/* =============================================================================== FILE: encoder.hpp CONTENTS: Encoder stuff PROGRAMMERS: martin.isenburg@rapidlasso.com - http://rapidlasso.com uday.karan@gmail.com - Hobu, Inc. COPYRIGHT: (c) 2007-2014, martin isenburg, rapidlasso - tools to catch reality (c) 2014, Uday Verma, Hobu, Inc. This is free software; you can redistribute and/or modify it under the terms of the Apache Public License 2.0 published by the Apache Software Foundation. See the COPYING file for more information. This software is distributed WITHOUT ANY WARRANTY and without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. CHANGE HISTORY: see header file =============================================================================== */ // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - // - // **************************** - // ARITHMETIC CODING EXAMPLES - // **************************** - // - // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - // - // Fast arithmetic coding implementation - // -> 32-bit variables, 32-bit product, periodic updates, table decoding - // - // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - // - // Version 1.00 - April 25, 2004 - // - // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - // - // WARNING - // ========= - // - // The only purpose of this program is to demonstrate the basic principles - // of arithmetic coding. The original version of this code can be found in - // Digital Signal Compression: Principles and Practice - // (Cambridge University Press, 2011, ISBN: 9780511984655) - // - // Copyright (c) 2019 by Amir Said (said@ieee.org) & - // William A. Pearlman (pearlw@ecse.rpi.edu) - // - // Redistribution and use in source and binary forms, with or without - // modification, are permitted provided that the following conditions are - // met: - // - // 1. Redistributions of source code must retain the above copyright notice, - // this list of conditions and the following disclaimer. - // - // 2. Redistributions in binary form must reproduce the above copyright - // notice, this list of conditions and the following disclaimer in the - // documentation and/or other materials provided with the distribution. - // - // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS - // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED - // TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A - // PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER - // OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, - // EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, - // PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR - // PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF - // LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING - // NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS - // SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. - // - // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - // - // A description of the arithmetic coding method used here is available in - // - // Lossless Compression Handbook, ed. K. Sayood - // Chapter 5: Arithmetic Coding (A. Said), pp. 101-152, Academic Press, 2003 - // - // A. Said, Introduction to Arithetic Coding Theory and Practice - // HP Labs report HPL-2004-76 - http://www.hpl.hp.com/techreports/ - // - // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - #ifndef __encoder_hpp__ #define __encoder_hpp__ #include #include "coderbase.hpp" namespace lazperf { namespace encoders { template struct arithmetic { public: arithmetic(TOutStream& out, bool valid = true) : outstream(out) { init(valid); } arithmetic(bool valid) : pOut(new TOutStream), outstream(*pOut) { init(valid); } arithmetic(const arithmetic& src) : pOut(new TOutStream(*src.pOut)), outstream(*pOut) { init(src); } ~arithmetic() { delete [] outbuffer; } void makeValid() { valid = true; } void done() { uint32_t init_base = base; // done encoding: set final data bytes bool another_byte = true; if (length > 2 * AC__MinLength) { base += AC__MinLength; // base offset length = AC__MinLength >> 1; // set new length for 1 more byte } else { base += AC__MinLength >> 1; // base offset length = AC__MinLength >> 9; // set new length for 2 more bytes another_byte = false; } if (init_base > base) // overflow = carry propagate_carry(); renorm_enc_interval(); // renormalization = output last bytes if (endbyte != endbuffer) { assert(outbyte < outbuffer + AC_BUFFER_SIZE); outstream.putBytes(outbuffer + AC_BUFFER_SIZE, AC_BUFFER_SIZE); } //ABELL - We control the buffer size. This calculation should never be // negative and we shouldn't need 64 bits. int64_t buffer_size = outbyte - outbuffer; if (buffer_size) outstream.putBytes(outbuffer, (uint32_t)buffer_size); // write two or three zero bytes to be in sync with the decoder's byte reads outstream.putByte(0); outstream.putByte(0); if (another_byte) outstream.putByte(0); } /* Encode a bit with modelling */ template void encodeBit(EntropyModel& m, uint32_t sym) { assert(sym <= 1); uint32_t x = m.bit_0_prob * (length >> BM__LengthShift); // product l x p0 // update interval if (sym == 0) { length = x; ++m.bit_0_count; } else { uint32_t init_base = base; base += x; length -= x; if (init_base > base) // overflow = carry propagate_carry(); } if (length < AC__MinLength) renorm_enc_interval(); // renormalization if (--m.bits_until_update == 0) m.update(); // periodic model update } /* Encode a symbol with modelling */ template void encodeSymbol(EntropyModel& m, uint32_t sym) { assert(sym <= m.last_symbol); uint32_t x, init_base = base; // compute products if (sym == m.last_symbol) { x = m.distribution[sym] * (length >> DM__LengthShift); base += x; // update interval length -= x; // no product needed } else { x = m.distribution[sym] * (length >>= DM__LengthShift); base += x; // update interval length = m.distribution[sym+1] * length - x; } if (init_base > base) propagate_carry(); // overflow = carry if (length < AC__MinLength) renorm_enc_interval(); // renormalization ++m.symbol_count[sym]; if (--m.symbols_until_update == 0) m.update(); // periodic model update } /* Encode a bit without modelling */ void writeBit(uint32_t sym) { assert(sym < 2); uint32_t init_base = base; base += sym * (length >>= 1); // new interval base and length if (init_base > base) propagate_carry(); // overflow = carry if (length < AC__MinLength) renorm_enc_interval(); // renormalization } void writeBits(uint32_t bits, uint32_t sym) { assert(bits && (bits <= 32) && (sym < (1u< 19) { writeShort(sym); sym = sym >> 16; bits = bits - 16; } uint32_t init_base = base; base += sym * (length >>= bits); // new interval base and length if (init_base > base) propagate_carry(); // overflow = carry if (length < AC__MinLength) renorm_enc_interval(); // renormalization } void writeByte(uint8_t sym) { uint32_t init_base = base; base += (uint32_t)(sym) * (length >>= 8); // new interval base and length if (init_base > base) propagate_carry(); // overflow = carry if (length < AC__MinLength) renorm_enc_interval(); // renormalization } void writeShort(uint16_t sym) { uint32_t init_base = base; base += (uint32_t)(sym) * (length >>= 16); // new interval base and length if (init_base > base) propagate_carry(); // overflow = carry if (length < AC__MinLength) renorm_enc_interval(); // renormalization } void writeInt(uint32_t sym) { writeShort((uint16_t)(sym & 0xFFFF)); // lower 16 bits writeShort((uint16_t)(sym >> 16)); // UPPER 16 bits } void writeFloat(float sym) /* danger in float reinterpretation */ { U32I32F32 u32i32f32; u32i32f32.f32 = sym; writeInt(u32i32f32.u32); } void writeInt64(uint64_t sym) { writeInt((uint32_t)(sym & 0xFFFFFFFF)); // lower 32 bits writeInt((uint32_t)(sym >> 32)); // UPPER 32 bits } void writeDouble(double sym) /* danger in float reinterpretation */ { U64I64F64 u64i64f64; u64i64f64.f64 = sym; writeInt64(u64i64f64.u64); } TOutStream& getOutStream() { return outstream; } uint32_t num_encoded() { return valid ? outstream.numBytesPut() : 0; } const uint8_t *encoded_bytes() { return valid ? outstream.data() : nullptr; } private: void init(bool v) { valid = v; outbuffer = new uint8_t[2*AC_BUFFER_SIZE]; endbuffer = outbuffer + 2 * AC_BUFFER_SIZE; base = 0; length = AC__MaxLength; outbyte = outbuffer; endbyte = endbuffer; } void init(const arithmetic& src) { valid = src.valid; outbuffer = new uint8_t[2*AC_BUFFER_SIZE]; endbuffer = outbuffer + 2 * AC_BUFFER_SIZE; base = src.base; length = src.length; outbyte = outbuffer + (src.outbyte - src.outbuffer); endbyte = outbuffer + (src.endbyte - src.outbuffer); } void propagate_carry() { uint8_t *b; if (outbyte == outbuffer) b = endbuffer - 1; else b = outbyte - 1; while (*b == 0xFFU) { *b = 0; if (b == outbuffer) b = endbuffer - 1; else b--; assert(outbuffer <= b); assert(b < endbuffer); assert(outbyte < endbuffer); } ++*b; } void renorm_enc_interval() { do { // output and discard top byte assert(outbuffer <= outbyte); assert(outbyte < endbuffer); assert(outbyte < endbyte); *outbyte++ = (uint8_t)(base >> 24); if (outbyte == endbyte) manage_outbuffer(); base <<= 8; } while ((length <<= 8) < AC__MinLength); // length multiplied by 256 } void manage_outbuffer() { if (outbyte == endbuffer) outbyte = outbuffer; outstream.putBytes(outbyte, AC_BUFFER_SIZE); endbyte = outbyte + AC_BUFFER_SIZE; assert(endbyte > outbyte); assert(outbyte < endbuffer); } arithmetic& operator = (const arithmetic&) = delete; private: uint8_t* outbuffer; uint8_t* endbuffer; uint8_t* outbyte; uint8_t* endbyte; uint32_t base, value, length; bool valid; std::unique_ptr pOut; TOutStream& outstream; }; } // namespace encoders } // namespace lazperf #endif // __encoder_hpp__