/// ------------------------------------------------------------------------------------
/// <summary>
/// Get the cached character associated with the given codepoint
/// </summary>
/// <param name="codepoint">The integer codepoint of the cached character.</param>
/// <returns>The character as a PUACharacter</returns>
/// ------------------------------------------------------------------------------------
public PUACharacter FindCachedIcuEntry(int codepoint)
{
CheckDisposed();
//? Need to have at least 4 digits in the string.
//? string codepointAsString = string.Format("{0:x4}", codepoint).ToUpper();
string codepointAsString = codepoint.ToString("x").ToUpper();
foreach (int[] range in m_cachedIcuRanges)
{
if (IsInRange(codepoint, range))
{
// If the range is not loaded, load it
if (range[2]++ == 0)
FillRange(range);
return (PUACharacter)m_cachedIcu.Find(codepointAsString, new UCDComparer());
}
}
// If we get here, we didn't find the codepoint
PUACharacter newIcuChar = new PUACharacter(codepointAsString);
// Load the character if it exists and has a name
if (newIcuChar.RefreshFromIcu(false))
{
// This can be called from PuaCharacterDlg where m_cachedIcu has not been initialized.
InitializeIcuCache();
m_cachedIcu.Insert(newIcuChar);
return newIcuChar;
}
return null;
}
/// ------------------------------------------------------------------------------------
/// <summary>
///
/// </summary>
/// ------------------------------------------------------------------------------------
private void FillRange(int[] range)
{
// Makes a new empty Icu Cache, if it doesn't already exist
InitializeIcuCache();
PUACharacter newIcuChar;
for (int codepoint = range[0]; codepoint <= range[1]; codepoint++)
{
// TODO: Make sure that leaving off leading zeros is okay.
newIcuChar = new PUACharacter(codepoint.ToString("x").ToUpper());
// Fill in the character from the ICU database.
newIcuChar.RefreshFromIcu(true);
// Add the character to the cache
m_cachedIcu.Insert(newIcuChar);
// REVIEW (TimS/EberhardB): We don't think we should call GC.Collect().
// If there seems to be a need for it, then we probably need to call
// Dispose() on one of the variables we want disposed.
if (codepoint % 0x100 == 0)
GC.Collect();
}
}
/// ------------------------------------------------------------------------------------
/// <summary>
/// Makes sure that if the specified codepoint is a custom PUA character, that that
/// character is added to the language definition's PUA collection.
/// </summary>
/// ------------------------------------------------------------------------------------
private static void UpdateLangDefPUACollection(int codepoint,
LanguageDefinition langDef, List<PUACharacter> customPuaCharacters)
{
// Go through all the custom defined PUA characters in ICU and make sure that if
// our codepoint is one of them, the language definition contains our codepoint
// in its collection of PUA characters.
foreach (PUACharacter customChar in customPuaCharacters)
{
string sCodePoint = codepoint.ToString("x4").ToUpperInvariant();
// Is our codepoint one of the custom PUA characters? If so,
// add it to the language definition's PUA collection.
if (sCodePoint == customChar.CodePoint)
{
PUACharacter puaChar = new PUACharacter(sCodePoint);
puaChar.RefreshFromIcu(false);
langDef.AddPuaDefinition(codepoint, puaChar.ToString());
}
}
}
/// ------------------------------------------------------------------------------------
/// <summary>
/// Returns a list of the user defined custom characters in ICU by scanning the Unicode data file.
/// </summary>
/// ------------------------------------------------------------------------------------
public static List<PUACharacter> GetDefinedCustomPUACharsFromICU()
{
List<PUACharacter> definedChars = new List<PUACharacter>();
string uniData = DirectoryFinder.GetIcuDirectory; // "c:\\work\\fw60\\distfiles\\Icu40\\icudt40l\\"
uniData = uniData.Substring(0, uniData.LastIndexOf(@"\icudt"));
uniData = Path.Combine(uniData, @"data\unidata\UnicodeData.txt");
StreamReader reader = new StreamReader(uniData, System.Text.Encoding.ASCII);
int chT = reader.Peek(); // force autodetection of encoding.
try
{
string line;
while((line = reader.ReadLine()) != null)
{
// skip entirely blank lines
if(line.Length <= 0)
continue;
if (line.IndexOf("User Added") != -1)
{
PUACharacter puaChar = new PUACharacter("0000");
string codepoint = line.Substring(0, line.IndexOf(';')).Trim();
puaChar.CodePoint = codepoint;
puaChar.RefreshFromIcu(false);
definedChars.Add(puaChar);
}
}
}
finally
{
reader.Close();
}
// REVIEW (TimS/EberhardB): We don't think we should call GC.Collect().
// If there seems to be a need for it, then we probably need to call
// Dispose() on one of the variables we want disposed.
GC.Collect();
return definedChars;
}
/// ------------------------------------------------------------------------------------
/// <summary>
/// Determines whether or not the specified codepoint is defined in ICU. If it is and
/// its a custom PUA character, then make sure the language definition's PUA character
/// collection is updated. Return false if the codepoint is not defined in ICU.
/// </summary>
/// ------------------------------------------------------------------------------------
private static bool IsCodePointDefined(int codepoint)
{
PUACharacter puaChar = new PUACharacter(codepoint);
try
{
// If the definition doesn't exist in ICU or the character has no name then
// it's undefined and shouldn't be allowed in the valid characters list.
if (!puaChar.RefreshFromIcu(false) || puaChar.Name.Length == 0)
return false;
// If the ICU category is one of the "Other" categories, then it's
// undefined and shouldn't be allowed in the valid characters list.
string ucdrep = puaChar.GeneralCategory.UcdRepresentation;
ucdrep = ucdrep.ToUpperInvariant();
// TODO: When tabs are supported, then allow them (they are in category Cc).
// See TE-3004.
if (ucdrep[0] == 'C' && ucdrep[1] != 'F')
return false;
}
catch
{
return false;
}
return true;
}
