lunece學習筆記之搜索部分

創建 一個SearcherUti類

 //精確匹配  搜索
 public void searchByTerm(String field,String name,int num) {
  try {
   IndexSearcher searcher = getSearcher();
   Query query = new TermQuery(new Term(field,name));
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 
 //字符範圍匹配 搜索
 public void searchByTermRange(String field,String start,String end,int num) {
  try {
   IndexSearcher searcher = getSearcher();
   Query query = new TermRangeQuery(field,start,end,true, true);
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 
 ///數字範圍匹配查詢 搜索
 public void searchByNumricRange(String field,int start,int end,int num) {
  try {
   IndexSearcher searcher = getSearcher();
   Query query = NumericRangeQuery.newIntRange(field,start, end,true,true);
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 
 //前綴搜索 以什麼打頭的 相似sql j%
 public void searchByPrefix(String field,String value,int num) {
  try {
   IndexSearcher searcher = getSearcher();
   Query query = new PrefixQuery(new Term(field,value));
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 //通配符查詢  表明任意 ?表明匹配一個字符
 public void searchByWildcard(String field,String value,int num) {
  try {
   IndexSearcher searcher = getSearcher();
   //在傳入的value中可使用通配符:?和*,?表示匹配一個字符，*表示匹配任意多個字符
   Query query = new WildcardQuery(new Term(field,value));
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 
 //能夠鏈接多個條件，可鏈接多個子查詢  搜索
 public void searchByBoolean(int num) {
  try {
   IndexSearcher searcher = getSearcher();
   BooleanQuery query = new BooleanQuery();
   /*
    * BooleanQuery能夠鏈接多個子查詢
    * Occur.MUST表示必須出現
    * Occur.SHOULD表示能夠出現
    * Occur.MUSE_NOT表示不能出現
    */
   query.add(new TermQuery(new Term("name","zhangsan")), Occur.MUST_NOT);
   query.add(new TermQuery(new Term("content","game")),Occur.SHOULD);
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 
 //短語查詢 搜索字符與字符之間中間的區域某個字符
 public void searchByPhrase(int num) {
  try {
   IndexSearcher searcher = getSearcher();
   PhraseQuery query = new PhraseQuery();
   query.setSlop(3);
   query.add(new Term("content","pingpeng"));
   //第一個Term
   query.add(new Term("content","i"));
   //產生距離以後的第二個Term
//   query.add(new Term("content","football"));
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 //模糊查詢 這個模糊查詢匹配一個字符出錯的 是又距離限制的 若有一組數據mike make mske 查詢會是三個都查詢出來
 public void searchByFuzzy(int num) {
  try {
   IndexSearcher searcher = getSearcher();
   FuzzyQuery query = new FuzzyQuery(new Term("name","mase"),0.4f,0);
   System.out.println(query.getPrefixLength());
   System.out.println(query.getMinSimilarity());
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date"));
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 
 //字符包含搜索最重要最經常使用  搜索任意數據包括like的字符,能夠修改改變搜索任意搜索域內容
 public void searchByQueryParse(Query query,int num) {
  try {
   IndexSearcher searcher = getSearcher();
   TopDocs tds = searcher.search(query, num);
   System.out.println("一共查詢了:"+tds.totalHits);
   for(ScoreDoc sd:tds.scoreDocs) {
    Document doc = searcher.doc(sd.doc);
    System.out.println(doc.get("id")+"---->"+
      doc.get("name")+"["+doc.get("email")+"]-->"+doc.get("id")+","+
      doc.get("attach")+","+doc.get("date")+"=="+sd.score);
   }
   searcher.close();
  } catch (CorruptIndexException e) {
   e.printStackTrace();
  } catch (IOException e) {
   e.printStackTrace();
  }
 }

 

測試類

@Test
 public void testCopyFiles() {
  try {
   File file = new File("d:/lucene/example/");
   for(File f:file.listFiles()) {
    String destFileName = FilenameUtils.getFullPath(f.getAbsolutePath())+
      FilenameUtils.getBaseName(f.getName())+".she";
    FileUtils.copyFile(f, new File(destFileName));
   }
  } catch (IOException e) {
   e.printStackTrace();
  }
 }
 
 @Test
 public void searchByTerm() {
  su.searchByTerm("content","i",3);
 }
 
 @Test
 public void searchByTermRange() {
  //查詢name以a開頭和s結尾的
//  su.searchByTermRange("name","a","s",10);
  //因爲attachs是數字類型，使用TermRange沒法查詢
  su.searchByTermRange("attach","2","10", 5);
 }
 
 @Test
 public void searchByNumRange() {
  su.searchByNumricRange("attach",2,10, 5);
 }
 
 @Test
 public void searchByPrefix() {
  su.searchByPrefix("content", "s", 10);
 }
 
 @Test
 public void searchByWildcard() {
  //匹配@itat.org結尾的全部字符
  su.searchByWildcard("email", "*@itat.org", 10);
  //匹配j開頭的有三個字符的name
  su.searchByWildcard("name", "j???", 10);
 }
 
 @Test
 public void searchByBoolean() {
  su.searchByBoolean(10);
 }
 
 @Test
 public void searchByPhrase() {
  su.searchByPhrase(10);
 }
 
 @Test
 public void searchByFuzzy() {
  su.searchByFuzzy(10);
 }
 
 @Test
 public void searchByQueryParse() throws ParseException {
  //一、建立QueryParser對象,默認搜索域爲content
  QueryParser parser = new QueryParser(Version.LUCENE_35, "content", new StandardAnalyzer(Version.LUCENE_35));
  //改變空格的默認操做符，如下能夠改爲AND
  //parser.setDefaultOperator(Operator.AND);
  //開啓第一個字符的通配符匹配，默認關閉由於效率不高
  parser.setAllowLeadingWildcard(true);
  //搜索content中包含有like的
  Query query = parser.parse("like");
  
  //有basketball或者football的，空格默認就是OR
  query = parser.parse("basketball football");
  
  //改變搜索域爲name爲mike
  //query = parser.parse("content:like");
  
  //一樣可使用*和?來進行通配符匹配
//  query = parser.parse("name:j*");
  
  //通配符默認不能放在首位
//  query = parser.parse("email:*@itat.org");
  
  //匹配name中沒有mike可是content中必須有football的，+和-要放置到域說明前面
  query = parser.parse("- name:mike + like");
  
  //匹配一個區間，注意:TO必須是大寫
  //query = parser.parse("id:[1 TO 6]");
  
  //閉區間匹配只會匹配到2
  //query = parser.parse("id:{1 TO 3}");
  
  //徹底匹配I Like Football的
  //query = parser.parse("\"I like football\"");
  
  //匹配I 和football之間有一個單詞距離的
  //query = parser.parse("\"I football\"~1");
  
  //模糊查詢
  //query = parser.parse("name:make~");
  
  //沒有辦法匹配數字範圍（本身擴展Parser）
  //query = parser.parse("attach:[2 TO 10]");
  su.searchByQueryParse(query, 10);
 }
 
 @Test
 public void indexFile() {
  FileIndexUtils.index(true);
 }
 
 @Test
 public void testSearchPage01() {
  su.searchPage("java", 2,20);
  System.out.println("-------------------------------");
//  su.searchNoPage("java");
  su.searchPageByAfter("java", 2,20);
 }
 
 @Test
 public void testSearchPage02() {
  su.searchPageByAfter("java", 3,20);
 }