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.snappy;
020
021import java.io.IOException;
022import java.io.InputStream;
023import java.io.PushbackInputStream;
024import java.util.Arrays;
025
026import org.apache.commons.compress.compressors.CompressorInputStream;
027import org.apache.commons.compress.utils.BoundedInputStream;
028import org.apache.commons.compress.utils.ByteUtils;
029import org.apache.commons.compress.utils.CountingInputStream;
030import org.apache.commons.compress.utils.IOUtils;
031import org.apache.commons.compress.utils.InputStreamStatistics;
032
033/**
034 * CompressorInputStream for the framing Snappy format.
035 *
036 * <p>Based on the "spec" in the version "Last revised: 2013-10-25"</p>
037 *
038 * @see <a href="https://github.com/google/snappy/blob/master/framing_format.txt">Snappy framing format description</a>
039 * @since 1.7
040 */
041public class FramedSnappyCompressorInputStream extends CompressorInputStream
042    implements InputStreamStatistics {
043
044    /**
045     * package private for tests only.
046     */
047    static final long MASK_OFFSET = 0xa282ead8L;
048
049    private static final int STREAM_IDENTIFIER_TYPE = 0xff;
050    static final int COMPRESSED_CHUNK_TYPE = 0;
051    private static final int UNCOMPRESSED_CHUNK_TYPE = 1;
052    private static final int PADDING_CHUNK_TYPE = 0xfe;
053    private static final int MIN_UNSKIPPABLE_TYPE = 2;
054    private static final int MAX_UNSKIPPABLE_TYPE = 0x7f;
055    private static final int MAX_SKIPPABLE_TYPE = 0xfd;
056
057    // used by FramedSnappyCompressorOutputStream as well
058    static final byte[] SZ_SIGNATURE = new byte[] { //NOSONAR
059        (byte) STREAM_IDENTIFIER_TYPE, // tag
060        6, 0, 0, // length
061        's', 'N', 'a', 'P', 'p', 'Y'
062    };
063
064    private long unreadBytes;
065    private final CountingInputStream countingStream;
066
067    /** The underlying stream to read compressed data from */
068    private final PushbackInputStream in;
069
070    /** The dialect to expect */
071    private final FramedSnappyDialect dialect;
072
073    private SnappyCompressorInputStream currentCompressedChunk;
074
075    // used in no-arg read method
076    private final byte[] oneByte = new byte[1];
077
078    private boolean endReached, inUncompressedChunk;
079
080    private int uncompressedBytesRemaining;
081    private long expectedChecksum = -1;
082    private final int blockSize;
083    private final PureJavaCrc32C checksum = new PureJavaCrc32C();
084
085    private final ByteUtils.ByteSupplier supplier = new ByteUtils.ByteSupplier() {
086        @Override
087        public int getAsByte() throws IOException {
088            return readOneByte();
089        }
090    };
091
092    /**
093     * Constructs a new input stream that decompresses
094     * snappy-framed-compressed data from the specified input stream
095     * using the {@link FramedSnappyDialect#STANDARD} dialect.
096     * @param in  the InputStream from which to read the compressed data
097     * @throws IOException if reading fails
098     */
099    public FramedSnappyCompressorInputStream(final InputStream in) throws IOException {
100        this(in, FramedSnappyDialect.STANDARD);
101    }
102
103    /**
104     * Constructs a new input stream that decompresses snappy-framed-compressed data
105     * from the specified input stream.
106     * @param in  the InputStream from which to read the compressed data
107     * @param dialect the dialect used by the compressed stream
108     * @throws IOException if reading fails
109     */
110    public FramedSnappyCompressorInputStream(final InputStream in,
111                                             final FramedSnappyDialect dialect)
112        throws IOException {
113        this(in, SnappyCompressorInputStream.DEFAULT_BLOCK_SIZE, dialect);
114    }
115
116    /**
117     * Constructs a new input stream that decompresses snappy-framed-compressed data
118     * from the specified input stream.
119     * @param in  the InputStream from which to read the compressed data
120     * @param blockSize the block size to use for the compressed stream
121     * @param dialect the dialect used by the compressed stream
122     * @throws IOException if reading fails
123     * @since 1.14
124     */
125    public FramedSnappyCompressorInputStream(final InputStream in,
126                                             final int blockSize,
127                                             final FramedSnappyDialect dialect)
128        throws IOException {
129        countingStream = new CountingInputStream(in);
130        this.in = new PushbackInputStream(countingStream, 1);
131        this.blockSize = blockSize;
132        this.dialect = dialect;
133        if (dialect.hasStreamIdentifier()) {
134            readStreamIdentifier();
135        }
136    }
137
138    /** {@inheritDoc} */
139    @Override
140    public int read() throws IOException {
141        return read(oneByte, 0, 1) == -1 ? -1 : oneByte[0] & 0xFF;
142    }
143
144    /** {@inheritDoc} */
145    @Override
146    public void close() throws IOException {
147        try {
148            if (currentCompressedChunk != null) {
149                currentCompressedChunk.close();
150                currentCompressedChunk = null;
151            }
152        } finally {
153            in.close();
154        }
155    }
156
157    /** {@inheritDoc} */
158    @Override
159    public int read(final byte[] b, final int off, final int len) throws IOException {
160        int read = readOnce(b, off, len);
161        if (read == -1) {
162            readNextBlock();
163            if (endReached) {
164                return -1;
165            }
166            read = readOnce(b, off, len);
167        }
168        return read;
169    }
170
171    /** {@inheritDoc} */
172    @Override
173    public int available() throws IOException {
174        if (inUncompressedChunk) {
175            return Math.min(uncompressedBytesRemaining,
176                            in.available());
177        } else if (currentCompressedChunk != null) {
178            return currentCompressedChunk.available();
179        }
180        return 0;
181    }
182
183    /**
184     * @since 1.17
185     */
186    @Override
187    public long getCompressedCount() {
188        return countingStream.getBytesRead() - unreadBytes;
189    }
190
191    /**
192     * Read from the current chunk into the given array.
193     *
194     * @return -1 if there is no current chunk or the number of bytes
195     * read from the current chunk (which may be -1 if the end of the
196     * chunk is reached).
197     */
198    private int readOnce(final byte[] b, final int off, final int len) throws IOException {
199        int read = -1;
200        if (inUncompressedChunk) {
201            final int amount = Math.min(uncompressedBytesRemaining, len);
202            if (amount == 0) {
203                return -1;
204            }
205            read = in.read(b, off, amount);
206            if (read != -1) {
207                uncompressedBytesRemaining -= read;
208                count(read);
209            }
210        } else if (currentCompressedChunk != null) {
211            final long before = currentCompressedChunk.getBytesRead();
212            read = currentCompressedChunk.read(b, off, len);
213            if (read == -1) {
214                currentCompressedChunk.close();
215                currentCompressedChunk = null;
216            } else {
217                count(currentCompressedChunk.getBytesRead() - before);
218            }
219        }
220        if (read > 0) {
221            checksum.update(b, off, read);
222        }
223        return read;
224    }
225
226    private void readNextBlock() throws IOException {
227        verifyLastChecksumAndReset();
228        inUncompressedChunk = false;
229        final int type = readOneByte();
230        if (type == -1) {
231            endReached = true;
232        } else if (type == STREAM_IDENTIFIER_TYPE) {
233            in.unread(type);
234            unreadBytes++;
235            pushedBackBytes(1);
236            readStreamIdentifier();
237            readNextBlock();
238        } else if (type == PADDING_CHUNK_TYPE
239                   || (type > MAX_UNSKIPPABLE_TYPE && type <= MAX_SKIPPABLE_TYPE)) {
240            skipBlock();
241            readNextBlock();
242        } else if (type >= MIN_UNSKIPPABLE_TYPE && type <= MAX_UNSKIPPABLE_TYPE) {
243            throw new IOException("unskippable chunk with type " + type
244                                  + " (hex " + Integer.toHexString(type) + ")"
245                                  + " detected.");
246        } else if (type == UNCOMPRESSED_CHUNK_TYPE) {
247            inUncompressedChunk = true;
248            uncompressedBytesRemaining = readSize() - 4 /* CRC */;
249            expectedChecksum = unmask(readCrc());
250        } else if (type == COMPRESSED_CHUNK_TYPE) {
251            final boolean expectChecksum = dialect.usesChecksumWithCompressedChunks();
252            final long size = readSize() - (expectChecksum ? 4L : 0L);
253            if (expectChecksum) {
254                expectedChecksum = unmask(readCrc());
255            } else {
256                expectedChecksum = -1;
257            }
258            currentCompressedChunk =
259                new SnappyCompressorInputStream(new BoundedInputStream(in, size), blockSize);
260            // constructor reads uncompressed size
261            count(currentCompressedChunk.getBytesRead());
262        } else {
263            // impossible as all potential byte values have been covered
264            throw new IOException("unknown chunk type " + type
265                                  + " detected.");
266        }
267    }
268
269    private long readCrc() throws IOException {
270        final byte[] b = new byte[4];
271        final int read = IOUtils.readFully(in, b);
272        count(read);
273        if (read != 4) {
274            throw new IOException("premature end of stream");
275        }
276        return ByteUtils.fromLittleEndian(b);
277    }
278
279    static long unmask(long x) {
280        // ugly, maybe we should just have used ints and deal with the
281        // overflow
282        x -= MASK_OFFSET;
283        x &= 0xffffFFFFL;
284        return ((x >> 17) | (x << 15)) & 0xffffFFFFL;
285    }
286
287    private int readSize() throws IOException {
288        return (int) ByteUtils.fromLittleEndian(supplier, 3);
289    }
290
291    private void skipBlock() throws IOException {
292        final int size = readSize();
293        final long read = IOUtils.skip(in, size);
294        count(read);
295        if (read != size) {
296            throw new IOException("premature end of stream");
297        }
298    }
299
300    private void readStreamIdentifier() throws IOException {
301        final byte[] b = new byte[10];
302        final int read = IOUtils.readFully(in, b);
303        count(read);
304        if (10 != read || !matches(b, 10)) {
305            throw new IOException("Not a framed Snappy stream");
306        }
307    }
308
309    private int readOneByte() throws IOException {
310        final int b = in.read();
311        if (b != -1) {
312            count(1);
313            return b & 0xFF;
314        }
315        return -1;
316    }
317
318    private void verifyLastChecksumAndReset() throws IOException {
319        if (expectedChecksum >= 0 && expectedChecksum != checksum.getValue()) {
320            throw new IOException("Checksum verification failed");
321        }
322        expectedChecksum = -1;
323        checksum.reset();
324    }
325
326    /**
327     * Checks if the signature matches what is expected for a .sz file.
328     *
329     * <p>.sz files start with a chunk with tag 0xff and content sNaPpY.</p>
330     *
331     * @param signature the bytes to check
332     * @param length    the number of bytes to check
333     * @return          true if this is a .sz stream, false otherwise
334     */
335    public static boolean matches(final byte[] signature, final int length) {
336
337        if (length < SZ_SIGNATURE.length) {
338            return false;
339        }
340
341        byte[] shortenedSig = signature;
342        if (signature.length > SZ_SIGNATURE.length) {
343            shortenedSig = new byte[SZ_SIGNATURE.length];
344            System.arraycopy(signature, 0, shortenedSig, 0, SZ_SIGNATURE.length);
345        }
346
347        return Arrays.equals(shortenedSig, SZ_SIGNATURE);
348    }
349
350}