Examples with Codec org.apache.lucene.codecs.Codec used on opensource projects

Example 6 with Codec

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

the class TestCodecs method testFixedPostings.

public void testFixedPostings() throws Throwable {
    final int NUM_TERMS = 100;
    final TermData[] terms = new TermData[NUM_TERMS];
    for (int i = 0; i < NUM_TERMS; i++) {
        final int[] docs = new int[] { i };
        final String text = Integer.toString(i, Character.MAX_RADIX);
        terms[i] = new TermData(text, docs, null);
    }
    final FieldInfos.Builder builder = new FieldInfos.Builder();
    final FieldData field = new FieldData("field", builder, terms, true, false);
    final FieldData[] fields = new FieldData[] { field };
    final FieldInfos fieldInfos = builder.finish();
    final Directory dir = newDirectory();
    Codec codec = Codec.getDefault();
    final SegmentInfo si = new SegmentInfo(dir, Version.LATEST, Version.LATEST, SEGMENT, 10000, false, codec, Collections.emptyMap(), StringHelper.randomId(), new HashMap<>(), null);
    this.write(si, fieldInfos, dir, fields);
    final FieldsProducer reader = codec.postingsFormat().fieldsProducer(new SegmentReadState(dir, si, fieldInfos, newIOContext(random())));
    final Iterator<String> fieldsEnum = reader.iterator();
    String fieldName = fieldsEnum.next();
    assertNotNull(fieldName);
    final Terms terms2 = reader.terms(fieldName);
    assertNotNull(terms2);
    final TermsEnum termsEnum = terms2.iterator();
    PostingsEnum postingsEnum = null;
    for (int i = 0; i < NUM_TERMS; i++) {
        final BytesRef term = termsEnum.next();
        assertNotNull(term);
        assertEquals(terms[i].text2, term.utf8ToString());
        // internal state:
        for (int iter = 0; iter < 2; iter++) {
            postingsEnum = TestUtil.docs(random(), termsEnum, postingsEnum, PostingsEnum.NONE);
            assertEquals(terms[i].docs[0], postingsEnum.nextDoc());
            assertEquals(DocIdSetIterator.NO_MORE_DOCS, postingsEnum.nextDoc());
        }
    }
    assertNull(termsEnum.next());
    for (int i = 0; i < NUM_TERMS; i++) {
        assertEquals(termsEnum.seekCeil(new BytesRef(terms[i].text2)), TermsEnum.SeekStatus.FOUND);
    }
    assertFalse(fieldsEnum.hasNext());
    reader.close();
    dir.close();
}

Example 7 with Codec

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

the class TestCodecs method testRandomPostings.

public void testRandomPostings() throws Throwable {
    final FieldInfos.Builder builder = new FieldInfos.Builder();
    final FieldData[] fields = new FieldData[NUM_FIELDS];
    for (int i = 0; i < NUM_FIELDS; i++) {
        final boolean omitTF = 0 == (i % 3);
        final boolean storePayloads = 1 == (i % 3);
        fields[i] = new FieldData(fieldNames[i], builder, this.makeRandomTerms(omitTF, storePayloads), omitTF, storePayloads);
    }
    final Directory dir = newDirectory();
    final FieldInfos fieldInfos = builder.finish();
    if (VERBOSE) {
        System.out.println("TEST: now write postings");
    }
    Codec codec = Codec.getDefault();
    final SegmentInfo si = new SegmentInfo(dir, Version.LATEST, Version.LATEST, SEGMENT, 10000, false, codec, Collections.emptyMap(), StringHelper.randomId(), new HashMap<>(), null);
    this.write(si, fieldInfos, dir, fields);
    if (VERBOSE) {
        System.out.println("TEST: now read postings");
    }
    final FieldsProducer terms = codec.postingsFormat().fieldsProducer(new SegmentReadState(dir, si, fieldInfos, newIOContext(random())));
    final Verify[] threads = new Verify[NUM_TEST_THREADS - 1];
    for (int i = 0; i < NUM_TEST_THREADS - 1; i++) {
        threads[i] = new Verify(si, fields, terms);
        threads[i].setDaemon(true);
        threads[i].start();
    }
    new Verify(si, fields, terms).run();
    for (int i = 0; i < NUM_TEST_THREADS - 1; i++) {
        threads[i].join();
        assert !threads[i].failed;
    }
    terms.close();
    dir.close();
}

Example 8 with Codec

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

the class TestDoc method merge.

private SegmentCommitInfo merge(Directory dir, SegmentCommitInfo si1, SegmentCommitInfo si2, String merged, boolean useCompoundFile) throws Exception {
    IOContext context = newIOContext(random(), new IOContext(new MergeInfo(-1, -1, false, -1)));
    SegmentReader r1 = new SegmentReader(si1, Version.LATEST.major, context);
    SegmentReader r2 = new SegmentReader(si2, Version.LATEST.major, context);
    final Codec codec = Codec.getDefault();
    TrackingDirectoryWrapper trackingDir = new TrackingDirectoryWrapper(si1.info.dir);
    final SegmentInfo si = new SegmentInfo(si1.info.dir, Version.LATEST, null, merged, -1, false, codec, Collections.emptyMap(), StringHelper.randomId(), new HashMap<>(), null);
    SegmentMerger merger = new SegmentMerger(Arrays.<CodecReader>asList(r1, r2), si, InfoStream.getDefault(), trackingDir, new FieldInfos.FieldNumbers(), context);
    MergeState mergeState = merger.merge();
    r1.close();
    r2.close();
    ;
    si.setFiles(new HashSet<>(trackingDir.getCreatedFiles()));
    if (useCompoundFile) {
        Collection<String> filesToDelete = si.files();
        codec.compoundFormat().write(dir, si, context);
        si.setUseCompoundFile(true);
        for (String name : filesToDelete) {
            si1.info.dir.deleteFile(name);
        }
    }
    return new SegmentCommitInfo(si, 0, -1L, -1L, -1L);
}

Example 9 with Codec

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

the class IndexWriter method mergeMiddle.

/** Does the actual (time-consuming) work of the merge,
   *  but without holding synchronized lock on IndexWriter
   *  instance */
private int mergeMiddle(MergePolicy.OneMerge merge, MergePolicy mergePolicy) throws IOException {
    merge.checkAborted();
    Directory mergeDirectory = config.getMergeScheduler().wrapForMerge(merge, directory);
    List<SegmentCommitInfo> sourceSegments = merge.segments;
    IOContext context = new IOContext(merge.getStoreMergeInfo());
    final TrackingDirectoryWrapper dirWrapper = new TrackingDirectoryWrapper(mergeDirectory);
    if (infoStream.isEnabled("IW")) {
        infoStream.message("IW", "merging " + segString(merge.segments));
    }
    merge.readers = new ArrayList<>(sourceSegments.size());
    // This is try/finally to make sure merger's readers are
    // closed:
    boolean success = false;
    try {
        int segUpto = 0;
        while (segUpto < sourceSegments.size()) {
            final SegmentCommitInfo info = sourceSegments.get(segUpto);
            // Hold onto the "live" reader; we will use this to
            // commit merged deletes
            final ReadersAndUpdates rld = readerPool.get(info, true);
            // Carefully pull the most recent live docs and reader
            SegmentReader reader;
            final Bits liveDocs;
            final int delCount;
            synchronized (this) {
                // Must sync to ensure BufferedDeletesStream cannot change liveDocs,
                // pendingDeleteCount and field updates while we pull a copy:
                reader = rld.getReaderForMerge(context);
                liveDocs = rld.getReadOnlyLiveDocs();
                delCount = rld.getPendingDeleteCount() + info.getDelCount();
                assert reader != null;
                assert rld.verifyDocCounts();
                if (infoStream.isEnabled("IW")) {
                    if (rld.getPendingDeleteCount() != 0) {
                        infoStream.message("IW", "seg=" + segString(info) + " delCount=" + info.getDelCount() + " pendingDelCount=" + rld.getPendingDeleteCount());
                    } else if (info.getDelCount() != 0) {
                        infoStream.message("IW", "seg=" + segString(info) + " delCount=" + info.getDelCount());
                    } else {
                        infoStream.message("IW", "seg=" + segString(info) + " no deletes");
                    }
                }
            }
            // make a new reader with updated live docs and del count.
            if (reader.numDeletedDocs() != delCount) {
                // beware of zombies
                assert delCount > reader.numDeletedDocs();
                SegmentReader newReader;
                synchronized (this) {
                    // We must also sync on IW here, because another thread could be writing
                    // new DV updates / remove old gen field infos files causing FNFE:
                    newReader = new SegmentReader(info, reader, liveDocs, info.info.maxDoc() - delCount);
                }
                boolean released = false;
                try {
                    rld.release(reader);
                    released = true;
                } finally {
                    if (!released) {
                        newReader.decRef();
                    }
                }
                reader = newReader;
            }
            merge.readers.add(reader);
            assert delCount <= info.info.maxDoc() : "delCount=" + delCount + " info.maxDoc=" + info.info.maxDoc() + " rld.pendingDeleteCount=" + rld.getPendingDeleteCount() + " info.getDelCount()=" + info.getDelCount();
            segUpto++;
        }
        //      System.out.println("[" + Thread.currentThread().getName() + "] IW.mergeMiddle: merging " + merge.getMergeReaders());
        // Let the merge wrap readers
        List<CodecReader> mergeReaders = new ArrayList<>();
        for (SegmentReader reader : merge.readers) {
            CodecReader wrappedReader = merge.wrapForMerge(reader);
            validateMergeReader(wrappedReader);
            mergeReaders.add(wrappedReader);
        }
        final SegmentMerger merger = new SegmentMerger(mergeReaders, merge.info.info, infoStream, dirWrapper, globalFieldNumberMap, context);
        merge.checkAborted();
        merge.mergeStartNS = System.nanoTime();
        // This is where all the work happens:
        if (merger.shouldMerge()) {
            merger.merge();
        }
        MergeState mergeState = merger.mergeState;
        assert mergeState.segmentInfo == merge.info.info;
        merge.info.info.setFiles(new HashSet<>(dirWrapper.getCreatedFiles()));
        if (infoStream.isEnabled("IW")) {
            if (merger.shouldMerge()) {
                String pauseInfo = merge.getMergeProgress().getPauseTimes().entrySet().stream().filter((e) -> e.getValue() > 0).map((e) -> String.format(Locale.ROOT, "%.1f sec %s", e.getValue() / 1000000000., e.getKey().name().toLowerCase(Locale.ROOT))).collect(Collectors.joining(", "));
                if (!pauseInfo.isEmpty()) {
                    pauseInfo = " (" + pauseInfo + ")";
                }
                long t1 = System.nanoTime();
                double sec = (t1 - merge.mergeStartNS) / 1000000000.;
                double segmentMB = (merge.info.sizeInBytes() / 1024. / 1024.);
                infoStream.message("IW", "merge codec=" + codec + " maxDoc=" + merge.info.info.maxDoc() + "; merged segment has " + (mergeState.mergeFieldInfos.hasVectors() ? "vectors" : "no vectors") + "; " + (mergeState.mergeFieldInfos.hasNorms() ? "norms" : "no norms") + "; " + (mergeState.mergeFieldInfos.hasDocValues() ? "docValues" : "no docValues") + "; " + (mergeState.mergeFieldInfos.hasProx() ? "prox" : "no prox") + "; " + (mergeState.mergeFieldInfos.hasProx() ? "freqs" : "no freqs") + "; " + (mergeState.mergeFieldInfos.hasPointValues() ? "points" : "no points") + "; " + String.format(Locale.ROOT, "%.1f sec%s to merge segment [%.2f MB, %.2f MB/sec]", sec, pauseInfo, segmentMB, segmentMB / sec));
            } else {
                infoStream.message("IW", "skip merging fully deleted segments");
            }
        }
        if (merger.shouldMerge() == false) {
            // Merge would produce a 0-doc segment, so we do nothing except commit the merge to remove all the 0-doc segments that we "merged":
            assert merge.info.info.maxDoc() == 0;
            commitMerge(merge, mergeState);
            return 0;
        }
        assert merge.info.info.maxDoc() > 0;
        // Very important to do this before opening the reader
        // because codec must know if prox was written for
        // this segment:
        //System.out.println("merger set hasProx=" + merger.hasProx() + " seg=" + merge.info.name);
        boolean useCompoundFile;
        synchronized (this) {
            // Guard segmentInfos
            useCompoundFile = mergePolicy.useCompoundFile(segmentInfos, merge.info, this);
        }
        if (useCompoundFile) {
            success = false;
            Collection<String> filesToRemove = merge.info.files();
            TrackingDirectoryWrapper trackingCFSDir = new TrackingDirectoryWrapper(mergeDirectory);
            try {
                createCompoundFile(infoStream, trackingCFSDir, merge.info.info, context);
                success = true;
            } catch (Throwable t) {
                synchronized (this) {
                    if (merge.isAborted()) {
                        // merged files:
                        if (infoStream.isEnabled("IW")) {
                            infoStream.message("IW", "hit merge abort exception creating compound file during merge");
                        }
                        return 0;
                    } else {
                        handleMergeException(t, merge);
                    }
                }
            } finally {
                if (success == false) {
                    if (infoStream.isEnabled("IW")) {
                        infoStream.message("IW", "hit exception creating compound file during merge");
                    }
                    // Safe: these files must exist
                    deleteNewFiles(merge.info.files());
                }
            }
            // So that, if we hit exc in deleteNewFiles (next)
            // or in commitMerge (later), we close the
            // per-segment readers in the finally clause below:
            success = false;
            synchronized (this) {
                // delete new non cfs files directly: they were never
                // registered with IFD
                deleteNewFiles(filesToRemove);
                if (merge.isAborted()) {
                    if (infoStream.isEnabled("IW")) {
                        infoStream.message("IW", "abort merge after building CFS");
                    }
                    // Safe: these files must exist
                    deleteNewFiles(merge.info.files());
                    return 0;
                }
            }
            merge.info.info.setUseCompoundFile(true);
        } else {
            // So that, if we hit exc in commitMerge (later),
            // we close the per-segment readers in the finally
            // clause below:
            success = false;
        }
        // Have codec write SegmentInfo.  Must do this after
        // creating CFS so that 1) .si isn't slurped into CFS,
        // and 2) .si reflects useCompoundFile=true change
        // above:
        boolean success2 = false;
        try {
            codec.segmentInfoFormat().write(directory, merge.info.info, context);
            success2 = true;
        } finally {
            if (!success2) {
                // Safe: these files must exist
                deleteNewFiles(merge.info.files());
            }
        }
        if (infoStream.isEnabled("IW")) {
            infoStream.message("IW", String.format(Locale.ROOT, "merged segment size=%.3f MB vs estimate=%.3f MB", merge.info.sizeInBytes() / 1024. / 1024., merge.estimatedMergeBytes / 1024 / 1024.));
        }
        final IndexReaderWarmer mergedSegmentWarmer = config.getMergedSegmentWarmer();
        if (poolReaders && mergedSegmentWarmer != null) {
            final ReadersAndUpdates rld = readerPool.get(merge.info, true);
            final SegmentReader sr = rld.getReader(IOContext.READ);
            try {
                mergedSegmentWarmer.warm(sr);
            } finally {
                synchronized (this) {
                    rld.release(sr);
                    readerPool.release(rld);
                }
            }
        }
        if (!commitMerge(merge, mergeState)) {
            // aborted
            return 0;
        }
        success = true;
    } finally {
        // an exc:
        if (success == false) {
            closeMergeReaders(merge, true);
        }
    }
    return merge.info.info.maxDoc();
}

Example 10 with Codec

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

the class ReadersAndUpdates method writeFieldUpdates.

// Writes field updates (new _X_N updates files) to the directory
public synchronized void writeFieldUpdates(Directory dir, DocValuesFieldUpdates.Container dvUpdates) throws IOException {
    assert Thread.holdsLock(writer);
    assert dvUpdates.any();
    // Do this so we can delete any created files on
    // exception; this saves all codecs from having to do
    // it:
    TrackingDirectoryWrapper trackingDir = new TrackingDirectoryWrapper(dir);
    final Map<Integer, Set<String>> newDVFiles = new HashMap<>();
    Set<String> fieldInfosFiles = null;
    FieldInfos fieldInfos = null;
    boolean success = false;
    try {
        final Codec codec = info.info.getCodec();
        // reader could be null e.g. for a just merged segment (from
        // IndexWriter.commitMergedDeletes).
        final SegmentReader reader = this.reader == null ? new SegmentReader(info, writer.segmentInfos.getIndexCreatedVersionMajor(), IOContext.READONCE) : this.reader;
        try {
            // clone FieldInfos so that we can update their dvGen separately from
            // the reader's infos and write them to a new fieldInfos_gen file
            FieldInfos.Builder builder = new FieldInfos.Builder(writer.globalFieldNumberMap);
            // clone FI.attributes as well FI.dvGen
            for (FieldInfo fi : reader.getFieldInfos()) {
                FieldInfo clone = builder.add(fi);
                // copy the stuff FieldInfos.Builder doesn't copy
                for (Entry<String, String> e : fi.attributes().entrySet()) {
                    clone.putAttribute(e.getKey(), e.getValue());
                }
                clone.setDocValuesGen(fi.getDocValuesGen());
            }
            // create new fields or update existing ones to have NumericDV type
            for (String f : dvUpdates.numericDVUpdates.keySet()) {
                FieldInfo fieldInfo = builder.getOrAdd(f);
                fieldInfo.setDocValuesType(DocValuesType.NUMERIC);
            }
            // create new fields or update existing ones to have BinaryDV type
            for (String f : dvUpdates.binaryDVUpdates.keySet()) {
                FieldInfo fieldInfo = builder.getOrAdd(f);
                fieldInfo.setDocValuesType(DocValuesType.BINARY);
            }
            fieldInfos = builder.finish();
            final DocValuesFormat docValuesFormat = codec.docValuesFormat();
            //          System.out.println("[" + Thread.currentThread().getName() + "] RLD.writeFieldUpdates: applying numeric updates; seg=" + info + " updates=" + numericFieldUpdates);
            handleNumericDVUpdates(fieldInfos, dvUpdates.numericDVUpdates, trackingDir, docValuesFormat, reader, newDVFiles);
            //        System.out.println("[" + Thread.currentThread().getName() + "] RAU.writeFieldUpdates: applying binary updates; seg=" + info + " updates=" + dvUpdates.binaryDVUpdates);
            handleBinaryDVUpdates(fieldInfos, dvUpdates.binaryDVUpdates, trackingDir, docValuesFormat, reader, newDVFiles);
            //        System.out.println("[" + Thread.currentThread().getName() + "] RAU.writeFieldUpdates: write fieldInfos; seg=" + info);
            fieldInfosFiles = writeFieldInfosGen(fieldInfos, trackingDir, docValuesFormat, codec.fieldInfosFormat());
        } finally {
            if (reader != this.reader) {
                //          System.out.println("[" + Thread.currentThread().getName() + "] RLD.writeLiveDocs: closeReader " + reader);
                reader.close();
            }
        }
        success = true;
    } finally {
        if (!success) {
            // Advance only the nextWriteFieldInfosGen and nextWriteDocValuesGen, so
            // that a 2nd attempt to write will write to a new file
            info.advanceNextWriteFieldInfosGen();
            info.advanceNextWriteDocValuesGen();
            // Delete any partially created file(s):
            for (String fileName : trackingDir.getCreatedFiles()) {
                IOUtils.deleteFilesIgnoringExceptions(dir, fileName);
            }
        }
    }
    // copy all the updates to mergingUpdates, so they can later be applied to the merged segment
    if (isMerging) {
        for (Entry<String, NumericDocValuesFieldUpdates> e : dvUpdates.numericDVUpdates.entrySet()) {
            DocValuesFieldUpdates updates = mergingDVUpdates.get(e.getKey());
            if (updates == null) {
                mergingDVUpdates.put(e.getKey(), e.getValue());
            } else {
                updates.merge(e.getValue());
            }
        }
        for (Entry<String, BinaryDocValuesFieldUpdates> e : dvUpdates.binaryDVUpdates.entrySet()) {
            DocValuesFieldUpdates updates = mergingDVUpdates.get(e.getKey());
            if (updates == null) {
                mergingDVUpdates.put(e.getKey(), e.getValue());
            } else {
                updates.merge(e.getValue());
            }
        }
    }
    // writing field updates succeeded
    assert fieldInfosFiles != null;
    info.setFieldInfosFiles(fieldInfosFiles);
    // were updated now.
    assert !newDVFiles.isEmpty();
    for (Entry<Integer, Set<String>> e : info.getDocValuesUpdatesFiles().entrySet()) {
        if (!newDVFiles.containsKey(e.getKey())) {
            newDVFiles.put(e.getKey(), e.getValue());
        }
    }
    info.setDocValuesUpdatesFiles(newDVFiles);
    // wrote new files, should checkpoint()
    writer.checkpoint();
    // if there is a reader open, reopen it to reflect the updates
    if (reader != null) {
        SegmentReader newReader = new SegmentReader(info, reader, liveDocs, info.info.maxDoc() - info.getDelCount() - pendingDeleteCount);
        boolean reopened = false;
        try {
            reader.decRef();
            reader = newReader;
            reopened = true;
        } finally {
            if (!reopened) {
                newReader.decRef();
            }
        }
    }
}