/*
* 将文件转换为文本
*/
private void FileConvert(object thread_payload)
{
//获取参数
string[] param = (string[])thread_payload;
int taskNo = int.Parse(param[0].ToString());
DirectoryInfo sourceFolder = new DirectoryInfo(param[1]);
DirectoryInfo textFolder = new DirectoryInfo(param[2]);
//判断源文件夹和目标文件夹是否存在
if (!sourceFolder.Exists || !textFolder.Exists)
{
runningEnv.UIContext.BeginInvoke(new Action(() =>
{
MessageBox.Show(runningEnv.UIContext, "数据损坏，任务失败！", "错误");
}));
return;
}
//获取源文件夹中的待转换文件
FileInfo[] fileInfo = sourceFolder.GetFiles();
//按本线程所分配的任务进行转换
int addNumo = int.Parse(runningEnv.SettingData.ConvertThreadCnt.ToString());
for (int FileInfoNo = taskNo; FileInfoNo < fileInfo.Length; FileInfoNo += addNumo)
{
FileInfo NextFile = fileInfo[FileInfoNo];
string path = sourceFolder.FullName + "\\" + NextFile.Name;
string dist_path = textFolder.FullName + "\\" + NextFile.Name + ".txt";
//文件已经被转换则忽略该文件
if (File.Exists(dist_path))
{
continue;
}
ConverterFactory converterFactory = new ConverterFactory();
try
{
//获取文件后缀
string file_type = path.ToLower().Split('.')[Math.Max(path.ToLower().Split('.').Length - 1, 0)];
//获取转换器
ConvertCore file_converter = converterFactory.GetConverter(file_type, runningEnv);
string text = "";
if (file_converter != null)//如果可以获得到转换器
{
//获取文本
text = file_converter.ConvertToString(path);
if (text.Length > RunningEnv.ProgramParam.MaxWords)
{
//舍弃过长的部分
text = text.Substring(0, RunningEnv.ProgramParam.MaxWords);
}
//写入目标路径
File.WriteAllText(dist_path, text, Encoding.GetEncoding("GBK"));
}
else
{
//没有获取到转换器则忽略
continue;
}
//如果转换后的文件不符合所设阈值则删除
if (new FileInfo(dist_path).Length <= runningEnv.CheckData.MinBytes || text.Length <= runningEnv.CheckData.MinWords)
{
if (File.Exists(dist_path))
{
File.Delete(dist_path);
}
throw new Exception();//并爆出一个异常
}
}
catch (Exception e)
{
runningEnv.CheckingData.ErrorPaperList.AddLast(NextFile.Name);
}
}
}
/*
* 将文件转换为文本
*/
private void FileConvert(object thread_payload)
{
//获取参数
string[] param = (string[])thread_payload;
int taskNo = int.Parse(param[0].ToString());
DirectoryInfo sourceFolder = new DirectoryInfo(param[1]);
DirectoryInfo textFolder = new DirectoryInfo(param[2]);
//判断源文件夹和目标文件夹是否存在
if (!sourceFolder.Exists || !textFolder.Exists)
{
runningEnv.UIContext.BeginInvoke(new Action(() =>
{
MessageBox.Show(runningEnv.UIContext, "数据损坏，任务失败！", "错误");
}));
return;
}
//获取源文件夹中的待转换文件
FileInfo[] fileInfo = Utils.GetFileInfoRecursion(sourceFolder).ToArray();
//按本线程所分配的任务进行转换
int addNumo = int.Parse(runningEnv.SettingData.ConvertThreadCnt.ToString());
for (int FileInfoNo = taskNo; FileInfoNo < fileInfo.Length; FileInfoNo += addNumo)
{
FileInfo NextFile = fileInfo[FileInfoNo];
string path = NextFile.FullName;
string real_dis_file_name = Regex.Replace(NextFile.Name, @"[^\u4e00-\u9fa5\u0022\《\》\（\）\—\；\，\。\“\”\！\#\\_\-\.\,\:\(\)\'\[\]\【\】\+\·\：\<\>\w]", string.Empty);
foreach (char rInvalidChar in Path.GetInvalidFileNameChars())
{
real_dis_file_name = real_dis_file_name.Replace(rInvalidChar.ToString(), string.Empty);
}
string dist_path = textFolder.FullName + Path.DirectorySeparatorChar + real_dis_file_name + ".txt";
//文件已经被转换则忽略该文件
if (File.Exists(dist_path))
{
continue;
}
ConverterFactory converterFactory = new ConverterFactory();
try
{
//获取文件后缀
string file_type = path.ToLower().Split('.')[Math.Max(path.ToLower().Split('.').Length - 1, 0)];
//获取转换器
ConvertCore file_converter = converterFactory.GetConverter(file_type, runningEnv);
string text = "";
if (file_converter != null)//如果可以获得到转换器
{
//获取文本
text = file_converter.ConvertToString(path, runningEnv.CheckData.Blocklist);
if (text != null && text.Length > 0)
{
if (text.Length > RunningEnv.ProgramParam.MaxWords)
{
//舍弃过长的部分
text = text.Substring(0, RunningEnv.ProgramParam.MaxWords);
}
//写入目标路径
File.WriteAllText(dist_path, text, Encoding.GetEncoding("GBK"));
}
}
else
{
//没有获取到转换器则忽略
continue;
}
//如果转换后的文件不符合所设阈值则删除
if (new FileInfo(dist_path).Length <= runningEnv.CheckData.MinBytes || text.Length <= runningEnv.CheckData.MinWords)
{
if (File.Exists(dist_path))
{
File.Delete(dist_path);
}
//并抛出一个异常
throw new Exception();
}
}
catch (Exception e)
{
runningEnv.CheckingData.ErrorPaperList.AddLast(NextFile.Name);
}
}
}
