private void SplitFileByLength(int length)
{
//Encoding encoding = null;
length = length - 100; //100个字的Buffer
//string str = FileOperationHelper.ReadFileContent(txbFilePath.Text, ref encoding, Encoding.UTF8);
Encoding encoding = FileOperationHelper.GetEncodingType(txbFilePath.Text);
string str = FileOperationHelper.ReadFile(txbFilePath.Text, encoding);
int fileIndex = 1;
do
{
if (str.Length == 0)
{
break;
}
string content = str.Substring(0, Math.Min(str.Length, length));
str = str.Substring(content.Length);
int i = Math.Min(str.IndexOf('\r'), str.IndexOf('\n'));
if (i != -1)
{
content += str.Substring(0, i + 2);
str = str.Substring(i + 2);
}
string newFile = GetWriteFilePath(fileIndex++);
FileOperationHelper.WriteFile(newFile, encoding, content);
rtbLogs.AppendText(newFile + "\r\n");
} while (true);
}
private void btnPreview_Click(object sender, EventArgs e)
{
var filePaths = txbFilePath.Text.Split(new char[] { '|' }, StringSplitOptions.RemoveEmptyEntries);
if (filePaths.Length == 0)
{
MessageBox.Show("未选择字幕文件");
return;
}
SubtitleFiles.Clear();
stOperator = SubtitleHelper.GetOperatorByFileName(filePaths[0]);
foreach (var filePath in filePaths)
{
var content = FileOperationHelper.ReadFile(filePath);
var srts = stOperator.Parse(content);
SubtitleFiles.Add(filePath, srts);
}
richTextBox1.Clear();
foreach (var subtitleFile in SubtitleFiles)
{
richTextBox1.AppendText("----" + subtitleFile.Key + "----\r\n");
richTextBox1.AppendText(stOperator.Subtitle2String(subtitleFile.Value));
}
}
public void BatchTest()
{
var txt = FileOperationHelper.ReadFile("erbi.txt");
foreach (var line in txt.Split('\n'))
{
var arr = line.Split(' ');
var code = arr[0];
for (var i = 1; i < arr.Length; i++)
{
var word = arr[i];
var codes = generater.GetCodeOfString(word);
try
{
if (!IsContain(codes, code))
{
Debug.WriteLine("Not Match:" + word + "\t" + code + " mycode:" +
CollectionHelper.ListToString(codes, " "));
}
}
catch (Exception ex)
{
Debug.WriteLine("Error:" + word + ";" + ex.Message);
}
}
}
}
private void btnPickupNewWords_Click(object sender, EventArgs e)
{
if (folderBrowserDialog1.ShowDialog() != DialogResult.OK)
{
return;
}
SentenceParse sentenceParse = new SentenceParse();
DirectoryInfo directoryInfo = new DirectoryInfo(folderBrowserDialog1.SelectedPath);
List <string> sentences = new List <string>();
foreach (FileInfo fileInfo in directoryInfo.GetFiles())
{
string filePath = fileInfo.FullName;
var txt = FileOperationHelper.ReadFile(filePath);
var stOperator = SubtitleHelper.GetOperatorByFileName(filePath);
var srts = stOperator.Parse(txt);
srts = stOperator.RemoveChinese(srts);
sentences.AddRange(srts.Bodies.Values.Select(l => l.EnglishText));
}
Splash.Show();
Splash.Status = "解析字幕中...";
IDictionary <string, VPreviewWord> previewWords = new Dictionary <string, VPreviewWord>();
foreach (var sentence in sentences)
{
var newWords = sentenceParse.Pickup(sentence);
foreach (KeyValuePair <string, string> keyValuePair in newWords)
{
string original = keyValuePair.Key;
string word = keyValuePair.Value;
if (previewWords.ContainsKey(original))
{
previewWords[original].Rank++;
}
else
{
var mean = sentenceParse.RemarkWord(sentence, word, original);
if (mean != null)
{
var wd = new VPreviewWord()
{
Word = mean.Word,
Rank = 1,
Sentence = sentence,
Mean =
mean.DefaultMean == null ? mean.Means[0].ToString() : mean.DefaultMean.ToString()
};
previewWords.Add(original, wd);
}
}
}
}
DisplayPreviewWords(previewWords.Values);
Splash.Close();
}
public void TestGetFileEncoding(string path, string encoding)
{
var e = FileOperationHelper.GetEncodingType(path);
Assert.AreEqual(e.ToString(), Encoding.GetEncoding(encoding).ToString());
var txt = FileOperationHelper.ReadFile(path);
Debug.WriteLine(txt);
}
public void TestGetFileEncoding(string path, string encoding)
{
path = GetFullPath(path);
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
var e = FileOperationHelper.GetEncodingType(path);
Assert.AreEqual(e.EncodingName, Encoding.GetEncoding(encoding).EncodingName);
var txt = FileOperationHelper.ReadFile(path);
Debug.WriteLine(txt);
}
public WordLibraryList Import(string path)
{
var tempUnzipFolder = FileOperationHelper.GetCurrentFolderPath();
FileOperationHelper.UnZip(path, tempUnzipFolder);
var tempFilePath = Path.Combine(FileOperationHelper.GetCurrentFolderPath(), "dictionary.txt");
string str = FileOperationHelper.ReadFile(tempFilePath, new UTF8Encoding(false));
File.Delete(tempFilePath);
return(ImportText(str));
}
public EmojiReplacer(string path)
{
string str = FileOperationHelper.ReadFile(path);
foreach (var line in str.Split(new[] { '\r', '\n' }, StringSplitOptions.RemoveEmptyEntries))
{
var arr = line.Split('\t');
var emoji = arr[0];
var word = arr[1];
mapping[word] = emoji;
}
}
private void ReadAndShowSubtitle()
{
var txt = FileOperationHelper.ReadFile(txbSubtitleFilePath.Text);
stOperator = SubtitleHelper.GetOperatorByFileName(txbSubtitleFilePath.Text);
var srts = stOperator.Parse(txt);
srts = stOperator.RemoveChinese(srts);
srts = stOperator.RemoveFormat(srts);
ShowSubtitleText(srts.Bodies.Values);
subtitle = srts;
}
private void btnParse_Click(object sender, EventArgs e)
{
var txt = FileOperationHelper.ReadFile(txbSubtitleFilePath.Text);
stOperator = SubtitleHelper.GetOperatorByFileName(txbSubtitleFilePath.Text);
var srts = stOperator.Parse(txt);
srts = stOperator.RemoveChinese(srts);
ShowSubtitleText(srts.Bodies);
subtitle = srts;
}
private void SplitFileByLine(int maxLine)
{
Encoding encoding = FileOperationHelper.GetEncodingType(txbFilePath.Text);
string str = FileOperationHelper.ReadFile(txbFilePath.Text, encoding);
string splitLineChar = "\r\n";
if (str.IndexOf(splitLineChar) < 0)
{
if (str.IndexOf('\r') > 0)
{
splitLineChar = "\r";
}
else if (str.IndexOf('\n') > 0)
{
splitLineChar = "\n";
}
else
{
MessageBox.Show("不能找到行分隔符");
return;
}
}
string[] list = str.Split(new[] { splitLineChar }, StringSplitOptions.RemoveEmptyEntries);
var fileContent = new StringBuilder();
int fileIndex = 1;
for (int i = 0; i < list.Length; i++)
{
fileContent.Append(list[i]);
fileContent.Append(splitLineChar);
if ((i + 1) % maxLine == 0 || i == list.Length - 1)
{
if (i != 0)
{
string newFile = GetWriteFilePath(fileIndex++);
FileOperationHelper.WriteFile(newFile, encoding, fileContent.ToString());
rtbLogs.AppendText(newFile + "\r\n");
fileContent = new StringBuilder();
}
}
}
}
/// <summary>
/// 读取外部的字典文件，覆盖系统默认字典
/// </summary>
/// <param name="dictionary"></param>
protected virtual void OverrideDictionary(IDictionary <char, IList <string> > dictionary)
{
var fileContent = FileOperationHelper.ReadFile("mb.txt");
if (fileContent != "")
{
foreach (string line in fileContent.Split(new[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries))
{
string[] arr = line.Split('\t');
if (arr[0].Length == 0)
{
continue;
}
char word = arr[0][0];
string code = arr[1];
var codes = code.Split(' ');
dictionary[word] = new List <string>(codes);//强行覆盖现有字典
}
}
}
private string GetMutiPinyin()
{
string path = ConstantString.PinyinLibPath;
var sb = new StringBuilder();
if (File.Exists(path))
{
string txt = FileOperationHelper.ReadFile(path);
var reg = new Regex(@"^('[a-z]+)+\s[\u4E00-\u9FA5]+$");
string[] lines = txt.Split(new[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < lines.Length; i++)
{
if (reg.IsMatch(lines[i]))
{
sb.Append(lines[i] + "\r\n");
}
}
}
sb.Append(Dictionaries.WordPinyin);
return(sb.ToString());
}
private void btnMergeWL_Click(object sender, EventArgs e)
{
string mainWL = FileOperationHelper.ReadFile(txbMainWLFile.Text);
Dictionary <string, List <string> > mainDict = ConvertTxt2Dictionary(mainWL);
string[] userFiles = txbUserWLFiles.Text.Split('|');
foreach (string userFile in userFiles)
{
string filePath = userFile.Trim();
string userTxt = FileOperationHelper.ReadFile(filePath);
Dictionary <string, List <string> > userDict = ConvertTxt2Dictionary(userTxt);
Merge2Dict(mainDict, userDict);
}
if (cbxSortByCode.Checked)
{
var keys = new List <string>(mainDict.Keys);
keys.Sort();
var sortedDict = new Dictionary <string, List <string> >();
foreach (string key in keys)
{
sortedDict.Add(key, mainDict[key]);
}
mainDict = sortedDict;
}
string result = Dict2String(mainDict);
richTextBox1.Text = result;
if (
MessageBox.Show("是否将合并的" + mainDict.Count + "条词库保存到本地硬盘上？", "是否保存", MessageBoxButtons.YesNo,
MessageBoxIcon.Question) == DialogResult.Yes)
{
if (saveFileDialog1.ShowDialog() == DialogResult.OK)
{
FileOperationHelper.WriteFile(saveFileDialog1.FileName, Encoding.Unicode, result);
}
}
}
public WordLibraryList Import(string path)
{
string str = FileOperationHelper.ReadFile(path, Encoding);
return(ImportText(str));
}
/// <summary>
/// 通过搜狗细胞词库txt内容构造词库对象
/// </summary>
/// <param name="str"></param>
/// <returns></returns>
public virtual WordLibraryList Import(string path)
{
string str = FileOperationHelper.ReadFile(path);
return(ImportText(str));
}
private void LoadData()
{
//try
//{
string path = GetText();
if (path == "")
{
MessageBox.Show("请选择一个搜狗词库的文件");
return;
}
string str = FileOperationHelper.ReadFile(path);
string[] lines = str.Split(new[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
int count = lines.Length;
var wls = new Dictionary <string, string>();
for (int i = 0; i < count; i++)
{
string line = lines[i];
if (line[0] == ';') //说明
{
continue;
}
string[] hzpy = line.Split(' ');
string py = hzpy[0];
string hz = hzpy[1];
if (NeedSave(hz, py))
{
//多音字做如下处理
if (!wls.ContainsKey(hz))
{
wls.Add(hz, py);
}
}
processString = i + "/" + count;
}
ShowTextMessage("开始载入现有的注音库");
string pylibString = FileOperationHelper.ReadFile(ConstantString.PinyinLibPath);
lines = pylibString.Split(new[] { "\r\n" }, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < lines.Length; i++)
{
string line = lines[i];
string[] hzpy = line.Split(' ');
string py = hzpy[0];
string hz = hzpy[1];
if (!wls.ContainsKey(hz))
{
wls.Add(hz, py);
}
processString = i + "/" + count;
}
ShowTextMessage("载入全部完成,开始去除重复");
Dictionary <string, string> rst = RemoveDuplicateWords(wls);
ShowTextMessage("去除重复完成，开始写入文件");
StreamWriter sw = FileOperationHelper.WriteFile(ConstantString.PinyinLibPath, Encoding.Unicode); //清空注音库文件
foreach (string key in rst.Keys)
{
string line = rst[key] + " " + key;
FileOperationHelper.WriteFileLine(sw, line);
}
sw.Close();
toolStripStatusLabel1.Text = "完成!";
ShowTextMessage("完成!");
timer1.Stop();
MessageBox.Show("完成!");
//}
//catch (Exception ex)
//{
// ShowTextErrorMessage(ex.Message);
//}
}