Examples with CharsRefBuilder org.apache.lucene.util.CharsRefBuilder used on opensource projects

Example 21 with CharsRefBuilder

use of org.apache.lucene.util.CharsRefBuilder in project lucene-solr by apache.

the class JaspellLookup method build.

@Override
public void build(InputIterator iterator) throws IOException {
    if (iterator.hasPayloads()) {
        throw new IllegalArgumentException("this suggester doesn't support payloads");
    }
    if (iterator.hasContexts()) {
        throw new IllegalArgumentException("this suggester doesn't support contexts");
    }
    count = 0;
    trie = new JaspellTernarySearchTrie();
    trie.setMatchAlmostDiff(editDistance);
    BytesRef spare;
    final CharsRefBuilder charsSpare = new CharsRefBuilder();
    while ((spare = iterator.next()) != null) {
        final long weight = iterator.weight();
        if (spare.length == 0) {
            continue;
        }
        charsSpare.copyUTF8Bytes(spare);
        trie.put(charsSpare.toString(), Long.valueOf(weight));
        count++;
    }
}

Example 22 with CharsRefBuilder

use of org.apache.lucene.util.CharsRefBuilder in project lucene-solr by apache.

the class TSTLookup method build.

@Override
public void build(InputIterator iterator) throws IOException {
    if (iterator.hasPayloads()) {
        throw new IllegalArgumentException("this suggester doesn't support payloads");
    }
    if (iterator.hasContexts()) {
        throw new IllegalArgumentException("this suggester doesn't support contexts");
    }
    root = new TernaryTreeNode();
    // make sure it's sorted and the comparator uses UTF16 sort order
    iterator = new SortedInputIterator(tempDir, tempFileNamePrefix, iterator, utf8SortedAsUTF16SortOrder);
    count = 0;
    ArrayList<String> tokens = new ArrayList<>();
    ArrayList<Number> vals = new ArrayList<>();
    BytesRef spare;
    CharsRefBuilder charsSpare = new CharsRefBuilder();
    while ((spare = iterator.next()) != null) {
        charsSpare.copyUTF8Bytes(spare);
        tokens.add(charsSpare.toString());
        vals.add(Long.valueOf(iterator.weight()));
        count++;
    }
    autocomplete.balancedTree(tokens.toArray(), vals.toArray(), 0, tokens.size() - 1, root);
}

Example 23 with CharsRefBuilder

use of org.apache.lucene.util.CharsRefBuilder in project lucene-solr by apache.

the class ExpandComponent method process.

@SuppressWarnings("unchecked")
@Override
public void process(ResponseBuilder rb) throws IOException {
    if (!rb.doExpand) {
        return;
    }
    SolrQueryRequest req = rb.req;
    SolrParams params = req.getParams();
    String field = params.get(ExpandParams.EXPAND_FIELD);
    String hint = null;
    if (field == null) {
        List<Query> filters = rb.getFilters();
        if (filters != null) {
            for (Query q : filters) {
                if (q instanceof CollapsingQParserPlugin.CollapsingPostFilter) {
                    CollapsingQParserPlugin.CollapsingPostFilter cp = (CollapsingQParserPlugin.CollapsingPostFilter) q;
                    field = cp.getField();
                    hint = cp.hint;
                }
            }
        }
    }
    if (field == null) {
        throw new IOException("Expand field is null.");
    }
    String sortParam = params.get(ExpandParams.EXPAND_SORT);
    String[] fqs = params.getParams(ExpandParams.EXPAND_FQ);
    String qs = params.get(ExpandParams.EXPAND_Q);
    int limit = params.getInt(ExpandParams.EXPAND_ROWS, 5);
    Sort sort = null;
    if (sortParam != null) {
        sort = SortSpecParsing.parseSortSpec(sortParam, rb.req).getSort();
    }
    Query query;
    if (qs == null) {
        query = rb.getQuery();
    } else {
        try {
            QParser parser = QParser.getParser(qs, req);
            query = parser.getQuery();
        } catch (Exception e) {
            throw new IOException(e);
        }
    }
    List<Query> newFilters = new ArrayList<>();
    if (fqs == null) {
        List<Query> filters = rb.getFilters();
        if (filters != null) {
            for (Query q : filters) {
                if (!(q instanceof CollapsingQParserPlugin.CollapsingPostFilter)) {
                    newFilters.add(q);
                }
            }
        }
    } else {
        try {
            for (String fq : fqs) {
                if (fq != null && fq.trim().length() != 0 && !fq.equals("*:*")) {
                    QParser fqp = QParser.getParser(fq, req);
                    newFilters.add(fqp.getQuery());
                }
            }
        } catch (Exception e) {
            throw new IOException(e);
        }
    }
    SolrIndexSearcher searcher = req.getSearcher();
    LeafReader reader = searcher.getSlowAtomicReader();
    SchemaField schemaField = searcher.getSchema().getField(field);
    FieldType fieldType = schemaField.getType();
    SortedDocValues values = null;
    long nullValue = 0L;
    if (fieldType instanceof StrField) {
        //Get The Top Level SortedDocValues
        if (CollapsingQParserPlugin.HINT_TOP_FC.equals(hint)) {
            Map<String, UninvertingReader.Type> mapping = new HashMap();
            mapping.put(field, UninvertingReader.Type.SORTED);
            UninvertingReader uninvertingReader = new UninvertingReader(new ReaderWrapper(searcher.getSlowAtomicReader(), field), mapping);
            values = uninvertingReader.getSortedDocValues(field);
        } else {
            values = DocValues.getSorted(reader, field);
        }
    } else {
        //Get the nullValue for the numeric collapse field
        String defaultValue = searcher.getSchema().getField(field).getDefaultValue();
        final NumberType numType = fieldType.getNumberType();
        // we don't need to handle invalid 64-bit field types here.
        if (defaultValue != null) {
            if (numType == NumberType.INTEGER) {
                nullValue = Long.parseLong(defaultValue);
            } else if (numType == NumberType.FLOAT) {
                nullValue = Float.floatToIntBits(Float.parseFloat(defaultValue));
            }
        } else if (NumberType.FLOAT.equals(numType)) {
            // Integer case already handled by nullValue defaulting to 0
            nullValue = Float.floatToIntBits(0.0f);
        }
    }
    FixedBitSet groupBits = null;
    LongHashSet groupSet = null;
    DocList docList = rb.getResults().docList;
    IntHashSet collapsedSet = new IntHashSet(docList.size() * 2);
    //Gather the groups for the current page of documents
    DocIterator idit = docList.iterator();
    int[] globalDocs = new int[docList.size()];
    int docsIndex = -1;
    while (idit.hasNext()) {
        globalDocs[++docsIndex] = idit.nextDoc();
    }
    Arrays.sort(globalDocs);
    Query groupQuery = null;
    /*
    * This code gathers the group information for the current page.
    */
    List<LeafReaderContext> contexts = searcher.getTopReaderContext().leaves();
    if (contexts.size() == 0) {
        //When no context is available we can skip the expanding
        return;
    }
    int currentContext = 0;
    int currentDocBase = contexts.get(currentContext).docBase;
    int nextDocBase = (currentContext + 1) < contexts.size() ? contexts.get(currentContext + 1).docBase : Integer.MAX_VALUE;
    IntObjectHashMap<BytesRef> ordBytes = null;
    if (values != null) {
        groupBits = new FixedBitSet(values.getValueCount());
        MultiDocValues.OrdinalMap ordinalMap = null;
        SortedDocValues[] sortedDocValues = null;
        LongValues segmentOrdinalMap = null;
        SortedDocValues currentValues = null;
        if (values instanceof MultiDocValues.MultiSortedDocValues) {
            ordinalMap = ((MultiDocValues.MultiSortedDocValues) values).mapping;
            sortedDocValues = ((MultiDocValues.MultiSortedDocValues) values).values;
            currentValues = sortedDocValues[currentContext];
            segmentOrdinalMap = ordinalMap.getGlobalOrds(currentContext);
        }
        int count = 0;
        ordBytes = new IntObjectHashMap<>();
        for (int i = 0; i < globalDocs.length; i++) {
            int globalDoc = globalDocs[i];
            while (globalDoc >= nextDocBase) {
                currentContext++;
                currentDocBase = contexts.get(currentContext).docBase;
                nextDocBase = (currentContext + 1) < contexts.size() ? contexts.get(currentContext + 1).docBase : Integer.MAX_VALUE;
                if (ordinalMap != null) {
                    currentValues = sortedDocValues[currentContext];
                    segmentOrdinalMap = ordinalMap.getGlobalOrds(currentContext);
                }
            }
            int contextDoc = globalDoc - currentDocBase;
            if (ordinalMap != null) {
                if (contextDoc > currentValues.docID()) {
                    currentValues.advance(contextDoc);
                }
                if (contextDoc == currentValues.docID()) {
                    int ord = currentValues.ordValue();
                    ++count;
                    BytesRef ref = currentValues.lookupOrd(ord);
                    ord = (int) segmentOrdinalMap.get(ord);
                    ordBytes.put(ord, BytesRef.deepCopyOf(ref));
                    groupBits.set(ord);
                    collapsedSet.add(globalDoc);
                }
            } else {
                if (globalDoc > values.docID()) {
                    values.advance(globalDoc);
                }
                if (globalDoc == values.docID()) {
                    int ord = values.ordValue();
                    ++count;
                    BytesRef ref = values.lookupOrd(ord);
                    ordBytes.put(ord, BytesRef.deepCopyOf(ref));
                    groupBits.set(ord);
                    collapsedSet.add(globalDoc);
                }
            }
        }
        if (count > 0 && count < 200) {
            try {
                groupQuery = getGroupQuery(field, count, ordBytes);
            } catch (Exception e) {
                throw new IOException(e);
            }
        }
    } else {
        groupSet = new LongHashSet(docList.size());
        NumericDocValues collapseValues = contexts.get(currentContext).reader().getNumericDocValues(field);
        int count = 0;
        for (int i = 0; i < globalDocs.length; i++) {
            int globalDoc = globalDocs[i];
            while (globalDoc >= nextDocBase) {
                currentContext++;
                currentDocBase = contexts.get(currentContext).docBase;
                nextDocBase = currentContext + 1 < contexts.size() ? contexts.get(currentContext + 1).docBase : Integer.MAX_VALUE;
                collapseValues = contexts.get(currentContext).reader().getNumericDocValues(field);
            }
            int contextDoc = globalDoc - currentDocBase;
            int valueDocID = collapseValues.docID();
            if (valueDocID < contextDoc) {
                valueDocID = collapseValues.advance(contextDoc);
            }
            long value;
            if (valueDocID == contextDoc) {
                value = collapseValues.longValue();
            } else {
                value = 0;
            }
            if (value != nullValue) {
                ++count;
                groupSet.add(value);
                collapsedSet.add(globalDoc);
            }
        }
        if (count > 0 && count < 200) {
            if (fieldType.isPointField()) {
                groupQuery = getPointGroupQuery(schemaField, count, groupSet);
            } else {
                groupQuery = getGroupQuery(field, fieldType, count, groupSet);
            }
        }
    }
    Collector collector;
    if (sort != null)
        sort = sort.rewrite(searcher);
    Collector groupExpandCollector = null;
    if (values != null) {
        //Get The Top Level SortedDocValues again so we can re-iterate:
        if (CollapsingQParserPlugin.HINT_TOP_FC.equals(hint)) {
            Map<String, UninvertingReader.Type> mapping = new HashMap();
            mapping.put(field, UninvertingReader.Type.SORTED);
            UninvertingReader uninvertingReader = new UninvertingReader(new ReaderWrapper(searcher.getSlowAtomicReader(), field), mapping);
            values = uninvertingReader.getSortedDocValues(field);
        } else {
            values = DocValues.getSorted(reader, field);
        }
        groupExpandCollector = new GroupExpandCollector(values, groupBits, collapsedSet, limit, sort);
    } else {
        groupExpandCollector = new NumericGroupExpandCollector(field, nullValue, groupSet, collapsedSet, limit, sort);
    }
    if (groupQuery != null) {
        //Limits the results to documents that are in the same group as the documents in the page.
        newFilters.add(groupQuery);
    }
    SolrIndexSearcher.ProcessedFilter pfilter = searcher.getProcessedFilter(null, newFilters);
    if (pfilter.postFilter != null) {
        pfilter.postFilter.setLastDelegate(groupExpandCollector);
        collector = pfilter.postFilter;
    } else {
        collector = groupExpandCollector;
    }
    if (pfilter.filter == null) {
        searcher.search(query, collector);
    } else {
        Query q = new BooleanQuery.Builder().add(query, Occur.MUST).add(pfilter.filter, Occur.FILTER).build();
        searcher.search(q, collector);
    }
    LongObjectMap<Collector> groups = ((GroupCollector) groupExpandCollector).getGroups();
    NamedList outMap = new SimpleOrderedMap();
    CharsRefBuilder charsRef = new CharsRefBuilder();
    for (LongObjectCursor<Collector> cursor : groups) {
        long groupValue = cursor.key;
        TopDocsCollector<?> topDocsCollector = TopDocsCollector.class.cast(cursor.value);
        TopDocs topDocs = topDocsCollector.topDocs();
        ScoreDoc[] scoreDocs = topDocs.scoreDocs;
        if (scoreDocs.length > 0) {
            int[] docs = new int[scoreDocs.length];
            float[] scores = new float[scoreDocs.length];
            for (int i = 0; i < docs.length; i++) {
                ScoreDoc scoreDoc = scoreDocs[i];
                docs[i] = scoreDoc.doc;
                scores[i] = scoreDoc.score;
            }
            DocSlice slice = new DocSlice(0, docs.length, docs, scores, topDocs.totalHits, topDocs.getMaxScore());
            if (fieldType instanceof StrField) {
                final BytesRef bytesRef = ordBytes.get((int) groupValue);
                fieldType.indexedToReadable(bytesRef, charsRef);
                String group = charsRef.toString();
                outMap.add(group, slice);
            } else {
                outMap.add(numericToString(fieldType, groupValue), slice);
            }
        }
    }
    rb.rsp.add("expanded", outMap);
}

Example 24 with CharsRefBuilder

use of org.apache.lucene.util.CharsRefBuilder in project lucene-solr by apache.

the class LukeRequestHandler method getDetailedFieldInfo.

// Get terribly detailed information about a particular field. This is a very expensive call, use it with caution
// especially on large indexes!
@SuppressWarnings("unchecked")
private static void getDetailedFieldInfo(SolrQueryRequest req, String field, SimpleOrderedMap<Object> fieldMap) throws IOException {
    SolrParams params = req.getParams();
    final int numTerms = params.getInt(NUMTERMS, DEFAULT_COUNT);
    // Something to collect the top N terms in.
    TopTermQueue tiq = new TopTermQueue(numTerms + 1);
    final CharsRefBuilder spare = new CharsRefBuilder();
    Terms terms = MultiFields.getTerms(req.getSearcher().getIndexReader(), field);
    if (terms == null) {
        // field does not exist
        return;
    }
    TermsEnum termsEnum = terms.iterator();
    BytesRef text;
    int[] buckets = new int[HIST_ARRAY_SIZE];
    while ((text = termsEnum.next()) != null) {
        ++tiq.distinctTerms;
        // This calculation seems odd, but it gives the same results as it used to.
        int freq = termsEnum.docFreq();
        int slot = 32 - Integer.numberOfLeadingZeros(Math.max(0, freq - 1));
        buckets[slot] = buckets[slot] + 1;
        if (numTerms > 0 && freq > tiq.minFreq) {
            spare.copyUTF8Bytes(text);
            String t = spare.toString();
            tiq.add(new TopTermQueue.TermInfo(new Term(field, t), termsEnum.docFreq()));
            if (tiq.size() > numTerms) {
                // if tiq full
                // remove lowest in tiq
                tiq.pop();
                tiq.minFreq = tiq.getTopTermInfo().docFreq;
            }
        }
    }
    tiq.histogram.add(buckets);
    fieldMap.add("distinct", tiq.distinctTerms);
    // Include top terms
    fieldMap.add("topTerms", tiq.toNamedList(req.getSearcher().getSchema()));
    // Add a histogram
    fieldMap.add("histogram", tiq.histogram.toNamedList());
}

Example 25 with CharsRefBuilder

use of org.apache.lucene.util.CharsRefBuilder in project lucene-solr by apache.

the class FieldType method toObject.

public Object toObject(SchemaField sf, BytesRef term) {
    final CharsRefBuilder ref = new CharsRefBuilder();
    indexedToReadable(term, ref);
    final IndexableField f = createField(sf, ref.toString());
    return toObject(f);
}