public StringBuilder ExecuteByContent(string content, ExecuteByLineDelegate objCallback, bool bIsSaveContent = false)
{
StringBuilder objContentBuilder = new StringBuilder();
double dFileCharLength = content.Length;
double dLoadCharLength = 0;
DateTime dtUpdateTime = DateTime.Now;
this.pgbScanLines.Maximum = Convert.ToInt32(dFileCharLength);
this.pgbScanLines.Minimum = 0;
this.pgbScanLines.Value = this.pgbScanLines.Minimum;
using (StringReader sr = new StringReader(content))
{
string line = null;
while ((line = sr.ReadLine()) != null)
{
Application.DoEvents();
if (bIsSaveContent)
{
objContentBuilder.AppendLine(line);
}
//if (!String.IsNullOrWhiteSpace(line)) line = Regex.Replace(line, @"(\p{C}+)|(<.*?>)", "");//去除可能的控制符、Html标签
if (String.IsNullOrWhiteSpace(line))
{
continue;
}
dLoadCharLength += line.Length;
this.pgbScanLines.Value = this.pgbScanLines.Maximum > dLoadCharLength?Convert.ToInt32(dLoadCharLength) : this.pgbScanLines.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(text.Length / (double)nmbReadSpeed.Value);
}
if (objCallback != null)
{
objCallback(text, gobjDataConfig, gobjOptionConfig);
}
}
}
}
this.pgbScanLines.Value = this.pgbScanLines.Maximum;
this.pgbScanLines.Value = this.pgbScanLines.Minimum;
return(objContentBuilder);
}
public void ExecuteByDesktop(ExecuteByLineDelegate objCallbackByLine, ExecuteByFileDelegate objCallbackByFile = null)
{
DateTime dtStartTime = DateTime.Now;
Encoding objEncoding = GetEncoding();
switch (this.tabCorpusSource.SelectedTab.Name)
{
case "tabCorpusPageFile":
{
if (radFileMode.Checked)
{
string sPathFile = this.tbFilePath.Text;
if (objCallbackByFile != null)
{
StringBuilder objContentBuilder = ExecuteByFile(sPathFile, objEncoding, objCallbackByLine, true);
objCallbackByFile(sPathFile, objContentBuilder, gobjDataConfig, gobjOptionConfig);
}
else
{
ExecuteByFile(sPathFile, objEncoding, objCallbackByLine);
}
}
if (radFloderMode.Checked)
{
if (Directory.Exists(this.tbFilePath.Text))
{
string[] objFileColl = Directory.GetFiles(this.tbFilePath.Text, "*.*", SearchOption.AllDirectories);
this.pgbScanFiles.Maximum = Convert.ToInt32(objFileColl.Count());
this.pgbScanFiles.Minimum = 0;
this.pgbScanFiles.Value = this.pgbScanFiles.Minimum;
foreach (string sPathFile in objFileColl)
{
if (objCallbackByFile != null)
{
StringBuilder objContentBuilder = ExecuteByFile(sPathFile, objEncoding, objCallbackByLine, true);
objCallbackByFile(sPathFile, objContentBuilder, gobjDataConfig, gobjOptionConfig);
}
else
{
ExecuteByFile(sPathFile, objEncoding, objCallbackByLine);
}
this.pgbScanFiles.Value += 1;
Application.DoEvents();
}
this.pgbScanFiles.Value = this.pgbScanFiles.Maximum;
this.pgbScanFiles.Value = this.pgbScanFiles.Minimum;
}
}
break;
}
case "tabCorpusPageContent":
{
string content = this.tbCorpusContent.Text;
if (objCallbackByFile != null)
{
StringBuilder objContentBuilder = ExecuteByContent(content, objCallbackByLine, true);
objCallbackByFile(String.Empty, objContentBuilder, gobjDataConfig, gobjOptionConfig);
}
else
{
ExecuteByContent(content, objCallbackByLine);
}
break;
}
default:
{
break;
}
}
AppendText(String.Format("完成，共用时{0}秒。", (DateTime.Now - dtStartTime).ToString()));
}
public StringBuilder ExecuteByFile(string sPathFile, Encoding objEncoding, ExecuteByLineDelegate objCallback, bool bIsSaveContent = false)
{
StringBuilder objContentBuilder = new StringBuilder();
if (File.Exists(sPathFile))
{
//this.AppendText(String.Format("【进行】{0}", sPathFile));
FileInfo info = new FileInfo(sPathFile);
double dFileCharLength = info.Length;
double dLoadCharLength = 0;
DateTime dtUpdateTime = DateTime.Now;
this.pgbScanLines.Maximum = Convert.ToInt32(Math.Min(Int32.MaxValue, dFileCharLength));
this.pgbScanLines.Minimum = 0;
this.pgbScanLines.Value = this.pgbScanLines.Minimum;
using (StreamReader sr = new StreamReader(sPathFile, objEncoding))
{
string line = null;
while ((line = sr.ReadLine()) != null)
{
Application.DoEvents();
if (bIsSaveContent)
{
objContentBuilder.AppendLine(line);
}
//if (!String.IsNullOrWhiteSpace(line)) line = Regex.Replace(line,@"(\p{C}+)|(<.*?>)","");//去除可能的控制符、Html标签
//if (!String.IsNullOrWhiteSpace(line)) line = Regex.Replace(line,@"(\p{C}+)","");//去除可能的控制符、Html标签
if (String.IsNullOrWhiteSpace(line))
{
continue;
}
dLoadCharLength += objEncoding.GetByteCount(line);
if (this.pgbScanLines.Maximum <= dLoadCharLength)
{
dFileCharLength = dFileCharLength - dLoadCharLength + 1;
this.pgbScanLines.Maximum = Convert.ToInt32(Math.Min(Int32.MaxValue, dFileCharLength));
this.pgbScanLines.Value = 0;
dLoadCharLength = 0;
}
this.pgbScanLines.Value = this.pgbScanLines.Maximum > dLoadCharLength?Convert.ToInt32(dLoadCharLength) : this.pgbScanLines.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(text.Length / (double)nmbReadSpeed.Value);
}
try
{
if (objCallback != null)
{
objCallback(text, gobjDataConfig, gobjOptionConfig);
}
}
catch (Exception ep)
{
AppendText(ep.Message);
}
}
}
}
this.pgbScanLines.Value = this.pgbScanLines.Maximum;
this.pgbScanLines.Value = this.pgbScanLines.Minimum;
}
return(objContentBuilder);
}
