欢迎投稿

今日深度:

solr QueryComponent,

solr QueryComponent,


/**
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */


package org.apache.solr.handler.component;


import java.io.IOException;
import java.net.URL;
import java.util.ArrayList;
import java.util.Collection;
import java.util.HashMap;
import java.util.Iterator;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.concurrent.Future;


import org.apache.commons.lang.exception.ExceptionUtils;
import org.apache.lucene.document.Field;
import org.apache.lucene.index.Term;
import org.apache.lucene.queryParser.ParseException;
import org.apache.lucene.search.BooleanQuery;
import org.apache.lucene.search.FieldComparator;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.ScoreDoc;
import org.apache.lucene.search.Sort;
import org.apache.lucene.search.SortField;
import org.apache.lucene.search.grouping.GroupDocs;
import org.apache.lucene.search.grouping.SearchGroup;
import org.apache.lucene.search.grouping.TopGroups;
import org.apache.solr.common.SolrDocument;
import org.apache.solr.common.SolrDocumentList;
import org.apache.solr.common.SolrException;
import org.apache.solr.common.params.CommonParams;
import org.apache.solr.common.params.GroupParams;
import org.apache.solr.common.params.ModifiableSolrParams;
import org.apache.solr.common.params.MoreLikeThisParams;
import org.apache.solr.common.params.ShardParams;
import org.apache.solr.common.params.SolrParams;
import org.apache.solr.common.util.NamedList;
import org.apache.solr.common.util.StrUtils;
import org.apache.solr.request.SolrQueryRequest;
import org.apache.solr.response.SolrQueryResponse;
import org.apache.solr.schema.FieldType;
import org.apache.solr.schema.SchemaField;
import org.apache.solr.search.DocIterator;
import org.apache.solr.search.DocList;
import org.apache.solr.search.DocListAndSet;
import org.apache.solr.search.DocSlice;
import org.apache.solr.search.Grouping;
import org.apache.solr.search.QParser;
import org.apache.solr.search.QParserPlugin;
import org.apache.solr.search.QueryParsing;
import org.apache.solr.search.SolrIndexReader;
import org.apache.solr.search.SolrIndexSearcher;
import org.apache.solr.search.SortSpec;
import org.apache.solr.search.grouping.CommandHandler;
import org.apache.solr.search.grouping.GroupingSpecification;
import org.apache.solr.search.grouping.distributed.ShardRequestFactory;
import org.apache.solr.search.grouping.distributed.ShardResponseProcessor;
import org.apache.solr.search.grouping.distributed.command.QueryCommand;
import org.apache.solr.search.grouping.distributed.command.SearchGroupsFieldCommand;
import org.apache.solr.search.grouping.distributed.command.TopGroupsFieldCommand;
import org.apache.solr.search.grouping.distributed.requestfactory.SearchGroupsRequestFactory;
import org.apache.solr.search.grouping.distributed.requestfactory.StoredFieldsShardRequestFactory;
import org.apache.solr.search.grouping.distributed.requestfactory.TopGroupsShardRequestFactory;
import org.apache.solr.search.grouping.distributed.responseprocessor.SearchGroupShardResponseProcessor;
import org.apache.solr.search.grouping.distributed.responseprocessor.StoredFieldsShardResponseProcessor;
import org.apache.solr.search.grouping.distributed.responseprocessor.TopGroupsShardResponseProcessor;
import org.apache.solr.search.grouping.distributed.shardresultserializer.SearchGroupsResultTransformer;
import org.apache.solr.search.grouping.distributed.shardresultserializer.TopGroupsResultTransformer;
import org.apache.solr.search.grouping.endresulttransformer.EndResultTransformer;
import org.apache.solr.search.grouping.endresulttransformer.GroupedEndResultTransformer;
import org.apache.solr.search.grouping.endresulttransformer.MainEndResultTransformer;
import org.apache.solr.search.grouping.endresulttransformer.SimpleEndResultTransformer;
import org.apache.solr.util.SolrPluginUtils;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;


/**
 * TODO!
 * 
 * @version $Id: QueryComponent.java 1338227 2012-05-14 14:38:37Z mvg $
 * @since solr 1.3
 */
public class QueryComponent extends SearchComponent
{
  public static final String COMPONENT_NAME = "query";
  
  @Override
  public void prepare(ResponseBuilder rb) throws IOException
  {


    SolrQueryRequest req = rb.req;
    SolrParams params = req.getParams();
    if (!params.getBool(COMPONENT_NAME, true)) {
      return;
    }
    SolrQueryResponse rsp = rb.rsp;


    // Set field flags
    String fl = params.get(CommonParams.FL);
    int fieldFlags = 0;
    if (fl != null) {
      fieldFlags |= SolrPluginUtils.setReturnFields(fl, rsp);
    }
    rb.setFieldFlags( fieldFlags );


    String defType = params.get(QueryParsing.DEFTYPE,QParserPlugin.DEFAULT_QTYPE);


    // get it from the response builder to give a different component a chance
    // to set it.
    String queryString = rb.getQueryString();
    if (queryString == null) {
      // this is the normal way it's set.
      queryString = params.get( CommonParams.Q );
      rb.setQueryString(queryString);
    }


    try {
      QParser parser = QParser.getParser(rb.getQueryString(), defType, req);
      Query q = parser.getQuery();
      if (q == null) {
        // normalize a null query to a query that matches nothing
        q = new BooleanQuery();        
      }
      rb.setQuery( q );
      rb.setSortSpec( parser.getSort(true) );
      rb.setQparser(parser);


      String[] fqs = req.getParams().getParams(CommonParams.FQ);
      if (fqs!=null && fqs.length!=0) {
        List<Query> filters = rb.getFilters();
        if (filters==null) {
          filters = new ArrayList<Query>(fqs.length);
        }
        for (String fq : fqs) {
          if (fq != null && fq.trim().length()!=0) {
            QParser fqp = QParser.getParser(fq, null, req);
            filters.add(fqp.getQuery());
          }
        }
        // only set the filters if they are not empty otherwise
        // fq=&someotherParam= will trigger all docs filter for every request 
        // if filter cache is disabled
        if (!filters.isEmpty()) {
          rb.setFilters( filters );
        }
      }
    } catch (ParseException e) {
      throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, e);
    }


    // TODO: temporary... this should go in a different component.
    String shards = params.get(ShardParams.SHARDS);
    if (shards != null) {
      List<String> lst = StrUtils.splitSmart(shards, ",", true);
      rb.shards = lst.toArray(new String[lst.size()]);
    }
    String shards_rows = params.get(ShardParams.SHARDS_ROWS);
    if(shards_rows != null) {
      rb.shards_rows = Integer.parseInt(shards_rows);
    }
    String shards_start = params.get(ShardParams.SHARDS_START);
    if(shards_start != null) {
      rb.shards_start = Integer.parseInt(shards_start);
    }


    boolean grouping = params.getBool(GroupParams.GROUP, false);
    if (!grouping) {
      return;
    }
    SolrIndexSearcher.QueryCommand cmd = rb.getQueryCommand();
    SolrIndexSearcher searcher = rb.req.getSearcher();
    GroupingSpecification groupingSpec = new GroupingSpecification();
    rb.setGroupingSpec(groupingSpec);


    //TODO: move weighting of sort
    Sort groupSort = searcher.weightSort(cmd.getSort());
    if (groupSort == null) {
      groupSort = Sort.RELEVANCE;
    }


    // groupSort defaults to sort
    String groupSortStr = params.get(GroupParams.GROUP_SORT);
    //TODO: move weighting of sort
    Sort sortWithinGroup = groupSortStr == null ?  groupSort : searcher.weightSort(QueryParsing.parseSort(groupSortStr, req));
    if (sortWithinGroup == null) {
      sortWithinGroup = Sort.RELEVANCE;
    }
    
    groupingSpec.setSortWithinGroup(sortWithinGroup);
    groupingSpec.setGroupSort(groupSort);


    String formatStr = params.get(GroupParams.GROUP_FORMAT, Grouping.Format.grouped.name());
    Grouping.Format responseFormat;
    try {
       responseFormat = Grouping.Format.valueOf(formatStr);
    } catch (IllegalArgumentException e) {
      throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, String.format("Illegal %s parameter", GroupParams.GROUP_FORMAT));
    }
    groupingSpec.setResponseFormat(responseFormat);


    groupingSpec.setFields(params.getParams(GroupParams.GROUP_FIELD));
    groupingSpec.setQueries(params.getParams(GroupParams.GROUP_QUERY));
    groupingSpec.setGroupOffset(params.getInt(GroupParams.GROUP_OFFSET, 0));
    groupingSpec.setGroupLimit(params.getInt(GroupParams.GROUP_LIMIT, 1));
    groupingSpec.setOffset(rb.getSortSpec().getOffset());
    groupingSpec.setLimit(rb.getSortSpec().getCount());
    groupingSpec.setIncludeGroupCount(params.getBool(GroupParams.GROUP_TOTAL_COUNT, false));
    groupingSpec.setMain(params.getBool(GroupParams.GROUP_MAIN, false));
    groupingSpec.setNeedScore((cmd.getFlags() & SolrIndexSearcher.GET_SCORES) != 0);
    groupingSpec.setTruncateGroups(params.getBool(GroupParams.GROUP_TRUNCATE, false));
  }


  /**
   * Actually run the query
   */
  @Override
  public void process(ResponseBuilder rb) throws IOException
  {
    SolrQueryRequest req = rb.req;
    SolrQueryResponse rsp = rb.rsp;
    SolrParams params = req.getParams();
    if (!params.getBool(COMPONENT_NAME, true)) {
      return;
    }
    SolrIndexSearcher searcher = req.getSearcher();


    if (rb.getQueryCommand().getOffset() < 0) {
      throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, "'start' parameter cannot be negative");
    }


    // -1 as flag if not set.
    long timeAllowed = (long)params.getInt( CommonParams.TIME_ALLOWED, -1 );


    // Optional: This could also be implemented by the top-level searcher sending
    // a filter that lists the ids... that would be transparent to
    // the request handler, but would be more expensive (and would preserve score
    // too if desired).
    String ids = params.get(ShardParams.IDS);
    if (ids != null) {
      SchemaField idField = req.getSchema().getUniqueKeyField();
      List<String> idArr = StrUtils.splitSmart(ids, ",", true);
      int[] luceneIds = new int[idArr.size()];
      int docs = 0;
      for (int i=0; i<idArr.size(); i++) {
        int id = req.getSearcher().getFirstMatch(
                new Term(idField.getName(), idField.getType().toInternal(idArr.get(i))));
        if (id >= 0)
          luceneIds[docs++] = id;
      }


      DocListAndSet res = new DocListAndSet();
      res.docList = new DocSlice(0, docs, luceneIds, null, docs, 0);
      if (rb.isNeedDocSet()) {
        // TODO: create a cache for this!
        List<Query> queries = new ArrayList<Query>();
        queries.add(rb.getQuery());
        List<Query> filters = rb.getFilters();
        if (filters != null) queries.addAll(filters);
        res.docSet = searcher.getDocSet(queries);
      }
      rb.setResults(res);
      rsp.add("response",rb.getResults().docList);
      return;
    }


    SolrIndexSearcher.QueryCommand cmd = rb.getQueryCommand();
    cmd.setTimeAllowed(timeAllowed);
    SolrIndexSearcher.QueryResult result = new SolrIndexSearcher.QueryResult();
      
    //
    // grouping / field collapsing
    //
    GroupingSpecification groupingSpec = rb.getGroupingSpec();
    if (groupingSpec != null) {
      try {
        boolean needScores = (cmd.getFlags() & SolrIndexSearcher.GET_SCORES) != 0;
        if (params.getBool(GroupParams.GROUP_DISTRIBUTED_FIRST, false)) {
          CommandHandler.Builder topsGroupsActionBuilder = new CommandHandler.Builder()
              .setQueryCommand(cmd)
              .setNeedDocSet(false) // Order matters here
              .setIncludeHitCount(true)
              .setSearcher(searcher);


          for (String field : groupingSpec.getFields()) {
            topsGroupsActionBuilder.addCommandField(new SearchGroupsFieldCommand.Builder()
                .setField(searcher.getSchema().getField(field))
                .setGroupSort(groupingSpec.getGroupSort())
                .setTopNGroups(cmd.getOffset() + cmd.getLen())
                .setIncludeGroupCount(groupingSpec.isIncludeGroupCount())
                .build()
            );
          }


          CommandHandler commandHandler = topsGroupsActionBuilder.build();
          commandHandler.execute();
          SearchGroupsResultTransformer serializer = new SearchGroupsResultTransformer(searcher);
          rsp.add("firstPhase", commandHandler.processResult(result, serializer));
          rsp.add("totalHitCount", commandHandler.getTotalHitCount());
          rb.setResult(result);
          return;
        } else if (params.getBool(GroupParams.GROUP_DISTRIBUTED_SECOND, false)) {
          CommandHandler.Builder secondPhaseBuilder = new CommandHandler.Builder()
              .setQueryCommand(cmd)
              .setTruncateGroups(groupingSpec.isTruncateGroups() && groupingSpec.getFields().length > 0)
              .setSearcher(searcher);


          for (String field : groupingSpec.getFields()) {
            String[] topGroupsParam = params.getParams(GroupParams.GROUP_DISTRIBUTED_TOPGROUPS_PREFIX + field);
            if (topGroupsParam == null) {
              topGroupsParam = new String[0];
            }


            List<SearchGroup<String>> topGroups = new ArrayList<SearchGroup<String>>(topGroupsParam.length);
            for (String topGroup : topGroupsParam) {
              SearchGroup<String> searchGroup = new SearchGroup<String>();
              if (!topGroup.equals(TopGroupsShardRequestFactory.GROUP_NULL_VALUE)) {
                searchGroup.groupValue = searcher.getSchema().getField(field).getType().readableToIndexed(topGroup);
              }
              topGroups.add(searchGroup);
            }


            secondPhaseBuilder.addCommandField(
                new TopGroupsFieldCommand.Builder()
                    .setField(searcher.getSchema().getField(field))
                    .setGroupSort(groupingSpec.getGroupSort())
                    .setSortWithinGroup(groupingSpec.getSortWithinGroup())
                    .setFirstPhaseGroups(topGroups)
                    .setMaxDocPerGroup(groupingSpec.getGroupOffset() + groupingSpec.getGroupLimit())
                    .setNeedScores(needScores)
                    .setNeedMaxScore(needScores)
                    .build()
            );
          }


          for (String query : groupingSpec.getQueries()) {
            secondPhaseBuilder.addCommandField(new QueryCommand.Builder()
                .setDocsToCollect(groupingSpec.getOffset() + groupingSpec.getLimit())
                .setSort(groupingSpec.getGroupSort())
                .setQuery(query, rb.req)
                .setDocSet(searcher)
                .build()
            );
          }


          CommandHandler commandHandler = secondPhaseBuilder.build();
          commandHandler.execute();
          TopGroupsResultTransformer serializer = new TopGroupsResultTransformer(rb);
          rsp.add("secondPhase", commandHandler.processResult(result, serializer));
          rb.setResult(result);
          return;
        }


        int maxDocsPercentageToCache = params.getInt(GroupParams.GROUP_CACHE_PERCENTAGE, 0);
        boolean cacheSecondPassSearch = maxDocsPercentageToCache >= 1 && maxDocsPercentageToCache <= 100;
        Grouping.TotalCount defaultTotalCount = groupingSpec.isIncludeGroupCount() ?
            Grouping.TotalCount.grouped : Grouping.TotalCount.ungrouped;
        int limitDefault = cmd.getLen(); // this is normally from "rows"
        Grouping grouping =
            new Grouping(searcher, result, cmd, cacheSecondPassSearch, maxDocsPercentageToCache, groupingSpec.isMain());
        grouping.setSort(groupingSpec.getGroupSort())
            .setGroupSort(groupingSpec.getSortWithinGroup())
            .setDefaultFormat(groupingSpec.getResponseFormat())
            .setLimitDefault(limitDefault)
            .setDefaultTotalCount(defaultTotalCount)
            .setDocsPerGroupDefault(groupingSpec.getGroupLimit())
            .setGroupOffsetDefault(groupingSpec.getGroupOffset())
            .setGetGroupedDocSet(groupingSpec.isTruncateGroups());


        if (groupingSpec.getFields() != null) {
          for (String field : groupingSpec.getFields()) {
            grouping.addFieldCommand(field, rb.req);
          }
        }


        if (groupingSpec.getQueries() != null) {
          for (String groupByStr : groupingSpec.getQueries()) {
            grouping.addQueryCommand(groupByStr, rb.req);
          }
        }


        if (rb.doHighlights || rb.isDebug() || params.getBool(MoreLikeThisParams.MLT, false)) {
          // we need a single list of the returned docs
          cmd.setFlags(SolrIndexSearcher.GET_DOCLIST);
        }


        grouping.execute();
        if (grouping.isSignalCacheWarning()) {
          rsp.add(
              "cacheWarning",
              String.format("Cache limit of %d percent relative to maxdoc has exceeded. Please increase cache size or disable caching.", maxDocsPercentageToCache)
          );
        }
        rb.setResult(result);


        if (grouping.mainResult != null) {
          rsp.add("response", grouping.mainResult);
          rsp.getToLog().add("hits", grouping.mainResult.matches());
        } else if (!grouping.getCommands().isEmpty()) { // Can never be empty since grouping.execute() checks for this.
          rsp.add("grouped", result.groupedResults);
          rsp.getToLog().add("hits", grouping.getCommands().get(0).getMatches());
        }
        return;
      } catch (ParseException e) {
        throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, e);
      }
    }


//modified by wangqiang406 2012-09-27
    //处理问题java.io.IOException: Stale NFS file handle: NIOFSIndexInput
    try {
searcher.search(result, cmd);
} catch (IOException e) {
final Logger logger = LoggerFactory.getLogger(QueryComponent.class);
logger.error("IOException,重新加载Searcher再查一次...");
logger.error(ExceptionUtils.getStackTrace(e));
searcher = req.getCore().getSearcher(true, true, new Future[1]).get();
searcher.search(result, cmd);
}
rb.setResult( result );


    rsp.add("response",rb.getResults().docList);
    rsp.getToLog().add("hits", rb.getResults().docList.matches());


    doFieldSortValues(rb, searcher);
    doPrefetch(rb);
  }


  protected void doFieldSortValues(ResponseBuilder rb, SolrIndexSearcher searcher) throws IOException
  {
    SolrQueryRequest req = rb.req;
    SolrQueryResponse rsp = rb.rsp;


    // The query cache doesn't currently store sort field values, and SolrIndexSearcher doesn't
    // currently have an option to return sort field values.  Because of this, we
    // take the documents given and re-derive the sort values.
    boolean fsv = req.getParams().getBool(ResponseBuilder.FIELD_SORT_VALUES,false);
    if(fsv){
      Sort sort = searcher.weightSort(rb.getSortSpec().getSort());
      SortField[] sortFields = sort==null ? new SortField[]{SortField.FIELD_SCORE} : sort.getSort();
      NamedList<List> sortVals = new NamedList<List>(); // order is important for the sort fields
      Field field = new Field("dummy", "", Field.Store.YES, Field.Index.NO); // a dummy Field


      SolrIndexReader reader = searcher.getReader();
      SolrIndexReader[] readers = reader.getLeafReaders();
      SolrIndexReader subReader = reader;
      if (readers.length==1) {
        // if there is a single segment, use that subReader and avoid looking up each time
        subReader = readers[0];
        readers=null;
      }
      int[] offsets = reader.getLeafOffsets();


      for (SortField sortField: sortFields) {
        int type = sortField.getType();
        if (type==SortField.SCORE || type==SortField.DOC) continue;


        FieldComparator comparator = null;
        FieldComparator comparators[] = (readers==null) ? null : new FieldComparator[readers.length];


        String fieldname = sortField.getField();
        FieldType ft = fieldname==null ? null : req.getSchema().getFieldTypeNoEx(fieldname);


        DocList docList = rb.getResults().docList;
        ArrayList<Object> vals = new ArrayList<Object>(docList.size());
        DocIterator it = rb.getResults().docList.iterator();


        int offset = 0;
        int idx = 0;


        while(it.hasNext()) {
          int doc = it.nextDoc();
          if (readers != null) {
            idx = SolrIndexReader.readerIndex(doc, offsets);
            subReader = readers[idx];
            offset = offsets[idx];
            comparator = comparators[idx];
          }


          if (comparator == null) {
            comparator = sortField.getComparator(1,0);
            comparator.setNextReader(subReader, offset);
            if (comparators != null)
              comparators[idx] = comparator;
          }


          doc -= offset;  // adjust for what segment this is in
          comparator.copy(0, doc);
          Object val = comparator.value(0);


          // Sortable float, double, int, long types all just use a string
          // comparator. For these, we need to put the type into a readable
          // format.  One reason for this is that XML can't represent all
          // string values (or even all unicode code points).
          // indexedToReadable() should be a no-op and should
          // thus be harmless anyway (for all current ways anyway)
          if (val instanceof String) {
            field.setValue((String)val);
            val = ft.toObject(field);
          }


          vals.add(val);
        }


        sortVals.add(fieldname, vals);
      }


      rsp.add("sort_values", sortVals);
    }
  }


  protected void doPrefetch(ResponseBuilder rb) throws IOException
  {
    SolrQueryRequest req = rb.req;
    SolrQueryResponse rsp = rb.rsp;
    //pre-fetch returned documents
    if (!req.getParams().getBool(ShardParams.IS_SHARD,false) && rb.getResults().docList != null && rb.getResults().docList.size()<=50) {
      // TODO: this may depend on the highlighter component (or other components?)
      SolrPluginUtils.optimizePreFetchDocs(rb.getResults().docList, rb.getQuery(), req, rsp);
    }
  }


  @Override  
  public int distributedProcess(ResponseBuilder rb) throws IOException {
    if (rb.grouping()) {
      return groupedDistributedProcess(rb);
    } else {
      return regularDistributedProcess(rb);
    }
  }


  private int groupedDistributedProcess(ResponseBuilder rb) {
    int nextStage = ResponseBuilder.STAGE_DONE;
    ShardRequestFactory shardRequestFactory = null;


    if (rb.stage < ResponseBuilder.STAGE_PARSE_QUERY) {
      nextStage = ResponseBuilder.STAGE_PARSE_QUERY;
    } else if (rb.stage == ResponseBuilder.STAGE_PARSE_QUERY) {
      createDistributedIdf(rb);
      nextStage = ResponseBuilder.STAGE_TOP_GROUPS;
    } else if (rb.stage < ResponseBuilder.STAGE_TOP_GROUPS) {
      nextStage = ResponseBuilder.STAGE_TOP_GROUPS;
    } else if (rb.stage == ResponseBuilder.STAGE_TOP_GROUPS) {
      shardRequestFactory = new SearchGroupsRequestFactory();
      nextStage = ResponseBuilder.STAGE_EXECUTE_QUERY;
    } else if (rb.stage < ResponseBuilder.STAGE_EXECUTE_QUERY) {
      nextStage = ResponseBuilder.STAGE_EXECUTE_QUERY;
    } else if (rb.stage == ResponseBuilder.STAGE_EXECUTE_QUERY) {
      shardRequestFactory = new TopGroupsShardRequestFactory();
      nextStage = ResponseBuilder.STAGE_GET_FIELDS;
    } else if (rb.stage < ResponseBuilder.STAGE_GET_FIELDS) {
      nextStage = ResponseBuilder.STAGE_GET_FIELDS;
    } else if (rb.stage == ResponseBuilder.STAGE_GET_FIELDS) {
      shardRequestFactory = new StoredFieldsShardRequestFactory();
      nextStage = ResponseBuilder.STAGE_DONE;
    }


    if (shardRequestFactory != null) {
      for (ShardRequest shardRequest : shardRequestFactory.constructRequest(rb)) {
        rb.addRequest(this, shardRequest);
      }
    }
    return nextStage;
  }


  private int regularDistributedProcess(ResponseBuilder rb) {
    if (rb.stage < ResponseBuilder.STAGE_PARSE_QUERY)
      return ResponseBuilder.STAGE_PARSE_QUERY;
    if (rb.stage == ResponseBuilder.STAGE_PARSE_QUERY) {
      createDistributedIdf(rb);
      return ResponseBuilder.STAGE_EXECUTE_QUERY;
    }
    if (rb.stage < ResponseBuilder.STAGE_EXECUTE_QUERY) return ResponseBuilder.STAGE_EXECUTE_QUERY;
    if (rb.stage == ResponseBuilder.STAGE_EXECUTE_QUERY) {
      createMainQuery(rb);
      return ResponseBuilder.STAGE_GET_FIELDS;
    }
    if (rb.stage < ResponseBuilder.STAGE_GET_FIELDS) return ResponseBuilder.STAGE_GET_FIELDS;
    if (rb.stage == ResponseBuilder.STAGE_GET_FIELDS) {
      createRetrieveDocs(rb);
      return ResponseBuilder.STAGE_DONE;
    }
    return ResponseBuilder.STAGE_DONE;
  }


  @Override
  public void handleResponses(ResponseBuilder rb, ShardRequest sreq) {
    if (rb.grouping()) {
      handleGroupedResponses(rb, sreq);
    } else {
      handleRegularResponses(rb, sreq);
    }
  }


  private void handleGroupedResponses(ResponseBuilder rb, ShardRequest sreq) {
    ShardResponseProcessor responseProcessor = null;
    if ((sreq.purpose & ShardRequest.PURPOSE_GET_TOP_GROUPS) != 0) {
      responseProcessor = new SearchGroupShardResponseProcessor();
    } else if ((sreq.purpose & ShardRequest.PURPOSE_GET_TOP_IDS) != 0) {
      responseProcessor = new TopGroupsShardResponseProcessor();
    } else if ((sreq.purpose & ShardRequest.PURPOSE_GET_FIELDS) != 0) {
      responseProcessor = new StoredFieldsShardResponseProcessor();
    }


    if (responseProcessor != null) {
      responseProcessor.process(rb, sreq);
    }
  }


  private void handleRegularResponses(ResponseBuilder rb, ShardRequest sreq) {
    if ((sreq.purpose & ShardRequest.PURPOSE_GET_TOP_IDS) != 0) {
      mergeIds(rb, sreq);
    }


    if ((sreq.purpose & ShardRequest.PURPOSE_GET_FIELDS) != 0) {
      returnFields(rb, sreq);
    }
  }


  @Override
  public void finishStage(ResponseBuilder rb) {
    if (rb.stage != ResponseBuilder.STAGE_GET_FIELDS) {
      return;
    }
    if (rb.grouping()) {
      groupedFinishStage(rb);
    } else {
      regularFinishStage(rb);
    }
  }


  private static final EndResultTransformer MAIN_END_RESULT_TRANSFORMER = new MainEndResultTransformer();
  private static final EndResultTransformer SIMPLE_END_RESULT_TRANSFORMER = new SimpleEndResultTransformer();


  @SuppressWarnings("unchecked")
  private void groupedFinishStage(final ResponseBuilder rb) {
    // To have same response as non-distributed request.
    GroupingSpecification groupSpec = rb.getGroupingSpec();
    if (rb.mergedTopGroups.isEmpty()) {
      for (String field : groupSpec.getFields()) {
        rb.mergedTopGroups.put(field, new TopGroups(null, null, 0, 0, new GroupDocs[]{}));
      }
      rb.resultIds = new HashMap<Object, ShardDoc>();
    }


    EndResultTransformer.SolrDocumentSource solrDocumentSource = new EndResultTransformer.SolrDocumentSource() {


      public SolrDocument retrieve(ScoreDoc doc) {
        ShardDoc solrDoc = (ShardDoc) doc;
        return rb.retrievedDocuments.get(solrDoc.id);
      }


    };
    EndResultTransformer endResultTransformer;
    if (groupSpec.isMain()) {
      endResultTransformer = MAIN_END_RESULT_TRANSFORMER;
    } else if (Grouping.Format.grouped == groupSpec.getResponseFormat()) {
      endResultTransformer = new GroupedEndResultTransformer(rb.req.getSearcher());
    } else if (Grouping.Format.simple == groupSpec.getResponseFormat() && !groupSpec.isMain()) {
      endResultTransformer = SIMPLE_END_RESULT_TRANSFORMER;
    } else {
      return;
    }
    Map<String, Object> combinedMap = new LinkedHashMap<String, Object>();
    combinedMap.putAll(rb.mergedTopGroups);
    combinedMap.putAll(rb.mergedQueryCommandResults);
    endResultTransformer.transform(combinedMap, rb, solrDocumentSource);
  }


  private void regularFinishStage(ResponseBuilder rb) {
    // We may not have been able to retrieve all the docs due to an
    // index change.  Remove any null documents.
    for (Iterator<SolrDocument> iter = rb._responseDocs.iterator(); iter.hasNext();) {
      if (iter.next() == null) {
        iter.remove();
        rb._responseDocs.setNumFound(rb._responseDocs.getNumFound()-1);
      }
    }


    rb.rsp.add("response", rb._responseDocs);
  }




  private void createDistributedIdf(ResponseBuilder rb) {
    // TODO
  }


  private void createMainQuery(ResponseBuilder rb) {
    ShardRequest sreq = new ShardRequest();
    sreq.purpose = ShardRequest.PURPOSE_GET_TOP_IDS;


    sreq.params = new ModifiableSolrParams(rb.req.getParams());
    // TODO: base on current params or original params?


    // don't pass through any shards param
    sreq.params.remove(ShardParams.SHARDS);


    // set the start (offset) to 0 for each shard request so we can properly merge
    // results from the start.
    if(rb.shards_start > -1) {
      // if the client set shards.start set this explicitly
      sreq.params.set(CommonParams.START,rb.shards_start);
    } else {
      sreq.params.set(CommonParams.START, "0");
    }
    // TODO: should we even use the SortSpec?  That's obtained from the QParser, and
    // perhaps we shouldn't attempt to parse the query at this level?
    // Alternate Idea: instead of specifying all these things at the upper level,
    // we could just specify that this is a shard request.
    if(rb.shards_rows > -1) {
      // if the client set shards.rows set this explicity
      sreq.params.set(CommonParams.ROWS,rb.shards_rows);
    } else {
      sreq.params.set(CommonParams.ROWS, rb.getSortSpec().getOffset() + rb.getSortSpec().getCount());
    }


    // in this first phase, request only the unique key field
    // and any fields needed for merging.
    sreq.params.set(ResponseBuilder.FIELD_SORT_VALUES,"true");


    if ( (rb.getFieldFlags() & SolrIndexSearcher.GET_SCORES)!=0 || rb.getSortSpec().includesScore()) {
      sreq.params.set(CommonParams.FL, rb.req.getSchema().getUniqueKeyField().getName() + ",score");
    } else {
      sreq.params.set(CommonParams.FL, rb.req.getSchema().getUniqueKeyField().getName());      
    }


    rb.addRequest(this, sreq);
  }










  private void mergeIds(ResponseBuilder rb, ShardRequest sreq) {
      SortSpec ss = rb.getSortSpec();
      Sort sort = ss.getSort();


      SortField[] sortFields = null;
      if(sort != null) sortFields = sort.getSort();
      else {
        sortFields = new SortField[]{SortField.FIELD_SCORE};
      }
 
      SchemaField uniqueKeyField = rb.req.getSchema().getUniqueKeyField();




      // id to shard mapping, to eliminate any accidental dups
      HashMap<Object,String> uniqueDoc = new HashMap<Object,String>();    


      // Merge the docs via a priority queue so we don't have to sort *all* of the
      // documents... we only need to order the top (rows+start)
      ShardFieldSortedHitQueue queue;
      queue = new ShardFieldSortedHitQueue(sortFields, ss.getOffset() + ss.getCount());


      long numFound = 0;
      Float maxScore=null;
      boolean partialResults = false;
      for (ShardResponse srsp : sreq.responses) {
        SolrDocumentList docs = (SolrDocumentList)srsp.getSolrResponse().getResponse().get("response");


        NamedList<?> responseHeader = (NamedList<?>)srsp.getSolrResponse().getResponse().get("responseHeader");
        if (responseHeader != null && Boolean.TRUE.equals(responseHeader.get("partialResults"))) {
          partialResults = true;
        }


        // calculate global maxScore and numDocsFound
        if (docs.getMaxScore() != null) {
          maxScore = maxScore==null ? docs.getMaxScore() : Math.max(maxScore, docs.getMaxScore());
        }
        numFound += docs.getNumFound();


        NamedList sortFieldValues = (NamedList)(srsp.getSolrResponse().getResponse().get("sort_values"));


        // go through every doc in this response, construct a ShardDoc, and
        // put it in the priority queue so it can be ordered.
        for (int i=0; i<docs.size(); i++) {
          SolrDocument doc = docs.get(i);
          Object id = doc.getFieldValue(uniqueKeyField.getName());


          String prevShard = uniqueDoc.put(id, srsp.getShard());
          if (prevShard != null) {
            // duplicate detected
            numFound--;


            // For now, just always use the first encountered since we can't currently
            // remove the previous one added to the priority queue.  If we switched
            // to the Java5 PriorityQueue, this would be easier.
            continue;
            // make which duplicate is used deterministic based on shard
            // if (prevShard.compareTo(srsp.shard) >= 0) {
            //  TODO: remove previous from priority queue
            //  continue;
            // }
          }


          ShardDoc shardDoc = new ShardDoc();
          shardDoc.id = id;
          shardDoc.shard = srsp.getShard();
          shardDoc.orderInShard = i;
          Object scoreObj = doc.getFieldValue("score");
          if (scoreObj != null) {
            if (scoreObj instanceof String) {
              shardDoc.score = Float.parseFloat((String)scoreObj);
            } else {
              shardDoc.score = (Float)scoreObj;
            }
          }


          shardDoc.sortFieldValues = sortFieldValues;


          queue.insertWithOverflow(shardDoc);
        } // end for-each-doc-in-response
      } // end for-each-response




      // The queue now has 0 -> queuesize docs, where queuesize <= start + rows
      // So we want to pop the last documents off the queue to get
      // the docs offset -> queuesize
      int resultSize = queue.size() - ss.getOffset();
      resultSize = Math.max(0, resultSize);  // there may not be any docs in range


      Map<Object,ShardDoc> resultIds = new HashMap<Object,ShardDoc>();
      for (int i=resultSize-1; i>=0; i--) {
        ShardDoc shardDoc = queue.pop();
        shardDoc.positionInResponse = i;
        // Need the toString() for correlation with other lists that must
        // be strings (like keys in highlighting, explain, etc)
        resultIds.put(shardDoc.id.toString(), shardDoc);
      }




      SolrDocumentList responseDocs = new SolrDocumentList();
      if (maxScore!=null) responseDocs.setMaxScore(maxScore);
      responseDocs.setNumFound(numFound);
      responseDocs.setStart(ss.getOffset());
      // size appropriately
      for (int i=0; i<resultSize; i++) responseDocs.add(null);


      // save these results in a private area so we can access them
      // again when retrieving stored fields.
      // TODO: use ResponseBuilder (w/ comments) or the request context?
      rb.resultIds = resultIds;
      rb._responseDocs = responseDocs;
      if (partialResults) {
        rb.rsp.getResponseHeader().add( "partialResults", Boolean.TRUE );
      }


  }


  private void createRetrieveDocs(ResponseBuilder rb) {


    // TODO: in a system with nTiers > 2, we could be passed "ids" here
    // unless those requests always go to the final destination shard


    // for each shard, collect the documents for that shard.
    HashMap<String, Collection<ShardDoc>> shardMap = new HashMap<String,Collection<ShardDoc>>();
    for (ShardDoc sdoc : rb.resultIds.values()) {
      Collection<ShardDoc> shardDocs = shardMap.get(sdoc.shard);
      if (shardDocs == null) {
        shardDocs = new ArrayList<ShardDoc>();
        shardMap.put(sdoc.shard, shardDocs);
      }
      shardDocs.add(sdoc);
    }


    SchemaField uniqueField = rb.req.getSchema().getUniqueKeyField();


    // Now create a request for each shard to retrieve the stored fields
    for (Collection<ShardDoc> shardDocs : shardMap.values()) {
      ShardRequest sreq = new ShardRequest();
      sreq.purpose = ShardRequest.PURPOSE_GET_FIELDS;


      sreq.shards = new String[] {shardDocs.iterator().next().shard};


      sreq.params = new ModifiableSolrParams();


      // add original params
      sreq.params.add( rb.req.getParams());


      // no need for a sort, we already have order
      sreq.params.remove(CommonParams.SORT);


      // we already have the field sort values
      sreq.params.remove(ResponseBuilder.FIELD_SORT_VALUES);
        
      // disable grouping
   sreq.params.remove("group");


      // make sure that the id is returned for correlation.
      String fl = sreq.params.get(CommonParams.FL);
      if (fl != null) {
         fl = fl.trim();
        // currently, "score" is synonymous with "*,score" so
        // don't add "id" if the fl is empty or "score" or it would change the meaning.
         if (fl.length()!=0 && !"score".equals(fl) && !"*".equals(fl)) {
           sreq.params.set(CommonParams.FL, fl+','+uniqueField.getName());
         }
      }      


      ArrayList<String> ids = new ArrayList<String>(shardDocs.size());
      for (ShardDoc shardDoc : shardDocs) {
        // TODO: depending on the type, we may need more tha a simple toString()?
        ids.add(shardDoc.id.toString());
      }
      sreq.params.add(ShardParams.IDS, StrUtils.join(ids, ','));


      rb.addRequest(this, sreq);
    }


  }




  private void returnFields(ResponseBuilder rb, ShardRequest sreq) {
    // Keep in mind that this could also be a shard in a multi-tiered system.
    // TODO: if a multi-tiered system, it seems like some requests
    // could/should bypass middlemen (like retrieving stored fields)
    // TODO: merge fsv to if requested


    if ((sreq.purpose & ShardRequest.PURPOSE_GET_FIELDS) != 0) {
      boolean returnScores = (rb.getFieldFlags() & SolrIndexSearcher.GET_SCORES) != 0;


      assert(sreq.responses.size() == 1);
      ShardResponse srsp = sreq.responses.get(0);
      SolrDocumentList docs = (SolrDocumentList)srsp.getSolrResponse().getResponse().get("response");


      String keyFieldName = rb.req.getSchema().getUniqueKeyField().getName();


      for (SolrDocument doc : docs) {
        Object id = doc.getFieldValue(keyFieldName);
        ShardDoc sdoc = rb.resultIds.get(id.toString());
        if (sdoc != null) {
          if (returnScores && sdoc.score != null) {
              doc.setField("score", sdoc.score);
          }
          rb._responseDocs.set(sdoc.positionInResponse, doc);
        }
      }      
    }
  }


  /////////////////////////////////////////////
  ///  SolrInfoMBean
  ////////////////////////////////////////////


  @Override
  public String getDescription() {
    return "query";
  }


  @Override
  public String getVersion() {
    return "$Revision: 1338227 $";
  }


  @Override
  public String getSourceId() {
    return "$Id: QueryComponent.java 1338227 2012-05-14 14:38:37Z mvg $";
  }


  @Override
  public String getSource() {
    return "$URL: https://svn.apache.org/repos/asf/lucene/dev/branches/lucene_solr_3_6/solr/core/src/java/org/apache/solr/handler/component/QueryComponent.java $";
  }


  @Override
  public URL[] getDocs() {
    return null;
  }
}

www.htsjk.Com true http://www.htsjk.com/solr/39065.html NewsArticle solr QueryComponent, /**  * Licensed to the Apache Software Foundation (ASF) under one or more  * contributor license agreements.  See the NOTICE file distributed with  * this work for additional information regarding copyright owners...
相关文章
    暂无相关文章
评论暂时关闭