/// <summary>
/// 搜索文档内容且精确匹配
/// </summary>
/// <param name="text"></param>
/// <returns></returns>
public List <Doc> SearchDocumentContentAndNotSplitWord(string text)
{
List <Doc> result = new List <Doc>();
try
{
bool isExistIndex = IndexReader.IndexExists(this._fsDir);
if (isExistIndex)
{
this._indexReader = IndexReader.Open(this._fsDir, false);
IndexSearcher searcher = new IndexSearcher(this._indexReader);
//搜索条件
BooleanQuery queryOr1 = new BooleanQuery();
PhraseQuery query1 = new PhraseQuery();
foreach (string word in SplitContent.SplitWords(text))
{
query1.Add(new Term(DocStruct.CONTENT, word));
queryOr1.Add(query1, BooleanClause.Occur.MUST);
}
MultiSearcher multiSearch = new MultiSearcher(new[] { searcher });
//TopScoreDocCollector盛放查询结果的容器
TopScoreDocCollector collector = TopScoreDocCollector.create(300, true);
multiSearch.Search(queryOr1, collector);
ScoreDoc[] docs = collector.TopDocs(0, collector.GetTotalHits()).scoreDocs.OrderByDescending(t => t.score).ToArray();
for (int i = 0; i < docs.Length; i++)
{
int docId = docs[i].doc; //得到查询结果文档的id（Lucene内部分配的id）
Document doc = searcher.Doc(docId); //根据文档id来获得文档对象Document
var d = new Doc();
d.Id = doc.Get(DocStruct.ID);
result.Add(d);
}
}
}
catch (Exception ex)
{
LogHelper.WriteError(this.GetType(), ex);
}
return(result);
}
/// <summary>
/// 搜索
/// </summary>
/// <param name="text"></param>
/// <returns></returns>
public List <Doc> Search(string text)
{
List <Doc> result = new List <Doc>();
try
{
bool isExistIndex = IndexReader.IndexExists(this._fsDir);
if (isExistIndex)
{
this._indexReader = IndexReader.Open(this._fsDir, false);
IndexSearcher searcher = new IndexSearcher(this._indexReader);
//搜索条件
BooleanQuery shouldQuery = new BooleanQuery();
BooleanQuery titleQuery = new BooleanQuery();
BooleanQuery contentQuery = new BooleanQuery();
BooleanQuery categoryQuery = new BooleanQuery();
//把用户输入的关键字进行分词
foreach (string word in SplitContent.SplitWords(text))
{
PhraseQuery query1 = new PhraseQuery();
PhraseQuery query2 = new PhraseQuery();
PhraseQuery query3 = new PhraseQuery();
query1.Add(new Term(DocStruct.TITLE, word));
query2.Add(new Term(DocStruct.CONTENT, word));
query3.Add(new Term(DocStruct.CATEGORY, word));
// 引号括起来的词语，必须出现
if (Regex.IsMatch(text, "^\".*\"$"))
{
titleQuery.Add(query1, BooleanClause.Occur.MUST);
contentQuery.Add(query2, BooleanClause.Occur.MUST);
categoryQuery.Add(query3, BooleanClause.Occur.MUST);
}
else
{
titleQuery.Add(query1, BooleanClause.Occur.SHOULD);
contentQuery.Add(query2, BooleanClause.Occur.SHOULD);
categoryQuery.Add(query3, BooleanClause.Occur.SHOULD);
}
}
shouldQuery.Add(titleQuery, BooleanClause.Occur.SHOULD);
shouldQuery.Add(contentQuery, BooleanClause.Occur.SHOULD);
shouldQuery.Add(categoryQuery, BooleanClause.Occur.SHOULD);
MultiSearcher multiSearch = new MultiSearcher(new[] { searcher });
//TopScoreDocCollector盛放查询结果的容器
TopScoreDocCollector collector = TopScoreDocCollector.create(300, true);
multiSearch.Search(shouldQuery, collector);
//TopDocs 指定0到GetTotalHits() 即所有查询结果中的文档 如果TopDocs(20,10)则意味着获取第20-30之间文档内容 达到分页的效果
ScoreDoc[] docs = collector.TopDocs(0, collector.GetTotalHits()).scoreDocs.OrderByDescending(t => t.score).ToArray();
for (int i = 0; i < docs.Length; i++)
{
int docId = docs[i].doc; //得到查询结果文档的id（Lucene内部分配的id）
Document doc = searcher.Doc(docId); //根据文档id来获得文档对象Document
var d = new Doc();
d.Id = doc.Get(DocStruct.ID);
result.Add(d);
}
//LogHelper.WriteInfo(this.GetType(), string.Format("Searched results count:{0}", docs.Length));
}
}
catch (Exception ex)
{
LogHelper.WriteError(this.GetType(), ex);
}
return(result);
}
public List <CodeIndex> Search(string text, string language, int size = 300)
{
List <CodeIndex> result = new List <CodeIndex>();
try
{
bool isExistIndex = this._codeIndexLucene.IndexExists();
if (isExistIndex && !text.IsNullOrEmpty())
{
IndexSearcher searcher = new IndexSearcher(this._codeIndexLucene.OpenReader());
BooleanQuery allQuery = new BooleanQuery();
allQuery.Add(new TermQuery(new Term(IndexField.Language, language)), BooleanClause.Occur.MUST);
//搜索条件
BooleanQuery shouldQuery = new BooleanQuery();
//把用户输入的关键字进行分词
foreach (string word in SplitContent.SplitWords(text))
{
PhraseQuery query1 = new PhraseQuery();
query1.Add(new Term(IndexField.SearchText, word));
shouldQuery.Add(query1, BooleanClause.Occur.SHOULD);
}
allQuery.Add(shouldQuery, BooleanClause.Occur.MUST);
MultiSearcher multiSearch = new MultiSearcher(new[] { searcher });
TopScoreDocCollector collector = TopScoreDocCollector.create(size, true);
multiSearch.Search(allQuery, collector);
ScoreDoc[] docs = collector.TopDocs(0, collector.GetTotalHits()).scoreDocs.OrderByDescending(t => t.score).ToArray();
for (int i = 0; i < docs.Length; i++)
{
int docId = docs[i].doc; //得到查询结果文档的id（Lucene内部分配的id）
Document doc = searcher.Doc(docId); //根据文档id来获得文档对象Document
var m = new CodeIndex
{
Id = doc.Get(IndexField.Id),
SearchText = doc.Get(IndexField.SearchText),
CodeBody = doc.Get(IndexField.CodeBody),
UserId = doc.Get(IndexField.UserId),
Language = doc.Get(IndexField.Language)
};
var codeDoc = this._codeFileLucene.GetDoc(m.FileId);
if (codeDoc != null)
{
m.FileContent = codeDoc.Get(nameof(CodeFile.FileContent));
}
result.Add(m);
// 当完全匹配时，只返回此条
if (text != null && m.SearchText != null && m.SearchText.Trim().ToLower() == text.Trim().ToLower())
{
break;
}
}
}
}
catch (Exception ex)
{
LogHelper.WriteError(this.GetType(), ex);
}
return(result);
}
