solr4.3纠错源码学习（一）——SpellCheckComponent

首先从solr的启动开始说，solr本身有很多component，在solr启动时回一个一个的调用，每个component都包含一个prepare()方法和一个process()

components的逐个调用在solr/core/src/java文件夹下的org.apache.solr.handler.component包SearchHandler.java

component构造如下图：

当然，我们要研究的纠错存在于Spellcheck Component。

在SearchHandler.java中定义了两个循环，对各个component的prepare()方法和process()方法进行调用，代码如下：

下面，我们进入spellcheckComponent

 在spellcheckComponent中，有两个主要方法：prepare()方法

public void prepare(ResponseBuilder rb) throws IOException {

    SolrParams params = rb.req.getParams();
    if (!params.getBool(COMPONENT_NAME, false)) {
      return;
    }
    SolrSpellChecker spellChecker = getSpellChecker(params);
    if (params.getBool(SPELLCHECK_BUILD, false)) {
      spellChecker.build(rb.req.getCore(), rb.req.getSearcher());
      rb.rsp.add("command", "build");
    } else if (params.getBool(SPELLCHECK_RELOAD, false)) {
      spellChecker.reload(rb.req.getCore(), rb.req.getSearcher());
      rb.rsp.add("command", "reload");
    }
  }

prepare

process()方法

public void process(ResponseBuilder rb) throws IOException {
              ......
    SolrParams params = rb.req.getParams();
    if (!params.getBool(COMPONENT_NAME, false) || spellCheckers.isEmpty()) {
      return;
    }
    boolean shardRequest = "true".equals(params.get(ShardParams.IS_SHARD));
    String q = params.get(SPELLCHECK_Q);//获取查询词
    SolrSpellChecker spellChecker = getSpellChecker(params);//获取参数，包括词典
    Collection<Token> tokens = null;
    
             .......

if (tokens != null && tokens.isEmpty() == false) {
      if (spellChecker != null) {
        int count = params.getInt(SPELLCHECK_COUNT, 1);
        boolean onlyMorePopular = params.getBool(SPELLCHECK_ONLY_MORE_POPULAR, DEFAULT_ONLY_MORE_POPULAR);
        System.out.println("onlyMorePopular"+onlyMorePopular);//onlyMorePopularfalse
        boolean extendedResults = params.getBool(SPELLCHECK_EXTENDED_RESULTS, false); 
        System.out.println("extendedResults"+extendedResults);//extendedResultsfalse
        boolean collate = params.getBool(SPELLCHECK_COLLATE, false);
        System.out.println("collate"+collate);//collatetrue
        float accuracy = params.getFloat(SPELLCHECK_ACCURACY, Float.MIN_VALUE);
        Integer alternativeTermCount = params.getInt(SpellingParams.SPELLCHECK_ALTERNATIVE_TERM_COUNT); 
        Integer maxResultsForSuggest = params.getInt(SpellingParams.SPELLCHECK_MAX_RESULTS_FOR_SUGGEST);
        ModifiableSolrParams customParams = new ModifiableSolrParams();//载入配置文件中的部分选项
        for (String checkerName : getDictionaryNames(params)) {
          System.out.println("getDictionaryNames(params):"+getDictionaryNames(params));//[Ljava.lang.String;@9bdb78
          customParams.add(getCustomParams(checkerName, params));        
        }//逐一查询词典名称，并将其放入customParams（用户参数）
        
        Integer hitsInteger = (Integer) rb.rsp.getToLog().get("hits");
        System.out.println(hitsInteger);//0
        long hits = 0;
        if (hitsInteger == null) {
          hits = rb.getNumberDocumentsFound();
        } else {
          hits = hitsInteger.longValue();
          System.out.println("hits"+hits);//hits0
        }
                      .......

boolean isCorrectlySpelled = hits > (maxResultsForSuggest==null ? 0 : maxResultsForSuggest);//判断词是否正确，返回isCorrectlySpelled
        NamedList suggestions = toNamedList(shardRequest, spellingResult, q,extendedResults, collate, isCorrectlySpelled);//提供建议词suggestions
        if (collate) {
          addCollationsToResponse(params, spellingResult, rb, q, suggestions, spellChecker.isSuggestionsMayOverlap());
        }
        NamedList response = new SimpleOrderedMap();
        response.add("suggestions", suggestions);
        System.out.println("suggestions"+suggestions);
        //suggestions{中国银航={numFound=1,startOffset=0,endOffset=4,suggestion=[中国银行]},collation={collationQuery=中国银行,hits=3,misspellingsAndCorrections={中国银航=中国银行}}}
        rb.rsp.add("spellcheck", response);//spellcheck，response的值：{suggestions={中国银航={numFound=1,startOffset=0,endOffset=4,suggestion=[中国银行]},collation={collationQuery=中国银行,hits=3,misspellingsAndCorrections={中国银航=中国银行}}}}
      } else {
        throw new SolrException(SolrException.ErrorCode.NOT_FOUND,
            "Specified dictionaries do not exist: " + getDictionaryNameAsSingleString(getDictionaryNames(params)));
      }
    }
  }

 

关键函数boolean isCorrectlySpelled = hits > (maxResultsForSuggest==null ? 0 : maxResultsForSuggest);//判断词是否正确，返回isCorrectlySpelled

NamedList suggestions = toNamedList(shardRequest, spellingResult, q,extendedResults, collate, isCorrectlySpelled);//提供建议词suggestions

 

跟踪：

在solr还没有启动之前，

• 先载入public void init(NamedList args)和 public void prepare(ResponseBuilder rb) 

• 在prepare（）方法中params.getBool(COMPONENT_NAME, false)为false；params.getBool(SPELLCHECK_BUILD, false)为false

• 再启动solr之后，进入browse之前，又调用到了prepare方法，载入了dictName——SpellCheckComponent.java:dictName:[Ljava.lang.String;

　　　 并输出webapp=/solr path=/browse params={} hits=415 status=0 QTime=1304091进入browse界面

当在browse界面中传入查询词q之后

 

转载于:https://www.cnblogs.com/sunshineKID/p/3459123.html