[solr] - defType - 查詢權重排序

Solr的defType有dismax/edismax兩種，這兩種的區別，可參見：http://blog.csdn.net/duck_genuine/article/details/8060026

 

下面示例用於演示以下場景：

有一網站，在用戶查詢的結果中，須要按這樣排序：

1. VIP的付費信息須要排在免費信息的前頭
2. 點擊率越高越靠前
3. 發佈時間越晚的越靠前

 

這樣的查詢排序使用普通的查詢結果的Order by是作不到的，必需使用solr的defType。

 

作法：

一、先看schema.xml的定義：

<?xml version="1.0" ?>
<schema name="sample5" version="1.1">

    <fieldtype name="string"  class="solr.StrField" sortMissingLast="true" omitNorms="true"/>
    <fieldType name="long" class="solr.TrieLongField" precisionStep="0" positionIncrementGap="0"/>
    <fieldType name="tdate" class="solr.TrieDateField" precisionStep="6" positionIncrementGap="0"/>
    <fieldType name="int" class="solr.TrieIntField" precisionStep="0" positionIncrementGap="0"/>
    <fieldType name="float" class="solr.TrieFloatField" precisionStep="0" positionIncrementGap="0"/>
    <fieldType name="double" class="solr.TrieDoubleField" precisionStep="0" positionIncrementGap="0"/>
    <fieldType name="boolean" class="solr.BoolField" sortMissingLast="true"/>
    <fieldtype name="binary" class="solr.BinaryField"/>
    <fieldType name="text_cn" class="solr.TextField">
        <analyzer type="index" class="org.wltea.analyzer.lucene.IKAnalyzer" useSmart="false" />
        <analyzer type="query" class="org.wltea.analyzer.lucene.IKAnalyzer" useSmart="true" />
        <analyzer>
            <tokenizer class="solr.KeywordTokenizerFactory"/>
            <filter class="solr.LowerCaseFilterFactory" ignoreCase="true"/>
        </analyzer>
    </fieldType>
    
    <!-- general -->
    <fields>
        <field name="id" type="long" indexed="true" stored="true" multiValued="false" required="true"/>
        <field name="subject" type="text_cn" indexed="true" stored="true" />
        <field name="content" type="text_cn" indexed="true" stored="true" />
        <field name="regionId" type="int" indexed="true" stored="true" />
        <field name="region" type="text_cn" indexed="true" stored="true" />
        <field name="categoryId" type="int" indexed="true" stored="true" />
        <field name="category" type="text_cn" indexed="true" stored="true" />
        <field name="price" type="float" indexed="true" stored="true" />
        <field name="createTime" type="tdate" indexed="true" stored="true" />
        <field name="point" type="long" indexed="true" stored="true" />
        <field name="vip" type="boolean" indexed="true" stored="true" />
        <field name="_version_" type="long" indexed="true" stored="true"/>
        <field name="searchText" type="text_cn" indexed="true" stored="false" multiValued="true" />
    </fields>
    
    <copyField source="subject" dest="searchText" />
    <copyField source="content" dest="searchText" />
    <copyField source="region" dest="searchText" />
    <copyField source="category" dest="searchText" />
    
     <!-- field to use to determine and enforce document uniqueness. -->
     <uniqueKey>id</uniqueKey>

     <!-- field for the QueryParser to use when an explicit fieldname is absent -->
     <defaultSearchField>searchText</defaultSearchField>

     <!-- SolrQueryParser configuration: defaultOperator="AND|OR" -->
     <solrQueryParser defaultOperator="AND"/>
</schema>

說明：

a)裏頭定義了一個copyField：searchText，此字段爲：subject+content+region+category，並把這個字段設置爲默認查詢字段。意思是查詢時，默認查詢四個字段的內容。

b)把solrQueryParser設置爲AND，事實上，大多狀況下，咱們是習慣使用AND爲條件查詢，而非OR

c)text_cn字段類型中的：useSmart

        <analyzer type="index" class="org.wltea.analyzer.lucene.IKAnalyzer" useSmart="false" />
        <analyzer type="query" class="org.wltea.analyzer.lucene.IKAnalyzer" useSmart="true" />

意思是：useSmart =true ，分詞器使用智能切分策略， =false則使用細粒度切分。詳細，可下載IK分詞器的源碼看看。

 

二、加入一個查詢Handler到solrconfig.xml的<config/>當中：

    <requestHandler name="/browse" class="solr.SearchHandler" default="true" >
        <lst name="defaults">
            <str name="defType">edismax</str>
            <str name="bf">
                sum(linear(vip,1000,0),linear(sqrt(log(linear(point,1,2))),100,0),sqrt(log(ms(createTime))))
            </str>
            <!--<str name="pf">
            searchText
            </str>
            <str name="qf">
            subject^1 content^0.8
            </str>-->
        </lst>
    </requestHandler>

說明：

a)上面的default="true"意思爲設置爲默認的查詢handler（記得把原standard中的default="true"刪除掉）

b)見已經被註釋的這段：

            <!--<str name="pf">
            searchText
            </str>
            <str name="qf">
            subject^1 content^0.8
            </str>-->

這是簡單的不使用bf的排序加權方式，能夠用於應付簡單的排序，具體pf/qf的使用，能夠上網上搜搜應用。這裏演示的功能相對「複雜」，不適用它。

c)見這句公式：

sum(linear(vip,1000,0),linear(sqrt(log(linear(point,1,2))),100,0),sqrt(log(ms(createTime))))

公式中的函數定義和意思，能夠參考： 

官方文檔：

http://wiki.apache.org/solr/FunctionQuery

中文說明：

http://mxsfengg.iteye.com/blog/352191

這裏的函數意思是：

• 若是是vip信息=值+1000，非vip信息=值+0
• 點擊率(point)的值範圍爲：50~500之間
• 發佈時間(createTime)值範圍爲：50之內

以上三個值相加得出最統權重分從高到低排序

 

三、Java bean:

package com.my.entity;

import java.util.Date;

import org.apache.solr.client.solrj.beans.Field;

public class Item {
    @Field
    private long id;
    @Field
    private String subject;
    @Field
    private String content;
    @Field
    private int regionId;
    @Field
    private int categoryId;
    @Field
    private float price;
    @Field
    private Date createTime;
    @Field
    private long point;
    @Field
    private boolean vip;
    
    public long getId() {
        return id;
    }
    public void setId(long id) {
        this.id = id;
    }
    public String getSubject() {
        return subject;
    }
    public void setSubject(String subject) {
        this.subject = subject;
    }
    public String getContent() {
        return content;
    }
    public void setContent(String content) {
        this.content = content;
    }
    public int getRegionId() {
        return regionId;
    }
    public void setRegionId(int regionId) {
        this.regionId = regionId;
    }
    public int getCategoryId() {
        return categoryId;
    }
    public void setCategoryId(int categoryId) {
        this.categoryId = categoryId;
    }
    public float getPrice() {
        return price;
    }
    public void setPrice(float price) {
        this.price = price;
    }
    public Date getCreateTime() {
        return createTime;
    }
    public void setCreateTime(Date createTime) {
        this.createTime = createTime;
    }
    public long getPoint() {
        return point;
    }
    public void setPoint(long point) {
        this.point = point;
    }
    public boolean isVip() {
        return vip;
    }
    public void setVip(boolean vip) {
        this.vip = vip;
    }
}

 

四、Java測試代碼：

package com.my.solr;

import java.io.IOException;
import java.util.ArrayList;
import java.util.Calendar;
import java.util.Date;
import java.util.HashMap;
import java.util.Iterator;
import java.util.List;

import org.apache.solr.client.solrj.SolrQuery;
import org.apache.solr.client.solrj.SolrQuery.ORDER;
import org.apache.solr.client.solrj.SolrQuery.SortClause;
import org.apache.solr.client.solrj.SolrServerException;
import org.apache.solr.client.solrj.impl.HttpSolrServer;
import org.apache.solr.client.solrj.impl.XMLResponseParser;
import org.apache.solr.client.solrj.response.FacetField;
import org.apache.solr.client.solrj.response.FacetField.Count;
import org.apache.solr.client.solrj.response.QueryResponse;
import org.apache.solr.common.params.AnalysisParams;
import org.apache.solr.common.params.CommonParams;
import org.apache.solr.common.params.FacetParams;
import org.apache.solr.common.util.NamedList;
import org.apache.solr.common.util.SimpleOrderedMap;

import com.my.entity.Item;

public class TestSolr {
    private static HashMap<Integer, String> mapRegion = new HashMap<Integer, String>();
    private static HashMap<Integer, String> mapCategory = new HashMap<Integer, String>();

    @SuppressWarnings("unchecked")
    public static void main(String[] args) throws IOException,
            SolrServerException {
        // ------------------------------------------------------
        // Set map
        // ------------------------------------------------------
        mapRegion.put(1, "羅湖區");
        mapRegion.put(2, "南山區");
        mapRegion.put(3, "龍崗區");
        mapRegion.put(4, "福田區");
        mapCategory.put(1, "單間");
        mapCategory.put(2, "2房1廳");
        mapCategory.put(3, "3房2廳");
        mapCategory.put(4, "1房1廳");
        
        String url = "http://localhost:8983/solr/sample5";
        HttpSolrServer core = new HttpSolrServer(url);
        core.setMaxRetries(1);
        core.setConnectionTimeout(5000);
        core.setParser(new XMLResponseParser()); // binary parser is used by
                                                    // default
        core.setSoTimeout(1000); // socket read timeout
        core.setDefaultMaxConnectionsPerHost(100);
        core.setMaxTotalConnections(100);
        core.setFollowRedirects(false); // defaults to false
        core.setAllowCompression(true);

        // ------------------------------------------------------
        // remove all data
        // ------------------------------------------------------
        core.deleteByQuery("*:*");
        List<Item> items = new ArrayList<Item>();
        items.add(makeItem(items.size() + 1, "龍城公寓一房一廳", "豪華城城公寓1房1廳，擰包入住", 1, 1, 1200f, 10, false));
        items.add(makeItem(items.size() + 1, "興新宿舍樓 1室0廳", " 中等裝修 招女性合租", 1, 1, 1000f, 11, false));
        items.add(makeItem(items.size() + 1, "西麗新屋村新宿舍樓單間", " 無敵裝修只招女性", 2, 1, 1000f, 2, true));
        items.add(makeItem(items.size() + 1, "大芬村信和愛琴居地鐵口2房1廳", " 地鐵口 + 出行便利=居家首選", 3, 2, 2000f, 5, false));
        items.add(makeItem(items.size() + 1, "龍崗富豪花園3房2廳出租", " 離地鐵口只要5分鐘，快來秒殺吧", 3, 3, 4500f, 21, true));
        items.add(makeItem(items.size() + 1, "海景房園3房2廳出租", "海景房園出租，無敵海景，能夠看到倫敦", 4, 3, 8500f, 12, false));
        items.add(makeItem(items.size() + 1, "天域花園1房1廳", "天域花園，男女不限，入住免水電一月", 2, 4, 1500f, 13, true));
        items.add(makeItem(items.size() + 1, "神同樣的漂亮，玉馨山莊3房2廳", "心動不如行動，擰包便可入住，來吧！", 1, 3, 9500f, 8, false));
        items.add(makeItem(items.size() + 1, "玉馨山莊2房1廳，情侶最愛", "宅男宅女快來吧只要2500，走過路過，別再錯過", 1, 2, 2500f, 5, false));
        items.add(makeItem(items.size() + 1, "天域花園3房2廳", "天域花園出租，都來看看，都來瞄瞄，3房出租只要7500.", 4, 3, 7500f, 6, true));
        items.add(makeItem(items.size() + 1, "深都花園出租3房2廳", "找愛乾淨的人氏，全新裝修", 4, 3, 5200f, 31, false));
        items.add(makeItem(items.size() + 1, "This is Mobile test", "haha Hello world!", 4, 3, 1200f, 31, false));
        core.addBeans(items);
        // commit
        core.commit();
        
        // ------------------------------------------------------
        // Set search text
        // ------------------------------------------------------
        String searchText = AnalysisSearchText(core, "出租花園"); //subject:*出租* && price:[1000 TO 8000]
        System.out.println("Search Text:" + searchText);
        
        // ------------------------------------------------------
        // Set query text
        // ------------------------------------------------------
        String queryText = searchText + "&& price:[1000 TO 8000]";
        System.out.println("Query Text:" + queryText);

        // ------------------------------------------------------
        // search
        // ------------------------------------------------------
        SolrQuery query = new SolrQuery();
        query.setQuery(queryText);
        query.setStart(0); // query的開始行數(分頁使用)
        query.setRows(100); // query的返回行數(分頁使用)
        query.setFacet(true); // 設置使用facet
        query.setFacetMinCount(0); // 設置facet最少的統計數量
        query.setFacetLimit(10); // facet結果的返回行數
        query.addFacetField("categoryId", "regionId"); // facet的字段
        query.setFacetSort(FacetParams.FACET_SORT_COUNT);
        //query.addSort(new SortClause("id", ORDER.asc)); // 排序
        query.setRequestHandler("/browse");
        QueryResponse response = core.query(query);
        List<Item> items_rep = response.getBeans(Item.class);
        List<FacetField> facetFields = response.getFacetFields();
        // 由於上面的start和rows均設置爲0，因此這裏不會有query結果輸出
        System.out.println("--------------------");
        System.out.println("Search result:");
        for (Item item : items_rep) {
            System.out.println("id=" + item.getId() + "\tsubject=" + item.getSubject()
                    + "\tregion=" + mapRegion.get(item.getRegionId())
                    + "\tcategory=" + mapCategory.get(item.getCategoryId())
                    + "\tprice=" + item.getPrice());
        }
        // 打印全部facet
        for (FacetField ff : facetFields) {
            System.out.println("--------------------");
            System.out.println("name=" + ff.getName() + "\tcount=" + ff.getValueCount());
            System.out.println("--------------------");
            switch (ff.getName()) {
            case "regionId":
                printOut(mapRegion, ff.getValues());
                break;
            case "categoryId":
                printOut(mapCategory, ff.getValues());
                break;
            }
        }
    }
    
    @SuppressWarnings({ "rawtypes" })
    private static void printOut(HashMap map, List<Count> counts) {
        for (Count count : counts) {
            System.out.println("name=" + map.get(Integer.parseInt(count.getName())) + "\tcount=" + count.getCount());
        }
        System.out.println("--------------------");
    }

    private static Item makeItem(long id, String subject, String content, int regionId, int categoryId, float price,
            long point, boolean vip) {
        Calendar cale = Calendar.getInstance();
        cale.setTime(new Date());
        cale.add(Calendar.DATE, (int)id);
        Item item = new Item();
        item.setId(id);
        item.setSubject(subject);
        item.setContent(content);
        item.setRegionId(regionId);
        item.setCategoryId(categoryId);
        item.setPrice(price);
        item.setCreateTime(cale.getTime());
        item.setPoint(point);
        item.setVip(vip);
        return item;
    }
    
    @SuppressWarnings("unchecked")
    /**
     * 從新將須要查詢的文本內容解析成分詞
     * @param core
     * @param searchText
     * @return
     * @throws SolrServerException
     */
    private static String AnalysisSearchText(HttpSolrServer core, String searchText) throws SolrServerException {
        StringBuilder strSearchText = new StringBuilder();
        final String STR_FIELD_TYPE = "text_cn";
        SolrQuery queryAnalysis = new SolrQuery();
        queryAnalysis.add(CommonParams.QT, "/analysis/field"); // query type
        queryAnalysis.add(AnalysisParams.FIELD_VALUE, searchText);
        queryAnalysis.add(AnalysisParams.FIELD_TYPE, STR_FIELD_TYPE);
        QueryResponse responseAnalysis = core.query(queryAnalysis);
        //對響應進行解析
        NamedList<Object> analysis = (NamedList<Object>) responseAnalysis.getResponse().get("analysis");// analysis node
        NamedList<Object> field_types = (NamedList<Object>) analysis.get("field_types");// field_types node
        NamedList<Object> fieldType = (NamedList<Object>) field_types.get(STR_FIELD_TYPE);// text_cn node
        NamedList<Object> index = (NamedList<Object>) fieldType.get("index");// index node
        List<SimpleOrderedMap<String>> list = (ArrayList<SimpleOrderedMap<String>>)index.get("org.wltea.analyzer.lucene.IKTokenizer");// tokenizer node
        // 在每一個詞條中間加上空格，爲每一個詞條進行或運算
        for(Iterator<SimpleOrderedMap<String>> iter = list.iterator(); iter.hasNext();)
        {
            strSearchText.append(iter.next().get("text") + " ");
        }
        return strSearchText.toString();
    }
}

說明：

a)AnalysisSearchText(...)方法：此方法會把須要查詢的語句先使用分詞分析，如上例子「出租花園」，調用AnalysisSearchText(...)後，會獲得「出租 花園」，會把兩個詞分拆成以空格分隔的字符串。否則solr會以「出租花園」總體作爲詞作查詢而得不到結果。

b)使用自定義的Handler，須要在代碼中加入這句：

query.setRequestHandler("/browse");

對應的是solrconfig.xml中的requestHandler的:/browse

 

五、運行結果：

 

或者使用solr的query查詢查看結果：