/// <summary>
/// Build a data table for all of an ASCII character's bestfit mappings.
/// </summary>
/// <param name="cAscii">The ASCII character to query on.</param>
/// <param name="lBestFit">Reference to a List you want to populate with data.</param>
/// <param name="sCharset">An optional charset to filter results by.
/// </param>
public void BuildBestfitTable(char cAscii,
ref List<BestFitMapping> lBestFit,
string sCharset = "")
{
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable<XElement> query;
if (String.IsNullOrEmpty(sCharset))
{
query = from mapping in XDocBestfit.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint
select mapping;
}
else
{
query = (from mapping in XDocBestfit.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint &&
(string)mapping.Element("Charset") == sCharset
select mapping);
}
var count = query.Count();
foreach (var item in query.Distinct())
{
BestFitMapping bf = new BestFitMapping();
UniChar uc2 = new UniChar();
bf.Ascii = item.Element("Ascii").Value;
bf.Unicode = item.Element("Unicode").Value;
bf.Character = uc2.ConvertCodePointToString(bf.Unicode);
bf.Charset = item.Element("Charset").Value;
bf.Name = item.Element("Name").Value;
lBestFit.Add(bf);
}
}
/// <summary>
/// Get all normalization mappings for an ASCII character. Optionally specify a specific transform
/// as a filter.
/// </summary>
/// <param name="cAscii">A single ASCII character.</param>
/// <param name="sTransform">Valid transformations specified as:
///
/// Simple_Lowercase_Mapping
/// Lowercase_Mapping
/// Simple_Case_Folding
/// cf
/// Simple_Uppercase_Mapping
/// Simple_Titlecase_Mapping
/// Uppercase_Mapping
/// Titlecase_Mapping
/// Decomposition_Mapping
/// FC_NFKC
/// </param>
/// <returns></returns>
public List <String> GetNormalizationMappings(char cAscii, string sTransform = "")
{
UnicodeMapping um = new UnicodeMapping();
// If something is passed in we don't know about then set transform to a wildcard
if (!um.transformations.Contains(sTransform))
{
sTransform = "";
}
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable <string> query;
// If a transformation wasn't specified, filter by the ASCII character
if (String.IsNullOrEmpty(sTransform))
{
query = (from mapping in XDocUnicode.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == uc.CodePoint
select mapping.Element("Unicode").Value);
}
// else filter by the transformation too
else
{
query = (from mapping in XDocUnicode.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == "0043" &&
(string)mapping.Element("Transform") == sTransform
select mapping.Element("Unicode").Value);
}
List <String> data = new List <string>();
foreach (var item in query.Distinct())
{
data.Add(item);
}
return(data);
}
/// <summary>
/// Send me an ASCII character and I'll return you a list of Unicode characters that
/// best fit map to it. Since you're not telling me a specific charset your're
/// interested in, I'm going to send you data for all of them.
/// </summary>
/// <param name="cAscii">The ASCII character to query on.</param>
/// <param name="sCharset">An option charset name to filter by, valid values include:
/// APL-ISO-IR-68
/// CP424
/// IBMGRAPH
/// US-ASCII-QUOTES
/// windows-1250
/// windows-1251
/// windows-1252
/// windows-1253
/// windows-1254
/// windows-1255
/// windows-1256
/// windows-1257
/// windows-1258
/// windows-874
/// CP864
/// CP037
/// CP1026
/// CP500
/// CP875
/// DINGBATS
/// KEYBOARD
/// SYMBOL
/// symbol
/// zdingbat
/// JAPANESE
/// GSM0338
/// </param>
/// <returns></returns>
public List <String> GetBestfitMappings(char cAscii, string sCharset = "")
{
BestFitMapping bm = new BestFitMapping();
// If an invald charset was entered then set it to the wildcard
if (!bm.charsets.Contains(sCharset))
{
sCharset = "";
}
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable <string> query;
// If a charset wasn't specified, filter by the ASCII character
if (String.IsNullOrEmpty(sCharset))
{
query = (from mapping in XDocBestfit.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == uc.CodePoint
select mapping.Element("Unicode").Value);
}
// else filter by the charset too
else
{
query = (from mapping in XDocBestfit.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == "0043" &&
(string)mapping.Element("Charset") == sCharset
select mapping.Element("Unicode").Value);
}
List <String> data = new List <string>();
foreach (var item in query.Distinct())
{
data.Add(item);
}
return(data);
}
public String GetExpandedUnicodeCharacter(char ascii)
{
IEnumerable <String> searchResults;
UniChar unicode = new UniChar();
unicode.CodePoint = unicode.GetCodePoint(ascii);
searchResults = (from mapping in XDocExpandedUnicode.Descendants("Mapping")
select mapping.Element("Unicode").Value);
List <String> data = new List <string>();
foreach (var item in searchResults.Distinct())
{
data.Add(item);
}
Random rand = new Random();
int size = rand.Next(data.Count());
return(data.ElementAtOrDefault(size));
}
/// <summary>
/// Build a data table for all of an ASCII character's bestfit mappings.
/// </summary>
/// <param name="cAscii">The ASCII character to query on.</param>
/// <param name="lTransformations">Reference to a List you want to populate with data.</param>
/// <param name="sTransform">An optional charset to filter results by.</param>
public void BuildTransformationsTable(char cAscii,
ref List <UnicodeMapping> lTransformations,
string sTransform = "")
{
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable <XElement> query;
if (String.IsNullOrEmpty(sTransform))
{
query = from mapping in xDocUnicode.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint
select mapping;
}
else
{
query = (from mapping in XDocUnicode.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint &&
(string)mapping.Element("Transform") == sTransform
select mapping);
}
var count = query.Count();
foreach (var item in query.Distinct())
{
UnicodeMapping um = new UnicodeMapping();
UniChar uc2 = new UniChar();
um.Ascii = item.Element("Ascii").Value;
um.Unicode = item.Element("Unicode").Value;
um.Character = uc2.ConvertCodePointToString(um.Unicode);
um.Transform = item.Element("Transform").Value;
um.Name = item.Element("Name").Value;
lTransformations.Add(um);
}
}
/// <summary>
/// Build a data table for all of an ASCII character's bestfit mappings.
/// </summary>
/// <param name="cAscii">The ASCII character to query on.</param>
/// <param name="lBestFit">Reference to a List you want to populate with data.</param>
/// <param name="sCharset">An optional charset to filter results by.
/// </param>
public void BuildBestfitTable(char cAscii,
ref List <BestFitMapping> lBestFit,
string sCharset = "")
{
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable <XElement> query;
if (String.IsNullOrEmpty(sCharset))
{
query = from mapping in XDocBestfit.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint
select mapping;
}
else
{
query = (from mapping in XDocBestfit.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint &&
(string)mapping.Element("Charset") == sCharset
select mapping);
}
var count = query.Count();
foreach (var item in query.Distinct())
{
BestFitMapping bf = new BestFitMapping();
UniChar uc2 = new UniChar();
bf.Ascii = item.Element("Ascii").Value;
bf.Unicode = item.Element("Unicode").Value;
bf.Character = uc2.ConvertCodePointToString(bf.Unicode);
bf.Charset = item.Element("Charset").Value;
bf.Name = item.Element("Name").Value;
lBestFit.Add(bf);
}
}
/// <summary>
/// Get all normalization mappings for an ASCII character. Optionally specify a specific transform
/// as a filter.
/// </summary>
/// <param name="cAscii">A single ASCII character.</param>
/// <param name="sTransform">Valid transformations specified as:
///
/// Simple_Lowercase_Mapping
/// Lowercase_Mapping
/// Simple_Case_Folding
/// cf
/// Simple_Uppercase_Mapping
/// Simple_Titlecase_Mapping
/// Uppercase_Mapping
/// Titlecase_Mapping
/// Decomposition_Mapping
/// FC_NFKC
/// </param>
/// <returns></returns>
public List<String> GetNormalizationMappings(char cAscii, string sTransform = "")
{
UnicodeMapping um = new UnicodeMapping();
// If something is passed in we don't know about then set transform to a wildcard
if (!um.transformations.Contains(sTransform))
{
sTransform = "";
}
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable<string> query;
// If a transformation wasn't specified, filter by the ASCII character
if (String.IsNullOrEmpty(sTransform))
{
query = (from mapping in XDocUnicode.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == uc.CodePoint
select mapping.Element("Unicode").Value);
}
// else filter by the transformation too
else
{
query = (from mapping in XDocUnicode.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == "0043" &&
(string)mapping.Element("Transform") == sTransform
select mapping.Element("Unicode").Value);
}
List<String> data = new List<string>();
foreach (var item in query.Distinct())
{
data.Add(item);
}
return data;
}
/// <summary>
/// Send me an ASCII character and I'll return you a list of Unicode characters that
/// best fit map to it. Since you're not telling me a specific charset your're
/// interested in, I'm going to send you data for all of them.
/// </summary>
/// <param name="cAscii">The ASCII character to query on.</param>
/// <param name="sCharset">An option charset name to filter by, valid values include:
/// APL-ISO-IR-68
/// CP424
/// IBMGRAPH
/// US-ASCII-QUOTES
/// windows-1250
/// windows-1251
/// windows-1252
/// windows-1253
/// windows-1254
/// windows-1255
/// windows-1256
/// windows-1257
/// windows-1258
/// windows-874
/// CP864
/// CP037
/// CP1026
/// CP500
/// CP875
/// DINGBATS
/// KEYBOARD
/// SYMBOL
/// symbol
/// zdingbat
/// JAPANESE
/// GSM0338
/// </param>
/// <returns></returns>
public List<String> GetBestfitMappings(char cAscii, string sCharset = "")
{
BestFitMapping bm = new BestFitMapping();
// If an invald charset was entered then set it to the wildcard
if (!bm.charsets.Contains(sCharset))
{
sCharset = "";
}
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable<string> query;
// If a charset wasn't specified, filter by the ASCII character
if (String.IsNullOrEmpty(sCharset))
{
query = (from mapping in XDocBestfit.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == uc.CodePoint
select mapping.Element("Unicode").Value);
}
// else filter by the charset too
else
{
query = (from mapping in XDocBestfit.Descendants("Mapping")
where
(string)mapping.Element("Ascii") == "0043" &&
(string)mapping.Element("Charset") == sCharset
select mapping.Element("Unicode").Value);
}
List<String> data = new List<string>();
foreach (var item in query.Distinct())
{
data.Add(item);
}
return data;
}
/// <summary>
/// Build a data table for all of an ASCII character's bestfit mappings.
/// </summary>
/// <param name="cAscii">The ASCII character to query on.</param>
/// <param name="lTransformations">Reference to a List you want to populate with data.</param>
/// <param name="sTransform">An optional charset to filter results by.</param>
public void BuildTransformationsTable(char cAscii,
ref List<UnicodeMapping> lTransformations,
string sTransform = "")
{
UniChar uc = new UniChar();
uc.CodePoint = uc.GetCodePoint(cAscii);
IEnumerable<XElement> query;
if (String.IsNullOrEmpty(sTransform))
{
query = from mapping in xDocUnicode.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint
select mapping;
}
else
{
query = (from mapping in XDocUnicode.Descendants("Mapping")
where (string)mapping.Element("Ascii") == uc.CodePoint &&
(string)mapping.Element("Transform") == sTransform
select mapping);
}
var count = query.Count();
foreach (var item in query.Distinct())
{
UnicodeMapping um = new UnicodeMapping();
UniChar uc2 = new UniChar();
um.Ascii = item.Element("Ascii").Value;
um.Unicode = item.Element("Unicode").Value;
um.Character = uc2.ConvertCodePointToString(um.Unicode);
um.Transform= item.Element("Transform").Value;
um.Name = item.Element("Name").Value;
lTransformations.Add(um);
}
}
private void buttonGetUnicode_Click(object sender, EventArgs e)
{
textBoxOutput.Text = "";
List<String> transforms = new List<string>();
transforms = Data.GetNormalizationMappings(Input);
string output = String.Empty;
foreach (string transform in transforms)
{
UniChar uc = new UniChar();
if (!String.IsNullOrEmpty(transform))
{
output += uc.ConvertCodePointToString(transform) + "\r\n";
}
}
textBoxOutput.Text = output;
// Fill DataGrid
List<UnicodeMapping> lTransformations = new List<UnicodeMapping>();
dataGridViewBestFit.DataSource = null;
Data.BuildTransformationsTable(Input, ref lTransformations, Transform);
dataGridViewBestFit.DataSource = lTransformations;
}
private void buttonGetBestfit_Click(object sender, EventArgs e)
{
textBoxOutput.Text = "";
List<String> bestfits = new List<string>();
bestfits = Data.GetBestfitMappings(Input);
string output = String.Empty;
foreach (string bestfit in bestfits)
{
UniChar uc = new UniChar();
if (!String.IsNullOrEmpty(bestfit))
{
output += uc.ConvertCodePointToString(bestfit) + "\r\n";
}
}
textBoxOutput.Text = output;
// Fill DataGrid
List<BestFitMapping> lBestfits = new List<BestFitMapping>();
dataGridViewBestFit.DataSource = null;
Data.BuildBestfitTable(Input,ref lBestfits, Charset);
dataGridViewBestFit.DataSource = lBestfits;
}
