private void btnSaveDictionary_Click(object sender, EventArgs e)
{
string floder = Path.GetFullPath(String.Format(@"{0}\dict", CachePathDAL.GetWorkSpacePath()));
if (!Directory.Exists(floder))
{
Directory.CreateDirectory(floder);
}
string filename1 = Path.GetFullPath(String.Format(@"{0}\dict\{1}", CachePathDAL.GetWorkSpacePath(), "CharBond.coll"));
AppendText(String.Format("请稍候，正在保存文件：{0}", filename1));
DictionaryDAL.ClearMemoryBondColl <string>(objCharBondColl);
SerialLib.SerializeBinary <MemoryBondColl <string> >(objCharBondColl, filename1);
string filename2 = Path.GetFullPath(String.Format(@"{0}\dict\{1}", CachePathDAL.GetWorkSpacePath(), "KeyWord.coll"));
AppendText(String.Format("请稍候，正在保存文件：{0}", filename2));
DictionaryDAL.ClearMemoryItemColl <string>(objKeyWordColl);
SerialLib.SerializeBinary <MemoryItemColl <string> >(objKeyWordColl, filename2);
AppendText("字典集保存完毕！");
}
private void CatchWordIndexColl()
{
objCharBondColl.MinuteOffsetSize = Convert.ToInt32(nmbReadSpeed.Value) * 60 * 60 * 24 * 6;
objKeyWordColl.MinuteOffsetSize = Convert.ToInt32(nmbReadSpeed.Value) * 60 * 60 * 24 * 6;
Encoding objEncoding = GetEncoding();
if (!String.IsNullOrWhiteSpace(this.tbCoreWordList.Text))
{
if (radSegment.Checked)
{
AppendText(SegmentBLL.ShowSegment(SegmentBLL.Segment(this.tbCoreWordList.Text, objCharBondColl, objKeyWordColl, 4, true, true)));
return;
}
}
DateTime dtStartTime = DateTime.Now;
if (radFileMode.Checked)
{
string sPathFile = this.tbFilePath.Text;
if (File.Exists(sPathFile))
{
this.AppendText(String.Format("【进行】{0}", sPathFile));
FileInfo info = new FileInfo(sPathFile);
double dFileCharLength = info.Length / 2 + 1;
double dLoadCharCount = 0;
double dTempCharCount = objCharBondColl.OffsetTotalCount;
DateTime dtUpdateTime = DateTime.Now;
this.progressBar1.Maximum = Convert.ToInt32(dFileCharLength);
this.progressBar1.Minimum = 0;
this.progressBar1.Value = this.progressBar1.Minimum;
using (StreamReader sr = new StreamReader(sPathFile, objEncoding))
{
string line = null;
while ((line = sr.ReadLine()) != null)
{
Application.DoEvents();
if (!String.IsNullOrWhiteSpace(line))
{
line = Regex.Replace(Regex.Replace(line, @"\p{C}+", ""), "<.*?>", "", RegexOptions.IgnoreCase | RegexOptions.Singleline); //去除可能的控制符、Html标签
}
if (String.IsNullOrWhiteSpace(line))
{
continue;
}
dLoadCharCount += line.Length;
dTempCharCount += line.Length;
this.progressBar1.Value = this.progressBar1.Maximum > dLoadCharCount?Convert.ToInt32(dLoadCharCount) : this.progressBar1.Maximum - 1;
string text = line; //这里可以再做一些需要特别处理的数据清洗，如多余的空格等
if (!String.IsNullOrEmpty(text))
{
if (this.chkDelayTime.Checked)
{
dtUpdateTime = dtPickerUpdateTime.Value.AddSeconds(dLoadCharCount / (double)nmbReadSpeed.Value);
}
if (radDictionary.Checked)
{
//当数据跑过一个周期的数据时清理一次邻键集、词库，避免内存空间不足
if (dTempCharCount > objCharBondColl.MinuteOffsetSize)
{
DictionaryDAL.ClearMemoryBondColl <string>(objCharBondColl);
DictionaryDAL.ClearMemoryItemColl <string>(objKeyWordColl);
dTempCharCount = 0;
}
WordDictBLL.UpdateKeyWordColl(text, objCharBondColl, objKeyWordColl);
}
if (radSegment.Checked)
{
AppendText(SegmentBLL.ShowSegment(SegmentBLL.Segment(text, objCharBondColl, objKeyWordColl, 4, true, true)));
}
}
}
}
this.progressBar1.Value = this.progressBar1.Maximum;
this.progressBar1.Value = this.progressBar1.Minimum;
if (this.chkDelayTime.Checked)
{
dtPickerUpdateTime.Value = dtPickerUpdateTime.Value.AddSeconds(dLoadCharCount / (double)nmbReadSpeed.Value);
}
}
}
if (radFloderMode.Checked)
{
if (Directory.Exists(this.tbFilePath.Text))
{
double dTempCharCount = objCharBondColl.OffsetTotalCount;
string[] objFileColl = Directory.GetFiles(this.tbFilePath.Text, "*.txt", SearchOption.AllDirectories);
foreach (string sPathFile in objFileColl)
{
if (File.Exists(sPathFile))
{
double dLoadCharCount = 0;
this.AppendText(String.Format("【进行】{0}", sPathFile));
FileInfo info = new FileInfo(sPathFile);
double dFileCharLength = info.Length / 2 + 1;
DateTime dtUpdateTime = DateTime.Now;
this.progressBar1.Maximum = Convert.ToInt32(dFileCharLength);
this.progressBar1.Minimum = 0;
this.progressBar1.Value = this.progressBar1.Minimum;
using (StreamReader sr = new StreamReader(sPathFile, objEncoding))
{
string line = null;
while ((line = sr.ReadLine()) != null)
{
Application.DoEvents();
if (!String.IsNullOrWhiteSpace(line))
{
line = Regex.Replace(Regex.Replace(line, @"\p{C}+", ""), "<.*?>", "", RegexOptions.IgnoreCase | RegexOptions.Singleline); //去除可能的控制符、Html标签
}
if (String.IsNullOrWhiteSpace(line))
{
continue;
}
dLoadCharCount += line.Length;
dTempCharCount += line.Length;
this.progressBar1.Value = this.progressBar1.Maximum > dLoadCharCount?Convert.ToInt32(dLoadCharCount) : this.progressBar1.Maximum - 1;
string text = line; //这里可以再做一些需要特别处理的数据清洗，如多余的空格等
if (!String.IsNullOrEmpty(text))
{
if (this.chkDelayTime.Checked)
{
dtUpdateTime = dtPickerUpdateTime.Value.AddSeconds(dLoadCharCount / (double)nmbReadSpeed.Value);
}
if (radDictionary.Checked)
{
//当数据跑过一个周期的数据时清理一次邻键集、词库，避免内存空间不足
if (dTempCharCount > objCharBondColl.MinuteOffsetSize)
{
DictionaryDAL.ClearMemoryBondColl <string>(objCharBondColl);
DictionaryDAL.ClearMemoryItemColl <string>(objKeyWordColl);
dTempCharCount = 0;
}
WordDictBLL.UpdateKeyWordColl(text, objCharBondColl, objKeyWordColl);
}
if (radSegment.Checked)
{
AppendText(SegmentBLL.ShowSegment(SegmentBLL.Segment(text, objCharBondColl, objKeyWordColl, 4, true, true)));
}
}
}
}
this.progressBar1.Value = this.progressBar1.Maximum;
this.progressBar1.Value = this.progressBar1.Minimum;
if (this.chkDelayTime.Checked)
{
dtPickerUpdateTime.Value = dtPickerUpdateTime.Value.AddSeconds(dLoadCharCount / (double)nmbReadSpeed.Value);
}
}
}
}
}
AppendText(String.Format("完成，共用时{0}秒。", (DateTime.Now - dtStartTime).ToString()));
}