| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338 |
- package com.uas.search.service.impl;
- import com.alibaba.fastjson.JSONObject;
- import com.uas.search.annotation.NotEmpty;
- import com.uas.search.constant.SearchConstants;
- import com.uas.search.constant.model.CollectField;
- import com.uas.search.constant.model.PageParams;
- import com.uas.search.constant.model.PageParams.FilterField;
- import com.uas.search.constant.model.SPage;
- import com.uas.search.exception.SearchException;
- import com.uas.search.grouping.DistinctGroupCollector;
- import com.uas.search.grouping.GoodsGroupCollector;
- import com.uas.search.model.*;
- import com.uas.search.service.SearchService;
- import com.uas.search.sort.StringFieldComparatorSource;
- import com.uas.search.util.CollectionUtils;
- import com.uas.search.util.DocumentToObjectUtils;
- import com.uas.search.util.SearchUtils;
- import com.uas.search.util.StringUtils;
- import org.apache.lucene.document.Document;
- import org.apache.lucene.index.Term;
- import org.apache.lucene.search.*;
- import org.apache.lucene.search.BooleanClause.Occur;
- import org.apache.lucene.search.SortField.Type;
- import org.slf4j.Logger;
- import org.slf4j.LoggerFactory;
- import org.springframework.stereotype.Service;
- import java.io.IOException;
- import java.net.URLDecoder;
- import java.util.*;
- import java.util.Map.Entry;
- /**
- * 搜索索引
- *
- * @author sunyj
- * @since 2016年8月5日 下午2:21:26
- */
- @Service
- public class SearchServiceImpl implements SearchService {
- /**
- * 获取联想词时返回的最大数目
- */
- private static final int SIMILAR_NUM = 20;
- private static Logger logger = LoggerFactory.getLogger(SearchServiceImpl.class);
- @Override
- public SPage<Long> getKindIds(String keyword, Integer page, Integer size) {
- List<Long> ids = new ArrayList<>();
- SPage<Document> documents = getKindDocuments(keyword, page, size);;
- SPage<Long> sPage = new SPage<Long>(documents.getTotalPage(), documents.getTotalElement(), documents.getPage(),
- documents.getSize(), documents.isFirst(), documents.isLast());
- for (Document document : documents.getContent()) {
- ids.add(Long.parseLong(document.get(SearchConstants.KIND_ID_FIELD)));
- }
- sPage.setContent(ids);
- return sPage;
- }
- @Override
- public SPage<Map<String, Object>> getKinds(String keyword, Integer page, Integer size) {
- List<Map<String, Object>> kinds = new ArrayList<>();
- SPage<Document> documents = getKindDocuments(keyword, page, size);
- SPage<Map<String, Object>> sPage = new SPage<>(documents.getTotalPage(),
- documents.getTotalElement(), documents.getPage(), documents.getSize(), documents.isFirst(),
- documents.isLast());
- for (Document document : documents.getContent()) {
- Map<String, Object> kind = new HashMap<>();
- kind.put("id", Long.parseLong(document.get(SearchConstants.KIND_ID_FIELD)));
- kind.put("nameCn", document.get(SearchConstants.KIND_NAMECN_FIELD));
- kinds.add(kind);
- }
- sPage.setContent(kinds);
- return sPage;
- }
- private SPage<Document> getKindDocuments(String keyword, Integer page, Integer size){
- if (SearchUtils.isKeywordInvalid(keyword)) {
- throw new SearchException("搜索关键词无效:" + keyword);
- }
- BooleanQuery booleanQuery = SearchUtils.getBooleanQuery(SearchConstants.KIND_NAMECN_FIELD, keyword);
- logger.info(booleanQuery.toString());
- return SearchUtils.getDocuments(SearchConstants.KIND_TABLE_NAME, booleanQuery, new Sort(sortKind(keyword)), page, size);
- }
- /**
- * @return 类目排序规则
- */
- private SortField[] sortKind(String keyword) {
- // 分数 > 访问量 > 搜索次数
- return new SortField[]{
- sortField(SearchConstants.KIND_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.KIND_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- new SortField(SearchConstants.KIND_NAMECN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword))
- };
- }
- /**
- * 构造 SortField
- *
- * @param field Name of field to sort by. Can be <code>null</code> if
- * <code>type</code> is SCORE or DOC.
- * @param type Type of values in the terms.
- * @param reverse True if natural order should be reversed.
- * @param missingValue Used for 'sortMissingFirst/Last'
- * @return SortField
- */
- private SortField sortField(String field, Type type, boolean reverse, Object missingValue) {
- SortField sortField = new SortField(field, type, reverse);
- sortField.setMissingValue(missingValue);
- return sortField;
- }
- @Override
- public SPage<Long> getBrandIds(String keyword, Integer page, Integer size) {
- List<Long> ids = new ArrayList<>();
- SPage<Document> documents = getBrandDocuments(keyword, page, size);
- SPage<Long> sPage = new SPage<Long>(documents.getTotalPage(), documents.getTotalElement(), documents.getPage(),
- documents.getSize(), documents.isFirst(), documents.isLast());
- for (Document document : documents.getContent()) {
- ids.add(Long.parseLong(document.get(SearchConstants.BRAND_ID_FIELD)));
- }
- sPage.setContent(ids);
- return sPage;
- }
- @Override
- public SPage<Map<String, Object>> getBrands(String keyword, Integer page, Integer size) {
- List<Map<String, Object>> brands = new ArrayList<>();
- SPage<Document> documents = getBrandDocuments(keyword, page, size);
- SPage<Map<String, Object>> sPage = new SPage<Map<String, Object>>(documents.getTotalPage(),
- documents.getTotalElement(), documents.getPage(), documents.getSize(), documents.isFirst(),
- documents.isLast());
- for (Document document : documents.getContent()) {
- Map<String, Object> brand = new HashMap<String, Object>();
- brand.put("id", Long.parseLong(document.get(SearchConstants.BRAND_ID_FIELD)));
- brand.put("uuid", document.get(SearchConstants.BRAND_UUID_FIELD));
- brand.put("nameCn", document.get(SearchConstants.BRAND_NAMECN_FIELD));
- brand.put("nameEn", document.get(SearchConstants.BRAND_NAMEEN_FIELD));
- brands.add(brand);
- }
- sPage.setContent(brands);
- return sPage;
- }
- private SPage<Document> getBrandDocuments(String keyword, Integer page, Integer size){
- if (SearchUtils.isKeywordInvalid(keyword)) {
- throw new SearchException("搜索关键词无效:" + keyword);
- }
- BooleanQuery booleanQuery = new BooleanQuery();
- booleanQuery.add(SearchUtils.getBooleanQuery(SearchConstants.BRAND_NAMECN_FIELD, keyword),
- BooleanClause.Occur.SHOULD);
- booleanQuery.add(SearchUtils.getBooleanQuery(SearchConstants.BRAND_NAMEEN_FIELD, keyword),
- BooleanClause.Occur.SHOULD);
- logger.info(booleanQuery.toString());
- return SearchUtils.getDocuments(SearchConstants.BRAND_TABLE_NAME, booleanQuery, new Sort(sortBrand(keyword)), page,
- size);
- }
- /**
- * @return 品牌排序规则
- */
- private SortField[] sortBrand(String keyword) {
- // 自定义排序 > 权重 > 访问量 > 搜索次数 > 分数
- // 分数排序放在最后,是因为有的中英文名称相同,分数翻倍,但实际匹配度并不高
- return new SortField[]{
- sortField(SearchConstants.BRAND_WEIGHT_FIELD, Type.DOUBLE, true, Double.MIN_VALUE),
- sortField(SearchConstants.BRAND_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.BRAND_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- new SortField(SearchConstants.BRAND_NAMEEN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.BRAND_NAMECN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword))
- };
- }
- @Override
- public Map<String, Object> getComponentIds(String keyword, PageParams pageParams) {
- Map<String, Object> searchComponentIds = getComponentIds(keyword, pageParams, null, null);
- return searchComponentIds;
- // TODO 对品牌、类目甚至拼音混合搜索(待完善)
- // int total = (int) searchComponentIds.get("total");
- // if (total != 0) {
- // return searchComponentIds;
- // }
- // List<Long> kindIds = getKindIds(keyword, Occur.SHOULD);
- // List<Long> brandIds = getBrandIds(keyword, Occur.SHOULD);
- // return getComponentIds(null, pageParams, kindIds, brandIds);
- }
- /**
- * 根据关键词搜索产品
- *
- * @param keyword
- * @param pageParams
- * @param kindIds
- * @param brandIds
- * @return
- * @throws SearchException
- */
- private Map<String, Object> getComponentIds(String keyword, PageParams pageParams, List<Long> kindIds,
- List<Long> brandIds) throws SearchException {
- // 因为器件、属性值的数据量远比类目、品牌大得多,而且器件搜索可能还需进行分页,
- // 所以涉及器件、属性值的搜索,大都不能像类目和品牌一样直接利用SearchUtils.getDocuments方法
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(SearchConstants.COMPONENT_TABLE_NAME);
- if (pageParams == null) {
- pageParams = new PageParams();
- }
- if (pageParams.getPage() <= 0)
- pageParams.setPage(1);
- if (pageParams.getSize() <= 0)
- pageParams.setSize(20);
- Map<String, Object> map = new HashMap<String, Object>();
- List<Long> ids = new ArrayList<Long>();
- try {
- BooleanQuery booleanQuery = new BooleanQuery();
- if (!SearchUtils.isKeywordInvalid(keyword)) {
- booleanQuery.add(setBoost(keyword), BooleanClause.Occur.MUST);
- }
- Map<FilterField, Object> filters = pageParams.getFilters();
- if (!CollectionUtils.isEmpty(filters)) {
- // 筛选类目
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_KINDID))) {
- String kindId = filters.get(FilterField.COMPONENT_KINDID).toString();
- TermQuery kindQuery = new TermQuery(new Term(SearchConstants.COMPONENT_KI_ID_FIELD, kindId));
- booleanQuery.add(kindQuery, BooleanClause.Occur.MUST);
- }
- // 筛选品牌
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_BRANDID))) {
- String brandId = filters.get(FilterField.COMPONENT_BRANDID).toString();
- TermQuery brandQuery = new TermQuery(new Term(SearchConstants.COMPONENT_BR_ID_FIELD, brandId));
- booleanQuery.add(brandQuery, BooleanClause.Occur.MUST);
- }
- // 库存不为0
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_HAS_RESERVE))) {
- Boolean isReserveNotEmpty = (Boolean) filters.get(FilterField.COMPONENT_HAS_RESERVE);
- if (isReserveNotEmpty) {
- booleanQuery.add(NumericRangeQuery.newDoubleRange(SearchConstants.COMPONENT_RESERVE_FIELD, 0.0,
- Double.MAX_VALUE, false, true), BooleanClause.Occur.MUST);
- }
- }
- // 现货、呆滞库存、样品数量不为0,取或的关系
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_HAS_SAMPLE))
- || !StringUtils.isEmpty(filters.get(FilterField.COMPONENT_HAS_ORIGINAL))
- || !StringUtils.isEmpty(filters.get(FilterField.COMPONENT_HAS_INACTION_STOCK))) {
- BooleanQuery booleanQuery2 = new BooleanQuery();
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_HAS_SAMPLE))) {
- booleanQuery2.add(NumericRangeQuery.newDoubleRange(SearchConstants.COMPONENT_SAMPLE_QTY_FIELD,
- 0.0, Double.MAX_VALUE, false, true), BooleanClause.Occur.SHOULD);
- }
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_HAS_ORIGINAL))) {
- booleanQuery2.add(NumericRangeQuery.newDoubleRange(SearchConstants.COMPONENT_ORIGINAL_QTY_FIELD,
- 0.0, Double.MAX_VALUE, false, true), BooleanClause.Occur.SHOULD);
- }
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_HAS_INACTION_STOCK))) {
- booleanQuery2.add(
- NumericRangeQuery.newDoubleRange(SearchConstants.COMPONENT_INACTION_STOCK_QTY_FIELD,
- 0.0, Double.MAX_VALUE, false, true),
- BooleanClause.Occur.SHOULD);
- }
- booleanQuery.add(booleanQuery2, Occur.MUST);
- }
- // 属性过滤
- if (!StringUtils.isEmpty(filters.get(FilterField.COMPONENT_PROPERTIES))) {
- JSONObject proJSON = JSONObject
- .parseObject(String.valueOf(filters.get(FilterField.COMPONENT_PROPERTIES)));
- for (String key : proJSON.keySet()) {
- String value = String.valueOf(proJSON.get(key));
- if (!StringUtils.isEmpty(value)) {
- if (!key.startsWith(SearchConstants.COMPONENT_PROPERTY_PREFIX)) {
- key = SearchConstants.COMPONENT_PROPERTY_PREFIX + key;
- }
- TermQuery propertyQuery = new TermQuery(new Term(key, value));
- booleanQuery.add(propertyQuery, BooleanClause.Occur.MUST);
- }
- }
- }
- }
- if (!CollectionUtils.isEmpty(kindIds)) {
- BooleanQuery booleanQuery2 = new BooleanQuery();
- for (Long id : kindIds) {
- booleanQuery2.add(new TermQuery(new Term(SearchConstants.COMPONENT_KI_ID_FIELD, id.toString())),
- Occur.SHOULD);
- }
- booleanQuery.add(booleanQuery2, Occur.MUST);
- }
- if (!CollectionUtils.isEmpty(brandIds)) {
- BooleanQuery booleanQuery2 = new BooleanQuery();
- for (Long id : brandIds) {
- booleanQuery2.add(new TermQuery(new Term(SearchConstants.COMPONENT_BR_ID_FIELD, id.toString())),
- Occur.SHOULD);
- }
- booleanQuery.add(booleanQuery2, Occur.MUST);
- }
- logger.info(booleanQuery.toString());
- Sort sort = new Sort(sortComponent(keyword));
- TopDocs hits;
- if (pageParams.getPage() > 1) {// 不是第一页
- TopDocs previousHits = indexSearcher.search(booleanQuery,
- (pageParams.getPage() - 1) * pageParams.getSize(), sort, true, false);
- ScoreDoc[] previousScoreDocs = previousHits.scoreDocs;
- ScoreDoc after = previousScoreDocs[previousScoreDocs.length - 1];
- hits = indexSearcher.searchAfter(after, booleanQuery, pageParams.getSize(), sort, true, false);
- } else {
- hits = indexSearcher.search(booleanQuery, pageParams.getSize(), sort, true, false);
- }
- ScoreDoc[] scoreDocs = hits.scoreDocs;
- for (ScoreDoc scoreDoc : scoreDocs) {
- // 数据量太大,需要指定将获取的数据(以免载入不必要的数据,降低速度)
- Set<String> fieldsToLoad = new HashSet<>();
- fieldsToLoad.add(SearchConstants.COMPONENT_ID_FIELD);
- Document document = indexSearcher.doc(scoreDoc.doc, fieldsToLoad);
- String componentId = document.get(SearchConstants.COMPONENT_ID_FIELD);
- ids.add(Long.parseLong(componentId));
- // System.out.println(componentId + "\t" + scoreDoc.score);
- }
- map.put("componentIds", ids);
- map.put("page", pageParams.getPage());
- map.put("size", pageParams.getSize());
- map.put("total", hits.totalHits);
- } catch (IOException e) {
- logger.error("", e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- return map;
- }
- /**
- * 同时搜索器件、类目、品牌,并设置boost
- *
- * @param keyword
- * @return
- */
- private Query setBoost(String keyword) {
- BooleanQuery booleanQuery = new BooleanQuery();
- PrefixQuery prefixQuery = new PrefixQuery(
- new Term(SearchConstants.COMPONENT_CODE_FIELD, keyword.toLowerCase()));
- prefixQuery.setBoost(100);
- booleanQuery.add(prefixQuery, BooleanClause.Occur.SHOULD);
- booleanQuery.add(createQuery(SearchConstants.COMPONENT_BR_NAMECN_FIELD, keyword, 10), BooleanClause.Occur.SHOULD);
- booleanQuery.add(createQuery(SearchConstants.COMPONENT_BR_NAMEEN_FIELD, keyword, 10), BooleanClause.Occur.SHOULD);
- booleanQuery.add(createQuery(SearchConstants.COMPONENT_KI_NAME_FIELD, keyword, 1), BooleanClause.Occur.SHOULD);
- return booleanQuery;
- }
- /**
- * @return 器件排序规则
- */
- private SortField[] sortComponent(String keyword) {
- // 分数 > 器件(访问量 > 搜索次数) > 品牌(权重 > 访问量 > 搜索次数) > 类目(访问量 > 搜索次数)
- return new SortField[]{
- sortField(SearchConstants.COMPONENT_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.COMPONENT_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.COMPONENT_BR_WEIGHT_FIELD, Type.DOUBLE, true, Double.MIN_VALUE),
- sortField(SearchConstants.COMPONENT_BR_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.COMPONENT_BR_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.COMPONENT_KI_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.COMPONENT_KI_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- new SortField(SearchConstants.COMPONENT_CODE_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.COMPONENT_BR_NAMEEN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.COMPONENT_BR_NAMECN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.COMPONENT_KI_NAME_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword))
- };
- }
- @Override
- public Set<Long> getKindIdsBySearchComponent(String keyword, String brandId) {
- Query filter = null;
- // 筛选品牌
- if (!StringUtils.isEmpty(brandId)) {
- filter = new TermQuery(new Term(SearchConstants.COMPONENT_BR_ID_FIELD, brandId));
- }
- return collectBySearchComponent(keyword, filter, SearchConstants.COMPONENT_KI_ID_FIELD).getValues();
- }
- @Override
- public Set<Map<String, Object>> getKindsBySearchComponent(String keyword, String brandId) {
- Query filter = null;
- // 筛选品牌
- if (!StringUtils.isEmpty(brandId)) {
- filter = new TermQuery(new Term(SearchConstants.COMPONENT_BR_ID_FIELD, brandId));
- }
- return collectBySearchComponent(keyword, filter, SearchConstants.KIND_ID_FIELD,
- new DistinctGroupCollector.CollectField(SearchConstants.KIND_ID_FIELD, "id"),
- new DistinctGroupCollector.CollectField(SearchConstants.KIND_NAMECN_FIELD, "nameCn"))
- .getCollectValues();
- }
- @Override
- public Set<Long> getBrandIdsBySearchComponent(String keyword, String kindId) {
- Query filter = null;
- // 筛选类目
- if (!StringUtils.isEmpty(kindId)) {
- filter = new TermQuery(new Term(SearchConstants.COMPONENT_KI_ID_FIELD, kindId));
- }
- return collectBySearchComponent(keyword, filter, SearchConstants.COMPONENT_BR_ID_FIELD).getValues();
- }
- @Override
- public Set<Map<String, Object>> getBrandsBySearchComponent(String keyword, String kindId) {
- Query filter = null;
- // 筛选类目
- if (!StringUtils.isEmpty(kindId)) {
- filter = new TermQuery(new Term(SearchConstants.COMPONENT_KI_ID_FIELD, kindId));
- }
- return collectBySearchComponent(keyword, filter, SearchConstants.COMPONENT_BR_ID_FIELD,
- new DistinctGroupCollector.CollectField(SearchConstants.COMPONENT_BR_ID_FIELD, "id"),
- new DistinctGroupCollector.CollectField(SearchConstants.COMPONENT_BR_UUID_FIELD, "uuid"),
- new DistinctGroupCollector.CollectField(SearchConstants.COMPONENT_BR_NAMECN_FIELD, "nameCn"))
- .getCollectValues();
- }
- /**
- * 搜索器件时统计指定信息
- *
- * @param keyword 关键词
- * @param filter 过滤条件
- * @param groupField 统计的单个字段(多为 id)
- * @param collectFields 统计的多个字段(详细信息),可为空
- * @return 统计信息
- */
- private DistinctGroupCollector collectBySearchComponent(String keyword, Query filter, String groupField, DistinctGroupCollector.CollectField... collectFields){
- if (SearchUtils.isKeywordInvalid(keyword)) {
- throw new SearchException("搜索关键词无效:" + keyword);
- }
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(SearchConstants.COMPONENT_TABLE_NAME);
- try {
- BooleanQuery booleanQuery = new BooleanQuery();
- keyword = URLDecoder.decode(keyword, "UTF-8");
- booleanQuery.add(setBoost(keyword), BooleanClause.Occur.MUST);
- if(filter != null){
- booleanQuery.add(filter, Occur.FILTER);
- }
- logger.info(booleanQuery.toString());
- DistinctGroupCollector collector = new DistinctGroupCollector(groupField, collectFields);
- indexSearcher.search(booleanQuery, collector);
- return collector;
- } catch (IOException e) {
- throw new IllegalStateException("统计失败", e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- }
- @Override
- public List<String> getSimilarKeywords(String keyword, Integer size) {
- size = size == null || size < 1 ? SIMILAR_NUM : size;
- List<String> result = new ArrayList<>();
- // 相似的器件原厂型号数量足够,直接返回
- List<String> componentCodes = getSimilarComponentCodes(keyword, size);
- result.addAll(componentCodes);
- removeDuplicate(result);
- if (result.size() == size) {
- return result;
- }
- // 获取相似类目
- List<String> kindNames = getSimilarKindNames(keyword, size);
- if (!CollectionUtils.isEmpty(kindNames)) {
- result.addAll(kindNames);
- removeDuplicate(result);
- // 如果总的数量超出SIMILAR_NUM,去除多余的元素
- if (result.size() > size) {
- removeElements(result, size);
- return result;
- }
- }
- // 获取相似品牌
- List<String> brandNames = getSimilarBrandNames(keyword, size);
- if (!CollectionUtils.isEmpty(brandNames)) {
- result.addAll(brandNames);
- removeDuplicate(result);
- if (result.size() > size) {
- removeElements(result, size);
- return result;
- }
- }
- return result;
- }
- @Override
- public List<Map<String, Object>> getSimilarComponents(String componentCode, Integer size) {
- size = size == null || size < 1 ? SIMILAR_NUM : size;
- if (SearchUtils.isKeywordInvalid(componentCode)) {
- throw new SearchException("输入无效:" + componentCode);
- }
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(SearchConstants.COMPONENT_TABLE_NAME);
- List<Map<String, Object>> components = new ArrayList<>();
- try {
- PrefixQuery prefixQuery = new PrefixQuery(
- new Term(SearchConstants.COMPONENT_CODE_FIELD, componentCode.toLowerCase()));
- logger.info(prefixQuery.toString());
- Sort sort = new Sort(new SortField(SearchConstants.COMPONENT_CODE_FIELD, new StringFieldComparatorSource(componentCode)));
- TopDocs hits = indexSearcher.search(prefixQuery, size, sort);
- ScoreDoc[] scoreDocs = hits.scoreDocs;
- for (ScoreDoc scoreDoc : scoreDocs) {
- Set<String> fieldsToLoad = new HashSet<>();
- fieldsToLoad.add(SearchConstants.COMPONENT_ID_FIELD);
- fieldsToLoad.add(SearchConstants.COMPONENT_UUID_FIELD);
- fieldsToLoad.add(SearchConstants.COMPONENT_CODE_FIELD);
- Document document = indexSearcher.doc(scoreDoc.doc, fieldsToLoad);
- Map<String, Object> map = new HashMap<>();
- map.put("id", Long.parseLong(document.get(SearchConstants.COMPONENT_ID_FIELD)));
- map.put("uuid", document.get(SearchConstants.COMPONENT_UUID_FIELD));
- map.put("code", document.get(SearchConstants.COMPONENT_CODE_FIELD));
- components.add(map);
- }
- } catch (IOException e) {
- logger.error("", e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- return components;
- }
- @Override
- public List<Map<String, Object>> getSimilarBrands(String brandName, Integer size) {
- size = size == null || size < 1 ? SIMILAR_NUM : size;
- if (SearchUtils.isKeywordInvalid(brandName)) {
- throw new SearchException("输入无效:" + brandName);
- }
- List<Map<String, Object>> brands = new ArrayList<Map<String, Object>>();
- // 品牌名称带有空格,并且中英文名并无一定顺序,因此对nameCn、nameEn均要搜索
- BooleanQuery booleanQuery = new BooleanQuery();
- // 搜索nameCn
- booleanQuery.add(SearchUtils.getBooleanQuery(SearchConstants.BRAND_NAMECN_FIELD, brandName),
- BooleanClause.Occur.SHOULD);
- // 搜索nameEn
- booleanQuery.add(SearchUtils.getBooleanQuery(SearchConstants.BRAND_NAMEEN_FIELD, brandName),
- BooleanClause.Occur.SHOULD);
- logger.info(booleanQuery.toString());
- Sort sort = new Sort(new SortField(SearchConstants.BRAND_NAMEEN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(brandName)),
- new SortField(SearchConstants.BRAND_NAMECN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(brandName)));
- List<Document> documents = SearchUtils.getDocuments(SearchConstants.BRAND_TABLE_NAME, booleanQuery, sort, null, size)
- .getContent();
- for (Document document : documents) {
- Map<String, Object> brand = new HashMap<>();
- brand.put("id", Long.parseLong(document.get(SearchConstants.BRAND_ID_FIELD)));
- brand.put("uuid", document.get(SearchConstants.BRAND_UUID_FIELD));
- brand.put("nameCn", document.get(SearchConstants.BRAND_NAMECN_FIELD));
- brand.put("nameEn", document.get(SearchConstants.BRAND_NAMEEN_FIELD));
- brands.add(brand);
- }
- return brands;
- }
- @Override
- public List<Map<String, Object>> getSimilarKinds(String kindName, Integer size) {
- size = size == null || size < 1 ? SIMILAR_NUM : size;
- return getSimilarKinds(kindName, null, null, size);
- }
- @Override
- public List<Map<String, Object>> getSimilarLeafKinds(String kindName, Integer size) {
- size = size == null || size < 1 ? SIMILAR_NUM : size;
- return getSimilarKinds(kindName, (short) 1, null, size);
- }
- @Override
- public List<Map<String, Object>> getSimilarKindsByLevel(String kindName, Short level, Integer size) {
- size = size == null || size < 1 ? SIMILAR_NUM : size;
- return getSimilarKinds(kindName, null, level, size);
- }
- /**
- * 根据输入的类目名获取联想词
- *
- * @param kindName
- * 类目名
- * @param isLeaf
- * 是否只获取末级类目
- * @param level
- * 指定的类目级别
- * @param size 指定的联想词数目
- * @return
- */
- private List<Map<String, Object>> getSimilarKinds(String kindName, Short isLeaf, Short level, Integer size) {
- size = size == null || size < 1 ? SIMILAR_NUM : size;
- if (SearchUtils.isKeywordInvalid(kindName)) {
- throw new SearchException("输入无效:" + kindName);
- }
- List<Map<String, Object>> kinds = new ArrayList<>();
- BooleanQuery booleanQuery = new BooleanQuery();
- booleanQuery.add(SearchUtils.getBooleanQuery(SearchConstants.KIND_NAMECN_FIELD, kindName),
- BooleanClause.Occur.MUST);
- if (isLeaf != null && isLeaf == 1) {
- booleanQuery.add(new TermQuery(new Term(SearchConstants.KIND_ISLEAF_FIELD, String.valueOf(isLeaf))),
- BooleanClause.Occur.MUST);
- } else {
- if (level != null && level > 0) {
- booleanQuery.add(new TermQuery(new Term(SearchConstants.KIND_LEVEL_FIELD, String.valueOf(level))),
- BooleanClause.Occur.MUST);
- }
- }
- logger.info(booleanQuery.toString());
- Sort sort = new Sort(new SortField(SearchConstants.KIND_NAMECN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(kindName)));
- List<Document> documents = SearchUtils.getDocuments(SearchConstants.KIND_TABLE_NAME, booleanQuery, sort, null, size).getContent();
- for (Document document : documents) {
- Map<String, Object> map = new HashMap<>();
- map.put("id", Long.parseLong(document.get(SearchConstants.KIND_ID_FIELD)));
- map.put("nameCn", document.get(SearchConstants.KIND_NAMECN_FIELD));
- map.put("isLeaf", Short.parseShort(document.get(SearchConstants.KIND_ISLEAF_FIELD)));
- map.put("level", Short.parseShort(document.get(SearchConstants.KIND_LEVEL_FIELD)));
- kinds.add(map);
- }
- return kinds;
- }
- @Override
- public List<Map<String, String>> getSimilarPropertyValues(Long kindId, Long propertyId, String keyword,
- Long topNum) {
- if (kindId == null || propertyId == null) {
- throw new SearchException("类目id和属性id不能为空");
- }
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(SearchConstants.COMPONENT_TABLE_NAME);
- String propertyIdString = String.valueOf(propertyId);
- if (!propertyIdString.startsWith(SearchConstants.COMPONENT_PROPERTY_PREFIX)) {
- propertyIdString = SearchConstants.COMPONENT_PROPERTY_PREFIX + propertyIdString;
- }
- propertyIdString = propertyIdString + SearchConstants.COMPONENT_PROPERTY_TOKENIZED_SUFFIX;
- if (keyword == null) {
- keyword = "";
- }
- if (topNum == null || topNum < 1) {
- topNum = (long) SIMILAR_NUM;
- }
- List<String> propertyValues = new ArrayList<>();
- try {
- BooleanQuery booleanQuery = new BooleanQuery();
- booleanQuery.add(new TermQuery(new Term(SearchConstants.COMPONENT_KI_ID_FIELD, String.valueOf(kindId))),
- BooleanClause.Occur.MUST);
- booleanQuery.add(new PrefixQuery(new Term(propertyIdString, keyword.toLowerCase())),
- BooleanClause.Occur.MUST);
- logger.info(booleanQuery.toString());
- // 如果只搜索topNum个结果,去除重复的属性值后,数目很可能是不够的
- TopDocs topDocs = indexSearcher.search(booleanQuery, SearchConstants.TOP_NUM);
- ScoreDoc[] scoreDocs = topDocs.scoreDocs;
- for (ScoreDoc scoreDoc : scoreDocs) {
- Set<String> fieldsToLoad = new HashSet<>();
- fieldsToLoad.add(propertyIdString);
- Document document = indexSearcher.doc(scoreDoc.doc, fieldsToLoad);
- String propertyValue = document.get(propertyIdString);
- if (!StringUtils.isEmpty(propertyValue) && !propertyValues.contains(propertyValue)) {
- propertyValues.add(propertyValue);
- }
- if (propertyValues.size() >= topNum) {
- break;
- }
- }
- } catch (IOException e) {
- logger.error("", e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- List<Map<String, String>> result = new ArrayList<>();
- for (String propertyValue : propertyValues) {
- Map<String, String> map = new HashMap<>();
- map.put("propertyValue", propertyValue);
- result.add(map);
- }
- return result;
- }
- /**
- * 根据输入获取相似的器件原厂型号
- *
- * @param componentCode
- * @param size 指定的联想词数目
- * @return
- */
- private List<String> getSimilarComponentCodes(String componentCode, Integer size) {
- return getSimilarValues(SearchConstants.COMPONENT_TABLE_NAME, SearchConstants.COMPONENT_CODE_FIELD,
- SearchConstants.COMPONENT_CODE_FIELD, componentCode.toLowerCase(), size);
- }
- /**
- * 根据输入获取相似的品牌名称
- *
- * @param brandName
- * @param size 指定的联想词数目
- * @return
- */
- private List<String> getSimilarBrandNames(String brandName, Integer size) {
- // 获取相似的中文品牌
- List<String> nameCns = getSimilarValues(SearchConstants.BRAND_TABLE_NAME, SearchConstants.BRAND_NAMECN_FIELD,
- SearchConstants.BRAND_NAMECN_UNTOKENIZED_FIELD, brandName, size);
- // 相似的中文品牌数量足够,直接返回
- if (nameCns != null && nameCns.size() == SIMILAR_NUM) {
- return nameCns;
- }
- List<String> names = nameCns;
- // 获取相似的英文品牌
- List<String> nameEns = getSimilarValues(SearchConstants.BRAND_TABLE_NAME, SearchConstants.BRAND_NAMEEN_FIELD,
- SearchConstants.BRAND_NAMEEN_UNTOKENIZED_FIELD, brandName, size);
- names.addAll(nameEns);
- return names;
- }
- /**
- * 根据输入获取相似的类目名称
- *
- * @param kindName
- * @param size 指定的联想词数目
- * @return
- */
- private List<String> getSimilarKindNames(String kindName, Integer size) {
- return getSimilarValues(SearchConstants.KIND_TABLE_NAME, SearchConstants.KIND_NAMECN_FIELD,
- SearchConstants.KIND_NAMECN_UNTOKENIZED_FIELD, kindName, size);
- }
- /**
- * 根据输入值获取该域相似的值
- *
- * @param tableName
- * @param field
- * @param keyword
- * @param size 指定的联想词数目
- * @return
- */
- private List<String> getSimilarValues(String tableName, String field, String sortField, String keyword, Integer size) {
- if (SearchUtils.isKeywordInvalid(keyword)) {
- throw new SearchException("输入无效:" + keyword);
- }
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(tableName);
- List<String> result = new ArrayList<>();
- try {
- Query query = SearchUtils.getBooleanQuery(field, keyword);
- logger.info(query.toString());
- Sort sort = new Sort(new SortField(sortField, new StringFieldComparatorSource(keyword)));
- TopDocs hits = indexSearcher.search(query, size, sort);
- ScoreDoc[] scoreDocs = hits.scoreDocs;
- for (ScoreDoc scoreDoc : scoreDocs) {
- Set<String> fieldsToLoad = new HashSet<>();
- fieldsToLoad.add(field);
- Document document = indexSearcher.doc(scoreDoc.doc, fieldsToLoad);
- result.add(document.get(field));
- // System.out.println(document.get(field) + "\t" +
- // scoreDoc.score);
- }
- } catch (IOException e) {
- logger.error("", e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- return result;
- }
- /**
- * 移除集合中重复的元素
- *
- * @param list
- * @return
- */
- private void removeDuplicate(List<String> list) {
- if (list == null) {
- return;
- }
- List<String> result = new ArrayList<>();
- for (String str : list) {
- if (!result.contains(str)) {
- result.add(str);
- }
- }
- list.removeAll(list);
- list.addAll(result);
- }
- /**
- * 删除集合内 startIndex(含)后的元素
- *
- * @param list
- * @param startIndex
- */
- private void removeElements(List<? extends String> list, int startIndex) {
- if (CollectionUtils.isEmpty(list)) {
- return;
- }
- int listsSize = list.size();
- for (int i = listsSize - 1; i >= startIndex; i--) {
- list.remove(i);
- }
- }
- @Override
- public Map<String, Object> getGoodsIds(String keyword, PageParams pageParams) throws SearchException {
- List<String> keywordFields = new ArrayList<>();
- // 先根据品牌搜索,品牌不存在再搜索型号等
- keywordFields.add(SearchConstants.GOODS_BR_NAME_CN_UNTOKENIZED_FIELD);
- keywordFields.add(SearchConstants.GOODS_BR_NAME_EN_UNTOKENIZED_FIELD);
- Map<String, Object> goodsIds = getGoodsIds(keyword, keywordFields, false, pageParams, false);
- if (CollectionUtils.isEmpty(goodsIds) || goodsIds.get("componentIds") == null
- || JSONObject.parseArray(goodsIds.get("componentIds").toString()).isEmpty()) {
- goodsIds = getGoodsIds(keyword, null, true, pageParams, true);
- }
- return goodsIds;
- }
- /**
- * @param keyword
- * @param keywordFields
- * 要查询的字段
- * @param tokenized
- * 是否分词
- * @param pageParams
- * @param recursivelyGet 是否递归获取(逐步降低精度,直到只匹配一个字符)
- * @return
- * @throws SearchException
- */
- private Map<String, Object> getGoodsIds(String keyword, List<String> keywordFields, Boolean tokenized,
- PageParams pageParams, Boolean recursivelyGet) throws SearchException {
- // 因为器件、属性值的数据量远比类目、品牌大得多,而且器件搜索可能还需进行分页,
- // 所以涉及器件、属性值的搜索,大都不能像类目和品牌一样直接利用SearchUtils.getDocuments方法
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(SearchConstants.GOODS_TABLE_NAME);
- if (pageParams == null) {
- pageParams = new PageParams();
- }
- if (pageParams.getPage() <= 0)
- pageParams.setPage(1);
- if (pageParams.getSize() <= 0)
- pageParams.setSize(20);
- Map<String, Object> map = new HashMap<String, Object>();
- List<Long> cmpIds = new ArrayList<>();
- List<Long> goIds = new ArrayList<>();
- try {
- BooleanQuery booleanQuery = queryGoods(keyword, keywordFields, tokenized);
- Map<FilterField, Object> filters = pageParams.getFilters();
- // 筛选状态
- if (!CollectionUtils.isEmpty(filters) && !StringUtils.isEmpty(filters.get(FilterField.GOODS_STATUS))) {
- filter(filters.get(FilterField.GOODS_STATUS), SearchConstants.GOODS_GO_STATUS_FIELD, booleanQuery);
- } else {
- // 未指定时,过滤默认状态
- filter(Arrays.asList(TradeGoods.VALID_STATUS), SearchConstants.GOODS_GO_STATUS_FIELD, booleanQuery);
- }
- if (!CollectionUtils.isEmpty(filters)) {
- // 筛选类目
- if (!StringUtils.isEmpty(filters.get(FilterField.GOODS_KINDID))) {
- filter(filters.get(FilterField.GOODS_KINDID), SearchConstants.GOODS_KI_ID_FIELD, booleanQuery);
- }
- // 筛选品牌
- if (!StringUtils.isEmpty(filters.get(FilterField.GOODS_BRANDID))) {
- filter(filters.get(FilterField.GOODS_BRANDID), SearchConstants.GOODS_BR_ID_FIELD, booleanQuery);
- }
- // 筛选货源
- if (!StringUtils.isEmpty(filters.get(FilterField.GOODS_STORE_TYPE))) {
- filter(filters.get(FilterField.GOODS_STORE_TYPE), SearchConstants.GOODS_ST_TYPE_FIELD, booleanQuery);
- }
- // 筛选货币
- if (!StringUtils.isEmpty(filters.get(FilterField.GOODS_CRNAME))) {
- filter(filters.get(FilterField.GOODS_CRNAME), SearchConstants.GOODS_CRNAME_FIELD, booleanQuery);
- }
- // 价格筛选
- Object minPriceRmb = filters.get(FilterField.GOODS_MINPRICERMB);
- Object maxPriceRmb = filters.get(FilterField.GOODS_MAXPRICERMB);
- Object minPriceUsd = filters.get(FilterField.GOODS_MINPRICEUSD);
- Object maxPriceUsd = filters.get(FilterField.GOODS_MAXPRICEUSD);
- // 筛选人民币价格
- if (!StringUtils.isEmpty(minPriceRmb) || !StringUtils.isEmpty(maxPriceRmb)) {
- Double minPrice = null;
- Double maxPrice = null;
- if (!StringUtils.isEmpty(minPriceRmb)) {
- minPrice = Double.valueOf(minPriceRmb.toString());
- }
- if (!StringUtils.isEmpty(maxPriceRmb)) {
- maxPrice = Double.valueOf(maxPriceRmb.toString());
- }
- booleanQuery.add(NumericRangeQuery.newDoubleRange(SearchConstants.GOODS_GO_MINPRICERMB_FIELD,
- minPrice, maxPrice, true, true), BooleanClause.Occur.FILTER);
- }
- // 筛选美元价格
- if (!StringUtils.isEmpty(minPriceUsd) || !StringUtils.isEmpty(maxPriceUsd)) {
- Double minPrice = null;
- Double maxPrice = null;
- if (!StringUtils.isEmpty(minPriceUsd)) {
- minPrice = Double.valueOf(minPriceUsd.toString());
- }
- if (!StringUtils.isEmpty(maxPriceUsd)) {
- maxPrice = Double.valueOf(maxPriceUsd.toString());
- }
- booleanQuery.add(NumericRangeQuery.newDoubleRange(SearchConstants.GOODS_GO_MINPRICEUSD_FIELD,
- minPrice, maxPrice, true, true), BooleanClause.Occur.FILTER);
- }
- }
- logger.info(booleanQuery.toString());
- // 排序
- // 自定义排序字段 > 批次(访问量) > 器件(访问量 > 搜索次数) > 品牌(权重 > 访问量 > 搜索次数) > 类目(访问量 > 搜索次数) > 分数
- SortField[] customSortFields = sortGoods(keyword);
- List<SortField> sortFieldList = new ArrayList<>();
- List<com.uas.search.constant.model.Sort> sorts = pageParams.getSort();
- if (sorts != null && !CollectionUtils.isEmpty(sorts)) {
- for (com.uas.search.constant.model.Sort s : sorts) {
- if (s.getField() == null) {
- throw new SearchException("排序字段不可为空:" + s);
- }
- switch (s.getField()) {
- // 价格
- case GO_RESERVE:
- sortFieldList.add(new SortField(SearchConstants.GOODS_GO_RESERVE_FIELD, Type.DOUBLE,
- s.isReverse()));
- break;
- // 人民币价格
- case GO_MINPRICERMB:
- sortFieldList.add(new SortField(SearchConstants.GOODS_GO_MINPRICERMB_FIELD, Type.DOUBLE,
- s.isReverse()));
- break;
- // 美元价格
- case GO_MINPRICEUSD:
- sortFieldList.add(new SortField(SearchConstants.GOODS_GO_MINPRICEUSD_FIELD, Type.DOUBLE,
- s.isReverse()));
- break;
- // 打分
- case GO_SEARCH:
- sortFieldList.addAll(Arrays.asList(customSortFields));
- break;
- default:
- throw new SearchException("不支持该排序方式:" + s.getField());
- }
- }
- } else{
- sortFieldList.addAll(Arrays.asList(customSortFields));
- }
- SortField[] sortFields = new SortField[sortFieldList.size()];
- sortFieldList.toArray(sortFields);
- Sort sort = new Sort(sortFields);
- TopDocs hits;
- if (pageParams.getPage() > 1) {// 不是第一页
- TopDocs previousHits = indexSearcher.search(booleanQuery,
- (pageParams.getPage() - 1) * pageParams.getSize(), sort, true, false);
- int totalHits = previousHits.totalHits;
- if ((pageParams.getPage() - 1) * pageParams.getSize() >= totalHits) {
- // 如果没有结果,则降低精度,直至 keyword 长度为 1
- if(recursivelyGet && totalHits < 1 && !SearchUtils.isKeywordInvalid(keyword) && keyword.length() > 1){
- return getGoodsIds(keyword.substring(0, keyword.length() - 1), keywordFields, tokenized, pageParams, recursivelyGet);
- }
- return map;
- }
- ScoreDoc[] previousScoreDocs = previousHits.scoreDocs;
- ScoreDoc after = previousScoreDocs[previousScoreDocs.length - 1];
- hits = indexSearcher.searchAfter(after, booleanQuery, pageParams.getSize(), sort, true, false);
- } else {
- hits = indexSearcher.search(booleanQuery, pageParams.getSize(), sort, true, false);
- }
- // 如果没有结果,则降低精度,直至 keyword 长度为 1
- if(recursivelyGet && hits.totalHits < 1 && !SearchUtils.isKeywordInvalid(keyword) && keyword.length() > 1){
- return getGoodsIds(keyword.substring(0, keyword.length() - 1), keywordFields, tokenized, pageParams, recursivelyGet);
- }
- // 数据量太大,需要指定将获取的数据(以免载入不必要的数据,降低速度)
- Set<String> fieldsToLoad = new HashSet<>();
- fieldsToLoad.add(SearchConstants.GOODS_CMP_ID_FIELD);
- fieldsToLoad.add(SearchConstants.GOODS_GO_ID_FIELD);
- ScoreDoc[] scoreDocs = hits.scoreDocs;
- for (ScoreDoc scoreDoc : scoreDocs) {
- Document document = indexSearcher.doc(scoreDoc.doc, fieldsToLoad);
- String cmpId = document.get(SearchConstants.GOODS_CMP_ID_FIELD);
- cmpIds.add(StringUtils.isEmpty(cmpId) ? null : Long.valueOf(cmpId));
- String goId = document.get(SearchConstants.GOODS_GO_ID_FIELD);
- goIds.add(StringUtils.isEmpty(goId) ? null : Long.valueOf(goId));
- // System.out.println(cmpId + "\t" + goId + "\t" +
- // scoreDoc.score);
- // System.out.println(indexSearcher.explain(booleanQuery,
- // scoreDoc.doc).toString());
- }
- map.put("componentIds", cmpIds);
- map.put("goodsIds", goIds);
- map.put("page", pageParams.getPage());
- map.put("size", pageParams.getSize());
- map.put("total", hits.totalHits);
- } catch (IOException e) {
- logger.error("", e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- return map;
- }
- /**
- * @return 批次排序规则
- */
- private SortField[] sortGoods(String keyword) {
- // 器件、非标型号自定义排序 > 批次(访问量) > 器件(自定义排序 > 访问量 > 搜索次数) > 品牌(自定义排序 > 权重 > 访问量 > 搜索次数) > 类目(访问量 > 搜索次数)
- return new SortField[]{
- sortField(SearchConstants.GOODS_GO_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.GOODS_CMP_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.GOODS_CMP_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.GOODS_BR_WEIGHT_FIELD, Type.DOUBLE, true, Double.MIN_VALUE),
- sortField(SearchConstants.GOODS_BR_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.GOODS_BR_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.GOODS_KI_VISIT_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- sortField(SearchConstants.GOODS_KI_SEARCH_COUNT_FIELD, Type.LONG, true, Long.MIN_VALUE),
- new SortField(SearchConstants.GOODS_CMP_CODE_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.GOODS_PR_PCMPCODE_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.GOODS_BR_NAME_EN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.GOODS_BR_NAME_CN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword, true)),
- new SortField(SearchConstants.GOODS_KI_NAME_CN_UNTOKENIZED_FIELD, new StringFieldComparatorSource(keyword))
- };
- }
- @Override
- public List<Map<String, Object>> collectBySearchGoods(String keyword, CollectField collectedField,
- Map<FilterField, Object> filters) {
- List<String> keywordFields = new ArrayList<>();
- // 先根据品牌搜索,品牌不存在再搜索型号等
- keywordFields.add(SearchConstants.GOODS_BR_NAME_CN_UNTOKENIZED_FIELD);
- keywordFields.add(SearchConstants.GOODS_BR_NAME_EN_UNTOKENIZED_FIELD);
- List<Map<String, Object>> result = collectBySearchGoods(keyword, keywordFields, false, collectedField, filters, false);
- if (CollectionUtils.isEmpty(result)) {
- result = collectBySearchGoods(keyword, null, true, collectedField, filters, true);
- }
- return result;
- }
- /**
- * @param keyword
- * @param keywordFields
- * 要查询的字段
- * @param tokenized
- * 是否分词
- * @param collectedField
- * @param filters
- * @param recursivelyGet 是否递归获取(逐步降低精度,直到只匹配一个字符)
- * @return
- */
- private List<Map<String, Object>> collectBySearchGoods(String keyword, List<String> keywordFields,
- Boolean tokenized, CollectField collectedField, Map<FilterField, Object> filters, Boolean recursivelyGet) {
- if (collectedField == null && CollectionUtils.isEmpty(filters)) {
- throw new SearchException("参数不合法:collectedField=" + collectedField + ", filter=" + filters);
- }
- // 与批次搜索的搜索词保持一致,最终有结果的 keyword 是相同的
- if(recursivelyGet != null && recursivelyGet){
- PageParams pageParams = new PageParams();
- pageParams.setFilters(filters);
- Map<String, Object> goodsIds = getGoodsIds(keyword, keywordFields, tokenized, pageParams, false);
- if(Integer.parseInt(goodsIds.get("total").toString()) < 1 && !SearchUtils.isKeywordInvalid(keyword) && keyword.length() > 1){
- return collectBySearchGoods(keyword.substring(0, keyword.length() - 1), keywordFields, tokenized, collectedField, filters, true);
- }
- }
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(SearchConstants.GOODS_TABLE_NAME);
- List<Map<String, Object>> result = new ArrayList<>();
- try {
- BooleanQuery booleanQuery = queryGoods(keyword, keywordFields, tokenized);
- // 筛选状态
- if (!CollectionUtils.isEmpty(filters) && !StringUtils.isEmpty(filters.get(FilterField.GOODS_STATUS))) {
- filter(filters.get(FilterField.GOODS_STATUS), SearchConstants.GOODS_GO_STATUS_FIELD, booleanQuery);
- } else {
- // 未指定时,过滤默认状态
- filter(Arrays.asList(TradeGoods.VALID_STATUS), SearchConstants.GOODS_GO_STATUS_FIELD, booleanQuery);
- }
- // 过滤
- Set<Entry<FilterField, Object>> entrySet = filters.entrySet();
- for (Entry<FilterField, Object> entry : entrySet) {
- switch (entry.getKey()) {
- case GOODS_KINDID:
- filter(entry.getValue(), SearchConstants.GOODS_KI_ID_FIELD, booleanQuery);
- break;
- case GOODS_BRANDID:
- filter(entry.getValue(), SearchConstants.GOODS_BR_ID_FIELD, booleanQuery);
- break;
- case GOODS_STORE_TYPE:
- filter(entry.getValue(), SearchConstants.GOODS_ST_TYPE_FIELD, booleanQuery);
- break;
- case GOODS_CRNAME:
- filter(entry.getValue(), SearchConstants.GOODS_CRNAME_FIELD, booleanQuery);
- break;
- default:
- throw new SearchException("不支持该过滤字段:" + entry.getKey());
- }
- }
- logger.info(booleanQuery.toString());
- // 统计
- String uniqueField;
- Set<String> fieldsToLoad = new HashSet<>();
- switch (collectedField) {
- case GOODS_KIND:
- uniqueField = SearchConstants.GOODS_KI_ID_FIELD;
- fieldsToLoad.add(SearchConstants.GOODS_KI_ID_FIELD);
- fieldsToLoad.add(SearchConstants.GOODS_KI_NAME_CN_FIELD);
- break;
- case GOODS_BRAND:
- uniqueField = SearchConstants.GOODS_BR_ID_FIELD;
- fieldsToLoad.add(SearchConstants.GOODS_BR_ID_FIELD);
- fieldsToLoad.add(SearchConstants.GOODS_BR_UUID_FIELD);
- fieldsToLoad.add(SearchConstants.GOODS_BR_NAME_CN_FIELD);
- fieldsToLoad.add(SearchConstants.GOODS_BR_NAME_EN_FIELD);
- break;
- case GOODS_STORE_TYPE:
- uniqueField = SearchConstants.GOODS_ST_TYPE_FIELD;
- fieldsToLoad.add(SearchConstants.GOODS_ST_TYPE_FIELD);
- break;
- case GOODS_CRNAME:
- uniqueField = SearchConstants.GOODS_CRNAME_FIELD;
- fieldsToLoad.add(SearchConstants.GOODS_CRNAME_FIELD);
- break;
- default:
- throw new SearchException("不支持该统计字段:" + collectedField);
- }
- GoodsGroupCollector collector = new GoodsGroupCollector(uniqueField, fieldsToLoad);
- indexSearcher.search(booleanQuery, collector);
- result = collector.getValues();
- } catch (IOException e) {
- logger.error("", e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- return result;
- }
- /**
- * 获取查询批次的query
- *
- * @param keyword
- * @param keywordFields
- * @param tokenized
- * @return
- */
- private BooleanQuery queryGoods(String keyword, List<String> keywordFields, Boolean tokenized) {
- BooleanQuery booleanQuery = new BooleanQuery();
- if (!SearchUtils.isKeywordInvalid(keyword)) {
- // 未指定搜索的字段,则采用默认搜索逻辑
- if (CollectionUtils.isEmpty(keywordFields)) {
- booleanQuery.add(setGoodsBoost(keyword), BooleanClause.Occur.MUST);
- } else {
- BooleanQuery booleanQuery2 = new BooleanQuery();
- for (String keywordField : keywordFields) {
- // 是否分词
- if (tokenized == null || !tokenized.booleanValue()) {
- booleanQuery2.add(new TermQuery(new Term(keywordField, keyword.toLowerCase())), BooleanClause.Occur.SHOULD);
- } else {
- booleanQuery2.add(SearchUtils.getBooleanQuery(keywordField, keyword), BooleanClause.Occur.SHOULD);
- }
- }
- booleanQuery.add(booleanQuery2, BooleanClause.Occur.MUST);
- }
- }
- return booleanQuery;
- }
- /**
- * 同时搜索器件、类目、品牌等,并设置boost
- */
- private Query setGoodsBoost(String keyword) {
- BooleanQuery booleanQuery = new BooleanQuery();
- // 原厂型号
- booleanQuery.add(createQuery(SearchConstants.GOODS_CMP_CODE_FIELD, keyword, 100), BooleanClause.Occur.SHOULD);
- // 非标
- booleanQuery.add(createQuery(SearchConstants.GOODS_PR_PCMPCODE_FIELD, keyword, 100), Occur.SHOULD);
- // 品牌
- booleanQuery.add(createQuery(SearchConstants.GOODS_BR_NAME_CN_FIELD, keyword, 10), BooleanClause.Occur.SHOULD);
- booleanQuery.add(createQuery(SearchConstants.GOODS_BR_NAME_EN_FIELD, keyword, 10), BooleanClause.Occur.SHOULD);
- // 类目
- booleanQuery.add(createQuery(SearchConstants.GOODS_KI_NAME_CN_FIELD, keyword, 1), BooleanClause.Occur.SHOULD);
- // 属性值
- booleanQuery.add(createQuery(SearchConstants.GOODS_CMP_DESCRIPTION_FIELD, keyword, 1), BooleanClause.Occur.SHOULD);
- return booleanQuery;
- }
- private BooleanQuery createQuery(String field, String keyword, float boost){
- return createQuery(field, keyword, false, boost);
- }
- private BooleanQuery createQuery(String field, String keyword, boolean useRegexpQuery, float boost){
- BooleanQuery booleanQuery = new BooleanQuery();
- if (StringUtils.isEmpty(field) || StringUtils.isEmpty(keyword)) {
- return booleanQuery;
- }
- // 根据空格分隔关键词,分隔的词取或的关系
- String[] array = keyword.split(" ");
- for(String str : array){
- if(!StringUtils.isEmpty(str)){
- booleanQuery.add(SearchUtils.getBooleanQuery(field, str, useRegexpQuery), Occur.SHOULD);
- }
- }
- booleanQuery.setBoost(boost);
- return booleanQuery;
- }
- /**
- * 过滤
- *
- * @param list
- * 过滤值列表
- * @param field
- * 过滤的字段
- * @param booleanQuery
- * 查询条件
- */
- @SuppressWarnings("unchecked")
- private void filter(Object list, String field, BooleanQuery booleanQuery) {
- List<Object> values;
- if (list instanceof List) {
- values = (List<Object>) list;
- }else{
- values = new ArrayList<>();
- values.add(list);
- }
- BooleanQuery booleanQuery2 = new BooleanQuery();
- for (Object value : values) {
- TermQuery query = new TermQuery(new Term(field, value.toString().toLowerCase()));
- booleanQuery2.add(query, BooleanClause.Occur.SHOULD);
- }
- booleanQuery.add(booleanQuery2, BooleanClause.Occur.FILTER);
- }
- @Override
- public Kind getKind(Long id) {
- return DocumentToObjectUtils.toKind(
- SearchUtils.getDocumentById(SearchConstants.KIND_TABLE_NAME, SearchConstants.KIND_ID_FIELD, id));
- }
- @Override
- public Brand getBrand(Long id) {
- return DocumentToObjectUtils.toBrand(
- SearchUtils.getDocumentById(SearchConstants.BRAND_TABLE_NAME, SearchConstants.BRAND_ID_FIELD, id));
- }
- @Override
- public Component getComponent(Long id) {
- return DocumentToObjectUtils.toComponent(SearchUtils.getDocumentById(SearchConstants.COMPONENT_TABLE_NAME,
- SearchConstants.COMPONENT_ID_FIELD, id));
- }
- @Override
- public Goods getGoods(String id) {
- return DocumentToObjectUtils.toGoods(
- SearchUtils.getDocumentById(SearchConstants.GOODS_TABLE_NAME, SearchConstants.GOODS_GO_ID_FIELD, id));
- }
- @Override
- public SPage<Object> getObjects(String tableName, String keyword, String field, Boolean tokenized, @NotEmpty("page") Integer page, @NotEmpty("size") Integer size) {
- if (keyword == null) {
- keyword = "";
- }
- if (field == null) {
- field = SearchUtils.getIdField(tableName);
- }
- if (tokenized == null) {
- tokenized = false;
- }
- IndexSearcher indexSearcher = SearchUtils.getIndexSearcher(tableName);
- SPage<Object> sPage = new SPage<>();
- try {
- Query query;
- if (tokenized) {
- query = SearchUtils.getBooleanQuery(field, keyword);
- } else {
- query = SearchUtils.getRegexpQuery(field, keyword);
- }
- // 分页信息
- if (page > 0) {
- sPage.setPage(page);
- } else {
- sPage.setPage(1);
- sPage.setFirst(true);
- }
- if (size > 0) {
- sPage.setSize(size);
- } else {
- sPage.setSize(20);
- }
- TopDocs topDocs;
- // 如果页码不为1
- if (sPage.getPage() > 1) {
- TopDocs previousTopDocs = indexSearcher.search(query, (sPage.getPage() - 1) * sPage.getSize());
- int totalHits = previousTopDocs.totalHits;
- ScoreDoc[] previousScoreDocs = previousTopDocs.scoreDocs;
- if ((sPage.getPage() - 1) * sPage.getSize() >= totalHits) {
- throw new SearchException("页码过大:元素总数量为" + totalHits);
- }
- topDocs = indexSearcher.searchAfter(previousScoreDocs[previousScoreDocs.length - 1], query,
- sPage.getSize());
- } else {
- sPage.setFirst(true);
- topDocs = indexSearcher.search(query, sPage.getSize());
- }
- int totalHits = topDocs.totalHits;
- // 设置总元素个数、页数等信息
- sPage.setTotalElement(totalHits);
- int totalPage = (int) Math.ceil(totalHits / (1.0 * sPage.getSize()));
- sPage.setTotalPage(totalPage);
- if (totalPage == sPage.getPage()) {
- sPage.setLast(true);
- }
- List<Object> content = new ArrayList<>();
- for (ScoreDoc scoreDoc : topDocs.scoreDocs) {
- Document document = indexSearcher.doc(scoreDoc.doc);
- content.add(DocumentToObjectUtils.toObject(document, tableName));
- }
- sPage.setContent(content);
- } catch (IOException e) {
- throw new SearchException(e).setDetailedMessage(e);
- } finally {
- SearchUtils.releaseIndexSearcher(indexSearcher);
- }
- return sPage;
- }
- }
|