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.IOException; 022import java.io.InputStream; 023import java.util.Arrays; 024 025import org.apache.commons.compress.compressors.CompressorInputStream; 026import org.apache.commons.compress.utils.BoundedInputStream; 027import org.apache.commons.compress.utils.ByteUtils; 028import org.apache.commons.compress.utils.ChecksumCalculatingInputStream; 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 LZ4 frame format. 035 * 036 * <p>Based on the "spec" in the version "1.5.1 (31/03/2015)"</p> 037 * 038 * @see <a href="http://lz4.github.io/lz4/lz4_Frame_format.html">LZ4 Frame Format Description</a> 039 * @since 1.14 040 * @NotThreadSafe 041 */ 042public class FramedLZ4CompressorInputStream extends CompressorInputStream 043 implements InputStreamStatistics { 044 045 // used by FramedLZ4CompressorOutputStream as well 046 static final byte[] LZ4_SIGNATURE = new byte[] { //NOSONAR 047 4, 0x22, 0x4d, 0x18 048 }; 049 private static final byte[] SKIPPABLE_FRAME_TRAILER = new byte[] { 050 0x2a, 0x4d, 0x18 051 }; 052 private static final byte SKIPPABLE_FRAME_PREFIX_BYTE_MASK = 0x50; 053 054 static final int VERSION_MASK = 0xC0; 055 static final int SUPPORTED_VERSION = 0x40; 056 static final int BLOCK_INDEPENDENCE_MASK = 0x20; 057 static final int BLOCK_CHECKSUM_MASK = 0x10; 058 static final int CONTENT_SIZE_MASK = 0x08; 059 static final int CONTENT_CHECKSUM_MASK = 0x04; 060 static final int BLOCK_MAX_SIZE_MASK = 0x70; 061 static final int UNCOMPRESSED_FLAG_MASK = 0x80000000; 062 063 // used in no-arg read method 064 private final byte[] oneByte = new byte[1]; 065 066 private final ByteUtils.ByteSupplier supplier = new ByteUtils.ByteSupplier() { 067 @Override 068 public int getAsByte() throws IOException { 069 return readOneByte(); 070 } 071 }; 072 073 private final CountingInputStream in; 074 private final boolean decompressConcatenated; 075 076 private boolean expectBlockChecksum; 077 private boolean expectBlockDependency; 078 private boolean expectContentSize; 079 private boolean expectContentChecksum; 080 081 private InputStream currentBlock; 082 private boolean endReached, inUncompressed; 083 084 // used for frame header checksum and content checksum, if present 085 private final XXHash32 contentHash = new XXHash32(); 086 087 // used for block checksum, if present 088 private final XXHash32 blockHash = new XXHash32(); 089 090 // only created if the frame doesn't set the block independence flag 091 private byte[] blockDependencyBuffer; 092 093 /** 094 * Creates a new input stream that decompresses streams compressed 095 * using the LZ4 frame format and stops after decompressing the 096 * first frame. 097 * @param in the InputStream from which to read the compressed data 098 * @throws IOException if reading fails 099 */ 100 public FramedLZ4CompressorInputStream(InputStream in) throws IOException { 101 this(in, false); 102 } 103 104 /** 105 * Creates a new input stream that decompresses streams compressed 106 * using the LZ4 frame format. 107 * @param in the InputStream from which to read the compressed data 108 * @param decompressConcatenated if true, decompress until the end 109 * of the input; if false, stop after the first LZ4 frame 110 * and leave the input position to point to the next byte 111 * after the frame stream 112 * @throws IOException if reading fails 113 */ 114 public FramedLZ4CompressorInputStream(InputStream in, boolean decompressConcatenated) throws IOException { 115 this.in = new CountingInputStream(in); 116 this.decompressConcatenated = decompressConcatenated; 117 init(true); 118 } 119 120 /** {@inheritDoc} */ 121 @Override 122 public int read() throws IOException { 123 return read(oneByte, 0, 1) == -1 ? -1 : oneByte[0] & 0xFF; 124 } 125 126 /** {@inheritDoc} */ 127 @Override 128 public void close() throws IOException { 129 try { 130 if (currentBlock != null) { 131 currentBlock.close(); 132 currentBlock = null; 133 } 134 } finally { 135 in.close(); 136 } 137 } 138 139 /** {@inheritDoc} */ 140 @Override 141 public int read(final byte[] b, final int off, final int len) throws IOException { 142 if (endReached) { 143 return -1; 144 } 145 int r = readOnce(b, off, len); 146 if (r == -1) { 147 nextBlock(); 148 if (!endReached) { 149 r = readOnce(b, off, len); 150 } 151 } 152 if (r != -1) { 153 if (expectBlockDependency) { 154 appendToBlockDependencyBuffer(b, off, r); 155 } 156 if (expectContentChecksum) { 157 contentHash.update(b, off, r); 158 } 159 } 160 return r; 161 } 162 163 /** 164 * @since 1.17 165 */ 166 @Override 167 public long getCompressedCount() { 168 return in.getBytesRead(); 169 } 170 171 private void init(boolean firstFrame) throws IOException { 172 if (readSignature(firstFrame)) { 173 readFrameDescriptor(); 174 nextBlock(); 175 } 176 } 177 178 private boolean readSignature(boolean firstFrame) throws IOException { 179 String garbageMessage = firstFrame ? "Not a LZ4 frame stream" : "LZ4 frame stream followed by garbage"; 180 final byte[] b = new byte[4]; 181 int read = IOUtils.readFully(in, b); 182 count(read); 183 if (0 == read && !firstFrame) { 184 // good LZ4 frame and nothing after it 185 endReached = true; 186 return false; 187 } 188 if (4 != read) { 189 throw new IOException(garbageMessage); 190 } 191 192 read = skipSkippableFrame(b); 193 if (0 == read && !firstFrame) { 194 // good LZ4 frame with only some skippable frames after it 195 endReached = true; 196 return false; 197 } 198 if (4 != read || !matches(b, 4)) { 199 throw new IOException(garbageMessage); 200 } 201 return true; 202 } 203 204 private void readFrameDescriptor() throws IOException { 205 int flags = readOneByte(); 206 if (flags == -1) { 207 throw new IOException("Premature end of stream while reading frame flags"); 208 } 209 contentHash.update(flags); 210 if ((flags & VERSION_MASK) != SUPPORTED_VERSION) { 211 throw new IOException("Unsupported version " + (flags >> 6)); 212 } 213 expectBlockDependency = (flags & BLOCK_INDEPENDENCE_MASK) == 0; 214 if (expectBlockDependency) { 215 if (blockDependencyBuffer == null) { 216 blockDependencyBuffer = new byte[BlockLZ4CompressorInputStream.WINDOW_SIZE]; 217 } 218 } else { 219 blockDependencyBuffer = null; 220 } 221 expectBlockChecksum = (flags & BLOCK_CHECKSUM_MASK) != 0; 222 expectContentSize = (flags & CONTENT_SIZE_MASK) != 0; 223 expectContentChecksum = (flags & CONTENT_CHECKSUM_MASK) != 0; 224 int bdByte = readOneByte(); 225 if (bdByte == -1) { // max size is irrelevant for this implementation 226 throw new IOException("Premature end of stream while reading frame BD byte"); 227 } 228 contentHash.update(bdByte); 229 if (expectContentSize) { // for now we don't care, contains the uncompressed size 230 byte[] contentSize = new byte[8]; 231 int skipped = IOUtils.readFully(in, contentSize); 232 count(skipped); 233 if (8 != skipped) { 234 throw new IOException("Premature end of stream while reading content size"); 235 } 236 contentHash.update(contentSize, 0, contentSize.length); 237 } 238 int headerHash = readOneByte(); 239 if (headerHash == -1) { // partial hash of header. 240 throw new IOException("Premature end of stream while reading frame header checksum"); 241 } 242 int expectedHash = (int) ((contentHash.getValue() >> 8) & 0xff); 243 contentHash.reset(); 244 if (headerHash != expectedHash) { 245 throw new IOException("frame header checksum mismatch."); 246 } 247 } 248 249 private void nextBlock() throws IOException { 250 maybeFinishCurrentBlock(); 251 long len = ByteUtils.fromLittleEndian(supplier, 4); 252 boolean uncompressed = (len & UNCOMPRESSED_FLAG_MASK) != 0; 253 int realLen = (int) (len & (~UNCOMPRESSED_FLAG_MASK)); 254 if (realLen == 0) { 255 verifyContentChecksum(); 256 if (!decompressConcatenated) { 257 endReached = true; 258 } else { 259 init(false); 260 } 261 return; 262 } 263 InputStream capped = new BoundedInputStream(in, realLen); 264 if (expectBlockChecksum) { 265 capped = new ChecksumCalculatingInputStream(blockHash, capped); 266 } 267 if (uncompressed) { 268 inUncompressed = true; 269 currentBlock = capped; 270 } else { 271 inUncompressed = false; 272 BlockLZ4CompressorInputStream s = new BlockLZ4CompressorInputStream(capped); 273 if (expectBlockDependency) { 274 s.prefill(blockDependencyBuffer); 275 } 276 currentBlock = s; 277 } 278 } 279 280 private void maybeFinishCurrentBlock() throws IOException { 281 if (currentBlock != null) { 282 currentBlock.close(); 283 currentBlock = null; 284 if (expectBlockChecksum) { 285 verifyChecksum(blockHash, "block"); 286 blockHash.reset(); 287 } 288 } 289 } 290 291 private void verifyContentChecksum() throws IOException { 292 if (expectContentChecksum) { 293 verifyChecksum(contentHash, "content"); 294 } 295 contentHash.reset(); 296 } 297 298 private void verifyChecksum(XXHash32 hash, String kind) throws IOException { 299 byte[] checksum = new byte[4]; 300 int read = IOUtils.readFully(in, checksum); 301 count(read); 302 if (4 != read) { 303 throw new IOException("Premature end of stream while reading " + kind + " checksum"); 304 } 305 long expectedHash = hash.getValue(); 306 if (expectedHash != ByteUtils.fromLittleEndian(checksum)) { 307 throw new IOException(kind + " checksum mismatch."); 308 } 309 } 310 311 private int readOneByte() throws IOException { 312 final int b = in.read(); 313 if (b != -1) { 314 count(1); 315 return b & 0xFF; 316 } 317 return -1; 318 } 319 320 private int readOnce(byte[] b, int off, int len) throws IOException { 321 if (inUncompressed) { 322 int cnt = currentBlock.read(b, off, len); 323 count(cnt); 324 return cnt; 325 } 326 BlockLZ4CompressorInputStream l = (BlockLZ4CompressorInputStream) currentBlock; 327 long before = l.getBytesRead(); 328 int cnt = currentBlock.read(b, off, len); 329 count(l.getBytesRead() - before); 330 return cnt; 331 } 332 333 private static boolean isSkippableFrameSignature(byte[] b) { 334 if ((b[0] & SKIPPABLE_FRAME_PREFIX_BYTE_MASK) != SKIPPABLE_FRAME_PREFIX_BYTE_MASK) { 335 return false; 336 } 337 for (int i = 1; i < 4; i++) { 338 if (b[i] != SKIPPABLE_FRAME_TRAILER[i - 1]) { 339 return false; 340 } 341 } 342 return true; 343 } 344 345 /** 346 * Skips over the contents of a skippable frame as well as 347 * skippable frames following it. 348 * 349 * <p>It then tries to read four more bytes which are supposed to 350 * hold an LZ4 signature and returns the number of bytes read 351 * while storing the bytes in the given array.</p> 352 */ 353 private int skipSkippableFrame(byte[] b) throws IOException { 354 int read = 4; 355 while (read == 4 && isSkippableFrameSignature(b)) { 356 long len = ByteUtils.fromLittleEndian(supplier, 4); 357 long skipped = IOUtils.skip(in, len); 358 count(skipped); 359 if (len != skipped) { 360 throw new IOException("Premature end of stream while skipping frame"); 361 } 362 read = IOUtils.readFully(in, b); 363 count(read); 364 } 365 return read; 366 } 367 368 private void appendToBlockDependencyBuffer(final byte[] b, final int off, int len) { 369 len = Math.min(len, blockDependencyBuffer.length); 370 if (len > 0) { 371 int keep = blockDependencyBuffer.length - len; 372 if (keep > 0) { 373 // move last keep bytes towards the start of the buffer 374 System.arraycopy(blockDependencyBuffer, len, blockDependencyBuffer, 0, keep); 375 } 376 // append new data 377 System.arraycopy(b, off, blockDependencyBuffer, keep, len); 378 } 379 } 380 381 /** 382 * Checks if the signature matches what is expected for a .lz4 file. 383 * 384 * <p>.lz4 files start with a four byte signature.</p> 385 * 386 * @param signature the bytes to check 387 * @param length the number of bytes to check 388 * @return true if this is a .sz stream, false otherwise 389 */ 390 public static boolean matches(final byte[] signature, final int length) { 391 392 if (length < LZ4_SIGNATURE.length) { 393 return false; 394 } 395 396 byte[] shortenedSig = signature; 397 if (signature.length > LZ4_SIGNATURE.length) { 398 shortenedSig = new byte[LZ4_SIGNATURE.length]; 399 System.arraycopy(signature, 0, shortenedSig, 0, LZ4_SIGNATURE.length); 400 } 401 402 return Arrays.equals(shortenedSig, LZ4_SIGNATURE); 403 } 404}