Search in sources :

Example 11 with AssertingCodec

use of org.apache.lucene.codecs.asserting.AssertingCodec in project lucene-solr by apache.

the class TestRuleSetupAndRestoreClassEnv method before.

@Override
protected void before() throws Exception {
    // we do this in beforeClass, because some tests currently disable it
    if (System.getProperty("solr.directoryFactory") == null) {
        System.setProperty("solr.directoryFactory", "org.apache.solr.core.MockDirectoryFactory");
    }
    // if verbose: print some debugging stuff about which codecs are loaded.
    if (VERBOSE) {
        System.out.println("Loaded codecs: " + Codec.availableCodecs());
        System.out.println("Loaded postingsFormats: " + PostingsFormat.availablePostingsFormats());
    }
    savedInfoStream = InfoStream.getDefault();
    final Random random = RandomizedContext.current().getRandom();
    final boolean v = random.nextBoolean();
    if (INFOSTREAM) {
        InfoStream.setDefault(new ThreadNameFixingPrintStreamInfoStream(System.out));
    } else if (v) {
        InfoStream.setDefault(new NullInfoStream());
    }
    Class<?> targetClass = RandomizedContext.current().getTargetClass();
    avoidCodecs = new HashSet<>();
    if (targetClass.isAnnotationPresent(SuppressCodecs.class)) {
        SuppressCodecs a = targetClass.getAnnotation(SuppressCodecs.class);
        avoidCodecs.addAll(Arrays.asList(a.value()));
    }
    savedCodec = Codec.getDefault();
    int randomVal = random.nextInt(11);
    if ("default".equals(TEST_CODEC)) {
        // just use the default, don't randomize
        codec = savedCodec;
    } else if (("random".equals(TEST_POSTINGSFORMAT) == false) || ("random".equals(TEST_DOCVALUESFORMAT) == false)) {
        // the user wired postings or DV: this is messy
        // refactor into RandomCodec....
        final PostingsFormat format;
        if ("random".equals(TEST_POSTINGSFORMAT)) {
            format = new AssertingPostingsFormat();
        } else if ("MockRandom".equals(TEST_POSTINGSFORMAT)) {
            format = new MockRandomPostingsFormat(new Random(random.nextLong()));
        } else {
            format = PostingsFormat.forName(TEST_POSTINGSFORMAT);
        }
        final DocValuesFormat dvFormat;
        if ("random".equals(TEST_DOCVALUESFORMAT)) {
            dvFormat = new AssertingDocValuesFormat();
        } else {
            dvFormat = DocValuesFormat.forName(TEST_DOCVALUESFORMAT);
        }
        codec = new AssertingCodec() {

            @Override
            public PostingsFormat getPostingsFormatForField(String field) {
                return format;
            }

            @Override
            public DocValuesFormat getDocValuesFormatForField(String field) {
                return dvFormat;
            }

            @Override
            public String toString() {
                return super.toString() + ": " + format.toString() + ", " + dvFormat.toString();
            }
        };
    } else if ("SimpleText".equals(TEST_CODEC) || ("random".equals(TEST_CODEC) && randomVal == 9 && LuceneTestCase.rarely(random) && !shouldAvoidCodec("SimpleText"))) {
        codec = new SimpleTextCodec();
    } else if ("CheapBastard".equals(TEST_CODEC) || ("random".equals(TEST_CODEC) && randomVal == 8 && !shouldAvoidCodec("CheapBastard") && !shouldAvoidCodec("Lucene41"))) {
        // we also avoid this codec if Lucene41 is avoided, since thats the postings format it uses.
        codec = new CheapBastardCodec();
    } else if ("Asserting".equals(TEST_CODEC) || ("random".equals(TEST_CODEC) && randomVal == 7 && !shouldAvoidCodec("Asserting"))) {
        codec = new AssertingCodec();
    } else if ("Compressing".equals(TEST_CODEC) || ("random".equals(TEST_CODEC) && randomVal == 6 && !shouldAvoidCodec("Compressing"))) {
        codec = CompressingCodec.randomInstance(random);
    } else if ("Lucene70".equals(TEST_CODEC) || ("random".equals(TEST_CODEC) && randomVal == 5 && !shouldAvoidCodec("Lucene70"))) {
        codec = new Lucene70Codec(RandomPicks.randomFrom(random, Lucene50StoredFieldsFormat.Mode.values()));
    } else if (!"random".equals(TEST_CODEC)) {
        codec = Codec.forName(TEST_CODEC);
    } else if ("random".equals(TEST_POSTINGSFORMAT)) {
        codec = new RandomCodec(random, avoidCodecs);
    } else {
        assert false;
    }
    Codec.setDefault(codec);
    // Initialize locale/ timezone.
    String testLocale = System.getProperty("tests.locale", "random");
    String testTimeZone = System.getProperty("tests.timezone", "random");
    // Always pick a random one for consistency (whether tests.locale was specified or not).
    savedLocale = Locale.getDefault();
    Locale randomLocale = randomLocale(random);
    locale = testLocale.equals("random") ? randomLocale : localeForLanguageTag(testLocale);
    Locale.setDefault(locale);
    savedTimeZone = TimeZone.getDefault();
    TimeZone randomTimeZone = randomTimeZone(random());
    timeZone = testTimeZone.equals("random") ? randomTimeZone : TimeZone.getTimeZone(testTimeZone);
    TimeZone.setDefault(timeZone);
    similarity = new RandomSimilarity(random());
    // Check codec restrictions once at class level.
    try {
        checkCodecRestrictions(codec);
    } catch (AssumptionViolatedException e) {
        System.err.println("NOTE: " + e.getMessage() + " Suppressed codecs: " + Arrays.toString(avoidCodecs.toArray()));
        throw e;
    }
    // We have "stickiness" so that sometimes all we do is vary the RAM buffer size, other times just the doc count to flush by, else both.
    // This way the assertMemory in DocumentsWriterFlushControl sometimes runs (when we always flush by RAM).
    LiveIWCFlushMode flushMode;
    switch(random().nextInt(3)) {
        case 0:
            flushMode = LiveIWCFlushMode.BY_RAM;
            break;
        case 1:
            flushMode = LiveIWCFlushMode.BY_DOCS;
            break;
        case 2:
            flushMode = LiveIWCFlushMode.EITHER;
            break;
        default:
            throw new AssertionError();
    }
    LuceneTestCase.setLiveIWCFlushMode(flushMode);
    initialized = true;
}
Also used : Locale(java.util.Locale) LuceneTestCase.randomLocale(org.apache.lucene.util.LuceneTestCase.randomLocale) AssumptionViolatedException(org.junit.internal.AssumptionViolatedException) CheapBastardCodec(org.apache.lucene.codecs.cheapbastard.CheapBastardCodec) DocValuesFormat(org.apache.lucene.codecs.DocValuesFormat) AssertingDocValuesFormat(org.apache.lucene.codecs.asserting.AssertingDocValuesFormat) AssertingPostingsFormat(org.apache.lucene.codecs.asserting.AssertingPostingsFormat) Random(java.util.Random) AssertingDocValuesFormat(org.apache.lucene.codecs.asserting.AssertingDocValuesFormat) RandomCodec(org.apache.lucene.index.RandomCodec) AssertingCodec(org.apache.lucene.codecs.asserting.AssertingCodec) LiveIWCFlushMode(org.apache.lucene.util.LuceneTestCase.LiveIWCFlushMode) RandomSimilarity(org.apache.lucene.search.similarities.RandomSimilarity) SimpleTextCodec(org.apache.lucene.codecs.simpletext.SimpleTextCodec) SuppressCodecs(org.apache.lucene.util.LuceneTestCase.SuppressCodecs) MockRandomPostingsFormat(org.apache.lucene.codecs.mockrandom.MockRandomPostingsFormat) LuceneTestCase.randomTimeZone(org.apache.lucene.util.LuceneTestCase.randomTimeZone) TimeZone(java.util.TimeZone) AssertingPostingsFormat(org.apache.lucene.codecs.asserting.AssertingPostingsFormat) MockRandomPostingsFormat(org.apache.lucene.codecs.mockrandom.MockRandomPostingsFormat) PostingsFormat(org.apache.lucene.codecs.PostingsFormat) Lucene70Codec(org.apache.lucene.codecs.lucene70.Lucene70Codec)

Example 12 with AssertingCodec

use of org.apache.lucene.codecs.asserting.AssertingCodec in project lucene-solr by apache.

the class TestIndexWriterExceptions2 method testBasics.

// just one thread, serial merge policy, hopefully debuggable
public void testBasics() throws Exception {
    // disable slow things: we don't rely upon sleeps here.
    Directory dir = newDirectory();
    if (dir instanceof MockDirectoryWrapper) {
        ((MockDirectoryWrapper) dir).setThrottling(MockDirectoryWrapper.Throttling.NEVER);
        ((MockDirectoryWrapper) dir).setUseSlowOpenClosers(false);
    }
    // log all exceptions we hit, in case we fail (for debugging)
    ByteArrayOutputStream exceptionLog = new ByteArrayOutputStream();
    PrintStream exceptionStream = new PrintStream(exceptionLog, true, "UTF-8");
    //PrintStream exceptionStream = System.out;
    // create lots of non-aborting exceptions with a broken analyzer
    final long analyzerSeed = random().nextLong();
    Analyzer analyzer = new Analyzer() {

        @Override
        protected TokenStreamComponents createComponents(String fieldName) {
            MockTokenizer tokenizer = new MockTokenizer(MockTokenizer.SIMPLE, false);
            // TODO: can we turn this on? our filter is probably too evil
            tokenizer.setEnableChecks(false);
            TokenStream stream = tokenizer;
            // emit some payloads
            if (fieldName.contains("payloads")) {
                stream = new MockVariableLengthPayloadFilter(new Random(analyzerSeed), stream);
            }
            stream = new CrankyTokenFilter(stream, new Random(analyzerSeed));
            return new TokenStreamComponents(tokenizer, stream);
        }
    };
    // create lots of aborting exceptions with a broken codec
    // we don't need a random codec, as we aren't trying to find bugs in the codec here.
    Codec inner = RANDOM_MULTIPLIER > 1 ? Codec.getDefault() : new AssertingCodec();
    Codec codec = new CrankyCodec(inner, new Random(random().nextLong()));
    IndexWriterConfig conf = newIndexWriterConfig(analyzer);
    // just for now, try to keep this test reproducible
    conf.setMergeScheduler(new SerialMergeScheduler());
    conf.setCodec(codec);
    int numDocs = atLeast(500);
    IndexWriter iw = new IndexWriter(dir, conf);
    try {
        boolean allowAlreadyClosed = false;
        for (int i = 0; i < numDocs; i++) {
            // TODO: add crankyDocValuesFields, etc
            Document doc = new Document();
            doc.add(newStringField("id", Integer.toString(i), Field.Store.NO));
            doc.add(new NumericDocValuesField("dv", i));
            doc.add(new BinaryDocValuesField("dv2", new BytesRef(Integer.toString(i))));
            doc.add(new SortedDocValuesField("dv3", new BytesRef(Integer.toString(i))));
            doc.add(new SortedSetDocValuesField("dv4", new BytesRef(Integer.toString(i))));
            doc.add(new SortedSetDocValuesField("dv4", new BytesRef(Integer.toString(i - 1))));
            doc.add(new SortedNumericDocValuesField("dv5", i));
            doc.add(new SortedNumericDocValuesField("dv5", i - 1));
            doc.add(newTextField("text1", TestUtil.randomAnalysisString(random(), 20, true), Field.Store.NO));
            // ensure we store something
            doc.add(new StoredField("stored1", "foo"));
            doc.add(new StoredField("stored1", "bar"));
            // ensure we get some payloads
            doc.add(newTextField("text_payloads", TestUtil.randomAnalysisString(random(), 6, true), Field.Store.NO));
            // ensure we get some vectors
            FieldType ft = new FieldType(TextField.TYPE_NOT_STORED);
            ft.setStoreTermVectors(true);
            doc.add(newField("text_vectors", TestUtil.randomAnalysisString(random(), 6, true), ft));
            doc.add(new IntPoint("point", random().nextInt()));
            doc.add(new IntPoint("point2d", random().nextInt(), random().nextInt()));
            if (random().nextInt(10) > 0) {
                // single doc
                try {
                    iw.addDocument(doc);
                    // we made it, sometimes delete our doc, or update a dv
                    int thingToDo = random().nextInt(4);
                    if (thingToDo == 0) {
                        iw.deleteDocuments(new Term("id", Integer.toString(i)));
                    } else if (thingToDo == 1) {
                        iw.updateNumericDocValue(new Term("id", Integer.toString(i)), "dv", i + 1L);
                    } else if (thingToDo == 2) {
                        iw.updateBinaryDocValue(new Term("id", Integer.toString(i)), "dv2", new BytesRef(Integer.toString(i + 1)));
                    }
                } catch (AlreadyClosedException ace) {
                    // OK: writer was closed by abort; we just reopen now:
                    assertTrue(iw.deleter.isClosed());
                    assertTrue(allowAlreadyClosed);
                    allowAlreadyClosed = false;
                    conf = newIndexWriterConfig(analyzer);
                    // just for now, try to keep this test reproducible
                    conf.setMergeScheduler(new SerialMergeScheduler());
                    conf.setCodec(codec);
                    iw = new IndexWriter(dir, conf);
                } catch (Exception e) {
                    if (e.getMessage() != null && e.getMessage().startsWith("Fake IOException")) {
                        exceptionStream.println("\nTEST: got expected fake exc:" + e.getMessage());
                        e.printStackTrace(exceptionStream);
                        allowAlreadyClosed = true;
                    } else {
                        Rethrow.rethrow(e);
                    }
                }
            } else {
                // block docs
                Document doc2 = new Document();
                doc2.add(newStringField("id", Integer.toString(-i), Field.Store.NO));
                doc2.add(newTextField("text1", TestUtil.randomAnalysisString(random(), 20, true), Field.Store.NO));
                doc2.add(new StoredField("stored1", "foo"));
                doc2.add(new StoredField("stored1", "bar"));
                doc2.add(newField("text_vectors", TestUtil.randomAnalysisString(random(), 6, true), ft));
                try {
                    iw.addDocuments(Arrays.asList(doc, doc2));
                    // we made it, sometimes delete our docs
                    if (random().nextBoolean()) {
                        iw.deleteDocuments(new Term("id", Integer.toString(i)), new Term("id", Integer.toString(-i)));
                    }
                } catch (AlreadyClosedException ace) {
                    // OK: writer was closed by abort; we just reopen now:
                    assertTrue(iw.deleter.isClosed());
                    assertTrue(allowAlreadyClosed);
                    allowAlreadyClosed = false;
                    conf = newIndexWriterConfig(analyzer);
                    // just for now, try to keep this test reproducible
                    conf.setMergeScheduler(new SerialMergeScheduler());
                    conf.setCodec(codec);
                    iw = new IndexWriter(dir, conf);
                } catch (Exception e) {
                    if (e.getMessage() != null && e.getMessage().startsWith("Fake IOException")) {
                        exceptionStream.println("\nTEST: got expected fake exc:" + e.getMessage());
                        e.printStackTrace(exceptionStream);
                        allowAlreadyClosed = true;
                    } else {
                        Rethrow.rethrow(e);
                    }
                }
            }
            if (random().nextInt(10) == 0) {
                // trigger flush:
                try {
                    if (random().nextBoolean()) {
                        DirectoryReader ir = null;
                        try {
                            ir = DirectoryReader.open(iw, random().nextBoolean(), false);
                            TestUtil.checkReader(ir);
                        } finally {
                            IOUtils.closeWhileHandlingException(ir);
                        }
                    } else {
                        iw.commit();
                    }
                    if (DirectoryReader.indexExists(dir)) {
                        TestUtil.checkIndex(dir);
                    }
                } catch (AlreadyClosedException ace) {
                    // OK: writer was closed by abort; we just reopen now:
                    assertTrue(iw.deleter.isClosed());
                    assertTrue(allowAlreadyClosed);
                    allowAlreadyClosed = false;
                    conf = newIndexWriterConfig(analyzer);
                    // just for now, try to keep this test reproducible
                    conf.setMergeScheduler(new SerialMergeScheduler());
                    conf.setCodec(codec);
                    iw = new IndexWriter(dir, conf);
                } catch (Exception e) {
                    if (e.getMessage() != null && e.getMessage().startsWith("Fake IOException")) {
                        exceptionStream.println("\nTEST: got expected fake exc:" + e.getMessage());
                        e.printStackTrace(exceptionStream);
                        allowAlreadyClosed = true;
                    } else {
                        Rethrow.rethrow(e);
                    }
                }
            }
        }
        try {
            iw.close();
        } catch (Exception e) {
            if (e.getMessage() != null && e.getMessage().startsWith("Fake IOException")) {
                exceptionStream.println("\nTEST: got expected fake exc:" + e.getMessage());
                e.printStackTrace(exceptionStream);
                try {
                    iw.rollback();
                } catch (Throwable t) {
                }
            } else {
                Rethrow.rethrow(e);
            }
        }
        dir.close();
    } catch (Throwable t) {
        System.out.println("Unexpected exception: dumping fake-exception-log:...");
        exceptionStream.flush();
        System.out.println(exceptionLog.toString("UTF-8"));
        System.out.flush();
        Rethrow.rethrow(t);
    }
    if (VERBOSE) {
        System.out.println("TEST PASSED: dumping fake-exception-log:...");
        System.out.println(exceptionLog.toString("UTF-8"));
    }
}
Also used : TokenStream(org.apache.lucene.analysis.TokenStream) CrankyTokenFilter(org.apache.lucene.analysis.CrankyTokenFilter) AlreadyClosedException(org.apache.lucene.store.AlreadyClosedException) Analyzer(org.apache.lucene.analysis.Analyzer) Document(org.apache.lucene.document.Document) CrankyCodec(org.apache.lucene.codecs.cranky.CrankyCodec) AssertingCodec(org.apache.lucene.codecs.asserting.AssertingCodec) Codec(org.apache.lucene.codecs.Codec) StoredField(org.apache.lucene.document.StoredField) Random(java.util.Random) SortedNumericDocValuesField(org.apache.lucene.document.SortedNumericDocValuesField) NumericDocValuesField(org.apache.lucene.document.NumericDocValuesField) SortedDocValuesField(org.apache.lucene.document.SortedDocValuesField) CrankyCodec(org.apache.lucene.codecs.cranky.CrankyCodec) BytesRef(org.apache.lucene.util.BytesRef) Directory(org.apache.lucene.store.Directory) MockDirectoryWrapper(org.apache.lucene.store.MockDirectoryWrapper) AssertingCodec(org.apache.lucene.codecs.asserting.AssertingCodec) PrintStream(java.io.PrintStream) ByteArrayOutputStream(java.io.ByteArrayOutputStream) BinaryDocValuesField(org.apache.lucene.document.BinaryDocValuesField) IntPoint(org.apache.lucene.document.IntPoint) AlreadyClosedException(org.apache.lucene.store.AlreadyClosedException) FieldType(org.apache.lucene.document.FieldType) MockTokenizer(org.apache.lucene.analysis.MockTokenizer) IntPoint(org.apache.lucene.document.IntPoint) SortedNumericDocValuesField(org.apache.lucene.document.SortedNumericDocValuesField) SortedSetDocValuesField(org.apache.lucene.document.SortedSetDocValuesField) MockVariableLengthPayloadFilter(org.apache.lucene.analysis.MockVariableLengthPayloadFilter)

Example 13 with AssertingCodec

use of org.apache.lucene.codecs.asserting.AssertingCodec in project lucene-solr by apache.

the class TestBinaryDocValuesUpdates method testDifferentDVFormatPerField.

public void testDifferentDVFormatPerField() throws Exception {
    // test relies on separate instances of "same thing"
    assert TestUtil.getDefaultDocValuesFormat() != TestUtil.getDefaultDocValuesFormat();
    Directory dir = newDirectory();
    IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
    conf.setCodec(new AssertingCodec() {

        @Override
        public DocValuesFormat getDocValuesFormatForField(String field) {
            return TestUtil.getDefaultDocValuesFormat();
        }
    });
    IndexWriter writer = new IndexWriter(dir, conf);
    Document doc = new Document();
    doc.add(new StringField("key", "doc", Store.NO));
    doc.add(new BinaryDocValuesField("bdv", toBytes(5L)));
    doc.add(new SortedDocValuesField("sorted", new BytesRef("value")));
    // flushed document
    writer.addDocument(doc);
    writer.commit();
    // in-memory document
    writer.addDocument(doc);
    writer.updateBinaryDocValue(new Term("key", "doc"), "bdv", toBytes(17L));
    writer.close();
    final DirectoryReader reader = DirectoryReader.open(dir);
    BinaryDocValues bdv = MultiDocValues.getBinaryValues(reader, "bdv");
    SortedDocValues sdv = MultiDocValues.getSortedValues(reader, "sorted");
    for (int i = 0; i < reader.maxDoc(); i++) {
        assertEquals(i, bdv.nextDoc());
        assertEquals(17, getValue(bdv));
        assertEquals(i, sdv.nextDoc());
        BytesRef term = sdv.binaryValue();
        assertEquals(new BytesRef("value"), term);
    }
    reader.close();
    dir.close();
}
Also used : AssertingCodec(org.apache.lucene.codecs.asserting.AssertingCodec) Document(org.apache.lucene.document.Document) BinaryDocValuesField(org.apache.lucene.document.BinaryDocValuesField) DocValuesFormat(org.apache.lucene.codecs.DocValuesFormat) AssertingDocValuesFormat(org.apache.lucene.codecs.asserting.AssertingDocValuesFormat) MockAnalyzer(org.apache.lucene.analysis.MockAnalyzer) StringField(org.apache.lucene.document.StringField) SortedDocValuesField(org.apache.lucene.document.SortedDocValuesField) BytesRef(org.apache.lucene.util.BytesRef) Directory(org.apache.lucene.store.Directory) NRTCachingDirectory(org.apache.lucene.store.NRTCachingDirectory)

Aggregations

AssertingCodec (org.apache.lucene.codecs.asserting.AssertingCodec)13 Document (org.apache.lucene.document.Document)10 Directory (org.apache.lucene.store.Directory)10 DocValuesFormat (org.apache.lucene.codecs.DocValuesFormat)9 MockAnalyzer (org.apache.lucene.analysis.MockAnalyzer)7 BinaryDocValuesField (org.apache.lucene.document.BinaryDocValuesField)7 NumericDocValuesField (org.apache.lucene.document.NumericDocValuesField)7 StringField (org.apache.lucene.document.StringField)7 BytesRef (org.apache.lucene.util.BytesRef)7 AssertingDocValuesFormat (org.apache.lucene.codecs.asserting.AssertingDocValuesFormat)5 SortedDocValuesField (org.apache.lucene.document.SortedDocValuesField)5 IndexWriterConfig (org.apache.lucene.index.IndexWriterConfig)5 PostingsFormat (org.apache.lucene.codecs.PostingsFormat)4 NRTCachingDirectory (org.apache.lucene.store.NRTCachingDirectory)4 Codec (org.apache.lucene.codecs.Codec)3 SortedNumericDocValuesField (org.apache.lucene.document.SortedNumericDocValuesField)3 SortedSetDocValuesField (org.apache.lucene.document.SortedSetDocValuesField)3 StoredField (org.apache.lucene.document.StoredField)3 IndexWriter (org.apache.lucene.index.IndexWriter)3 ArrayList (java.util.ArrayList)2