private async void TextBox_data_MouseDoubleClick(object sender, MouseButtonEventArgs e)
{
var fileName = MyLib.Open();
if (fileName == null)
{
return;
}
var encodingSelected = MessageBox.Show("\"Yes\" to use UTF-8\r\n\"No\" to use UTF-16 (Unicode)", "", MessageBoxButton.YesNo);
if (encodingSelected == MessageBoxResult.None)
{
return;
}
Log.Assert(encodingSelected == MessageBoxResult.Yes || encodingSelected == MessageBoxResult.No);
var encoding = encodingSelected == MessageBoxResult.Yes ? Encoding.UTF8 : Encoding.Unicode;
int dataPreprocessMethodId = radioPanel_newData.SelectedIndex;
bool debugMode = (bool)checkBox_debugMode.IsChecked;
await stackPanel_tasksQueue.EnqueueTaskAsync($"Read Data {fileName} {encoding} {debugMode} {dataPreprocessMethodId}", new Func <Task>(async() =>
{
mainData = await Task.Run(() => TextProcess.ReadTextStream(new FileStream(fileName, FileMode.Open, FileAccess.Read), encoding, debugMode));
await Task.Run(() => TextProcess.Process(ref mainData, dataPreprocessMethodId));
Log.Write(" OK");
}));
}
private async void Button_new_Click(object sender, RoutedEventArgs e)
{
var fileName = MyLib.Open();
if (fileName == null)
{
return;
}
var encodingSelected = MessageBox.Show("\"Yes\" to use UTF-8\r\n\"No\" to use UTF-16 (Unicode)", "", MessageBoxButton.YesNo);
if (encodingSelected == MessageBoxResult.None)
{
return;
}
Log.Assert(encodingSelected == MessageBoxResult.Yes || encodingSelected == MessageBoxResult.No);
var encoding = encodingSelected == MessageBoxResult.Yes ? Encoding.UTF8 : Encoding.Unicode;
int maxWordLength = int.Parse(inputField_data["maxWordLength"].Text);
bool debugMode = (bool)checkBox_debugMode.IsChecked;
int processMethod = radioPanel_newData.SelectedIndex;
await stackPanel_tasksQueue.EnqueueTaskAsync($"BuildDataAsync({fileName},{encoding},{processMethod},{maxWordLength},{debugMode})", new Func <Task>(async() => await Task.Run(() => BuildData(
new FileStream(fileName, FileMode.Open, FileAccess.Read),
encoding,
processMethod,
maxWordLength,
debugMode))));
}
public void AddField(string title, string content)
{
MyLib.Assert(!textBoxes.ContainsKey(title));
var textBox = new TextBox {
Text = content
};
textBoxes.Add(title, textBox);
this.Children.Add(new Grid
{
ColumnDefinitions =
{
new ColumnDefinition {
Width = new GridLength(1, GridUnitType.Star)
},
new ColumnDefinition {
Width = new GridLength(1, GridUnitType.Star)
}
},
Children =
{
new Label {
Content = title
}.Set(0, 0),
textBox.Set(0, 1)
}
});
}
public static void Process(ref string data, int methodId)
{
var sb = new StringBuilder();
switch (methodId)
{
case 0: return;
case 1: //non-chinese => space
{
int percentage = -1;
long progress = 0, total_progress = data.Length;
char last_char = '\0';
foreach (char c in data)
{
if (++progress * 100 / total_progress > percentage)
{
Log.WriteLine($"Preprocessing {++percentage}%: non-chinese => space");
}
if (!MyLib.IsChinese(c))
{
if (last_char != ' ')
{
sb.Append(last_char = ' ');
}
}
else
{
sb.Append(last_char = c);
}
}
data = sb.ToString();
return;
}
case 2: //non-chinese => removed
{
int percentage = -1;
long progress = 0, total_progress = data.Length;
char last_char = '\0';
foreach (char c in data)
{
if (++progress * 100 / total_progress > percentage)
{
Log.WriteLine($"Preprocessing {++percentage}%: non-chinese => space");
}
if (MyLib.IsChinese(c))
{
sb.Append(last_char = c);
}
}
data = sb.ToString();
return;
}
default: throw new Exception($"Unknown methodId: {methodId}");
}
}
void InitializeButtons(List <Tuple <string, Action> > bs)
{
bs.Add(new Tuple <string, Action>("Convert Encoding", async() =>
{
Encoding encodingSource, encodingTarget;
var encodingList = GetProperties(typeof(Encoding)).Select(v => v as Encoding).ToList();
if ((encodingSource = await Select(encodingList, $"Source Encoding ({encodingList.Count})")) == null ||
(encodingTarget = await Select(encodingList, $"Target Encoding ({encodingList.Count})")) == null)
{
return;
}
await MyLib.OpenSave(async(fs, ss) =>
{
try
{
Log.Indent();
Log.WriteLine($"Converting: {encodingSource} → {encodingTarget}");
using (var reader = new StreamReader(fs, encodingSource))
{
using (var writer = new StreamWriter(ss, false, encodingTarget))
{
const int bufLen = 1 << 20;
var buf = new char[bufLen];
long progress = 0, total_progress = reader.BaseStream.Length;
DateTime time = DateTime.Now;
for (int n; (n = await reader.ReadAsync(buf, 0, buf.Length)) > 0;)
{
await writer.WriteAsync(buf, 0, n);
progress += n;
if ((DateTime.Now - time).TotalSeconds > 0.5)
{
Log.WriteLine($"Converting: {(double)progress * 100 / total_progress}% {encodingSource} → {encodingTarget}");
time = DateTime.Now;
}
}
}
}
Log.WriteLine($"Convert OK: {encodingSource} → {encodingTarget}");
}
finally { Log.Unindent(); }
});
}));
bs.Add(new Tuple <string, Action>("Convert Trie to WordList (file to file)", async() =>
{
await MyLib.OpenSave((fs, ss) => Log.SubTask(async() =>
{
Log.WriteLine($"Reading... file size: {fs.Length}");
var trie = new Trie();
await Task.Run(() => trie.Load(new FileStream(fs, FileMode.Open, FileAccess.Read)));
Log.WriteLine("Exporting...");
await trie.ExportList(new FileStream(ss, FileMode.Create, FileAccess.Write));
Log.WriteLine($"Done");
}));
}));
}
private async void Button_exportList_Click(object sender, RoutedEventArgs e)
{
var fileName = MyLib.Save();
if (fileName == null)
{
return;
}
await stackPanel_tasksQueue.EnqueueTaskAsync($"Export Word List {fileName}", new Func <Task>(async() =>
{
Log.WriteLine("Exporting Word List...");
await trie.ExportList(new FileStream(fileName, FileMode.Create, FileAccess.Write));
Log.Write(" OK");
}));
}
private async void Button_load_Click(object sender, RoutedEventArgs e)
{
var fileName = MyLib.Open();
if (fileName == null)
{
return;
}
await stackPanel_tasksQueue.EnqueueTaskAsync($"Read Trie {fileName}", new Func <Task>(async() =>
{
Log.WriteLine("Loading trie...");
await Task.Run(() => trie.Load(new FileStream(fileName, FileMode.Open, FileAccess.Read)));
Log.Write(" OK");
}));
}
private async void Button_save_Click(object sender, RoutedEventArgs e)
{
var fileName = MyLib.Save();
if (fileName == null)
{
return;
}
await stackPanel_tasksQueue.EnqueueTaskAsync($"Save Trie {fileName}", new Func <Task>(async() =>
{
Log.WriteLine("Saving trie...");
await Task.Run(() => trie.Save(new FileStream(fileName, FileMode.Create, FileAccess.Write)));
Log.Write(" OK");
}));
}
private async void Button_wordsPerCount_Click(object sender, RoutedEventArgs e)
{
int targetWordLength = int.Parse(Microsoft.VisualBasic.Interaction.InputBox("targetWordLength?", "", "4"));
var fileName = MyLib.Save();
if (fileName == null)
{
return;
}
await stackPanel_tasksQueue.EnqueueTaskAsync($"Words per Count {fileName} {targetWordLength}", async() => await Task.Run(() =>
{
long percent = -1, progress = 0, total_progress = trie.Size;
int wordLength = 0;
SortedDictionary <long, long> ans = new SortedDictionary <long, long>();
for (int i = 1; i <= 10; i++)
{
ans[i] = 0;
}
trie.Traverse(c => wordLength++, () => wordLength--, cnt =>
{
if (++progress * 100 / total_progress > percent)
{
Log.WriteLine($"Words per Count... {++percent}% {ans[1]} {ans[2]} {ans[3]} {ans[4]} {ans[5]} | {ans[6]} {ans[7]} {ans[8]} {ans[9]} {ans[10]}");
}
if (wordLength != targetWordLength)
{
return;
}
if (!ans.ContainsKey(cnt))
{
ans[cnt] = 0;
}
ans[cnt]++;
});
Log.WriteLine($"Writing...");
using (StreamWriter writer = new StreamWriter(fileName, false, Encoding.UTF8))
{
writer.WriteLine("Count,Words");
foreach (var p in ans)
{
writer.WriteLine($"{p.Key},{p.Value}");
}
}
Log.Write(" OK");
}));
}
