Example usage for org.apache.lucene.codecs.perfield PerFieldPostingsFormat PER_FIELD_SUFFIX_KEY

Introduction

In this page you can find the example usage for org.apache.lucene.codecs.perfield PerFieldPostingsFormat PER_FIELD_SUFFIX_KEY.

Prototype

String PER_FIELD_SUFFIX_KEY

To view the source code for org.apache.lucene.codecs.perfield PerFieldPostingsFormat PER_FIELD_SUFFIX_KEY.

Click Source Link

Document

FieldInfo attribute name used to store the segment suffix name for each field.

Usage

From source file:perf.DiskUsage.java

License:Apache License

static Set<FieldStats> analyzeFields(SegmentReader reader) throws Exception {
    Map<String, FieldStats> stats = new HashMap<>();
    Map<String, String> dvSuffixes = new HashMap<>();
    Map<String, String> postingsSuffixes = new HashMap<>();
    for (FieldInfo field : reader.getFieldInfos()) {
        FieldStats fieldStats = new FieldStats(field.name);
        stats.put(field.name, fieldStats);
        Map<String, String> attributes = field.attributes();
        if (attributes != null) {
            String postingsSuffix = attributes.get(PerFieldPostingsFormat.PER_FIELD_SUFFIX_KEY);
            if (postingsSuffix != null) {
                postingsSuffixes.put(postingsSuffix, field.name);
            }//www  . jav  a  2s  .c om
            String dvSuffix = attributes.get(PerFieldDocValuesFormat.PER_FIELD_SUFFIX_KEY);
            if (dvSuffix != null) {
                dvSuffixes.put(dvSuffix, field.name);
            }
        }

        Bits docsWithField = reader.getDocsWithField(field.name);
        if (docsWithField != null) {
            int count = 0;
            for (int docID = 0; docID < reader.maxDoc(); docID++) {
                if (docsWithField.get(docID)) {
                    count++;
                }
            }
            fieldStats.docCountWithField = count;
        }
    }

    Directory directory = reader.directory();
    for (String file : directory.listAll()) {
        String suffix = parseSuffix(file);
        long bytes = directory.fileLength(file);
        if (suffix != null) {
            switch (IndexFileNames.getExtension(file)) {
            case "dvd":
            case "dvm":
                stats.get(dvSuffixes.get(suffix)).dvBytes += bytes;
                break;
            case "tim":
            case "tip":
                stats.get(postingsSuffixes.get(suffix)).termsBytes += bytes;
                break;
            case "doc":
                stats.get(postingsSuffixes.get(suffix)).postingsBytes += bytes;
                break;
            case "pos":
            case "pay":
                stats.get(postingsSuffixes.get(suffix)).proxBytes += bytes;
                break;
            default:
                throw new AssertionError("unexpected suffixed file: " + file);
            }
        } else {
            // not a per-field file, but we can hackishly do this for the points case.
            if ("dii".equals(IndexFileNames.getExtension(file))) {
                System.err.println(
                        "retrieving per-field point usage, if you see a scary corruption error, its probably just this tool!!!!");
                try (ChecksumIndexInput in = directory.openChecksumInput(file, IOContext.READONCE)) {
                    // fail hard if its not exactly the version we do this hack for.
                    CodecUtil.checkIndexHeader(in, "Lucene60PointsFormatMeta", 0, 0,
                            reader.getSegmentInfo().info.getId(), "");
                    int fieldCount = in.readVInt();
                    // strangely, bkd offsets are not in any guaranteed order
                    TreeMap<Long, String> offsetToField = new TreeMap<>();
                    for (int i = 0; i < fieldCount; i++) {
                        int field = in.readVInt();
                        long offset = in.readVLong();
                        offsetToField.put(offset, reader.getFieldInfos().fieldInfo(field).name);
                    }
                    // now we can traverse in order
                    long previousOffset = 0;
                    for (Map.Entry<Long, String> entry : offsetToField.entrySet()) {
                        long offset = entry.getKey();
                        String field = entry.getValue();
                        stats.get(field).pointsBytes += (offset - previousOffset);
                        previousOffset = offset;
                    }
                    CodecUtil.checkFooter(in);
                }
            }
        }
    }

    return new TreeSet<FieldStats>(stats.values());
}

---

From source file:perf.DiskUsage.java

License:Apache License

static Set<FieldStats> analyzeFields(SegmentReader reader) throws Exception {
    Map<String, FieldStats> stats = new HashMap<>();
    Map<String, String> dvSuffixes = new HashMap<>();
    Map<String, String> postingsSuffixes = new HashMap<>();
    for (FieldInfo field : reader.getFieldInfos()) {
        FieldStats fieldStats = new FieldStats(field.name);
        stats.put(field.name, fieldStats);
        Map<String, String> attributes = field.attributes();
        if (attributes != null) {
            String postingsSuffix = attributes.get(PerFieldPostingsFormat.PER_FIELD_SUFFIX_KEY);
            if (postingsSuffix != null) {
                postingsSuffixes.put(postingsSuffix, field.name);
            }//from w  w w .  ja  v a2s.  co m
            String dvSuffix = attributes.get(PerFieldDocValuesFormat.PER_FIELD_SUFFIX_KEY);
            if (dvSuffix != null) {
                dvSuffixes.put(dvSuffix, field.name);
            }
        }

        DocIdSetIterator docsWithField;
        switch (field.getDocValuesType()) {
        case NUMERIC:
            docsWithField = reader.getNumericDocValues(field.name);
            break;
        case BINARY:
            docsWithField = reader.getBinaryDocValues(field.name);
            break;
        case SORTED:
            docsWithField = reader.getSortedDocValues(field.name);
            break;
        case SORTED_NUMERIC:
            docsWithField = reader.getSortedNumericDocValues(field.name);
            break;
        case SORTED_SET:
            docsWithField = reader.getSortedSetDocValues(field.name);
            break;
        case NONE:
            docsWithField = null;
            break;
        default:
            docsWithField = null;
            break;
        }

        if (docsWithField != null) {
            int count = 0;
            while (docsWithField.nextDoc() != DocIdSetIterator.NO_MORE_DOCS) {
                count++;
            }
            fieldStats.docCountWithField = count;
        }
    }

    Directory directory = reader.directory();
    for (String file : directory.listAll()) {
        String suffix = parseSuffix(file);
        long bytes = directory.fileLength(file);
        if (suffix != null) {
            switch (IndexFileNames.getExtension(file)) {
            case "dvd":
            case "dvm":
                stats.get(dvSuffixes.get(suffix)).dvBytes += bytes;
                break;
            case "tim":
            case "tip":
                stats.get(postingsSuffixes.get(suffix)).termsBytes += bytes;
                break;
            case "doc":
                stats.get(postingsSuffixes.get(suffix)).postingsBytes += bytes;
                break;
            case "pos":
            case "pay":
                stats.get(postingsSuffixes.get(suffix)).proxBytes += bytes;
                break;
            default:
                throw new AssertionError("unexpected suffixed file: " + file);
            }
        } else {
            // not a per-field file, but we can hackishly do this for the points case.
            if ("dii".equals(IndexFileNames.getExtension(file))) {
                System.err.println(
                        "retrieving per-field point usage, if you see a scary corruption error, its probably just this tool!!!!");
                try (ChecksumIndexInput in = directory.openChecksumInput(file, IOContext.READONCE)) {
                    // fail hard if its not exactly the version we do this hack for.
                    CodecUtil.checkIndexHeader(in, "Lucene60PointsFormatMeta", 0, 0,
                            reader.getSegmentInfo().info.getId(), "");
                    int fieldCount = in.readVInt();
                    // strangely, bkd offsets are not in any guaranteed order
                    TreeMap<Long, String> offsetToField = new TreeMap<>();
                    for (int i = 0; i < fieldCount; i++) {
                        int field = in.readVInt();
                        long offset = in.readVLong();
                        offsetToField.put(offset, reader.getFieldInfos().fieldInfo(field).name);
                    }
                    // now we can traverse in order
                    long previousOffset = 0;
                    for (Map.Entry<Long, String> entry : offsetToField.entrySet()) {
                        long offset = entry.getKey();
                        String field = entry.getValue();
                        stats.get(field).pointsBytes += (offset - previousOffset);
                        previousOffset = offset;
                    }
                    CodecUtil.checkFooter(in);
                }
            }
        }
    }

    return new TreeSet<FieldStats>(stats.values());
}

---