1 // -*- mode: c++ -*- 2 3 // Copyright (c) 2010, Google Inc. 4 // All rights reserved. 5 // 6 // Redistribution and use in source and binary forms, with or without 7 // modification, are permitted provided that the following conditions are 8 // met: 9 // 10 // * Redistributions of source code must retain the above copyright 11 // notice, this list of conditions and the following disclaimer. 12 // * Redistributions in binary form must reproduce the above 13 // copyright notice, this list of conditions and the following disclaimer 14 // in the documentation and/or other materials provided with the 15 // distribution. 16 // * Neither the name of Google Inc. nor the names of its 17 // contributors may be used to endorse or promote products derived from 18 // this software without specific prior written permission. 19 // 20 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 21 // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 22 // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 23 // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 24 // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 25 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 26 // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 27 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 28 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 29 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 30 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 31 32 // Original author: Jim Blandy <jimb@mozilla.com> <jimb@red-bean.com> 33 34 // byte_cursor.h: Classes for parsing values from a buffer of bytes. 35 // The ByteCursor class provides a convenient interface for reading 36 // fixed-size integers of arbitrary endianness, being thorough about 37 // checking for buffer overruns. 38 39 #ifndef COMMON_BYTE_CURSOR_H_ 40 #define COMMON_BYTE_CURSOR_H_ 41 42 #include <assert.h> 43 #include <stdint.h> 44 #include <stdlib.h> 45 #include <string.h> 46 #include <string> 47 48 namespace google_breakpad { 49 50 // A buffer holding a series of bytes. 51 struct ByteBuffer { ByteBufferByteBuffer52 ByteBuffer() : start(0), end(0) { } ByteBufferByteBuffer53 ByteBuffer(const uint8_t *set_start, size_t set_size) 54 : start(set_start), end(set_start + set_size) { } ~ByteBufferByteBuffer55 ~ByteBuffer() { }; 56 57 // Equality operators. Useful in unit tests, and when we're using 58 // ByteBuffers to refer to regions of a larger buffer. 59 bool operator==(const ByteBuffer &that) const { 60 return start == that.start && end == that.end; 61 } 62 bool operator!=(const ByteBuffer &that) const { 63 return start != that.start || end != that.end; 64 } 65 66 // Not C++ style guide compliant, but this definitely belongs here. SizeByteBuffer67 size_t Size() const { 68 assert(start <= end); 69 return end - start; 70 } 71 72 const uint8_t *start, *end; 73 }; 74 75 // A cursor pointing into a ByteBuffer that can parse numbers of various 76 // widths and representations, strings, and data blocks, advancing through 77 // the buffer as it goes. All ByteCursor operations check that accesses 78 // haven't gone beyond the end of the enclosing ByteBuffer. 79 class ByteCursor { 80 public: 81 // Create a cursor reading bytes from the start of BUFFER. By default, the 82 // cursor reads multi-byte values in little-endian form. 83 ByteCursor(const ByteBuffer *buffer, bool big_endian = false) buffer_(buffer)84 : buffer_(buffer), here_(buffer->start), 85 big_endian_(big_endian), complete_(true) { } 86 87 // Accessor and setter for this cursor's endianness flag. big_endian()88 bool big_endian() const { return big_endian_; } set_big_endian(bool big_endian)89 void set_big_endian(bool big_endian) { big_endian_ = big_endian; } 90 91 // Accessor and setter for this cursor's current position. The setter 92 // returns a reference to this cursor. here()93 const uint8_t *here() const { return here_; } set_here(const uint8_t * here)94 ByteCursor &set_here(const uint8_t *here) { 95 assert(buffer_->start <= here && here <= buffer_->end); 96 here_ = here; 97 return *this; 98 } 99 100 // Return the number of bytes available to read at the cursor. Available()101 size_t Available() const { return size_t(buffer_->end - here_); } 102 103 // Return true if this cursor is at the end of its buffer. AtEnd()104 bool AtEnd() const { return Available() == 0; } 105 106 // When used as a boolean value this cursor converts to true if all 107 // prior reads have been completed, or false if we ran off the end 108 // of the buffer. 109 operator bool() const { return complete_; } 110 111 // Read a SIZE-byte integer at this cursor, signed if IS_SIGNED is true, 112 // unsigned otherwise, using the cursor's established endianness, and set 113 // *RESULT to the number. If we read off the end of our buffer, clear 114 // this cursor's complete_ flag, and store a dummy value in *RESULT. 115 // Return a reference to this cursor. 116 template<typename T> Read(size_t size,bool is_signed,T * result)117 ByteCursor &Read(size_t size, bool is_signed, T *result) { 118 if (CheckAvailable(size)) { 119 T v = 0; 120 if (big_endian_) { 121 for (size_t i = 0; i < size; i++) 122 v = (v << 8) + here_[i]; 123 } else { 124 // This loop condition looks weird, but size_t is unsigned, so 125 // decrementing i after it is zero yields the largest size_t value. 126 for (size_t i = size - 1; i < size; i--) 127 v = (v << 8) + here_[i]; 128 } 129 if (is_signed && size < sizeof(T)) { 130 size_t sign_bit = (T)1 << (size * 8 - 1); 131 v = (v ^ sign_bit) - sign_bit; 132 } 133 here_ += size; 134 *result = v; 135 } else { 136 *result = (T) 0xdeadbeef; 137 } 138 return *this; 139 } 140 141 // Read an integer, using the cursor's established endianness and 142 // *RESULT's size and signedness, and set *RESULT to the number. If we 143 // read off the end of our buffer, clear this cursor's complete_ flag. 144 // Return a reference to this cursor. 145 template<typename T> 146 ByteCursor &operator>>(T &result) { 147 bool T_is_signed = (T)-1 < 0; 148 return Read(sizeof(T), T_is_signed, &result); 149 } 150 151 // Copy the SIZE bytes at the cursor to BUFFER, and advance this 152 // cursor to the end of them. If we read off the end of our buffer, 153 // clear this cursor's complete_ flag, and set *POINTER to NULL. 154 // Return a reference to this cursor. Read(uint8_t * buffer,size_t size)155 ByteCursor &Read(uint8_t *buffer, size_t size) { 156 if (CheckAvailable(size)) { 157 memcpy(buffer, here_, size); 158 here_ += size; 159 } 160 return *this; 161 } 162 163 // Set STR to a copy of the '\0'-terminated string at the cursor. If the 164 // byte buffer does not contain a terminating zero, clear this cursor's 165 // complete_ flag, and set STR to the empty string. Return a reference to 166 // this cursor. CString(std::string * str)167 ByteCursor &CString(std::string *str) { 168 const uint8_t *end 169 = static_cast<const uint8_t *>(memchr(here_, '\0', Available())); 170 if (end) { 171 str->assign(reinterpret_cast<const char *>(here_), end - here_); 172 here_ = end + 1; 173 } else { 174 str->clear(); 175 here_ = buffer_->end; 176 complete_ = false; 177 } 178 return *this; 179 } 180 181 // Like CString(STR), but extract the string from a fixed-width buffer 182 // LIMIT bytes long, which may or may not contain a terminating '\0' 183 // byte. Specifically: 184 // 185 // - If there are not LIMIT bytes available at the cursor, clear the 186 // cursor's complete_ flag and set STR to the empty string. 187 // 188 // - Otherwise, if the LIMIT bytes at the cursor contain any '\0' 189 // characters, set *STR to a copy of the bytes before the first '\0', 190 // and advance the cursor by LIMIT bytes. 191 // 192 // - Otherwise, set *STR to a copy of those LIMIT bytes, and advance the 193 // cursor by LIMIT bytes. CString(std::string * str,size_t limit)194 ByteCursor &CString(std::string *str, size_t limit) { 195 if (CheckAvailable(limit)) { 196 const uint8_t *end 197 = static_cast<const uint8_t *>(memchr(here_, '\0', limit)); 198 if (end) 199 str->assign(reinterpret_cast<const char *>(here_), end - here_); 200 else 201 str->assign(reinterpret_cast<const char *>(here_), limit); 202 here_ += limit; 203 } else { 204 str->clear(); 205 } 206 return *this; 207 } 208 209 // Set *POINTER to point to the SIZE bytes at the cursor, and advance 210 // this cursor to the end of them. If SIZE is omitted, don't move the 211 // cursor. If we read off the end of our buffer, clear this cursor's 212 // complete_ flag, and set *POINTER to NULL. Return a reference to this 213 // cursor. 214 ByteCursor &PointTo(const uint8_t **pointer, size_t size = 0) { 215 if (CheckAvailable(size)) { 216 *pointer = here_; 217 here_ += size; 218 } else { 219 *pointer = NULL; 220 } 221 return *this; 222 } 223 224 // Skip SIZE bytes at the cursor. If doing so would advance us off 225 // the end of our buffer, clear this cursor's complete_ flag, and 226 // set *POINTER to NULL. Return a reference to this cursor. Skip(size_t size)227 ByteCursor &Skip(size_t size) { 228 if (CheckAvailable(size)) 229 here_ += size; 230 return *this; 231 } 232 233 private: 234 // If there are at least SIZE bytes available to read from the buffer, 235 // return true. Otherwise, set here_ to the end of the buffer, set 236 // complete_ to false, and return false. CheckAvailable(size_t size)237 bool CheckAvailable(size_t size) { 238 if (Available() >= size) { 239 return true; 240 } else { 241 here_ = buffer_->end; 242 complete_ = false; 243 return false; 244 } 245 } 246 247 // The buffer we're reading bytes from. 248 const ByteBuffer *buffer_; 249 250 // The next byte within buffer_ that we'll read. 251 const uint8_t *here_; 252 253 // True if we should read numbers in big-endian form; false if we 254 // should read in little-endian form. 255 bool big_endian_; 256 257 // True if we've been able to read all we've been asked to. 258 bool complete_; 259 }; 260 261 } // namespace google_breakpad 262 263 #endif // COMMON_BYTE_CURSOR_H_ 264