doclib.cs

实现文本的特征提取

using System;
using System.Data;
using System.Data.SqlClient;

namespace Njnu.DAL
{
	/// <summary>
	/// DocLib 的摘要说明。
	/// </summary>
	public class DocLib
	{
		/// <summary>
		/// 添加Doc文档，文档编号时用
		/// </summary>
		/// <param name="doc">文档对象</param>
		/// <param name="tableName">待插入的表名</param>
		/// <returns></returns>
		public bool AddDoc(Common.Doc doc, string tableName)
		{
			SqlParameter[] parameters = {};
			int effectCount = 0;
			string str = "insert " + tableName + "(docID, cateID, docPath) values('" + doc.DocID + "','" + doc.CateID + "','" + doc.Path + "')";
			using(SqlDatabase sqlDB = new SqlDatabase(str, parameters))
			{
				effectCount = (int)sqlDB.Run();
			}
			return ( ( effectCount==1?true:false) );
		}

		//设置文档已处理
		public bool SetDone(Common.Doc doc, string tableName)
		{
			SqlParameter[] parameters = {};
			int effectCount = 0;
			string str = "update " + tableName + " set IsDone = 1 where docID = '" + doc.DocID +"'";
			using(SqlDatabase sqlDB = new SqlDatabase(str, parameters))
			{
				effectCount = (int)sqlDB.Run();
			}
			return ( ( effectCount==1?true:false) );
		}
	
		//获取未处理过的文档
		public DataTable GetAllDoc(string tableName)
		{
			DataTable dt = new DataTable();
			SqlParameter[] parameters = {};
			
			string str = "select * from " + tableName + " where isDone = 0";
			using(SqlDatabase sqlDB = new SqlDatabase(str, parameters))
			{
				sqlDB.Run(dt);
			}
			return dt;
		}

		//获取所有测试文档
		public DataTable GetAllTestDoc()
		{
			DataTable dt = new DataTable();
			SqlParameter[] parameters = {};
			
			string str = "select * from DocInforLib_test";
			using(SqlDatabase sqlDB = new SqlDatabase(str, parameters))
			{
				sqlDB.Run(dt);
			}
			return dt;
		}

		//获取文档总数，在tf*idf公式中用到
		public int GetDocCount()
		{
			DataTable dt = new DataTable();
			SqlParameter[] parameters = {};
			
			string str = "select count(docID) from DocInforLib";
			using(SqlDatabase sqlDB = new SqlDatabase(str, parameters))
			{
				sqlDB.Run(dt);
			}
			return Convert.ToInt32( dt.Rows[0][0] );
		}

		//获取某一个类别的文档总数，在合并特征项时用到
		public int GetDocCountByCateID(string cateID)
		{
			DataTable dt = new DataTable();
			SqlParameter[] parameters = {};
			
			string str = "select count(docID) from DocInforLib where cateID = '" + cateID + "'";
			using(SqlDatabase sqlDB = new SqlDatabase(str, parameters))
			{
				sqlDB.Run(dt);
			}
			return Convert.ToInt32( dt.Rows[0][0] );
		}

		//获取测试文档的文件名
		public string GetDocName(string docID)
		{
			DataTable dt = new DataTable();
			SqlParameter[] parameters = {};
			
			string str = "select docPath from DocInforLib_test where docID = '" + docID + "'";
			using(SqlDatabase sqlDB = new SqlDatabase(str, parameters))
			{
				sqlDB.Run(dt);
			}
			int i = dt.Rows[0][0].ToString().LastIndexOf("\\");
			int a =dt.Rows[0][0].ToString().Length;
			string docPath = dt.Rows[0][0].ToString();
			string docName = docPath.Substring(i, a-i); 
			return docName;
		}
	}
}