From 52c08b0282aacb44c2e75cefeae5626a7aa1b21a Mon Sep 17 00:00:00 2001 From: pacien Date: Fri, 23 Nov 2018 12:31:11 +0100 Subject: Implement bitstream --- src/bitstream.nim | 112 +++++++++++++++++++++++++++++++++++++++++++ src/integers.nim | 24 ++++++++++ tests/tbitstream.nim | 131 +++++++++++++++++++++++++++++++++++++++++++++++++++ tests/tintegers.nim | 28 +++++++++++ 4 files changed, 295 insertions(+) create mode 100644 src/bitstream.nim create mode 100644 src/integers.nim create mode 100644 tests/tbitstream.nim create mode 100644 tests/tintegers.nim diff --git a/src/bitstream.nim b/src/bitstream.nim new file mode 100644 index 0000000..81401ce --- /dev/null +++ b/src/bitstream.nim @@ -0,0 +1,112 @@ +# "à-la-gzip" gzip-like LZSS compressor +# Copyright (C) 2018 Pacien TRAN-GIRARD +# +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU Affero General Public License as +# published by the Free Software Foundation, either version 3 of the +# License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Affero General Public License for more details. +# +# You should have received a copy of the GNU Affero General Public License +# along with this program. If not, see . + +import streams +import integers + +# Stream functions + +proc newEIO(msg: string): ref IOError = + new(result) + result.msg = msg + +proc read[T](s: Stream, t: typedesc[T]): T = + if readData(s, addr(result), sizeof(T)) != sizeof(T): + raise newEIO("cannot read from stream") + +proc peek[T](s: Stream, t: typedesc[T]): T = + if peekData(s, addr(result), sizeof(T)) != sizeof(T): + raise newEIO("cannot read from stream") + +# Utils + +iterator chunks*(totalBitLength: int, chunkType: typedesc[SomeInteger]): tuple[index: int, chunkBitLength: int] = + let chunkBitLength = sizeof(chunkType) * wordBitLength + let wordCount = totalBitLength div chunkBitLength + for i in 0..<(wordCount): yield (i, chunkBitLength) + let remainder = totalBitLength mod chunkBitLength + if remainder > 0: yield (wordCount, remainder) + +# BitStream + +type BitStream* = ref object + stream: Stream + bitOffset: int + writeBuffer: uint8 + +proc bitStream*(stream: Stream): BitStream = + BitStream(stream: stream, bitOffset: 0, writeBuffer: 0) + +proc flush*(bitStream: BitStream) = + if bitStream.bitOffset == 0: return + bitStream.stream.write(bitStream.writeBuffer) + bitStream.stream.flush() + (bitStream.bitOffset, bitStream.writeBuffer) = (0, 0'u8) + +proc atEnd*(bitStream: BitStream): bool = + bitStream.stream.atEnd() + +proc readBits*[T: SomeUnsignedInt](bitStream: BitStream, bits: int, to: typedesc[T]): T = + let targetBitLength = sizeof(T) * wordBitLength + if bits < 0 or bits > targetBitLength: + raise newException(RangeError, "invalid bit length") + elif bits == 0: + result = 0 + elif bits < targetBitLength - bitStream.bitOffset: + result = bitStream.stream.peek(T) shl (targetBitLength - bits - bitStream.bitOffset) shr (targetBitLength - bits) + elif bits == targetBitLength - bitStream.bitOffset: + result = bitStream.stream.read(T) shl (targetBitLength - bits - bitStream.bitOffset) shr (targetBitLength - bits) + else: + let rightBits = targetBitLength - bitStream.bitOffset + let leftBits = bits - rightBits + let right = bitStream.stream.read(T) shr bitStream.bitOffset + let left = bitStream.stream.peek(T) shl (targetBitLength - leftBits) shr (targetBitLength - bits) + result = left or right + bitStream.bitOffset = (bitStream.bitOffset + bits) mod wordBitLength + +proc readBool*(bitStream: BitStream): bool = + bitStream.readBits(1, uint8) != 0 + +proc readSeq*[T: SomeUnsignedInt](bitStream: BitStream, bitLength: int, to: typedesc[T]): tuple[bitLength: int, data: seq[T]] = + result = (0, newSeqOfCap[T](bitLength /^ (sizeof(T) * wordBitLength))) + for _, chunkBitLength in chunks(bitLength, T): + if bitStream.atEnd(): return + result.bitLength += chunkBitLength + result.data.add(bitStream.readBits(chunkBitLength, T)) + +proc writeBits*(bitStream: BitStream, bits: int, value: SomeUnsignedInt) = + let valueContainerBitLength = sizeof(value) * wordBitLength + if bits < 0 or bits > valueContainerBitLength: + raise newException(RangeError, "invalid bit length") + var bitsToWrite = bits + if bitsToWrite + bitStream.bitOffset >= wordBitLength: + bitStream.stream.write(truncateToUint8(value shl bitStream.bitOffset) or bitStream.writeBuffer) + bitsToWrite -= wordBitLength - bitStream.bitOffset + (bitStream.bitOffset, bitStream.writeBuffer) = (0, 0'u8) + while bitsToWrite >= wordBitLength: + bitStream.stream.write(truncateToUint8(value shr (bits - bitsToWrite))) + bitsToWrite -= wordBitLength + if bitsToWrite > 0: + let left = truncateToUint8((value shl (valueContainerBitLength - bits)) shr (valueContainerBitLength - bitsToWrite)) + bitStream.writeBuffer = (left shl bitStream.bitOffset) or bitStream.writeBuffer + bitStream.bitOffset = (bitStream.bitOffset + bitsToWrite) mod wordBitLength + +proc writeBool*(bitStream: BitStream, value: bool) = + bitStream.writeBits(1, value.uint8) + +proc writeSeq*[T: SomeUnsignedInt](bitStream: BitStream, bitLength: int, data: seq[T]) = + for i, chunkBitLength in chunks(bitLength, T): + bitStream.writeBits(chunkBitLength, data[i]) diff --git a/src/integers.nim b/src/integers.nim new file mode 100644 index 0000000..1b9121c --- /dev/null +++ b/src/integers.nim @@ -0,0 +1,24 @@ +# "à-la-gzip" gzip-like LZSS compressor +# Copyright (C) 2018 Pacien TRAN-GIRARD +# +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU Affero General Public License as +# published by the Free Software Foundation, either version 3 of the +# License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Affero General Public License for more details. +# +# You should have received a copy of the GNU Affero General Public License +# along with this program. If not, see . + +const wordBitLength* = 8 +const wordBitMask* = 0b1111_1111'u8 + +proc `/^`*[T: Natural](x, y: T): T = + (x + y - 1) div y + +proc truncateToUint8*(x: SomeUnsignedInt): uint8 = + (x and wordBitMask).uint8 diff --git a/tests/tbitstream.nim b/tests/tbitstream.nim new file mode 100644 index 0000000..2292049 --- /dev/null +++ b/tests/tbitstream.nim @@ -0,0 +1,131 @@ +# "à-la-gzip" gzip-like LZSS compressor +# Copyright (C) 2018 Pacien TRAN-GIRARD +# +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU Affero General Public License as +# published by the Free Software Foundation, either version 3 of the +# License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Affero General Public License for more details. +# +# You should have received a copy of the GNU Affero General Public License +# along with this program. If not, see . + +import unittest, streams, sugar, sequtils +import integers, bitstream + +suite "bitstream": + test "chunks iterator": + check toSeq(chunks(70, uint32)) == @[(0, 32), (1, 32), (2, 6)] + check toSeq(chunks(32, uint16)) == @[(0, 16), (1, 16)] + + test "flush": + let stream = newStringStream() + defer: stream.close() + let bitStream = stream.bitStream() + + bitStream.writeBool(true) + stream.setPosition(0) + expect IOError: discard stream.peekUint8() + + bitStream.flush() + stream.setPosition(0) + check stream.readUint8() == 0x01'u8 + check stream.atEnd() + + bitStream.flush() + check stream.atEnd() + + test "readBool": + let stream = newStringStream() + defer: stream.close() + stream.write(0b1001_1111'u8) + stream.write(0b0110_0000'u8) + stream.setPosition(0) + + let bitStream = stream.bitStream() + check lc[bitStream.readBool() | (_ <- 0..<16), bool] == @[ + true, true, true, true, true, false, false, true, + false, false, false, false, false, true, true, false] + + expect IOError: discard bitStream.readBool() + check bitStream.atEnd() + + test "readBits": + let stream = newStringStream() + defer: stream.close() + stream.write(0xF00F'u16) + stream.write(0x0FFF'u16) + stream.setPosition(0) + + let bitStream = stream.bitStream() + check bitStream.readBits(8, uint8) == 0x0F'u8 + check bitStream.readBits(16, uint16) == 0xFFF0'u16 + check bitStream.readBits(8, uint8) == 0x0F'u8 + + expect RangeError: discard bitStream.readBits(9, uint8) + expect IOError: discard bitStream.readBits(16, uint16) + check bitStream.atEnd() + + test "readSeq": + let stream = newStringStream() + defer: stream.close() + stream.write(0x0F00_F0FF_F0F0_F0F0'u64) + stream.setPosition(0) + + let bitStream = stream.bitStream() + check bitStream.readSeq(32, uint16) == (32, @[0xF0F0'u16, 0xF0F0]) + check bitStream.readSeq(40, uint8) == (32, @[0xFF'u8, 0xF0, 0x00, 0x0F]) + check bitStream.atEnd() + + test "writeBool": + let stream = newStringStream() + defer: stream.close() + + let bitStream = stream.bitStream() + let booleanValues = @[ + true, true, true, true, true, false, false, true, + false, false, false, false, false, true, true, false, + true, true, false, true] + for b in booleanValues: bitStream.writeBool(b) + bitStream.flush() + + stream.setPosition(0) + check stream.readUint8() == 0b1001_1111'u8 + check stream.readUint8() == 0b0110_0000'u8 + check stream.readUint8() == 0b0000_1011'u8 + expect IOError: discard stream.readUint8() + check stream.atEnd() + + test "writeBits": + let stream = newStringStream() + defer: stream.close() + + let bitStream = stream.bitStream() + bitStream.writeBits(4, 0xF00F'u16) + bitStream.writeBits(16, 0xF00F'u16) + bitStream.writeBits(16, 0xFFFF'u16) + bitStream.flush() + + stream.setPosition(0) + check stream.readUint16() == 0x00FF'u16 + check stream.readUint16() == 0xFFFF'u16 + check stream.readUint8() == 0x0F'u8 + expect IOError: discard stream.readUint8() + check stream.atEnd() + + test "writeSeq": + let stream = newStringStream() + defer: stream.close() + + let bitStream = stream.bitStream() + bitStream.writeSeq(32, @[0xF0F0'u16, 0xF0F0]) + bitStream.writeSeq(28, @[0xFF'u8, 0xF0, 0x00, 0xFF]) + bitStream.flush() + + stream.setPosition(0) + check stream.readUint64() == 0x0F00_F0FF_F0F0_F0F0'u64 + check stream.atEnd() diff --git a/tests/tintegers.nim b/tests/tintegers.nim new file mode 100644 index 0000000..720677e --- /dev/null +++ b/tests/tintegers.nim @@ -0,0 +1,28 @@ +# "à-la-gzip" gzip-like LZSS compressor +# Copyright (C) 2018 Pacien TRAN-GIRARD +# +# This program is free software: you can redistribute it and/or modify +# it under the terms of the GNU Affero General Public License as +# published by the Free Software Foundation, either version 3 of the +# License, or (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Affero General Public License for more details. +# +# You should have received a copy of the GNU Affero General Public License +# along with this program. If not, see . + +import unittest +import integers + +suite "integers": + test "Round-up integer division": + check 42 /^ 2 == 21 + check 43 /^ 2 == 22 + + test "truncateToUint8": + check truncateToUint8(0xFA'u8) == 0xFA'u8 + check truncateToUint8(0x00FA'u16) == 0xFA'u8 + check truncateToUint8(0xFFFA'u16) == 0xFA'u8 -- cgit v1.2.3