internal void FixCasing(Subtitle subtitle, string language)
{
var namesList = new NamesList(Configuration.DictionariesFolder, language, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
var namesEtc = namesList.GetAllNames();
// Longer names must be first
namesEtc.Sort((s1, s2) => s2.Length.CompareTo(s1.Length));
string lastLine = string.Empty;
foreach (Paragraph p in subtitle.Paragraphs)
{
p.Text = FixCasing(p.Text, lastLine, namesEtc);
// fix casing of English alone i to I
if (radioButtonNormal.Checked && language.StartsWith("en", StringComparison.Ordinal))
{
p.Text = FixEnglishAloneILowerToUpper(p.Text);
}
lastLine = p.Text;
}
}
private void MakeSureNamesListIsLoaded()
{
if (_namesEtcList == null)
{
_namesEtcList = new List<string>();
string languageTwoLetterCode = Utilities.AutoDetectGoogleLanguage(Subtitle);
// Will contains both one word names and multi names
var namesList = new NamesList(Configuration.DictionariesFolder, languageTwoLetterCode, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
_namesEtcList = namesList.GetAllNames();
}
}
private void LoadNamesEtc(string language, bool reloadListBox)
{
var task = Task.Factory.StartNew(() =>
{
// names etc
var namesList = new NamesList(Configuration.DictionariesFolder, language, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
_wordListNamesEtc = namesList.GetAllNames();
_wordListNamesEtc.Sort();
return _wordListNamesEtc;
});
if (reloadListBox)
{
// reload the listbox on a continuation ui thead
var uiContext = TaskScheduler.FromCurrentSynchronizationContext();
task.ContinueWith(originalTask =>
{
listBoxNamesEtc.BeginUpdate();
listBoxNamesEtc.Items.Clear();
foreach (var name in originalTask.Result)
{
listBoxNamesEtc.Items.Add(name);
}
listBoxNamesEtc.EndUpdate();
}, uiContext);
}
}
private void FindAllNames()
{
string language = LanguageAutoDetect.AutoDetectLanguageName("en_US", _subtitle);
if (string.IsNullOrEmpty(language))
language = "en_US";
var namesList = new NamesList(Configuration.DictionariesFolder, language, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
// Will contains both one word names and multi names
var namesEtcList = namesList.GetAllNames();
if (language.StartsWith("en", StringComparison.Ordinal))
{
namesEtcList.Remove("Black");
namesEtcList.Remove("Bill");
namesEtcList.Remove("Long");
namesEtcList.Remove("Don");
}
var sb = new StringBuilder();
foreach (Paragraph p in _subtitle.Paragraphs)
sb.AppendLine(p.Text);
string text = HtmlUtil.RemoveHtmlTags(sb.ToString());
string textToLower = text.ToLower();
listViewNames.BeginUpdate();
foreach (string name in namesEtcList)
{
int startIndex = textToLower.IndexOf(name.ToLower(), StringComparison.Ordinal);
if (startIndex >= 0)
{
while (startIndex >= 0 && startIndex < text.Length &&
textToLower.Substring(startIndex).Contains(name.ToLower()) && name.Length > 1 && name != name.ToLower())
{
bool startOk = (startIndex == 0) || (text[startIndex - 1] == ' ') || (text[startIndex - 1] == '-') ||
(text[startIndex - 1] == '"') || (text[startIndex - 1] == '\'') || (text[startIndex - 1] == '>') ||
(Environment.NewLine.EndsWith(text[startIndex - 1].ToString(CultureInfo.InvariantCulture)));
if (startOk)
{
int end = startIndex + name.Length;
bool endOk = end <= text.Length;
if (endOk)
endOk = end == text.Length || ExpectedEndChars.Contains(text[end]);
if (endOk && text.Substring(startIndex, name.Length) != name) // do not add names where casing already is correct
{
if (!_usedNames.Contains(name))
{
_usedNames.Add(name);
AddToListViewNames(name);
break; // break while
}
}
}
startIndex = textToLower.IndexOf(name.ToLower(), startIndex + 2, StringComparison.Ordinal);
}
}
}
listViewNames.EndUpdate();
groupBoxNames.Text = string.Format(Configuration.Settings.Language.ChangeCasingNames.NamesFoundInSubtitleX, listViewNames.Items.Count);
}