001/*
002 * Licensed to the Apache Software Foundation (ASF) under one
003 * or more contributor license agreements.  See the NOTICE file
004 * distributed with this work for additional information
005 * regarding copyright ownership.  The ASF licenses this file
006 * to you under the Apache License, Version 2.0 (the
007 * "License"); you may not use this file except in compliance
008 * with the License.  You may obtain a copy of the License at
009 *
010 * http://www.apache.org/licenses/LICENSE-2.0
011 *
012 * Unless required by applicable law or agreed to in writing,
013 * software distributed under the License is distributed on an
014 * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
015 * KIND, either express or implied.  See the License for the
016 * specific language governing permissions and limitations
017 * under the License.
018 */
019package org.apache.commons.compress.compressors.lz4;
020
021import java.io.ByteArrayOutputStream;
022import java.io.IOException;
023import java.io.OutputStream;
024
025import org.apache.commons.compress.compressors.CompressorOutputStream;
026import org.apache.commons.compress.utils.ByteUtils;
027
028/**
029 * CompressorOutputStream for the LZ4 frame format.
030 *
031 * <p>Based on the "spec" in the version "1.5.1 (31/03/2015)"</p>
032 *
033 * @see <a href="http://lz4.github.io/lz4/lz4_Frame_format.html">LZ4 Frame Format Description</a>
034 * @since 1.14
035 * @NotThreadSafe
036 */
037public class FramedLZ4CompressorOutputStream extends CompressorOutputStream {
038
039    private static final byte[] END_MARK = new byte[4];
040
041    // used in one-arg write method
042    private final byte[] oneByte = new byte[1];
043
044    private final byte[] blockData;
045    private final OutputStream out;
046    private final Parameters params;
047    private boolean finished;
048    private int currentIndex;
049
050    // used for frame header checksum and content checksum, if requested
051    private final XXHash32 contentHash = new XXHash32();
052    // used for block checksum, if requested
053    private final XXHash32 blockHash;
054
055    // only created if the config requires block dependency
056    private final byte[] blockDependencyBuffer;
057    private int collectedBlockDependencyBytes;
058
059    /**
060     * The block sizes supported by the format.
061     */
062    public enum BlockSize {
063        /** Block size of 64K */
064        K64(64 * 1024, 4),
065        /** Block size of 256K */
066        K256(256 * 1024, 5),
067        /** Block size of 1M */
068        M1(1024 * 1024, 6),
069        /** Block size of 4M */
070        M4(4096 * 1024, 7);
071
072        private final int size, index;
073        BlockSize(final int size, final int index) {
074            this.size = size;
075            this.index = index;
076        }
077        int getSize() {
078            return size;
079        }
080        int getIndex() {
081            return index;
082        }
083    }
084
085    /**
086     * Parameters of the LZ4 frame format.
087     */
088    public static class Parameters {
089        private final BlockSize blockSize;
090        private final boolean withContentChecksum, withBlockChecksum, withBlockDependency;
091        private final org.apache.commons.compress.compressors.lz77support.Parameters lz77params;
092
093        /**
094         * The default parameters of 4M block size, enabled content
095         * checksum, disabled block checksums and independent blocks.
096         *
097         * <p>This matches the defaults of the lz4 command line utility.</p>
098         */
099        public static final Parameters DEFAULT = new Parameters(BlockSize.M4, true, false, false);
100
101        /**
102         * Sets up custom a custom block size for the LZ4 stream but
103         * otherwise uses the defaults of enabled content checksum,
104         * disabled block checksums and independent blocks.
105         * @param blockSize the size of a single block.
106         */
107        public Parameters(final BlockSize blockSize) {
108            this(blockSize, true, false, false);
109        }
110        /**
111         * Sets up custom a custom block size for the LZ4 stream but
112         * otherwise uses the defaults of enabled content checksum,
113         * disabled block checksums and independent blocks.
114         * @param blockSize the size of a single block.
115         * @param lz77params parameters used to fine-tune compression,
116         * in particular to balance compression ratio vs compression
117         * speed.
118         */
119        public Parameters(final BlockSize blockSize,
120            final org.apache.commons.compress.compressors.lz77support.Parameters lz77params) {
121            this(blockSize, true, false, false, lz77params);
122        }
123        /**
124         * Sets up custom parameters for the LZ4 stream.
125         * @param blockSize the size of a single block.
126         * @param withContentChecksum whether to write a content checksum
127         * @param withBlockChecksum whether to write a block checksum.
128         * Note that block checksums are not supported by the lz4
129         * command line utility
130         * @param withBlockDependency whether a block may depend on
131         * the content of a previous block. Enabling this may improve
132         * compression ratio but makes it impossible to decompress the
133         * output in parallel.
134         */
135        public Parameters(final BlockSize blockSize, final boolean withContentChecksum, final boolean withBlockChecksum,
136            final boolean withBlockDependency) {
137            this(blockSize, withContentChecksum, withBlockChecksum, withBlockDependency,
138                 BlockLZ4CompressorOutputStream.createParameterBuilder().build());
139        }
140
141        /**
142         * Sets up custom parameters for the LZ4 stream.
143         * @param blockSize the size of a single block.
144         * @param withContentChecksum whether to write a content checksum
145         * @param withBlockChecksum whether to write a block checksum.
146         * Note that block checksums are not supported by the lz4
147         * command line utility
148         * @param withBlockDependency whether a block may depend on
149         * the content of a previous block. Enabling this may improve
150         * compression ratio but makes it impossible to decompress the
151         * output in parallel.
152         * @param lz77params parameters used to fine-tune compression,
153         * in particular to balance compression ratio vs compression
154         * speed.
155         */
156        public Parameters(final BlockSize blockSize, final boolean withContentChecksum, final boolean withBlockChecksum,
157                final boolean withBlockDependency,
158                final org.apache.commons.compress.compressors.lz77support.Parameters lz77params) {
159            this.blockSize = blockSize;
160            this.withContentChecksum = withContentChecksum;
161            this.withBlockChecksum = withBlockChecksum;
162            this.withBlockDependency = withBlockDependency;
163            this.lz77params = lz77params;
164        }
165
166        @Override
167        public String toString() {
168            return "LZ4 Parameters with BlockSize " + blockSize + ", withContentChecksum " + withContentChecksum
169                + ", withBlockChecksum " + withBlockChecksum + ", withBlockDependency " + withBlockDependency;
170        }
171    }
172
173    /**
174     * Constructs a new output stream that compresses data using the
175     * LZ4 frame format using the default block size of 4MB.
176     * @param out the OutputStream to which to write the compressed data
177     * @throws IOException if writing the signature fails
178     */
179    public FramedLZ4CompressorOutputStream(final OutputStream out) throws IOException {
180        this(out, Parameters.DEFAULT);
181    }
182
183    /**
184     * Constructs a new output stream that compresses data using the
185     * LZ4 frame format using the given block size.
186     * @param out the OutputStream to which to write the compressed data
187     * @param params the parameters to use
188     * @throws IOException if writing the signature fails
189     */
190    public FramedLZ4CompressorOutputStream(final OutputStream out, final Parameters params) throws IOException {
191        this.params = params;
192        blockData = new byte[params.blockSize.getSize()];
193        this.out = out;
194        blockHash = params.withBlockChecksum ? new XXHash32() : null;
195        out.write(FramedLZ4CompressorInputStream.LZ4_SIGNATURE);
196        writeFrameDescriptor();
197        blockDependencyBuffer = params.withBlockDependency
198            ? new byte[BlockLZ4CompressorInputStream.WINDOW_SIZE]
199            : null;
200    }
201
202    @Override
203    public void write(final int b) throws IOException {
204        oneByte[0] = (byte) (b & 0xff);
205        write(oneByte);
206    }
207
208    @Override
209    public void write(final byte[] data, int off, int len) throws IOException {
210        if (params.withContentChecksum) {
211            contentHash.update(data, off, len);
212        }
213        final int blockDataLength = blockData.length;
214        if (currentIndex + len > blockDataLength) {
215            flushBlock();
216            while (len > blockDataLength) {
217                System.arraycopy(data, off, blockData, 0, blockDataLength);
218                off += blockDataLength;
219                len -= blockDataLength;
220                currentIndex = blockDataLength;
221                flushBlock();
222            }
223        }
224        System.arraycopy(data, off, blockData, currentIndex, len);
225        currentIndex += len;
226    }
227
228    @Override
229    public void close() throws IOException {
230        try {
231            finish();
232        } finally {
233            out.close();
234        }
235    }
236
237    /**
238     * Compresses all remaining data and writes it to the stream,
239     * doesn't close the underlying stream.
240     * @throws IOException if an error occurs
241     */
242    public void finish() throws IOException {
243        if (!finished) {
244            if (currentIndex > 0) {
245                flushBlock();
246            }
247            writeTrailer();
248            finished = true;
249        }
250    }
251
252    private void writeFrameDescriptor() throws IOException {
253        int flags = FramedLZ4CompressorInputStream.SUPPORTED_VERSION;
254        if (!params.withBlockDependency) {
255            flags |= FramedLZ4CompressorInputStream.BLOCK_INDEPENDENCE_MASK;
256        }
257        if (params.withContentChecksum) {
258            flags |= FramedLZ4CompressorInputStream.CONTENT_CHECKSUM_MASK;
259        }
260        if (params.withBlockChecksum) {
261            flags |= FramedLZ4CompressorInputStream.BLOCK_CHECKSUM_MASK;
262        }
263        out.write(flags);
264        contentHash.update(flags);
265        final int bd = (params.blockSize.getIndex() << 4) & FramedLZ4CompressorInputStream.BLOCK_MAX_SIZE_MASK;
266        out.write(bd);
267        contentHash.update(bd);
268        out.write((int) ((contentHash.getValue() >> 8) & 0xff));
269        contentHash.reset();
270    }
271
272    private void flushBlock() throws IOException {
273        final boolean withBlockDependency = params.withBlockDependency;
274        final ByteArrayOutputStream baos = new ByteArrayOutputStream();
275        try (BlockLZ4CompressorOutputStream o = new BlockLZ4CompressorOutputStream(baos, params.lz77params)) {
276            if (withBlockDependency) {
277                o.prefill(blockDependencyBuffer, blockDependencyBuffer.length - collectedBlockDependencyBytes,
278                    collectedBlockDependencyBytes);
279            }
280            o.write(blockData, 0, currentIndex);
281        }
282        if (withBlockDependency) {
283            appendToBlockDependencyBuffer(blockData, 0, currentIndex);
284        }
285        final byte[] b = baos.toByteArray();
286        if (b.length > currentIndex) { // compression increased size, maybe beyond blocksize
287            ByteUtils.toLittleEndian(out, currentIndex | FramedLZ4CompressorInputStream.UNCOMPRESSED_FLAG_MASK,
288                4);
289            out.write(blockData, 0, currentIndex);
290            if (params.withBlockChecksum) {
291                blockHash.update(blockData, 0, currentIndex);
292            }
293        } else {
294            ByteUtils.toLittleEndian(out, b.length, 4);
295            out.write(b);
296            if (params.withBlockChecksum) {
297                blockHash.update(b, 0, b.length);
298            }
299        }
300        if (params.withBlockChecksum) {
301            ByteUtils.toLittleEndian(out, blockHash.getValue(), 4);
302            blockHash.reset();
303        }
304        currentIndex = 0;
305    }
306
307    private void writeTrailer() throws IOException {
308        out.write(END_MARK);
309        if (params.withContentChecksum) {
310            ByteUtils.toLittleEndian(out, contentHash.getValue(), 4);
311        }
312    }
313
314    private void appendToBlockDependencyBuffer(final byte[] b, final int off, int len) {
315        len = Math.min(len, blockDependencyBuffer.length);
316        if (len > 0) {
317            final int keep = blockDependencyBuffer.length - len;
318            if (keep > 0) {
319                // move last keep bytes towards the start of the buffer
320                System.arraycopy(blockDependencyBuffer, len, blockDependencyBuffer, 0, keep);
321            }
322            // append new data
323            System.arraycopy(b, off, blockDependencyBuffer, keep, len);
324            collectedBlockDependencyBytes = Math.min(collectedBlockDependencyBytes + len,
325                blockDependencyBuffer.length);
326        }
327    }
328
329}
330