaryStreamReader(const BinaryStreamReader &Other) |
| 41 | : Stream(Other.Stream), Offset(Other.Offset) {} |
| 42 | |
| 43 | BinaryStreamReader &operator=(const BinaryStreamReader &Other) { |
| 44 | Stream = Other.Stream; |
| 45 | Offset = Other.Offset; |
| 46 | return *this; |
| 47 | } |
| 48 | |
| 49 | virtual ~BinaryStreamReader() {} |
| 50 | |
| 51 | /// Read as much as possible from the underlying string at the current offset |
| 52 | /// without invoking a copy, and set \p Buffer to the resulting data slice. |
| 53 | /// Updates the stream's offset to point after the newly read data. |
| 54 | /// |
| 55 | /// \returns a success error code if the data was successfully read, otherwise |
| 56 | /// returns an appropriate error code. |
| 57 | Error readLongestContiguousChunk(ArrayRef<uint8_t> &Buffer); |
| 58 | |
| 59 | /// Read \p Size bytes from the underlying stream at the current offset and |
| 60 | /// and set \p Buffer to the resulting data slice. Whether a copy occurs |
| 61 | /// depends on the implementation of the underlying stream. Updates the |
| 62 | /// stream's offset to point after the newly read data. |
| 63 | /// |
| 64 | /// \returns a success error code if the data was successfully read, otherwise |
| 65 | /// returns an appropriate error code. |
| 66 | Error readBytes(ArrayRef<uint8_t> &Buffer, uint32_t Size); |
| 67 | |
| 68 | /// Read an integer of the specified endianness into \p Dest and update the |
| 69 | /// stream's offset. The data is always copied from the stream's underlying |
| 70 | /// buffer into \p Dest. Updates the stream's offset to point after the newly |
| 71 | /// read data. |
| 72 | /// |
| 73 | /// \returns a success error code if the data was successfully read, otherwise |
| 74 | /// returns an appropriate error code. |
| 75 | template <typename T> Error readInteger(T &Dest) { |
| 76 | static_assert(std::is_integral<T>::value, |
| 77 | "Cannot call readInteger with non-integral value!"); |
| 78 | |
| 79 | ArrayRef<uint8_t> Bytes; |
| 80 | if (auto EC = readBytes(Bytes, sizeof(T))) |
| 81 | return EC; |
| 82 | |
| 83 | Dest = llvm::support::endian::read<T, llvm::support::unaligned>( |
| 84 | Bytes.data(), Stream.getEndian()); |
| 85 | return Error::success(); |
| 86 | } |
| 87 | |
| 88 | /// Similar to readInteger. |
| 89 | template <typename T> Error readEnum(T &Dest) { |
| 90 | static_assert(std::is_enum<T>::value, |
| 91 | "Cannot call readEnum with non-enum value!"); |
| 92 | typename std::underlying_type<T>::type N; |
| 93 | if (auto EC = readInteger(N)) |
| 94 | return EC; |
| 95 | Dest = static_cast<T>(N); |
| 96 | return Error::success(); |
| 97 | } |
| 98 | |
| 99 | /// Read a null terminated string from \p Dest. Whether a copy occurs depends |
| 100 | /// on the implementation of the underlying stream. Updates the stream's |
| 101 | /// offset to point after the newly read data. |
| 102 | /// |
| 103 | /// \returns a success error code if the data was successfully read, otherwise |
| 104 | /// returns an appropriate error code. |
| 105 | Error readCString(StringRef &Dest); |
| 106 | |
| 107 | /// Similar to readCString, however read a null-terminated UTF16 string |
| 108 | /// instead. |
| 109 | /// |
| 110 | /// \returns a success error code if the data was successfully read, otherwise |
| 111 | /// returns an appropriate error code. |
| 112 | Error readWideString(ArrayRef<UTF16> &Dest); |
| 113 | |
| 114 | /// Read a \p Length byte string into \p Dest. Whether a copy occurs depends |
| 115 | /// on the implementation of the underlying stream. Updates the stream's |
| 116 | /// offset to point after the newly read data. |
| 117 | /// |
| 118 | /// \returns a success error code if the data was successfully read, otherwise |
| 119 | /// returns an appropriate error code. |
| 120 | Error readFixedString(StringRef &Dest, uint32_t Length); |
| 121 | |
| 122 | /// Read the entire remainder of the underlying stream into \p Ref. This is |
| 123 | /// equivalent to calling getUnderlyingStream().slice(Offset). Updates the |
| 124 | /// stream's offset to point to the end of the stream. Never causes a copy. |
| 125 | /// |
| 126 | /// \returns a success error code if the data was successfully read, otherwise |
| 127 | /// returns an appropriate error code. |
| 128 | Error readStreamRef(BinaryStreamRef &Ref); |
| 129 | |
| 130 | /// Read \p Length bytes from the underlying stream into \p Ref. This is |
| 131 | /// equivalent to calling getUnderlyingStream().slice(Offset, Length). |
| 132 | /// Updates the stream's offset to point after the newly read object. Never |
| 133 | /// causes a copy. |
| 134 | /// |
| 135 | /// \returns a success error code if the data was successfully read, otherwise |
| 136 | /// returns an appropriate error code. |
| 137 | Error readStreamRef(BinaryStreamRef &Ref, uint32_t Length); |
| 138 | |
| 139 | /// Read \p Length bytes from the underlying stream into \p Stream. This is |
| 140 | /// equivalent to calling getUnderlyingStream().slice(Offset, Length). |
| 141 | /// Updates the stream's offset to point after the newly read object. Never |
| 142 | /// causes a copy. |
| 143 | /// |
| 144 | /// \returns a success error code if the data was successfully read, otherwise |
| 145 | /// returns an appropriate error code. |
| 146 | Error readSubstream(BinarySubstreamRef &Stream, uint32_t Size); |
| 147 | |
| 148 | /// Get a pointer to an object of type T from the underlying stream, as if by |
| 149 | /// memcpy, and store the result into \p Dest. It is up to the caller to |
| 150 | /// ensure that objects of type T can be safely treated in this manner. |
| 151 | /// Updates the stream's offset to point after the newly read object. Whether |
| 152 | /// a copy occurs depends upon the implementation of the underlying |
| 153 | /// stream. |
| 154 | /// |
| 155 | /// \returns a success error code if the data was successfully read, otherwise |
| 156 | /// returns an appropriate error code. |
| 157 | template <typename T> Error readObject(const T *&Dest) { |
| 158 | ArrayRef<uint8_t> Buffer; |
| 159 | if (auto EC = readBytes(Buffer, sizeof(T))) |
| 160 | return EC; |
| 161 | Dest = reinterpret_cast<const T *>(Buffer.data()); |
| 162 | return Error::success(); |
| 163 | } |
| 164 | |
| 165 | /// Get a reference to a \p NumElements element array of objects of type T |
| 166 | /// from the underlying stream as if by memcpy, and store the resulting array |
| 167 | /// slice into \p array. It is up to the caller to ensure that objects of |
| 168 | /// type T can be safely treated in this manner. Updates the stream's offset |
| 169 | /// to point after the newly read object. Whether a copy occurs depends upon |
| 170 | /// the implementation of the underlying stream. |
| 171 | /// |
| 172 | /// \returns a success error code if the data was successfully read, otherwise |
| 173 | /// returns an appropriate error code. |
| 174 | template <typename T> |
| 175 | Error readArray(ArrayRef<T> &Array, uint32_t NumElements) { |
| 176 | ArrayRef<uint8_t> Bytes; |
| 177 | if (NumElements == 0) { |
| 178 | Array = ArrayRef<T>(); |
| 179 | return Error::success(); |
| 180 | } |
| 181 | |
| 182 | if (NumElements > UINT32_MAX / sizeof(T)) |
| 183 | return make_error<BinaryStreamError>( |
| 184 | stream_error_code::invalid_array_size); |
| 185 | |
| 186 | if (auto EC = readBytes(Bytes, NumElements * sizeof(T))) |
| 187 | return EC; |
| 188 | |
| 189 | assert(alignmentAdjustment(Bytes.data(), alignof(T)) == 0 && |
| 190 | "Reading at invalid alignment!"); |
| 191 | |
| 192 | Array = ArrayRef<T>(reinterpret_cast<const T *>(Bytes.data()), NumElements); |
| 193 | return Error::success(); |
| 194 | } |
| 195 | |
| 196 | /// Read a VarStreamArray of size \p Size bytes and store the result into |
| 197 | /// \p Array. Updates the stream's offset to point after the newly read |
| 198 | /// array. Never causes a copy (although iterating the elements of the |
| 199 | /// VarStreamArray may, depending upon the implementation of the underlying |
| 200 | /// stream). |
| 201 | /// |
| 202 | /// \returns a success error code if the data was successfully read, otherwise |
| 203 | /// returns an appropriate error code. |
| 204 | template <typename T, typename U> |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame] | 205 | Error readArray(VarStreamArray<T, U> &Array, uint32_t Size, |
| 206 | uint32_t Skew = 0) { |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 207 | BinaryStreamRef S; |
| 208 | if (auto EC = readStreamRef(S, Size)) |
| 209 | return EC; |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame] | 210 | Array.setUnderlyingStream(S, Skew); |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 211 | return Error::success(); |
| 212 | } |
| 213 | |
| 214 | /// Read a FixedStreamArray of \p NumItems elements and store the result into |
| 215 | /// \p Array. Updates the stream's offset to point after the newly read |
| 216 | /// array. Never causes a copy (although iterating the elements of the |
| 217 | /// FixedStreamArray may, depending upon the implementation of the underlying |
| 218 | /// stream). |
| 219 | /// |
| 220 | /// \returns a success error code if the data was successfully read, otherwise |
| 221 | /// returns an appropriate error code. |
| 222 | template <typename T> |
| 223 | Error readArray(FixedStreamArray<T> &Array, uint32_t NumItems) { |
| 224 | if (NumItems == 0) { |
| 225 | Array = FixedStreamArray<T>(); |
| 226 | return Error::success(); |
| 227 | } |
| 228 | |
| 229 | if (NumItems > UINT32_MAX / sizeof(T)) |
| 230 | return make_error<BinaryStreamError>( |
| 231 | stream_error_code::invalid_array_size); |
| 232 | |
| 233 | BinaryStreamRef View; |
| 234 | if (auto EC = readStreamRef(View, NumItems * sizeof(T))) |
| 235 | return EC; |
| 236 | |
| 237 | Array = FixedStreamArray<T>(View); |
| 238 | return Error::success(); |
| 239 | } |
| 240 | |
| 241 | bool empty() const { return bytesRemaining() == 0; } |
| 242 | void setOffset(uint32_t Off) { Offset = Off; } |
| 243 | uint32_t getOffset() const { return Offset; } |
| 244 | uint32_t getLength() const { return Stream.getLength(); } |
| 245 | uint32_t bytesRemaining() const { return getLength() - getOffset(); } |
| 246 | |
| 247 | /// Advance the stream's offset by \p Amount bytes. |
| 248 | /// |
| 249 | /// \returns a success error code if at least \p Amount bytes remain in the |
| 250 | /// stream, otherwise returns an appropriate error code. |
| 251 | Error skip(uint32_t Amount); |
| 252 | |
| 253 | /// Examine the next byte of the underlying stream without advancing the |
| 254 | /// stream's offset. If the stream is empty the behavior is undefined. |
| 255 | /// |
| 256 | /// \returns the next byte in the stream. |
| 257 | uint8_t peek() const; |
| 258 | |
| 259 | Error padToAlignment(uint32_t Align); |
| 260 | |
| 261 | std::pair<BinaryStreamReader, BinaryStreamReader> |
| 262 | split(uint32_t Offset) const; |
| 263 | |
| 264 | private: |
| 265 | BinaryStreamRef Stream; |
| 266 | uint32_t Offset = 0; |
| 267 | }; |
| 268 | } // namespace llvm |
| 269 | |
| 270 | #endif // LLVM_SUPPORT_BINARYSTREAMREADER_H |