private void UpdateAllCharsRefinerSettingsUI()
{
FilteringCharacterCategory cats = m_allCharactersRefinerSettings.NotIgnoredCategories;
cbRefineDigits.Checked = ((FilteringCharacterCategory.ArabicDigit & cats) == FilteringCharacterCategory.ArabicDigit);
cbRefineErab.Checked = ((FilteringCharacterCategory.Erab & cats) == FilteringCharacterCategory.Erab);
cbRefineHalfSpaceChar.Checked = ((FilteringCharacterCategory.HalfSpace & cats) == FilteringCharacterCategory.HalfSpace);
cbRefineKaaf.Checked = ((FilteringCharacterCategory.Kaaf & cats) == FilteringCharacterCategory.Kaaf);
cbRefineYaa.Checked = ((FilteringCharacterCategory.Yaa & cats) == FilteringCharacterCategory.Yaa);
cbRemoveHalfSpaces.Checked = m_allCharactersRefinerSettings.RefineHalfSpacePositioning;
cbRefineAndNormalizeHeYe.Checked = m_allCharactersRefinerSettings.NormalizeHeYe;
cbConvertLongHeYeToShort.Checked = m_allCharactersRefinerSettings.ConvertLongHeYeToShort;
cbConvertShortHeYeToLong.Checked = m_allCharactersRefinerSettings.ConvertShortHeYeToLong;
tetLetterToIgnore.Text = "";
listViewIgnoreList.Items.Clear();
if (listViewIgnoreList.Items.Count <= 0) // if the form has been ewnewed with an empty list only
{
foreach (char c in m_allCharactersRefinerSettings.GetIgnoreListAsString())
{
AddLetterToIgnoreList(c, true);
}
}
}
/// <summary>
/// Adds a sequence of character codes to a filtering category
/// </summary>
/// <param name="category">The filtering category.</param>
/// <param name="codes">The codes.</param>
private void AddCodesToCategory(FilteringCharacterCategory category, params int[] codes)
{
foreach (int code in codes)
{
m_dicCharCategories.Add(code, category);
}
}
/// <summary>
/// Filters the char and returns the string for its filtered (i.e. standardized) equivalant.
/// The string may contain 0, 1, or more characters.
/// If the length of the string is 0, then the character should have been left out.
/// If the length of the string is 1, then the character might be left intact or replaced with another character.
/// If the length of the string is more than 1, then there have been no 1-character replacement for this character.
/// It is replaced with 2 or more characters. e.g. some fonts have encoded Tashdid, and Tanvin in one character.
/// To make it standard this character is replaced with 2 characters, one for Tashdid, and the other for Tanvin.
/// </summary>
/// <param name="ch">The character to filter.</param>
/// <param name="ignoreList">The characters to be ignored.</param>
/// <param name="ignoreCats">The filtering categories to be ignored.</param>
/// <returns></returns>
public string FilterChar(char ch, HashSet <char> ignoreList, FilteringCharacterCategory ignoreCats)
{
if (ignoreList.Contains(ch))
{
return(ch.ToString());
}
return(FilterChar(ch, ignoreCats));
}
/// <summary>
/// Filters every character of the string and returns the filtered string. To see how each character is
/// filtered see: <see cref="FilterChar(char)"/>
/// </summary>
/// <param name="str">The string to be filtered.</param>
/// <param name="ignoreList">The characters to be ignored.</param>
/// <param name="ignoreCats">The filtering categories to be ignored.</param>
/// <returns></returns>
public string FilterString(string str, HashSet <char> ignoreList, FilteringCharacterCategory ignoreCats)
{
var sb = new StringBuilder();
foreach (char c in str)
{
sb.Append(FilterChar(c, ignoreList, ignoreCats));
}
return(sb.ToString());
}
public CharacterRefinementVerifier(AllCharactersRefinerSettings settings)
: base(0, 1)
{
m_settings = settings;
ResetStats();
ResetCurStepStats();
m_filteringCategory = settings.GetIgnoredCategories();
m_ignoreList = settings.GetIgnoreList();
}
/// <summary>
/// Filters the char and returns the string for its filtered (i.e. standardized) equivalant.
/// The string may contain 0, 1, or more characters.
/// If the length of the string is 0, then the character should have been left out.
/// If the length of the string is 1, then the character might be left intact or replaced with another character.
/// If the length of the string is more than 1, then there have been no 1-character replacement for this character.
/// It is replaced with 2 or more characters. e.g. some fonts have encoded Tashdid, and Tanvin in one character.
/// To make it standard this character is replaced with 2 characters, one for Tashdid, and the other for Tanvin.
/// </summary>
/// <param name="ch">The character to filter.</param>
/// <param name="ignoreCats">The filtering categories to be ignored.</param>
/// <returns></returns>
public string FilterChar(char ch, FilteringCharacterCategory ignoreCats)
{
int nCh = Convert.ToInt32(ch);
FilteringCharacterCategory cat;
if (m_dicCharCategories.TryGetValue(nCh, out cat))
{
return((cat & ignoreCats) == cat?ch.ToString() : FilterChar(ch));
}
return(FilterChar(ch));
}
public void SetOffFilteringCategory(FilteringCharacterCategory cat)
{
NotIgnoredCategories = (NotIgnoredCategories & (~cat));
}
public void SetFilteringCategory(FilteringCharacterCategory cat)
{
NotIgnoredCategories = (NotIgnoredCategories | cat);
}
/// <summary>
/// Filters the given string with the given options while returning the filtered string as well as filtering statistics.
/// </summary>
/// <param name="str">The string to filter.</param>
/// <param name="ignoreList">list of characters to ignore.</param>
/// <param name="ignoreCats">The character-categories to ignore.</param>
/// <returns></returns>
public FilterResultsWithStats FilterStringWithStats(string str, HashSet <char> ignoreList, FilteringCharacterCategory ignoreCats)
{
var sbResult = new StringBuilder(str.Length);
int numLetters = 0;
int numErabs = 0;
int numDigits = 0;
int numHlafSpaces = 0;
foreach (var ch in str)
{
if (ignoreList != null && ignoreList.Contains(ch))
{
sbResult.Append(ch);
}
int nCh = Convert.ToInt32(ch);
string filterResult;
// if there's something to filter
if (m_dicCharFilterings.TryGetValue(nCh, out filterResult))
{
// try to find the filtering category
FilteringCharacterCategory cat;
if (m_dicCharCategories.TryGetValue(nCh, out cat))
{
if ((cat & ignoreCats) == cat)
{
sbResult.Append(ch);
}
else
{
sbResult.Append(filterResult);
switch (cat)
{
case FilteringCharacterCategory.ArabicDigit:
numDigits++;
break;
case FilteringCharacterCategory.Erab:
numErabs++;
break;
case FilteringCharacterCategory.HalfSpace:
numHlafSpaces++;
break;
case FilteringCharacterCategory.Kaaf:
numLetters++;
break;
case FilteringCharacterCategory.Yaa:
numLetters++;
break;
}
}
}
else // if there's no filtering category specified for the given character
{
numLetters++;
}
}
else // there's no need to filter the given character
{
sbResult.Append(ch);
}
} // end of foreach char in str
return(new FilterResultsWithStats(sbResult.ToString(), numLetters, numDigits, numErabs, numHlafSpaces));
}
