public Statistic()
{
initDict();
output = new StatisticOutput();
output.time = 0;
output.feat = 0;
output.title = "";
output.discount = 0;
output.url = "";
output.description = "";
output.keywordList = new Hashtable();
}
public Statistic()
{
initDict();
output = new StatisticOutput();
output.time = 0;
output.feat = 0;
output.title = "";
output.discount = 0;
output.url = "";
output.description = "";
output.keywordList = new Hashtable();
}
public Statistic(oneUrl item)
{
initDict();
output = new StatisticOutput();
output.time = item.time;
output.feat = item.feat;
output.discount = 100 - Convert.ToInt32(10 * Convert.ToDouble(item.discount));
output.title = item.seller + "-" + item.deal_title;
output.description = item.description;
output.keywordList = new Hashtable();
splitwords(output.title, TITLE_WEIGHT);
splitwords(item.other, OTHER_WEIGHT);
splitwords(item.description, DESCRIPTION_WEIGHT);
}
public Statistic(oneUrl item)
{
initDict();
output = new StatisticOutput();
output.time = item.time;
output.feat = item.feat;
output.discount = 100 - Convert.ToInt32(10 * Convert.ToDouble(item.discount));
output.title = item.seller + "-" + item.deal_title;
output.description = item.description;
output.keywordList = new Hashtable();
splitwords(output.title, TITLE_WEIGHT);
splitwords(item.other, OTHER_WEIGHT);
splitwords(item.description, DESCRIPTION_WEIGHT);
}
public StatisticOutput statistic(oneUrl item)
{//统计item中存储的团购条目信息
Console.WriteLine("Computing the statisitc data of the item...");
StatisticOutput oneOutput = new StatisticOutput();
oneOutput.url = item.url;
oneOutput.time = item.time;
oneOutput.feat = item.feat;
oneOutput.discount = 100 - Convert.ToInt32(10 * Convert.ToDouble(item.discount));
oneOutput.title = item.seller + "-" + item.deal_title;
oneOutput.description = item.description;
oneOutput.keywordList = new Hashtable();
splitwords(ref oneOutput, oneOutput.title, TITLE_WEIGHT);
splitwords(ref oneOutput, item.other, OTHER_WEIGHT);
splitwords(ref oneOutput, item.description, DESCRIPTION_WEIGHT);
Console.WriteLine("Statistic ready.");
return(oneOutput);
}
private static void splitwords(ref StatisticOutput temOutput, string sentence, int weight)
{//分词函数，作用同上，但分出的词加入tempOutput的关键词列表中
List <T_WordInfo> words = m_SimpleDictSeg.SegmentToWordInfos(sentence);
for (int i = 0; i < words.Count; i++)
{
//判断文本是否进入
if (words[i] == null)
{
continue;
}
if (temOutput.keywordList.ContainsKey(words[i].Word))
{
temOutput.keywordList[words[i].Word] = (int)temOutput.keywordList[words[i].Word] + weight;
}
else
{
temOutput.keywordList.Add(words[i].Word, weight);
}
}
}
private bool addOneUrl(int pos)
{//添加第pos条团购条目信息
string query;
int link_id;
MD5 md5 = new MD5CryptoServiceProvider();;
//XmlNode oneUrl = lashouDocument.DocumentElement.ChildNodes[pos];
oneUrl oneUrl = xmlparse.getOneUrl(pos);
StatisticOutput sta = staContext.statistic(oneUrl);
Console.WriteLine("Insert the data to the database....");
query = "insert into links (site_id, url, title, description) values(5, \"" +
sta.url + "\",\"" + sta.title + "\",\"" + sta.description + "\")";
cmd = new MySQLDriverCS.MySQLCommand(query, conn);
cmd.ExecuteNonQuery();
cmd.Dispose();
query = "select link_id from links where url = \"" + sta.url + "\"";
cmd = new MySQLDriverCS.MySQLCommand(query, conn);
IDataReader dt = cmd.ExecuteReader();
link_id = Convert.ToInt32(dt.GetSchemaTable().Rows[0]["link_id"]);
foreach (DictionaryEntry de in sta.keywordList)
{
int keyword_id = 0;
query = "select keyword_id from keywords where keyword = \"" + de.Key + "\"";
cmd = new MySQLDriverCS.MySQLCommand(query, conn);
dt = cmd.ExecuteReader();
if (dt.GetSchemaTable().Rows.Count == 0)
{
query = "insert into keywords (keyword) values (\"" + de.Key + "\")";
cmd = new MySQLDriverCS.MySQLCommand(query, conn);
cmd.ExecuteNonQuery();
cmd.Dispose();
query = "select keyword_id from keywords where keyword = \"" + de.Key + "\"";
cmd = new MySQLDriverCS.MySQLCommand(query, conn);
dt = cmd.ExecuteReader();
keyword_id = Convert.ToInt32(dt.GetSchemaTable().Rows[0]["keyword_id"]);
}
else
{
keyword_id = Convert.ToInt32(dt.GetSchemaTable().Rows[0]["keyword_id"]);
}
byte[] hash = md5.ComputeHash(System.Text.Encoding.Default.GetBytes(de.Key.ToString()));
string flag;
switch (hash[0] >> 4)
{
case 10:
flag = "a";
break;
case 11:
flag = "b";
break;
case 12:
flag = "c";
break;
case 13:
flag = "d";
break;
case 14:
flag = "e";
break;
case 15:
flag = "f";
break;
default:
flag = (hash[0] >> 4).ToString();
break;
}
query = "insert into link_keyword" + flag + "(link_id, keyword_id, weight, time, feat, discount) values (" +
link_id + "," + keyword_id + "," + de.Value + "," + sta.time + "," + sta.feat + "," +
sta.discount + ")";
cmd = new MySQLDriverCS.MySQLCommand(query, conn);
cmd.ExecuteNonQuery();
cmd.Dispose();
}
Console.WriteLine("Write one link sucessfully!");
return(true);
}
private static void splitwords(ref StatisticOutput temOutput, string sentence, int weight)
{
//分词函数，作用同上，但分出的词加入tempOutput的关键词列表中
List<T_WordInfo> words = m_SimpleDictSeg.SegmentToWordInfos(sentence);
for (int i = 0; i < words.Count; i++)
{
//判断文本是否进入
if (words[i] == null)
continue;
if (temOutput.keywordList.ContainsKey(words[i].Word))
{
temOutput.keywordList[words[i].Word] = (int)temOutput.keywordList[words[i].Word] + weight;
}
else
{
temOutput.keywordList.Add(words[i].Word, weight);
}
}
}
public StatisticOutput statistic(oneUrl item)
{
//统计item中存储的团购条目信息
Console.WriteLine("Computing the statisitc data of the item...");
StatisticOutput oneOutput = new StatisticOutput();
oneOutput.url = item.url;
oneOutput.time = item.time;
oneOutput.feat = item.feat;
oneOutput.discount = 100 - Convert.ToInt32(10 * Convert.ToDouble(item.discount));
oneOutput.title = item.seller + "-" + item.deal_title;
oneOutput.description = item.description;
oneOutput.keywordList = new Hashtable();
splitwords(ref oneOutput, oneOutput.title, TITLE_WEIGHT);
splitwords(ref oneOutput, item.other, OTHER_WEIGHT);
splitwords(ref oneOutput, item.description, DESCRIPTION_WEIGHT);
Console.WriteLine("Statistic ready.");
return oneOutput;
}