public void NamesListNotInList()
{
// Arrange
var namesList = new NamesList(Directory.GetCurrentDirectory(), "en", false, null);
// Act
var exists = namesList.GetNames().Contains("JonesASDFLKJCKJXFLKJSLDKFJASDF");
// Assert
Assert.IsFalse(exists);
}
public void NamesListRemove()
{
// Arrange
var namesList = new NamesList(Directory.GetCurrentDirectory(), "en", false, null);
namesList.Add("Jones123");
// Act
namesList.Remove("Jones123");
// Assert
Assert.IsFalse(namesList.GetNames().Contains("Jones123"));
}
public void NamesListIsInNamesEtcMultiWordList()
{
// Arrange
var namesList = new NamesList(Directory.GetCurrentDirectory(), "en", false, null);
// Act
namesList.Add("Charlie Parker123");
var exists = namesList.IsInNamesEtcMultiWordList("This is Charlie Parker123!", "Charlie Parker123");
// Assert
Assert.IsTrue(exists);
}
public void NamesListAddMultiWord()
{
// Arrange
var namesList = new NamesList(Directory.GetCurrentDirectory(), "en", false, null);
// Act
namesList.Add("Charlie Parker123");
var exists = namesList.GetMultiNames().Contains("Charlie Parker123");
// Assert
Assert.IsTrue(exists);
}
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 ButtonOkClick(object sender, EventArgs e)
{
if (string.IsNullOrWhiteSpace(textBoxAddName.Text))
{
return;
}
NewName = textBoxAddName.Text.RemoveControlCharacters().Trim();
string languageName = null;
_language = Configuration.Settings.Language.Main;
if (!string.IsNullOrEmpty(Configuration.Settings.General.SpellCheckLanguage))
{
languageName = Configuration.Settings.General.SpellCheckLanguage;
}
else
{
List<string> list = Utilities.GetDictionaryLanguages();
if (list.Count > 0)
{
string name = list[0];
int start = name.LastIndexOf('[');
int end = name.LastIndexOf(']');
if (start > 0 && end > start)
{
start++;
name = name.Substring(start, end - start);
languageName = name;
}
else
{
MessageBox.Show(string.Format(_language.InvalidLanguageNameX, name));
return;
}
}
}
languageName = LanguageAutoDetect.AutoDetectLanguageName(languageName, _subtitle);
if (comboBoxDictionaries.Items.Count > 0)
{
string name = comboBoxDictionaries.SelectedItem.ToString();
int start = name.LastIndexOf('[');
int end = name.LastIndexOf(']');
if (start >= 0 && end > start)
{
start++;
name = name.Substring(start, end - start);
languageName = name;
}
}
if (string.IsNullOrEmpty(languageName))
languageName = "en_US";
var namesList = new NamesList(Configuration.DictionariesFolder, languageName, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
if (namesList.Add(textBoxAddName.Text))
DialogResult = DialogResult.OK;
else
DialogResult = DialogResult.Cancel;
}
public SpellCheckWordLists(string dictionaryFolder, string languageName, IDoSpell doSpell)
{
if (languageName == null)
throw new NullReferenceException("languageName");
if (doSpell == null)
throw new NullReferenceException("doSpell");
_languageName = languageName;
_doSpell = doSpell;
_namesList = new NamesList(Configuration.DictionariesFolder, languageName, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
_namesEtcList = _namesList.GetNames();
var namesEtcMultiWordList = _namesList.GetMultiNames();
foreach (string namesItem in _namesEtcList)
_namesEtcListUppercase.Add(namesItem.ToUpper());
if (languageName.StartsWith("en_", StringComparison.OrdinalIgnoreCase))
{
foreach (string namesItem in _namesEtcList)
{
if (!namesItem.EndsWith('s'))
{
_namesEtcListWithApostrophe.Add(namesItem + "'s");
_namesEtcListWithApostrophe.Add(namesItem + "’s");
}
else if (!namesItem.EndsWith('\''))
{
_namesEtcListWithApostrophe.Add(namesItem + "'");
}
}
}
_userWordList = new List<string>();
_userPhraseList = new List<string>();
if (File.Exists(dictionaryFolder + languageName + "_user.xml"))
{
var userWordDictionary = new XmlDocument();
userWordDictionary.Load(dictionaryFolder + languageName + "_user.xml");
if (userWordDictionary.DocumentElement != null)
{
var xmlNodeList = userWordDictionary.DocumentElement.SelectNodes("word");
if (xmlNodeList != null)
{
foreach (XmlNode node in xmlNodeList)
{
string word = node.InnerText.Trim().ToLower();
if (word.Contains(' '))
_userPhraseList.Add(word);
else
_userWordList.Add(word);
}
}
}
}
// Add names/userdic with "." or " " or "-"
_wordsWithDashesOrPeriods = new List<string>();
_wordsWithDashesOrPeriods.AddRange(namesEtcMultiWordList);
foreach (string name in _namesEtcList)
{
if (name.Contains(PeriodAndDash))
_wordsWithDashesOrPeriods.Add(name);
}
foreach (string word in _userWordList)
{
if (word.Contains(PeriodAndDash))
_wordsWithDashesOrPeriods.Add(word);
}
_wordsWithDashesOrPeriods.AddRange(_userPhraseList);
}
public bool AddName(string word)
{
if (word == null || word.Length <= 1 || _namesEtcList.Contains(word))
return false;
_namesEtcList.Add(word);
_namesEtcListUppercase.Add(word.ToUpper());
if (_languageName.StartsWith("en_", StringComparison.Ordinal) && !word.EndsWith('s'))
{
_namesEtcList.Add(word + "s");
_namesEtcListUppercase.Add(word.ToUpper() + "S");
}
if (!word.EndsWith('s'))
{
_namesEtcListWithApostrophe.Add(word + "'s");
_namesEtcListUppercase.Add(word.ToUpper() + "'S");
}
if (!word.EndsWith('\''))
_namesEtcListWithApostrophe.Add(word + "'");
var namesList = new NamesList(Configuration.DictionariesFolder, _languageName, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
namesList.Add(word);
return true;
}
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 ButtonRemoveNameEtcClick(object sender, EventArgs e)
{
if (listBoxNamesEtc.SelectedIndices.Count == 0)
return;
string language = GetCurrentWordListLanguage();
int index = listBoxNamesEtc.SelectedIndex;
string text = listBoxNamesEtc.Items[index].ToString();
int itemsToRemoveCount = listBoxNamesEtc.SelectedIndices.Count;
if (!string.IsNullOrEmpty(language) && index >= 0)
{
DialogResult result;
if (itemsToRemoveCount == 1)
result = MessageBox.Show(string.Format(Configuration.Settings.Language.Settings.RemoveX, text), "Subtitle Edit", MessageBoxButtons.YesNo);
else
result = MessageBox.Show(string.Format(Configuration.Settings.Language.Main.DeleteXLinesPrompt, itemsToRemoveCount), "Subtitle Edit", MessageBoxButtons.YesNo);
if (result == DialogResult.Yes)
{
int removeCount = 0;
var namesList = new NamesList(Configuration.DictionariesFolder, language, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
for (int idx = listBoxNamesEtc.SelectedIndices.Count - 1; idx >= 0; idx--)
{
index = listBoxNamesEtc.SelectedIndices[idx];
text = listBoxNamesEtc.Items[index].ToString();
namesList.Remove(text);
removeCount++;
listBoxNamesEtc.Items.RemoveAt(index);
}
if (removeCount > 0)
{
LoadNamesEtc(language, true); // reload
if (index < listBoxNamesEtc.Items.Count)
listBoxNamesEtc.SelectedIndex = index;
else if (listBoxNamesEtc.Items.Count > 0)
listBoxNamesEtc.SelectedIndex = index - 1;
listBoxNamesEtc.Focus();
buttonRemoveNameEtc.Enabled = false;
return;
}
if (removeCount < itemsToRemoveCount && Configuration.Settings.WordLists.UseOnlineNamesEtc && !string.IsNullOrEmpty(Configuration.Settings.WordLists.NamesEtcUrl))
{
MessageBox.Show(Configuration.Settings.Language.Settings.CannotUpdateNamesEtcOnline);
return;
}
if (removeCount == 0)
MessageBox.Show(Configuration.Settings.Language.Settings.WordNotFound);
}
}
}
private void ButtonAddNamesEtcClick(object sender, EventArgs e)
{
string language = GetCurrentWordListLanguage();
string text = textBoxNameEtc.Text.RemoveControlCharacters().Trim();
if (!string.IsNullOrEmpty(language) && text.Length > 1 && !_wordListNamesEtc.Contains(text))
{
var namesList = new NamesList(Configuration.DictionariesFolder, language, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
namesList.Add(text);
LoadNamesEtc(language, true);
labelStatus.Text = string.Format(Configuration.Settings.Language.Settings.WordAddedX, text);
textBoxNameEtc.Text = string.Empty;
textBoxNameEtc.Focus();
for (int i = 0; i < listBoxNamesEtc.Items.Count; i++)
{
if (listBoxNamesEtc.Items[i].ToString() == text)
{
listBoxNamesEtc.SelectedIndex = i;
int top = i - 5;
if (top < 0)
top = 0;
listBoxNamesEtc.TopIndex = top;
break;
}
}
}
else
{
MessageBox.Show(Configuration.Settings.Language.Settings.WordAlreadyExists);
}
}
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 LoadDictionaries(string dictionaryFolder, string dictionary)
{
_changeAllDictionary = new Dictionary<string, string>();
_skipAllList = new List<string>();
_namesList = new NamesList(Configuration.DictionariesFolder, _languageName, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
_namesEtcList = _namesList.GetNames();
_namesEtcMultiWordList = _namesList.GetMultiNames();
foreach (string namesItem in _namesEtcList)
_namesEtcListUppercase.Add(namesItem.ToUpper());
if (_languageName.StartsWith("en_", StringComparison.OrdinalIgnoreCase))
{
foreach (string namesItem in _namesEtcList)
{
if (!namesItem.EndsWith('s'))
{
_namesEtcListWithApostrophe.Add(namesItem + "'s");
_namesEtcListWithApostrophe.Add(namesItem + "’s");
}
else if (!namesItem.EndsWith('\''))
{
_namesEtcListWithApostrophe.Add(namesItem + "'");
}
}
}
_userWordList = new List<string>();
_userPhraseList = new List<string>();
if (File.Exists(dictionaryFolder + _languageName + "_user.xml"))
{
var userWordDictionary = new XmlDocument();
userWordDictionary.Load(dictionaryFolder + _languageName + "_user.xml");
foreach (XmlNode node in userWordDictionary.DocumentElement.SelectNodes("word"))
{
string word = node.InnerText.Trim().ToLower();
if (word.Contains(' '))
_userPhraseList.Add(word);
else
_userWordList.Add(word);
}
}
// Add names/userdic with "." or " " or "-"
_wordsWithDashesOrPeriods = new List<string>();
_wordsWithDashesOrPeriods.AddRange(_namesEtcMultiWordList);
foreach (string name in _namesEtcList)
{
if (name.Contains(new[] { '.', '-' }))
_wordsWithDashesOrPeriods.Add(name);
}
foreach (string word in _userWordList)
{
if (word.Contains(new[] { '.', '-' }))
_wordsWithDashesOrPeriods.Add(word);
}
_wordsWithDashesOrPeriods.AddRange(_userPhraseList);
_changeAllDictionary = new Dictionary<string, string>();
LoadHunspell(dictionary);
}
public void DoAction(SpellCheckAction action)
{
switch (action)
{
case SpellCheckAction.Change:
_noOfChangedWords++;
_mainWindow.CorrectWord(_prefix + ChangeWord + _postfix, _currentParagraph, _prefix + _currentWord + _postfix, ref _firstChange, _wordsIndex);
break;
case SpellCheckAction.ChangeAll:
_noOfChangedWords++;
if (!_changeAllDictionary.ContainsKey(_currentWord))
_changeAllDictionary.Add(_currentWord, ChangeWord);
_mainWindow.CorrectWord(_prefix + ChangeWord + _postfix, _currentParagraph, _prefix + _currentWord + _postfix, ref _firstChange, -1);
break;
case SpellCheckAction.Skip:
_noOfSkippedWords++;
break;
case SpellCheckAction.SkipAll:
_noOfSkippedWords++;
_skipAllList.Add(ChangeWord.ToUpper());
if (ChangeWord.EndsWith('\'') || ChangeWord.StartsWith('\''))
_skipAllList.Add(ChangeWord.ToUpper().Trim('\''));
break;
case SpellCheckAction.AddToDictionary:
if (_userWordList.IndexOf(ChangeWord) < 0)
{
_noOfAddedWords++;
string s = ChangeWord.Trim().ToLower();
if (s.Contains(' '))
_userPhraseList.Add(s);
else
_userWordList.Add(s);
Utilities.AddToUserDictionary(s, _languageName);
}
break;
case SpellCheckAction.AddToNamesEtc:
if (ChangeWord.Length > 1 && !_namesEtcList.Contains(ChangeWord))
{
_namesEtcList.Add(ChangeWord);
_namesEtcListUppercase.Add(ChangeWord.ToUpper());
if (_languageName.StartsWith("en_", StringComparison.Ordinal) && !ChangeWord.EndsWith('s'))
{
_namesEtcList.Add(ChangeWord + "s");
_namesEtcListUppercase.Add(ChangeWord.ToUpper() + "S");
}
if (!ChangeWord.EndsWith('s'))
{
_namesEtcListWithApostrophe.Add(ChangeWord + "'s");
_namesEtcListUppercase.Add(ChangeWord.ToUpper() + "'S");
}
if (!ChangeWord.EndsWith('\''))
_namesEtcListWithApostrophe.Add(ChangeWord + "'");
var namesList = new NamesList(Configuration.DictionariesFolder, _languageName, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
namesList.Add(ChangeWord);
}
break;
case SpellCheckAction.ChangeWholeText:
_mainWindow.ShowStatus(string.Format(Configuration.Settings.Language.Main.SpellCheckChangedXToY, _currentParagraph.Text.Replace(Environment.NewLine, " "), ChangeWholeText.Replace(Environment.NewLine, " ")));
_currentParagraph.Text = ChangeWholeText;
_mainWindow.ChangeWholeTextMainPart(ref _noOfChangedWords, ref _firstChange, _currentIndex, _currentParagraph);
break;
}
labelActionInfo.Text = string.Empty;
PrepareNextWord();
CheckActions();
}
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);
}
private void LoadSpellingDictionariesViaDictionaryFileName(string threeLetterIsoLanguageName, CultureInfo culture, string dictionaryFileName, bool resetSkipList)
{
_fiveLetterWordListLanguageName = Path.GetFileNameWithoutExtension(dictionaryFileName);
if (_fiveLetterWordListLanguageName.Length > 5)
_fiveLetterWordListLanguageName = _fiveLetterWordListLanguageName.Substring(0, 5);
string dictionary = Utilities.DictionaryFolder + _fiveLetterWordListLanguageName;
if (resetSkipList)
{
_wordSkipList = new HashSet<string> { Configuration.Settings.Tools.MusicSymbol, "*", "%", "#", "+", "$" };
}
// Load names etc list (names/noise words)
_namesList = new NamesList(Configuration.DictionariesFolder, _fiveLetterWordListLanguageName, Configuration.Settings.WordLists.UseOnlineNamesEtc, Configuration.Settings.WordLists.NamesEtcUrl);
_namesEtcList = _namesList.GetNames();
_namesEtcMultiWordList = _namesList.GetMultiNames();
_namesEtcListUppercase = new HashSet<string>();
foreach (string name in _namesEtcList)
_namesEtcListUppercase.Add(name.ToUpper());
_namesEtcListWithApostrophe = new HashSet<string>();
if (threeLetterIsoLanguageName.Equals("eng", StringComparison.OrdinalIgnoreCase))
{
foreach (string namesItem in _namesEtcList)
{
if (!namesItem.EndsWith('s'))
_namesEtcListWithApostrophe.Add(namesItem + "'s");
else
_namesEtcListWithApostrophe.Add(namesItem + "'");
}
}
// Load user words
_userWordList = new HashSet<string>();
_userWordListXmlFileName = Utilities.LoadUserWordList(_userWordList, _fiveLetterWordListLanguageName);
// Find abbreviations
_abbreviationList = new HashSet<string>();
foreach (string name in _namesEtcList)
{
if (name.EndsWith('.'))
_abbreviationList.Add(name);
}
if (threeLetterIsoLanguageName.Equals("eng", StringComparison.OrdinalIgnoreCase))
{
if (!_abbreviationList.Contains("a.m."))
_abbreviationList.Add("a.m.");
if (!_abbreviationList.Contains("p.m."))
_abbreviationList.Add("p.m.");
if (!_abbreviationList.Contains("o.r."))
_abbreviationList.Add("o.r.");
}
foreach (string name in _userWordList)
{
if (name.EndsWith('.'))
_abbreviationList.Add(name);
}
// Load Hunspell spell checker
try
{
if (!File.Exists(dictionary + ".dic"))
{
var fileMatches = Directory.GetFiles(Utilities.DictionaryFolder, _fiveLetterWordListLanguageName + "*.dic");
if (fileMatches.Length > 0)
dictionary = fileMatches[0].Substring(0, fileMatches[0].Length - 4);
}
if (_hunspell != null)
_hunspell.Dispose();
_hunspell = Hunspell.GetHunspell(dictionary);
IsDictionaryLoaded = true;
_spellCheckDictionaryName = dictionary;
DictionaryCulture = culture;
}
catch
{
IsDictionaryLoaded = false;
}
}