?

上文solr源碼分析之數據導入DataImporter追溯中提到了solr的工作流程，其核心是各種handler。

?

?handler定義了各種search Component，

@Overridepublic void handleRequestBody(SolrQueryRequest req, SolrQueryResponse rsp) throws Exception{List<SearchComponent> components = getComponents();ResponseBuilder rb = new ResponseBuilder(req, rsp, components); }

然后調用組件的prepare方法：

if (timer == null) {// non-debugging prepare phasefor( SearchComponent c : components ) {c.prepare(rb);}} else {// debugging prepare phaseRTimer subt = timer.sub( "prepare" );for( SearchComponent c : components ) {rb.setTimer( subt.sub( c.getName() ) );c.prepare(rb);rb.getTimer().stop();}subt.stop();}

再調用組件的process方法：

if(!rb.isDebug()) {// Processfor( SearchComponent c : components ) {c.process(rb);}}else {// ProcessRTimer subt = timer.sub( "process" );for( SearchComponent c : components ) {rb.setTimer( subt.sub( c.getName() ) );c.process(rb);rb.getTimer().stop();}subt.stop();// add the timing infoif (rb.isDebugTimings()) {rb.addDebugInfo("timing", timer.asNamedList() );}}

從源碼上來分析一下search Component，searchComponent的結構如下：

?以QueryComponent為例，理解一下它的工作原理。

prepare方法獲取查詢語句：

@Overridepublic 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 ReturnFields returnFields = new SolrReturnFields( req );rsp.setReturnFields( returnFields );int flags = 0;if (returnFields.wantsScore()) {flags |= SolrIndexSearcher.GET_SCORES;}rb.setFieldFlags( flags );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 nothingq = new MatchNoDocsQuery();}rb.setQuery( q );String rankQueryString = rb.req.getParams().get(CommonParams.RQ);if(rankQueryString != null) {QParser rqparser = QParser.getParser(rankQueryString, defType, req);Query rq = rqparser.getQuery();if(rq instanceof RankQuery) {RankQuery rankQuery = (RankQuery)rq;rb.setRankQuery(rankQuery);MergeStrategy mergeStrategy = rankQuery.getMergeStrategy();if(mergeStrategy != null) {rb.addMergeStrategy(mergeStrategy);if(mergeStrategy.handlesMergeFields()) {rb.mergeFieldHandler = mergeStrategy;}}} else {throw new SolrException(SolrException.ErrorCode.BAD_REQUEST,"rq parameter must be a RankQuery");}}rb.setSortSpec( parser.getSort(true) );rb.setQparser(parser);final String cursorStr = rb.req.getParams().get(CursorMarkParams.CURSOR_MARK_PARAM);if (null != cursorStr) {final CursorMark cursorMark = new CursorMark(rb.req.getSchema(),rb.getSortSpec());cursorMark.parseSerializedTotem(cursorStr);rb.setCursorMark(cursorMark);}String[] fqs = req.getParams().getParams(CommonParams.FQ);if (fqs!=null && fqs.length!=0) {List<Query> filters = rb.getFilters();// if filters already exists, make a copy instead of modifying the originalfilters = filters == null ? new ArrayList<Query>(fqs.length) : new ArrayList<>(filters);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 disabledif (!filters.isEmpty()) {rb.setFilters( filters );}}} catch (SyntaxError e) {throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, e);}if (params.getBool(GroupParams.GROUP, false)) {prepareGrouping(rb);} else {//Validate only in case of non-grouping search.if(rb.getSortSpec().getCount() < 0) {throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, "'rows' parameter cannot be negative");}}//Input validation.if (rb.getQueryCommand().getOffset() < 0) {throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, "'start' parameter cannot be negative");}}

process方法執行查詢：

/*** Actually run the query*/@Overridepublic void process(ResponseBuilder rb) throws IOException{LOG.debug("process: {}", rb.req.getParams());SolrQueryRequest req = rb.req;SolrParams params = req.getParams();if (!params.getBool(COMPONENT_NAME, true)) {return;}SolrIndexSearcher searcher = req.getSearcher();StatsCache statsCache = req.getCore().getStatsCache();int purpose = params.getInt(ShardParams.SHARDS_PURPOSE, ShardRequest.PURPOSE_GET_TOP_IDS);if ((purpose & ShardRequest.PURPOSE_GET_TERM_STATS) != 0) {statsCache.returnLocalStats(rb, searcher);return;}// check if we need to update the local copy of global dfsif ((purpose & ShardRequest.PURPOSE_SET_TERM_STATS) != 0) {// retrieve from request and update local cache statsCache.receiveGlobalStats(req);}SolrQueryResponse rsp = rb.rsp;IndexSchema schema = searcher.getSchema();// 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 = schema.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 = searcher.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<>();queries.add(rb.getQuery());List<Query> filters = rb.getFilters();if (filters != null) queries.addAll(filters);res.docSet = searcher.getDocSet(queries);}rb.setResults(res);ResultContext ctx = new ResultContext();ctx.docs = rb.getResults().docList;ctx.query = null; // anything?rsp.add("response", ctx);return;}// -1 as flag if not set.long timeAllowed = params.getLong(CommonParams.TIME_ALLOWED, -1L);if (null != rb.getCursorMark() && 0 < timeAllowed) {// fundamentally incompatiblethrow new SolrException(SolrException.ErrorCode.BAD_REQUEST, "Can not search using both " +CursorMarkParams.CURSOR_MARK_PARAM + " and " + CommonParams.TIME_ALLOWED);}SolrIndexSearcher.QueryCommand cmd = rb.getQueryCommand();cmd.setTimeAllowed(timeAllowed);req.getContext().put(SolrIndexSearcher.STATS_SOURCE, statsCache.get(req));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(schema.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()) {SchemaField schemaField = schema.getField(field);String[] topGroupsParam = params.getParams(GroupParams.GROUP_DISTRIBUTED_TOPGROUPS_PREFIX + field);if (topGroupsParam == null) {topGroupsParam = new String[0];}List<SearchGroup<BytesRef>> topGroups = new ArrayList<>(topGroupsParam.length);for (String topGroup : topGroupsParam) {SearchGroup<BytesRef> searchGroup = new SearchGroup<>();if (!topGroup.equals(TopGroupsShardRequestFactory.GROUP_NULL_VALUE)) {searchGroup.groupValue = new BytesRef(schemaField.getType().readableToIndexed(topGroup));}topGroups.add(searchGroup);}secondPhaseBuilder.addCommandField(new TopGroupsFieldCommand.Builder().setField(schemaField).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.getFunctions() != null) {for (String groupByStr : groupingSpec.getFunctions()) {grouping.addFunctionCommand(groupByStr, 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(Locale.ROOT, "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) {ResultContext ctx = new ResultContext();ctx.docs = grouping.mainResult;ctx.query = null; // TODO? add the query?rsp.add("response", ctx);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 (SyntaxError e) {throw new SolrException(SolrException.ErrorCode.BAD_REQUEST, e);}}// normal search result searcher.search(result, cmd);rb.setResult(result);ResultContext ctx = new ResultContext();ctx.docs = rb.getResults().docList;ctx.query = rb.getQuery();rsp.add("response", ctx);rsp.getToLog().add("hits", rb.getResults().docList.matches());if ( ! rb.req.getParams().getBool(ShardParams.IS_SHARD,false) ) {if (null != rb.getNextCursorMark()) {rb.rsp.add(CursorMarkParams.CURSOR_MARK_NEXT,rb.getNextCursorMark().getSerializedTotem());}}if(rb.mergeFieldHandler != null) {rb.mergeFieldHandler.handleMergeFields(rb, searcher);} else {doFieldSortValues(rb, searcher);}doPrefetch(rb);}

?

轉載于:https://www.cnblogs.com/davidwang456/p/4757337.html

總結

以上是生活随笔為你收集整理的solr源码分析之searchComponent的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。