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

Example 16 with ChecksumIndexInput

use of org.apache.lucene.store.ChecksumIndexInput in project crate by crate.

the class CorruptionUtils method corruptFile.

/**
 * Corrupts a random file at a random position
 */
public static void corruptFile(Random random, Path... files) throws IOException {
    assertTrue("files must be non-empty", files.length > 0);
    final Path fileToCorrupt = RandomPicks.randomFrom(random, files);
    assertTrue(fileToCorrupt + " is not a file", Files.isRegularFile(fileToCorrupt));
    try (Directory dir = FSDirectory.open(fileToCorrupt.toAbsolutePath().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)) {
            long maxPosition = raf.size();
            final int position = random.nextInt((int) Math.min(Integer.MAX_VALUE, maxPosition));
            corruptAt(fileToCorrupt, raf, position);
        }
        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();
        }
        // we need to add assumptions here that the checksums actually really don't match there is a small chance to get collisions
        // in the checksum which is ok though....
        StringBuilder msg = new StringBuilder();
        msg.append("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()).append(" length: ");
        msg.append(dir.fileLength(fileToCorrupt.getFileName().toString()));
        logger.info("Checksum {}", msg);
        assumeTrue("Checksum collision - " + msg.toString(), // collision
        checksumAfterCorruption != checksumBeforeCorruption || // checksum corrupted
        actualChecksumAfterCorruption != checksumBeforeCorruption);
        assertThat("no file corrupted", fileToCorrupt, notNullValue());
    }
}

Example 17 with ChecksumIndexInput

use of org.apache.lucene.store.ChecksumIndexInput in project crate by crate.

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 18 with ChecksumIndexInput

use of org.apache.lucene.store.ChecksumIndexInput in project lucene-solr by apache.

the class SimpleTextFieldsReader method readFields.

private TreeMap<String, Long> readFields(IndexInput in) throws IOException {
    ChecksumIndexInput input = new BufferedChecksumIndexInput(in);
    BytesRefBuilder scratch = new BytesRefBuilder();
    TreeMap<String, Long> fields = new TreeMap<>();
    while (true) {
        SimpleTextUtil.readLine(input, scratch);
        if (scratch.get().equals(END)) {
            SimpleTextUtil.checkFooter(input);
            return fields;
        } else if (StringHelper.startsWith(scratch.get(), FIELD)) {
            String fieldName = new String(scratch.bytes(), FIELD.length, scratch.length() - FIELD.length, StandardCharsets.UTF_8);
            fields.put(fieldName, input.getFilePointer());
        }
    }
}

Example 19 with ChecksumIndexInput

use of org.apache.lucene.store.ChecksumIndexInput in project lucene-solr by apache.

the class SimpleTextSegmentInfoFormat method read.

@Override
public SegmentInfo read(Directory directory, String segmentName, byte[] segmentID, IOContext context) throws IOException {
    BytesRefBuilder scratch = new BytesRefBuilder();
    String segFileName = IndexFileNames.segmentFileName(segmentName, "", SimpleTextSegmentInfoFormat.SI_EXTENSION);
    try (ChecksumIndexInput input = directory.openChecksumInput(segFileName, context)) {
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_VERSION);
        final Version version;
        try {
            version = Version.parse(readString(SI_VERSION.length, scratch));
        } catch (ParseException pe) {
            throw new CorruptIndexException("unable to parse version string: " + pe.getMessage(), input, pe);
        }
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_MIN_VERSION);
        Version minVersion;
        try {
            String versionString = readString(SI_MIN_VERSION.length, scratch);
            if (versionString.equals("null")) {
                minVersion = null;
            } else {
                minVersion = Version.parse(versionString);
            }
        } catch (ParseException pe) {
            throw new CorruptIndexException("unable to parse version string: " + pe.getMessage(), input, pe);
        }
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_DOCCOUNT);
        final int docCount = Integer.parseInt(readString(SI_DOCCOUNT.length, scratch));
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_USECOMPOUND);
        final boolean isCompoundFile = Boolean.parseBoolean(readString(SI_USECOMPOUND.length, scratch));
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_NUM_DIAG);
        int numDiag = Integer.parseInt(readString(SI_NUM_DIAG.length, scratch));
        Map<String, String> diagnostics = new HashMap<>();
        for (int i = 0; i < numDiag; i++) {
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_DIAG_KEY);
            String key = readString(SI_DIAG_KEY.length, scratch);
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_DIAG_VALUE);
            String value = readString(SI_DIAG_VALUE.length, scratch);
            diagnostics.put(key, value);
        }
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_NUM_ATT);
        int numAtt = Integer.parseInt(readString(SI_NUM_ATT.length, scratch));
        Map<String, String> attributes = new HashMap<>(numAtt);
        for (int i = 0; i < numAtt; i++) {
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_ATT_KEY);
            String key = readString(SI_ATT_KEY.length, scratch);
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_ATT_VALUE);
            String value = readString(SI_ATT_VALUE.length, scratch);
            attributes.put(key, value);
        }
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_NUM_FILES);
        int numFiles = Integer.parseInt(readString(SI_NUM_FILES.length, scratch));
        Set<String> files = new HashSet<>();
        for (int i = 0; i < numFiles; i++) {
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_FILE);
            String fileName = readString(SI_FILE.length, scratch);
            files.add(fileName);
        }
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_ID);
        final byte[] id = Arrays.copyOfRange(scratch.bytes(), SI_ID.length, scratch.length());
        if (!Arrays.equals(segmentID, id)) {
            throw new CorruptIndexException("file mismatch, expected: " + StringHelper.idToString(segmentID) + ", got: " + StringHelper.idToString(id), input);
        }
        SimpleTextUtil.readLine(input, scratch);
        assert StringHelper.startsWith(scratch.get(), SI_SORT);
        final int numSortFields = Integer.parseInt(readString(SI_SORT.length, scratch));
        SortField[] sortField = new SortField[numSortFields];
        for (int i = 0; i < numSortFields; ++i) {
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_SORT_FIELD);
            final String field = readString(SI_SORT_FIELD.length, scratch);
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_SORT_TYPE);
            final String typeAsString = readString(SI_SORT_TYPE.length, scratch);
            final SortField.Type type;
            SortedSetSelector.Type selectorSet = null;
            SortedNumericSelector.Type selectorNumeric = null;
            switch(typeAsString) {
                case "string":
                    type = SortField.Type.STRING;
                    break;
                case "long":
                    type = SortField.Type.LONG;
                    break;
                case "int":
                    type = SortField.Type.INT;
                    break;
                case "double":
                    type = SortField.Type.DOUBLE;
                    break;
                case "float":
                    type = SortField.Type.FLOAT;
                    break;
                case "multi_valued_string":
                    type = SortField.Type.STRING;
                    selectorSet = readSetSelector(input, scratch);
                    break;
                case "multi_valued_long":
                    type = SortField.Type.LONG;
                    selectorNumeric = readNumericSelector(input, scratch);
                    break;
                case "multi_valued_int":
                    type = SortField.Type.INT;
                    selectorNumeric = readNumericSelector(input, scratch);
                    break;
                case "multi_valued_double":
                    type = SortField.Type.DOUBLE;
                    selectorNumeric = readNumericSelector(input, scratch);
                    break;
                case "multi_valued_float":
                    type = SortField.Type.FLOAT;
                    selectorNumeric = readNumericSelector(input, scratch);
                    break;
                default:
                    throw new CorruptIndexException("unable to parse sort type string: " + typeAsString, input);
            }
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_SORT_REVERSE);
            final boolean reverse = Boolean.parseBoolean(readString(SI_SORT_REVERSE.length, scratch));
            SimpleTextUtil.readLine(input, scratch);
            assert StringHelper.startsWith(scratch.get(), SI_SORT_MISSING);
            final String missingLastAsString = readString(SI_SORT_MISSING.length, scratch);
            final Object missingValue;
            switch(type) {
                case STRING:
                    switch(missingLastAsString) {
                        case "null":
                            missingValue = null;
                            break;
                        case "first":
                            missingValue = SortField.STRING_FIRST;
                            break;
                        case "last":
                            missingValue = SortField.STRING_LAST;
                            break;
                        default:
                            throw new CorruptIndexException("unable to parse missing string: " + typeAsString, input);
                    }
                    break;
                case LONG:
                    switch(missingLastAsString) {
                        case "null":
                            missingValue = null;
                            break;
                        default:
                            missingValue = Long.parseLong(missingLastAsString);
                            break;
                    }
                    break;
                case INT:
                    switch(missingLastAsString) {
                        case "null":
                            missingValue = null;
                            break;
                        default:
                            missingValue = Integer.parseInt(missingLastAsString);
                            break;
                    }
                    break;
                case DOUBLE:
                    switch(missingLastAsString) {
                        case "null":
                            missingValue = null;
                            break;
                        default:
                            missingValue = Double.parseDouble(missingLastAsString);
                            break;
                    }
                    break;
                case FLOAT:
                    switch(missingLastAsString) {
                        case "null":
                            missingValue = null;
                            break;
                        default:
                            missingValue = Float.parseFloat(missingLastAsString);
                            break;
                    }
                    break;
                default:
                    throw new AssertionError();
            }
            if (selectorSet != null) {
                sortField[i] = new SortedSetSortField(field, reverse);
            } else if (selectorNumeric != null) {
                sortField[i] = new SortedNumericSortField(field, type, reverse);
            } else {
                sortField[i] = new SortField(field, type, reverse);
            }
            if (missingValue != null) {
                sortField[i].setMissingValue(missingValue);
            }
        }
        Sort indexSort = sortField.length == 0 ? null : new Sort(sortField);
        SimpleTextUtil.checkFooter(input);
        SegmentInfo info = new SegmentInfo(directory, version, minVersion, segmentName, docCount, isCompoundFile, null, Collections.unmodifiableMap(diagnostics), id, Collections.unmodifiableMap(attributes), indexSort);
        info.setFiles(files);
        return info;
    }
}

Example 20 with ChecksumIndexInput

use of org.apache.lucene.store.ChecksumIndexInput in project lucene-solr by apache.

the class SimpleTextStoredFieldsReader method readIndex.

// we don't actually write a .fdx-like index, instead we read the 
// stored fields file in entirety up-front and save the offsets 
// so we can seek to the documents later.
private void readIndex(int size) throws IOException {
    ChecksumIndexInput input = new BufferedChecksumIndexInput(in);
    offsets = new long[size];
    int upto = 0;
    while (!scratch.get().equals(END)) {
        SimpleTextUtil.readLine(input, scratch);
        if (StringHelper.startsWith(scratch.get(), DOC)) {
            offsets[upto] = input.getFilePointer();
            upto++;
        }
    }
    SimpleTextUtil.checkFooter(input);
    assert upto == offsets.length;
}