Search in sources :

Example 16 with StandardTocReader

use of org.apache.nifi.provenance.toc.StandardTocReader in project nifi by apache.

the class TestEventIdFirstSchemaRecordReaderWriter method testContentClaimUnchanged.

@Test
public void testContentClaimUnchanged() throws IOException {
    final File journalFile = new File("target/storage/" + UUID.randomUUID().toString() + "/testSimpleWrite.gz");
    final File tocFile = TocUtil.getTocFile(journalFile);
    final TocWriter tocWriter = new StandardTocWriter(tocFile, false, false);
    final RecordWriter writer = createWriter(journalFile, tocWriter, true, 8192);
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("filename", "1.txt");
    attributes.put("uuid", UUID.randomUUID().toString());
    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(TestUtil.createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");
    builder.setPreviousContentClaim("container-1", "section-1", "identifier-1", 1L, 1L);
    builder.setCurrentContentClaim("container-1", "section-1", "identifier-1", 1L, 1L);
    final ProvenanceEventRecord record = builder.build();
    writer.writeHeader(1L);
    writer.writeRecord(record);
    writer.close();
    final TocReader tocReader = new StandardTocReader(tocFile);
    try (final FileInputStream fis = new FileInputStream(journalFile);
        final RecordReader reader = createReader(fis, journalFile.getName(), tocReader, 2048)) {
        assertEquals(0, reader.getBlockIndex());
        reader.skipToBlock(0);
        final StandardProvenanceEventRecord recovered = reader.nextRecord();
        assertNotNull(recovered);
        assertEquals("nifi://unit-test", recovered.getTransitUri());
        assertEquals("container-1", recovered.getPreviousContentClaimContainer());
        assertEquals("container-1", recovered.getContentClaimContainer());
        assertEquals("section-1", recovered.getPreviousContentClaimSection());
        assertEquals("section-1", recovered.getContentClaimSection());
        assertEquals("identifier-1", recovered.getPreviousContentClaimIdentifier());
        assertEquals("identifier-1", recovered.getContentClaimIdentifier());
        assertEquals(1L, recovered.getPreviousContentClaimOffset().longValue());
        assertEquals(1L, recovered.getContentClaimOffset().longValue());
        assertEquals(1L, recovered.getPreviousFileSize().longValue());
        assertEquals(1L, recovered.getContentClaimOffset().longValue());
        assertNull(reader.nextRecord());
    }
    FileUtils.deleteFile(journalFile.getParentFile(), true);
}
Also used : TocReader(org.apache.nifi.provenance.toc.TocReader) StandardTocReader(org.apache.nifi.provenance.toc.StandardTocReader) HashMap(java.util.HashMap) StandardTocReader(org.apache.nifi.provenance.toc.StandardTocReader) RecordReader(org.apache.nifi.provenance.serialization.RecordReader) FileInputStream(java.io.FileInputStream) StandardTocWriter(org.apache.nifi.provenance.toc.StandardTocWriter) RecordWriter(org.apache.nifi.provenance.serialization.RecordWriter) StandardTocWriter(org.apache.nifi.provenance.toc.StandardTocWriter) TocWriter(org.apache.nifi.provenance.toc.TocWriter) File(java.io.File) Test(org.junit.Test)

Example 17 with StandardTocReader

use of org.apache.nifi.provenance.toc.StandardTocReader in project nifi by apache.

the class TestSchemaRecordReaderWriter method testFieldAddedToSchema.

@Test
public void testFieldAddedToSchema() throws IOException {
    final RecordField unitTestField = new SimpleRecordField("Unit Test Field", FieldType.STRING, Repetition.EXACTLY_ONE);
    final Consumer<List<RecordField>> schemaModifier = fields -> fields.add(unitTestField);
    final Map<RecordField, Object> toAdd = new HashMap<>();
    toAdd.put(unitTestField, "hello");
    try (final ByteArraySchemaRecordWriter writer = createSchemaWriter(schemaModifier, toAdd)) {
        writer.writeHeader(1L);
        writer.writeRecord(createEvent());
        writer.writeRecord(createEvent());
    }
    try (final InputStream in = new FileInputStream(journalFile);
        final TocReader tocReader = new StandardTocReader(tocFile);
        final RecordReader reader = createReader(in, journalFile.getName(), tocReader, 10000)) {
        for (int i = 0; i < 2; i++) {
            final StandardProvenanceEventRecord event = reader.nextRecord();
            assertNotNull(event);
            assertEquals("1234", event.getComponentId());
            assertEquals(ProvenanceEventType.RECEIVE, event.getEventType());
            assertNotNull(event.getUpdatedAttributes());
            assertFalse(event.getUpdatedAttributes().isEmpty());
        }
    }
}
Also used : StandardTocWriter(org.apache.nifi.provenance.toc.StandardTocWriter) TocReader(org.apache.nifi.provenance.toc.TocReader) Record(org.apache.nifi.repository.schema.Record) ByteArrayOutputStream(java.io.ByteArrayOutputStream) HashMap(java.util.HashMap) Callable(java.util.concurrent.Callable) RecordWriter(org.apache.nifi.provenance.serialization.RecordWriter) ArrayList(java.util.ArrayList) NopTocWriter(org.apache.nifi.provenance.toc.NopTocWriter) RecordSchema(org.apache.nifi.repository.schema.RecordSchema) StandardTocReader(org.apache.nifi.provenance.toc.StandardTocReader) TocWriter(org.apache.nifi.provenance.toc.TocWriter) FieldType(org.apache.nifi.repository.schema.FieldType) DataOutputStream(java.io.DataOutputStream) Map(java.util.Map) TocUtil(org.apache.nifi.provenance.toc.TocUtil) Repetition(org.apache.nifi.repository.schema.Repetition) Before(org.junit.Before) OutputStream(java.io.OutputStream) NullOutputStream(org.apache.nifi.stream.io.NullOutputStream) Assert.assertNotNull(org.junit.Assert.assertNotNull) EventFieldNames(org.apache.nifi.provenance.schema.EventFieldNames) RecordField(org.apache.nifi.repository.schema.RecordField) Assert.assertTrue(org.junit.Assert.assertTrue) IOException(java.io.IOException) Test(org.junit.Test) FileInputStream(java.io.FileInputStream) UUID(java.util.UUID) File(java.io.File) FieldMapRecord(org.apache.nifi.repository.schema.FieldMapRecord) TimeUnit(java.util.concurrent.TimeUnit) Consumer(java.util.function.Consumer) AtomicLong(java.util.concurrent.atomic.AtomicLong) List(java.util.List) Assert.assertNull(org.junit.Assert.assertNull) EventRecord(org.apache.nifi.provenance.schema.EventRecord) Ignore(org.junit.Ignore) Assert.assertFalse(org.junit.Assert.assertFalse) ProvenanceEventSchema(org.apache.nifi.provenance.schema.ProvenanceEventSchema) RecordReader(org.apache.nifi.provenance.serialization.RecordReader) SimpleRecordField(org.apache.nifi.repository.schema.SimpleRecordField) Assert(org.junit.Assert) Assert.assertEquals(org.junit.Assert.assertEquals) InputStream(java.io.InputStream) TocReader(org.apache.nifi.provenance.toc.TocReader) StandardTocReader(org.apache.nifi.provenance.toc.StandardTocReader) RecordField(org.apache.nifi.repository.schema.RecordField) SimpleRecordField(org.apache.nifi.repository.schema.SimpleRecordField) HashMap(java.util.HashMap) StandardTocReader(org.apache.nifi.provenance.toc.StandardTocReader) FileInputStream(java.io.FileInputStream) InputStream(java.io.InputStream) RecordReader(org.apache.nifi.provenance.serialization.RecordReader) FileInputStream(java.io.FileInputStream) SimpleRecordField(org.apache.nifi.repository.schema.SimpleRecordField) ArrayList(java.util.ArrayList) List(java.util.List) Test(org.junit.Test)

Example 18 with StandardTocReader

use of org.apache.nifi.provenance.toc.StandardTocReader in project nifi by apache.

the class EventFileCompressor method run.

@Override
public void run() {
    while (!shutdown) {
        File uncompressedEventFile = null;
        try {
            final long start = System.nanoTime();
            uncompressedEventFile = filesToCompress.poll(1, TimeUnit.SECONDS);
            if (uncompressedEventFile == null || shutdown) {
                continue;
            }
            File outputFile = null;
            long bytesBefore = 0L;
            StandardTocReader tocReader = null;
            File tmpTocFile = null;
            eventFileManager.obtainReadLock(uncompressedEventFile);
            try {
                StandardTocWriter tocWriter = null;
                final File tocFile = TocUtil.getTocFile(uncompressedEventFile);
                try {
                    tocReader = new StandardTocReader(tocFile);
                } catch (final IOException e) {
                    logger.error("Failed to read TOC File {}", tocFile, e);
                    continue;
                }
                bytesBefore = uncompressedEventFile.length();
                try {
                    outputFile = new File(uncompressedEventFile.getParentFile(), uncompressedEventFile.getName() + ".gz");
                    try {
                        tmpTocFile = new File(tocFile.getParentFile(), tocFile.getName() + ".tmp");
                        tocWriter = new StandardTocWriter(tmpTocFile, true, false);
                        compress(uncompressedEventFile, tocReader, outputFile, tocWriter);
                        tocWriter.close();
                    } catch (final IOException ioe) {
                        logger.error("Failed to compress {} on rollover", uncompressedEventFile, ioe);
                    }
                } finally {
                    CloseableUtil.closeQuietly(tocReader, tocWriter);
                }
            } finally {
                eventFileManager.releaseReadLock(uncompressedEventFile);
            }
            eventFileManager.obtainWriteLock(uncompressedEventFile);
            try {
                // Attempt to delete the input file and associated toc file
                if (uncompressedEventFile.delete()) {
                    if (tocReader != null) {
                        final File tocFile = tocReader.getFile();
                        if (!tocFile.delete()) {
                            logger.warn("Failed to delete {}; this file should be cleaned up manually", tocFile);
                        }
                        if (tmpTocFile != null) {
                            tmpTocFile.renameTo(tocFile);
                        }
                    }
                } else {
                    logger.warn("Failed to delete {}; this file should be cleaned up manually", uncompressedEventFile);
                }
            } finally {
                eventFileManager.releaseWriteLock(uncompressedEventFile);
            }
            final long millis = TimeUnit.NANOSECONDS.toMillis(System.nanoTime() - start);
            final long bytesAfter = outputFile.length();
            final double reduction = 100 * (1 - (double) bytesAfter / (double) bytesBefore);
            final String reductionTwoDecimals = String.format("%.2f", reduction);
            logger.debug("Successfully compressed Provenance Event File {} in {} millis from {} to {}, a reduction of {}%", uncompressedEventFile, millis, FormatUtils.formatDataSize(bytesBefore), FormatUtils.formatDataSize(bytesAfter), reductionTwoDecimals);
        } catch (final InterruptedException e) {
            Thread.currentThread().interrupt();
            return;
        } catch (final Exception e) {
            logger.error("Failed to compress {}", uncompressedEventFile, e);
        }
    }
}
Also used : StandardTocWriter(org.apache.nifi.provenance.toc.StandardTocWriter) StandardTocReader(org.apache.nifi.provenance.toc.StandardTocReader) IOException(java.io.IOException) File(java.io.File) IOException(java.io.IOException)

Aggregations

StandardTocReader (org.apache.nifi.provenance.toc.StandardTocReader)18 File (java.io.File)17 StandardTocWriter (org.apache.nifi.provenance.toc.StandardTocWriter)17 Test (org.junit.Test)16 FileInputStream (java.io.FileInputStream)15 RecordWriter (org.apache.nifi.provenance.serialization.RecordWriter)15 TocReader (org.apache.nifi.provenance.toc.TocReader)15 RecordReader (org.apache.nifi.provenance.serialization.RecordReader)14 TocWriter (org.apache.nifi.provenance.toc.TocWriter)14 HashMap (java.util.HashMap)8 InputStream (java.io.InputStream)6 IOException (java.io.IOException)4 ArrayList (java.util.ArrayList)4 ByteArrayOutputStream (java.io.ByteArrayOutputStream)3 DataOutputStream (java.io.DataOutputStream)3 EventRecord (org.apache.nifi.provenance.schema.EventRecord)3 NopTocWriter (org.apache.nifi.provenance.toc.NopTocWriter)3 RecordField (org.apache.nifi.repository.schema.RecordField)3 RecordSchema (org.apache.nifi.repository.schema.RecordSchema)3 SimpleRecordField (org.apache.nifi.repository.schema.SimpleRecordField)3