/// <summary>
/// 搜索可能喜欢的文档
/// </summary>
/// <param name="pageNo"></param>
/// <param name="pageLen"></param>
/// <returns></returns>
public ArrayList SearchMaybeLike(string title, int pageNo, int pageLen,int docId)
{
string connStr = Helper.ConfigHelper.HubbleConnStr;
DataSet ds;
ArrayList rList = new ArrayList();
using (HubbleConnection conn = new HubbleConnection(connStr))
{
#region hubble
conn.Open();
GetAnalyzerName(conn, TableName);
string wordssplitbyspace;
HubbleCommand matchCmd = new HubbleCommand(conn);
string matchString = matchCmd.GetKeywordAnalyzerStringFromServer(
TableName, "Title", title, int.MaxValue, out wordssplitbyspace);
HubbleDataAdapter da = new HubbleDataAdapter();
da.SelectCommand = new HubbleCommand(
string.Format(
"select between @begin to @end * from {0} where IsAudit=1 and ( Description match @matchString or title^2 match @matchString or tags^2 match @matchString ) "
+ " and DocumentId<>@docid order by score desc"
, TableName),
conn);
da.SelectCommand.Parameters.Add("@begin", pageNo * pageLen);
da.SelectCommand.Parameters.Add("@end", (pageNo + 1) * pageLen - 1);
da.SelectCommand.Parameters.Add("@matchString", matchString);
da.SelectCommand.Parameters.Add("@docid",docId);
da.SelectCommand.CacheTimeout = 0;
ds = new DataSet();
HubbleCommand cmd = da.SelectCommand;
ds = cmd.Query(0);
long[] docids = new long[ds.Tables[0].Rows.Count];
int i = 0;
foreach (System.Data.DataRow row in ds.Tables[0].Rows)
{
docids[i++] = (long)row["DocId"];
}
rList.AddRange(DataToList(ds, title, false));
#endregion
}
return rList;
}
private string descAnalyzerName; //描述分析器名称
public ArrayList Search(string keyWords, int pageNo, int pageLen, out int recCount, out long elapsedMilliseconds)
{
string connStr = Helper.ConfigHelper.HubbleConnStr;
System.Diagnostics.Stopwatch watch = new System.Diagnostics.Stopwatch();
watch.Start();
DataSet ds;
using (HubbleConnection conn = new HubbleConnection(connStr))
{
#region hubble
conn.Open();
GetAnalyzerName(conn, TableName);
string wordssplitbyspace;
HubbleCommand matchCmd = new HubbleCommand(conn);
string matchString = matchCmd.GetKeywordAnalyzerStringFromServer(
TableName, "Title", keyWords, int.MaxValue, out wordssplitbyspace);
HubbleDataAdapter da = new HubbleDataAdapter();
string sqlTemplate = string.Empty;
if (string.IsNullOrEmpty(this.DocType))
sqlTemplate = "select between @begin to @end * from {0} where IsAudit=1 and ( Description match @matchString or title^2 match @matchString or tags^2 match @matchString ) order by score desc";
else
sqlTemplate = "select between @begin to @end * from {0} where IsAudit=1 and DocType='" + this.DocType + "' and ( Description match @matchString or title^2 match @matchString or tags^2 match @matchString ) order by score desc";
da.SelectCommand = new HubbleCommand( string.Format(sqlTemplate,TableName),conn);
da.SelectCommand.Parameters.Add("@begin",pageNo * pageLen);
da.SelectCommand.Parameters.Add("@end", (pageNo+1) * pageLen - 1);
da.SelectCommand.Parameters.Add("@matchString", matchString);
da.SelectCommand.CacheTimeout = 0;
ds = new DataSet();
HubbleCommand cmd = da.SelectCommand;
ds = cmd.Query(0);
long[] docids = new long[ds.Tables[0].Rows.Count];
int i = 0;
//foreach (System.Data.DataRow row in ds.Tables[0].Rows)
//{
// docids[i++] = (long)row["DocumentId"];
//}
#endregion
}
recCount = ds.Tables[0].MinimumCapacity;
ArrayList rList = DataToList(ds, keyWords,true);
watch.Stop();
elapsedMilliseconds = watch.ElapsedMilliseconds;
return rList;
}
/// <summary>
/// 获取格式化后的关键字
/// eg. '要出发旅行网' --> '要出发^rank^0 旅行网^rank^1'
/// 其中，rank为分词后词性的等级，position为词的位置，为计分提供参数
/// </summary>
/// <param name="keyWords">搜索的关键字</param>
/// <param name="tableName">搜索Hubble中对应的表（索引）</param>
/// <param name="fieldName">被搜索的字段（如果是多字段搜索，只放其中一个字段）</param>
/// <returns></returns>
public string GetKeywordAnalyzerStringFromat(string keyWords, string tableName, string fieldName)
{
HubbleCommand matchCmd = CreateHubbleCommand();
string wordssplitbyspace;
string matchString;
try
{
matchString = matchCmd.GetKeywordAnalyzerStringFromServer(tableName,
fieldName, keyWords, int.MaxValue, out wordssplitbyspace);
}
catch (Exception ex)
{
throw new Exception(ex.Message);
}
return matchString;
}
/// <summary>
/// 搜索可能喜欢的文档
/// </summary>
/// <param name="pageNo"></param>
/// <param name="pageLen"></param>
/// <returns></returns>
public ArrayList SearchMaybeLike(string title, int pageNo, int pageLen,int docId)
{
string connStr = Helper.ConfigHelper.HubbleConnStr;
DataSet ds;
ArrayList rList = new ArrayList();
using (HubbleConnection conn = new HubbleConnection(connStr))
{
#region hubble
conn.Open();
GetAnalyzerName(conn, TableName);
string wordssplitbyspace;
HubbleCommand matchCmd = new HubbleCommand(conn);
string matchString = matchCmd.GetKeywordAnalyzerStringFromServer(
TableName, "Title", title, int.MaxValue, out wordssplitbyspace);
HubbleDataAdapter da = new HubbleDataAdapter();
da.SelectCommand = new HubbleCommand(
string.Format(
"select between @begin to @end * from {0} where IsAudit=1 and ( Description match @matchString or title^2 match @matchString or tags^2 match @matchString ) "
+ " and DocumentId<>@docid order by score desc"
, TableName),
conn);
da.SelectCommand.Parameters.Add("@begin", pageNo * pageLen);
da.SelectCommand.Parameters.Add("@end", (pageNo + 1) * pageLen - 1);
da.SelectCommand.Parameters.Add("@matchString", matchString);
da.SelectCommand.Parameters.Add("@docid",docId);
da.SelectCommand.CacheTimeout = 0;
ds = new DataSet();
HubbleCommand cmd = da.SelectCommand;
ds = cmd.Query(0);
long[] docids = new long[ds.Tables[0].Rows.Count];
int i = 0;
foreach (System.Data.DataRow row in ds.Tables[0].Rows)
{
docids[i++] = (long)row["DocId"];
}
rList.AddRange(DataToList(ds, title, false));
#endregion
}
return rList;
}
public ArrayList Search(string keyWords, int pageNo, int pageLen, out int recCount, out long elapsedMilliseconds)
{
string connStr = Helper.ConfigHelper.HubbleConnStr;
System.Diagnostics.Stopwatch watch = new System.Diagnostics.Stopwatch();
watch.Start();
DataSet ds;
using (HubbleConnection conn = new HubbleConnection(connStr))
{
#region hubble
conn.Open();
GetAnalyzerName(conn, TableName);
string wordssplitbyspace;
HubbleCommand matchCmd = new HubbleCommand(conn);
string matchString = matchCmd.GetKeywordAnalyzerStringFromServer(
TableName, "Title", keyWords, int.MaxValue, out wordssplitbyspace);
HubbleDataAdapter da = new HubbleDataAdapter();
string sqlTemplate = string.Empty;
if (string.IsNullOrEmpty(this.DocType))
sqlTemplate = "select between @begin to @end * from {0} where IsAudit=1 and ( Description match @matchString or title^2 match @matchString or tags^2 match @matchString ) order by score desc";
else
sqlTemplate = "select between @begin to @end * from {0} where IsAudit=1 and DocType='" + this.DocType + "' and ( Description match @matchString or title^2 match @matchString or tags^2 match @matchString ) order by score desc";
da.SelectCommand = new HubbleCommand( string.Format(sqlTemplate,TableName),conn);
da.SelectCommand.Parameters.Add("@begin",pageNo * pageLen);
da.SelectCommand.Parameters.Add("@end", (pageNo+1) * pageLen - 1);
da.SelectCommand.Parameters.Add("@matchString", matchString);
da.SelectCommand.CacheTimeout = 0;
ds = new DataSet();
HubbleCommand cmd = da.SelectCommand;
ds = cmd.Query(0);
long[] docids = new long[ds.Tables[0].Rows.Count];
int i = 0;
//foreach (System.Data.DataRow row in ds.Tables[0].Rows)
//{
// docids[i++] = (long)row["DocumentId"];
//}
#endregion
}
recCount = ds.Tables[0].MinimumCapacity;
ArrayList rList = DataToList(ds, keyWords,true);
watch.Stop();
elapsedMilliseconds = watch.ElapsedMilliseconds;
return rList;
}
public static List <TNews> Search(String indexDir, string searchType, String q, int pageLen, int pageNo, string sortBy,
out int recCount, out long elapsedMilliseconds, out string sql)
{
List <TNews> result = new List <TNews>();
string keywords = q;
//string matchString = GetKeyWordsSplit(q, new PanGuAnalyzer(), out wordssplitbyspace);
System.Configuration.ConnectionStringSettings connString =
System.Web.Configuration.WebConfigurationManager.ConnectionStrings["News"];
System.Diagnostics.Stopwatch sw = new System.Diagnostics.Stopwatch();
string connectString = connString.ConnectionString;
System.Data.DataSet ds;
//System.Data.DataTable titleWordsPositions;
//System.Data.DataTable contentWordsPositions;
sw.Start();
using (HubbleAsyncConnection conn = new HubbleAsyncConnection(connectString))
{
conn.Open();
GetAnalyzerName(conn, "News");
if (string.IsNullOrEmpty(sortBy))
{
sortBy = "score";
}
string wordssplitbyspace;
HubbleCommand matchCmd = new HubbleCommand(conn);
string matchString = matchCmd.GetKeywordAnalyzerStringFromServer("News",
"Content", keywords, int.MaxValue, out wordssplitbyspace);
//HubbleCommand cmd = new HubbleCommand("select between {0} to {1} * from News where content match {2} or title^2 match {2} order by " + sortBy,
// conn, (pageNo - 1) * pageLen, pageNo * pageLen - 1, matchString);
HubbleDataAdapter adapter = new HubbleDataAdapter();
if (searchType == "Precise")
{
adapter.SelectCommand = new HubbleCommand("select between @begin to @end * from News where content contains @matchString or title^2 contains @matchString order by " + sortBy,
conn);
}
else if (searchType == "Fuzzy")
{
adapter.SelectCommand = new HubbleCommand("select between @begin to @end * from News where content match @matchString or title^2 match @matchString order by " + sortBy,
conn);
}
else if (searchType == "Like")
{
adapter.SelectCommand = new HubbleCommand("select between @begin to @end * from News where content like @likeString or title^2 like @likeString order by " + sortBy,
conn);
}
else
{
throw new ArgumentException(string.Format("Invalid search type: {0}", searchType));
}
adapter.SelectCommand.Parameters.Add("@begin", (pageNo - 1) * pageLen);
adapter.SelectCommand.Parameters.Add("@end", pageNo * pageLen - 1);
adapter.SelectCommand.Parameters.Add("@matchString", matchString);
adapter.SelectCommand.Parameters.Add("@likeString", "*" + q.Trim() + "*");
adapter.SelectCommand.CacheTimeout = CacheTimeout;
sql = adapter.SelectCommand.Sql;
ds = new System.Data.DataSet();
//adapter.Fill(ds);
HubbleCommand cmd = adapter.SelectCommand;
ds = cmd.Query(CacheTimeout);
long[] docids = new long[ds.Tables[0].Rows.Count];
int i = 0;
foreach (System.Data.DataRow row in ds.Tables[0].Rows)
{
docids[i++] = (long)row["DocId"];
}
//titleWordsPositions = cmd.GetWordsPositions(wordssplitbyspace, "News", "Title", docids, int.MaxValue);
//contentWordsPositions = cmd.GetWordsPositions(wordssplitbyspace, "News", "Content", docids, int.MaxValue);
}
recCount = ds.Tables[0].MinimumCapacity;
foreach (System.Data.DataRow row in ds.Tables[0].Rows)
{
TNews news = new TNews();
news.Title = row["Title"].ToString();
news.Content = row["Content"].ToString();
news.Url = row["Url"].ToString();
news.Time = (DateTime)row["Time"];
SimpleHTMLFormatter simpleHTMLFormatter =
new SimpleHTMLFormatter("<font color=\"red\">", "</font>");
Highlighter titleHighlighter;
Highlighter contentHighlighter;
if (_TitleAnalyzerName.Equals("PanGuSegment", StringComparison.CurrentCultureIgnoreCase))
{
titleHighlighter =
new Highlighter(simpleHTMLFormatter, new PanGuAnalyzer());
}
else if (_TitleAnalyzerName.Equals("EnglishAnalyzer", StringComparison.CurrentCultureIgnoreCase))
{
titleHighlighter = new Highlighter(simpleHTMLFormatter, new Hubble.Core.Analysis.EnglishAnalyzer());
}
else
{
titleHighlighter = new Highlighter(simpleHTMLFormatter, new Hubble.Core.Analysis.SimpleAnalyzer());
}
if (_ContentAnalyzerName.Equals("PanGuSegment", StringComparison.CurrentCultureIgnoreCase))
{
contentHighlighter =
new Highlighter(simpleHTMLFormatter, new PanGuAnalyzer());
}
else if (_ContentAnalyzerName.Equals("EnglishAnalyzer", StringComparison.CurrentCultureIgnoreCase))
{
contentHighlighter = new Highlighter(simpleHTMLFormatter, new Hubble.Core.Analysis.EnglishAnalyzer());
}
else
{
contentHighlighter = new Highlighter(simpleHTMLFormatter, new Hubble.Core.Analysis.SimpleAnalyzer());
}
titleHighlighter.FragmentSize = 50;
contentHighlighter.FragmentSize = 50;
//news.Abstract = highlighter.GetBestFragment(contentWordsPositions, news.Content, (long)row["DocId"]);
//news.TitleHighLighter = highlighter.GetBestFragment(titleWordsPositions, news.Title, (long)row["DocId"]);
news.Abstract = contentHighlighter.GetBestFragment(keywords, news.Content);
news.TitleHighLighter = titleHighlighter.GetBestFragment(keywords, news.Title);
if (string.IsNullOrEmpty(news.TitleHighLighter))
{
news.TitleHighLighter = news.Title;
}
result.Add(news);
}
sw.Stop();
elapsedMilliseconds = sw.ElapsedMilliseconds;
return(result);
//QueryParser queryParser = new QueryParser("contents", new PanGuAnalyzer(true));
//Query query = queryParser.Parse(q);
//QueryParser titleQueryParser = new QueryParser("title", new PanGuAnalyzer(true));
//Query titleQuery = titleQueryParser.Parse(q);
//BooleanQuery bq = new BooleanQuery();
//bq.Add(query, BooleanClause.Occur.SHOULD);
//bq.Add(titleQuery, BooleanClause.Occur.SHOULD);
//Hits hits = search.Search(bq);
//List<TNews> result = new List<TNews>();
//recCount = hits.Length();
//int i = (pageNo - 1) * pageLen;
//while (i < recCount && result.Count < pageLen)
//{
// TNews news = null;
// try
// {
// //// 高亮显示设置
// ////TermQuery tQuery = new TermQuery(new Term("contents", q));
// //SimpleHTMLFormatter simpleHTMLFormatter = new SimpleHTMLFormatter("<font color=\"red\">", "</font>");
// //Highlighter highlighter = new Highlighter(simpleHTMLFormatter, new QueryScorer(query));
// ////关键内容显示大小设置
// //highlighter.SetTextFragmenter(new SimpleFragmenter(50));
// ////取出高亮显示内容
// //Lucene.Net.Analysis.KTDictSeg.KTDictSegAnalyzer analyzer = new Lucene.Net.Analysis.KTDictSeg.KTDictSegAnalyzer();
// //TokenStream tokenStream = analyzer.TokenStream("contents", new StringReader(news.Content));
// //news.Abstract = highlighter.GetBestFragment(tokenStream, news.Content);
// }
// catch (Exception e)
// {
// Console.WriteLine(e.Message);
// }
// finally
// {
// result.Add(news);
// i++;
// }
//}
//search.Close();
//return result;
}