Examples with IndexInput org.apache.lucene.store.IndexInput used on opensource projects

Example 1 with IndexInput

use of org.apache.lucene.store.IndexInput in project elasticsearch by elastic.

the class MetaDataStateFormat method read.

/**
     * Reads the state from a given file and compares the expected version against the actual version of
     * the state.
     */
public final T read(NamedXContentRegistry namedXContentRegistry, Path file) throws IOException {
    try (Directory dir = newDirectory(file.getParent())) {
        try (IndexInput indexInput = dir.openInput(file.getFileName().toString(), IOContext.DEFAULT)) {
            // We checksum the entire file before we even go and parse it. If it's corrupted we barf right here.
            CodecUtil.checksumEntireFile(indexInput);
            final int fileVersion = CodecUtil.checkHeader(indexInput, STATE_FILE_CODEC, MIN_COMPATIBLE_STATE_FILE_VERSION, STATE_FILE_VERSION);
            final XContentType xContentType = XContentType.values()[indexInput.readInt()];
            if (fileVersion == STATE_FILE_VERSION_ES_2X_AND_BELOW) {
                // format version 0, wrote a version that always came from the content state file and was never used
                // version currently unused
                indexInput.readLong();
            }
            long filePointer = indexInput.getFilePointer();
            long contentSize = indexInput.length() - CodecUtil.footerLength() - filePointer;
            try (IndexInput slice = indexInput.slice("state_xcontent", filePointer, contentSize)) {
                try (XContentParser parser = XContentFactory.xContent(xContentType).createParser(namedXContentRegistry, new InputStreamIndexInput(slice, contentSize))) {
                    return fromXContent(parser);
                }
            }
        } catch (CorruptIndexException | IndexFormatTooOldException | IndexFormatTooNewException ex) {
            // we trick this into a dedicated exception with the original stacktrace
            throw new CorruptStateException(ex);
        }
    }
}

Example 2 with IndexInput

use of org.apache.lucene.store.IndexInput in project elasticsearch by elastic.

the class MetaDataStateFormatTests method corruptFile.

public static void corruptFile(Path file, Logger logger) throws IOException {
    Path fileToCorrupt = file;
    try (SimpleFSDirectory dir = new SimpleFSDirectory(fileToCorrupt.getParent())) {
        long checksumBeforeCorruption;
        try (IndexInput input = dir.openInput(fileToCorrupt.getFileName().toString(), IOContext.DEFAULT)) {
            checksumBeforeCorruption = CodecUtil.retrieveChecksum(input);
        }
        try (FileChannel raf = FileChannel.open(fileToCorrupt, StandardOpenOption.READ, StandardOpenOption.WRITE)) {
            raf.position(randomIntBetween(0, (int) Math.min(Integer.MAX_VALUE, raf.size() - 1)));
            long filePointer = raf.position();
            ByteBuffer bb = ByteBuffer.wrap(new byte[1]);
            raf.read(bb);
            bb.flip();
            byte oldValue = bb.get(0);
            byte newValue = (byte) ~oldValue;
            bb.put(0, newValue);
            raf.write(bb, filePointer);
            logger.debug("Corrupting file {} --  flipping at position {} from {} to {} ", fileToCorrupt.getFileName().toString(), filePointer, Integer.toHexString(oldValue), Integer.toHexString(newValue));
        }
        long checksumAfterCorruption;
        long actualChecksumAfterCorruption;
        try (ChecksumIndexInput input = dir.openChecksumInput(fileToCorrupt.getFileName().toString(), IOContext.DEFAULT)) {
            assertThat(input.getFilePointer(), is(0L));
            // one long is the checksum... 8 bytes
            input.seek(input.length() - 8);
            checksumAfterCorruption = input.getChecksum();
            actualChecksumAfterCorruption = input.readLong();
        }
        StringBuilder msg = new StringBuilder();
        msg.append("Checksum before: [").append(checksumBeforeCorruption).append("]");
        msg.append(" after: [").append(checksumAfterCorruption).append("]");
        msg.append(" checksum value after corruption: ").append(actualChecksumAfterCorruption).append("]");
        msg.append(" file: ").append(fileToCorrupt.getFileName().toString()).append(" length: ").append(dir.fileLength(fileToCorrupt.getFileName().toString()));
        logger.debug("{}", msg.toString());
        assumeTrue("Checksum collision - " + msg.toString(), // collision
        checksumAfterCorruption != checksumBeforeCorruption || // checksum corrupted
        actualChecksumAfterCorruption != checksumBeforeCorruption);
    }
}

Example 3 with IndexInput

use of org.apache.lucene.store.IndexInput in project elasticsearch by elastic.

the class StoreTests method testCheckIntegrity.

public void testCheckIntegrity() throws IOException {
    Directory dir = newDirectory();
    long luceneFileLength = 0;
    try (IndexOutput output = dir.createOutput("lucene_checksum.bin", IOContext.DEFAULT)) {
        int iters = scaledRandomIntBetween(10, 100);
        for (int i = 0; i < iters; i++) {
            BytesRef bytesRef = new BytesRef(TestUtil.randomRealisticUnicodeString(random(), 10, 1024));
            output.writeBytes(bytesRef.bytes, bytesRef.offset, bytesRef.length);
            luceneFileLength += bytesRef.length;
        }
        CodecUtil.writeFooter(output);
        luceneFileLength += CodecUtil.footerLength();
    }
    final long luceneChecksum;
    try (IndexInput indexInput = dir.openInput("lucene_checksum.bin", IOContext.DEFAULT)) {
        assertEquals(luceneFileLength, indexInput.length());
        luceneChecksum = CodecUtil.retrieveChecksum(indexInput);
    }
    dir.close();
}

Example 4 with IndexInput

use of org.apache.lucene.store.IndexInput in project elasticsearch by elastic.

the class StoreTests method testVerifyingIndexInput.

public void testVerifyingIndexInput() throws IOException {
    Directory dir = newDirectory();
    IndexOutput output = dir.createOutput("foo.bar", IOContext.DEFAULT);
    int iters = scaledRandomIntBetween(10, 100);
    for (int i = 0; i < iters; i++) {
        BytesRef bytesRef = new BytesRef(TestUtil.randomRealisticUnicodeString(random(), 10, 1024));
        output.writeBytes(bytesRef.bytes, bytesRef.offset, bytesRef.length);
    }
    CodecUtil.writeFooter(output);
    output.close();
    // Check file
    IndexInput indexInput = dir.openInput("foo.bar", IOContext.DEFAULT);
    long checksum = CodecUtil.retrieveChecksum(indexInput);
    indexInput.seek(0);
    IndexInput verifyingIndexInput = new Store.VerifyingIndexInput(dir.openInput("foo.bar", IOContext.DEFAULT));
    readIndexInputFullyWithRandomSeeks(verifyingIndexInput);
    Store.verify(verifyingIndexInput);
    assertThat(checksum, equalTo(((ChecksumIndexInput) verifyingIndexInput).getChecksum()));
    IOUtils.close(indexInput, verifyingIndexInput);
    // Corrupt file and check again
    corruptFile(dir, "foo.bar", "foo1.bar");
    verifyingIndexInput = new Store.VerifyingIndexInput(dir.openInput("foo1.bar", IOContext.DEFAULT));
    readIndexInputFullyWithRandomSeeks(verifyingIndexInput);
    try {
        Store.verify(verifyingIndexInput);
        fail("should be a corrupted index");
    } catch (CorruptIndexException | IndexFormatTooOldException | IndexFormatTooNewException ex) {
    // ok
    }
    IOUtils.close(verifyingIndexInput);
    IOUtils.close(dir);
}

Example 5 with IndexInput

use of org.apache.lucene.store.IndexInput in project elasticsearch by elastic.

the class StoreTests method testChecksumCorrupted.

public void testChecksumCorrupted() throws IOException {
    Directory dir = newDirectory();
    IndexOutput output = dir.createOutput("foo.bar", IOContext.DEFAULT);
    int iters = scaledRandomIntBetween(10, 100);
    for (int i = 0; i < iters; i++) {
        BytesRef bytesRef = new BytesRef(TestUtil.randomRealisticUnicodeString(random(), 10, 1024));
        output.writeBytes(bytesRef.bytes, bytesRef.offset, bytesRef.length);
    }
    output.writeInt(CodecUtil.FOOTER_MAGIC);
    output.writeInt(0);
    String checksum = Store.digestToString(output.getChecksum());
    // write a wrong checksum to the file
    output.writeLong(output.getChecksum() + 1);
    output.close();
    IndexInput indexInput = dir.openInput("foo.bar", IOContext.DEFAULT);
    indexInput.seek(0);
    BytesRef ref = new BytesRef(scaledRandomIntBetween(1, 1024));
    long length = indexInput.length();
    IndexOutput verifyingOutput = new Store.LuceneVerifyingIndexOutput(new StoreFileMetaData("foo1.bar", length, checksum), dir.createOutput("foo1.bar", IOContext.DEFAULT));
    // we write the checksum in the try / catch block below
    length -= 8;
    while (length > 0) {
        if (random().nextInt(10) == 0) {
            verifyingOutput.writeByte(indexInput.readByte());
            length--;
        } else {
            int min = (int) Math.min(length, ref.bytes.length);
            indexInput.readBytes(ref.bytes, ref.offset, min);
            verifyingOutput.writeBytes(ref.bytes, ref.offset, min);
            length -= min;
        }
    }
    try {
        BytesRef checksumBytes = new BytesRef(8);
        checksumBytes.length = 8;
        indexInput.readBytes(checksumBytes.bytes, checksumBytes.offset, checksumBytes.length);
        if (randomBoolean()) {
            verifyingOutput.writeBytes(checksumBytes.bytes, checksumBytes.offset, checksumBytes.length);
        } else {
            for (int i = 0; i < checksumBytes.length; i++) {
                verifyingOutput.writeByte(checksumBytes.bytes[i]);
            }
        }
        fail("should be a corrupted index");
    } catch (CorruptIndexException | IndexFormatTooOldException | IndexFormatTooNewException ex) {
    // ok
    }
    IOUtils.close(indexInput, verifyingOutput, dir);
}