2 * Copyright (c) 2013 David Chisnall
5 * This software was developed by SRI International and the University of
6 * Cambridge Computer Laboratory under DARPA/AFRL contract (FA8750-10-C-0237)
7 * ("CTSRD"), as part of the DARPA CRASH research programme.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 #ifndef _INPUT_BUFFER_HH_
34 #define _INPUT_BUFFER_HH_
43 typedef std::unique_ptr<expression> expression_ptr;
47 * Class encapsulating the input file. Can be used as a const char*, but has
48 * range checking. Attempting to access anything out of range will return a 0
49 * byte. The input buffer can be cheaply copied, without copying the
50 * underlying memory, however it is the user's responsibility to ensure that
51 * such copies do not persist beyond the lifetime of the underlying memory.
53 * This also contains methods for reporting errors and for consuming the token
60 * The buffer. This class doesn't own the buffer, but the
61 * mmap_input_buffer subclass does.
65 * The size of the buffer.
70 * Parse an expression. If `stopAtParen` is set, then only parse a number
71 * or a parenthetical expression, otherwise assume that either is the
72 * left-hand side of a binary expression and try to parse the right-hand
75 expression_ptr parse_expression(bool stopAtParen=false);
77 * Parse a binary expression, having already parsed the right-hand side.
79 expression_ptr parse_binary_expression(expression_ptr lhs);
81 * The current place in the buffer where we are reading. This class
82 * keeps a separate size, pointer, and cursor so that we can move
83 * forwards and backwards and still have checks that we haven't fallen
88 * Private constructor. This is used to create input buffers that
89 * refer to the same memory, but have different cursors.
91 input_buffer(const char* b, int s, int c) : buffer(b), size(s),
94 * Reads forward past any spaces. The DTS format is not whitespace
95 * sensitive and so we want to scan past whitespace when reading it.
100 * Return whether all input has been consumed.
102 bool finished() { return cursor >= size; }
104 * Virtual destructor. Does nothing, but exists so that subclasses
105 * that own the memory can run cleanup code for deallocating it.
107 virtual ~input_buffer() {};
109 * Constructs an empty buffer.
111 input_buffer() : buffer(0), size(0), cursor(0) {}
113 * Constructs a new buffer with a specified memory region and size.
115 input_buffer(const char* b, int s) : buffer(b), size(s), cursor(0){}
117 * Returns a new input buffer referring into this input, clamped to the
118 * specified size. If the requested buffer would fall outside the
119 * range of this one, then it returns an empty buffer.
121 * The returned buffer shares the same underlying storage as the
122 * original. This is intended to be used for splitting up the various
123 * sections of a device tree blob. Requesting a size of 0 will give a
124 * buffer that extends to the end of the available memory.
126 input_buffer buffer_from_offset(int offset, int s=0);
128 * Returns true if this buffer has no unconsumed space in it.
132 return cursor >= size;
135 * Dereferencing operator, allows the buffer to be treated as a char*
136 * and dereferenced to give a character. This returns a null byte if
137 * the cursor is out of range.
139 inline char operator*()
141 if (cursor >= size) { return '\0'; }
142 if (cursor < 0) { return '\0'; }
143 return buffer[cursor];
146 * Array subscripting operator, returns a character at the specified
147 * index offset from the current cursor. The offset may be negative,
148 * to reread characters that have already been read. If the current
149 * cursor plus offset is outside of the range, this returns a nul
152 inline char operator[](int offset)
154 if (cursor + offset >= size) { return '\0'; }
155 if (cursor + offset < 0) { return '\0'; }
156 return buffer[cursor + offset];
159 * Increments the cursor, iterating forward in the buffer.
161 inline input_buffer &operator++()
167 * Cast to char* operator. Returns a pointer into the buffer that can
168 * be used for constructing strings.
170 inline operator const char*()
172 if (cursor >= size) { return 0; }
173 if (cursor < 0) { return 0; }
174 return &buffer[cursor];
177 * Consumes a character. Moves the cursor one character forward if the
178 * next character matches the argument, returning true. If the current
179 * character does not match the argument, returns false.
181 inline bool consume(char c)
191 * Consumes a string. If the (null-terminated) string passed as the
192 * argument appears in the input, advances the cursor to the end and
193 * returns true. Returns false if the string does not appear at the
194 * current point in the input.
196 bool consume(const char *str);
198 * Reads an integer in base 8, 10, or 16. Returns true and advances
199 * the cursor to the end of the integer if the cursor points to an
200 * integer, returns false and does not move the cursor otherwise.
202 * The parsed value is returned via the argument.
204 bool consume_integer(unsigned long long &outInt);
206 * Reads an arithmetic expression (containing any of the normal C
207 * operators), evaluates it, and returns the result.
209 bool consume_integer_expression(unsigned long long &outInt);
211 * Template function that consumes a binary value in big-endian format
212 * from the input stream. Returns true and advances the cursor if
213 * there is a value of the correct size. This function assumes that
214 * all values must be natively aligned, and so advances the cursor to
215 * the correct alignment before reading.
218 bool consume_binary(T &out)
221 int type_size = sizeof(T);
222 if (cursor % type_size != 0)
224 align = type_size - (cursor % type_size);
226 if (size < cursor + align + type_size)
231 assert(cursor % type_size == 0);
233 for (int i=0 ; i<type_size ; ++i)
236 out |= (((T)buffer[cursor++]) & 0xff);
241 * Consumes two hex digits and return the resulting byte via the first
242 * argument. If the next two characters are hex digits, returns true
243 * and advances the cursor. If not, then returns false and leaves the
246 bool consume_hex_byte(uint8_t &outByte);
248 * Advances the cursor to the start of the next token, skipping
249 * comments and whitespace. If the cursor already points to the start
250 * of a token, then this function does nothing.
252 input_buffer &next_token();
254 * Prints a message indicating the location of a parse error.
256 void parse_error(const char *msg);
259 * Dumps the current cursor value and the unconsumed values in the
260 * input buffer to the standard error. This method is intended solely
267 * Explicit specialisation for reading a single byte.
270 inline bool input_buffer::consume_binary(uint8_t &out)
272 if (size < cursor + 1)
276 out = buffer[cursor++];
281 * Subclass of input_buffer that mmap()s a file and owns the resulting memory.
282 * When this object is destroyed, the memory is unmapped.
284 struct mmap_input_buffer : public input_buffer
287 * Constructs a new buffer from the file passed in as a file
290 mmap_input_buffer(int fd);
292 * Unmaps the buffer, if one exists.
294 virtual ~mmap_input_buffer();
297 * Input buffer read from standard input. This is used for reading device tree
298 * blobs and source from standard input. It reads the entire input into
299 * malloc'd memory, so will be very slow for large inputs. DTS and DTB files
300 * are very rarely more than 10KB though, so this is probably not a problem.
302 struct stream_input_buffer : public input_buffer
305 * The buffer that will store the data read from the standard input.
309 * Constructs a new buffer from the standard input.
311 stream_input_buffer();
316 #endif // !_INPUT_BUFFER_HH_