public void ConvertAsciiToUnicode()
{
CheckDisposed();
string encFileName = string.Empty;
EncConverters converters = new EncConverters();
try
{
// Define an encoding converter
StreamReader reader = new StreamReader(System.Reflection.Assembly.GetExecutingAssembly().GetManifestResourceStream(
"SIL.FieldWorks.FDO.Scripture.EncTest.map"));
encFileName = Path.Combine(Path.GetTempPath(), "test.map");
StreamWriter writer = new StreamWriter(encFileName);
writer.Write(reader.ReadToEnd());
reader.Close();
writer.Close();
converters.Add("MyConverter", encFileName, ConvType.Legacy_to_from_Unicode, string.Empty,
string.Empty, ProcessTypeFlags.UnicodeEncodingConversion);
using (TempSFFileMaker fileMaker = new TempSFFileMaker())
{
string fileName = fileMaker.CreateFile("ROM",
new string[] {
@"\mt 0123456789",
"\\s \u0081\u009a\u0096\u00b5",
@"\c 1",
@"\v 1"},
Encoding.GetEncoding(28591), false);
m_settings.AddFile(fileName, ImportDomain.Main, null, 0);
// Set the vernacular WS to use the MyConverter encoder
LgWritingSystem wsVern =
new LgWritingSystem(Cache, Cache.LangProject.DefaultVernacularWritingSystem);
wsVern.LegacyMapping = "MyConverter";
ISCTextEnum textEnum = GetTextEnum(ImportDomain.Main,
new ScrReference(45, 0, 0, Paratext.ScrVers.English), new ScrReference(45, 1, 1, Paratext.ScrVers.English));
ISCTextSegment textSeg = textEnum.Next();
Assert.IsNotNull(textSeg, "Unable to read segment");
Assert.AreEqual(@"\id", textSeg.Marker);
Assert.AreEqual("ROM ", textSeg.Text);
textSeg = textEnum.Next();
Assert.IsNotNull(textSeg, "Unable to read segment");
Assert.AreEqual(@"\mt", textSeg.Marker);
Assert.AreEqual("\u0966\u0967\u0968\u0969\u096a\u096b\u096c\u096d\u096e\u096f ", textSeg.Text);
textSeg = textEnum.Next();
Assert.IsNotNull(textSeg, "Unable to read segment");
Assert.AreEqual(@"\s", textSeg.Marker);
Assert.AreEqual("\u0492\u043a\u2013\u04e9 ", textSeg.Text);
textSeg = textEnum.Next();
Assert.IsNotNull(textSeg, "Unable to read segment");
Assert.AreEqual(@"\c", textSeg.Marker);
Assert.AreEqual(@" ", textSeg.Text);
textSeg = textEnum.Next();
Assert.IsNotNull(textSeg, "Unable to read segment");
Assert.AreEqual(@"\v", textSeg.Marker);
Assert.AreEqual(@" ", textSeg.Text);
}
}
finally
{
converters.Remove("MyConverter");
try
{
File.Delete(encFileName);
}
catch {}
}
}
private void buttonAddNewProject_Click(object sender, System.EventArgs e)
{
// in case the Add New project button was default, change it to the OK button.
this.AcceptButton = this.buttonOK;
// get the delimiter for word boundaries and disallow the /"/ character
m_strWordBoundaryDelimiter = this.textBoxWordBoundaryDelimiter.Text;
if( m_strWordBoundaryDelimiter.IndexOf('"') != -1 )
{
MessageBox.Show("Can't use the double-quote character for the word boundary delimiter",SpellingFixerEC.cstrCaption);
return;
}
bool bRewriteCCTable = false;
string strPunctuation = EncodePunctuationForCC(this.textBoxAddlPunctuation.Text);
if(strPunctuation == null )
return; // it had a bad character
else if( strPunctuation != m_strNonWordCharacters )
{
// this means the file must be re-written
bRewriteCCTable = true;
m_strNonWordCharacters = strPunctuation;
}
if( (!m_bLegacy) != this.checkBoxUnicode.Checked )
{
m_bLegacy = !this.checkBoxUnicode.Checked;
bRewriteCCTable = true;
}
// check for existing EncConverter with this same project information
string strCCTableSpec = null;
string strPartialName = this.textBoxNewProjectName.Text;
string strEncConverterName = FullName(strPartialName);
EncConverters aECs = new EncConverters();
IEncConverter aEC = aECs[strEncConverterName];
if( aEC != null )
{
// if we're *not* in edit mode
if( this.buttonAddNewProject.Text == cstrAddNewProjectButtonText )
{
if( MessageBox.Show(String.Format("A project already exists by the name {0}. Click 'Yes' to overwrite", this.textBoxNewProjectName.Text),SpellingFixerEC.cstrCaption, MessageBoxButtons.YesNoCancel) == DialogResult.Yes)
{
// take it out of the check box list
checkedListBoxProjects.Items.Remove(strPartialName);
strCCTableSpec = aEC.ConverterIdentifier;
if( File.Exists(strCCTableSpec) )
{
File.Delete(strCCTableSpec);
strCCTableSpec = null;
}
// remove the existing one and we'll add a new one next
aECs.Remove(aEC.Name);
aEC = null;
}
else
return;
}
else // edit mode
{
if( MessageBox.Show(String.Format("Do you want to update the '{0}' project?", this.textBoxNewProjectName.Text),SpellingFixerEC.cstrCaption, MessageBoxButtons.YesNoCancel) == DialogResult.Yes)
{
// take it out of the check box list
checkedListBoxProjects.Items.Remove(strPartialName);
// save the spec so that we don't make one below
strCCTableSpec = aEC.ConverterIdentifier;
// the remove the converter since we'll add it back again next
aECs.Remove(aEC.Name);
aEC = null;
}
else
{
// reset this in case we were just editing it.
ResetNewProjectLook();
return;
}
}
}
// if we're aren't using the old cc table, then...
if( strCCTableSpec == null )
{
// now add it (put it in the normal 'MapsTables' folder in \pf\cf\sil\...)
string strMapsTableDir = GetMapTableFolderPath;
strCCTableSpec = strMapsTableDir + @"\" + strEncConverterName + ".cct";
if( File.Exists(strCCTableSpec) )
{
// the converter doesn't exist, but a file with the name we would have
// given it does... ask the user if they want to overwrite it.
// TODO: this doesn't allow for complete flexibility. It might be nicer to
// allow for any arbitrary name, but not if noone complains.
if( MessageBox.Show(String.Format("A file exists by the name {0}. Click 'Yes' to overwrite", strCCTableSpec),SpellingFixerEC.cstrCaption, MessageBoxButtons.YesNoCancel) == DialogResult.Yes)
{
File.Delete(strCCTableSpec);
#if WriteOnAdd
// if the user goes to add the first record, it'd be better if the file didn't exist because now we do some
// preliminary testing whether the CC table already changes a word before adding a new one, but this causes
// a non-trapable error if there are no rules in the file. So just create it when it is actually needed
CreateCCTable(strCCTableSpec,strEncConverterName,m_strNonWordCharacters, !this.m_bLegacy);
#endif
}
}
#if WriteOnAdd
else
CreateCCTable(strCCTableSpec,strEncConverterName,m_strNonWordCharacters, !this.m_bLegacy);
#endif
bRewriteCCTable = false;
}
// now add the EncConverter
// TODO: EncConverters needs a new interface to get the defining encodingID from
// a FontName (so we can use it in this call) just like we can 'try' to get the
// code page given a font name (see 'CodePage' below)
ConvType eConvType = (m_bLegacy)
? ConvType.Legacy_to_Legacy : ConvType.Unicode_to_Unicode;
aECs.Add(strEncConverterName,strCCTableSpec,eConvType,null,null,SpellingFixerEC.SFProcessType);
Font font = null;
try
{
font = new Font(comboBoxFont.SelectedItem.ToString(),Convert.ToSingle(listBoxFontSize.SelectedItem));
}
catch
{
MessageBox.Show("Couldn't create the selected font. Contact support");
return;
}
// add this 'displaying font' information to the converter as properties/attributes
ECAttributes aECAttrs = aECs.Attributes(strEncConverterName,AttributeType.Converter);
aECAttrs.Add(SpellingFixerEC.cstrAttributeFontToUse,font.Name);
aECAttrs.Add(SpellingFixerEC.cstrAttributeFontSizeToUse, font.Size);
aECAttrs.Add(SpellingFixerEC.cstrAttributeWordBoundaryDelimiter, m_strWordBoundaryDelimiter);
aECAttrs.Add(SpellingFixerEC.cstrAttributeNonWordChars, m_strNonWordCharacters);
// if it's not Unicode, then we need a code page in order to convert from wide to
// narrow (when writing to the file).
int cp = 0;
if( m_bLegacy )
{
// try to get the code page from EncConverters
try
{
cp = aECs.CodePage(font.Name);
}
catch
{
// if it fails, it means we don't have a mapping, so add one here.
// TODO: it would be nice to have an encoding, but I'm loath to query
// the user for it here since it isn't extremely relevant to this app.
cp = Convert.ToInt32(this.textBoxCP.Text);
aECs.AddFont(font.Name,cp,null);
}
}
if(bRewriteCCTable) // we are going to continue using the old file... so we must re-write it.
{
// if it was legacy encoded, then we need to convert the data to narrow using
// the code page the user specified (or we got out of the repository)
Encoding enc = null;
if( m_bLegacy )
{
if (cp == EncConverters.cnSymbolFontCodePage)
cp = EncConverters.cnIso8859_1CodePage;
enc = Encoding.GetEncoding(cp);
}
else
enc = new UTF8Encoding();
DataTable myTable;
if( SpellingFixerEC.InitializeDataTableFromCCTable(strCCTableSpec, enc, m_strWordBoundaryDelimiter, out myTable) )
{
ReWriteCCTableHeader(strCCTableSpec,m_strNonWordCharacters,enc);
SpellingFixerEC.AppendCCTableFromDataTable(strCCTableSpec, enc, m_strWordBoundaryDelimiter, m_strNonWordCharacters, myTable);
}
}
// finally, add the new project to the now-visible checkbox list
this.buttonOK.Enabled = checkedListBoxProjects.Visible = true;
ClearClickedItems();
checkedListBoxProjects.Items.Add(strPartialName,CheckState.Checked);
// reset this in case we were just editing it.
ResetNewProjectLook();
}
