public override Encoding GetEncoding(int codepage)
{
if (codepage < 0 || codepage > 65535)
return null;
if (codepage == 0)
{
// Retrieve the system default non-unicode code page if possible, or return null,
// giving the rest of the EncodingProviders a chance to return a default.
int systemDefaultCodePage = SystemDefaultCodePage;
return systemDefaultCodePage != 0 ?
GetEncoding(systemDefaultCodePage) :
null;
}
Encoding result = null;
_cacheLock.EnterUpgradeableReadLock();
try
{
if (_encodings.TryGetValue(codepage, out result))
return result;
int i = BaseCodePageEncoding.GetCodePageByteSize(codepage);
if (i == 1)
{
result = new SBCSCodePageEncoding(codepage);
}
else if (i == 2)
{
result = new DBCSCodePageEncoding(codepage);
}
else
{
result = GetEncodingRare(codepage);
if (result == null)
return null;
}
_cacheLock.EnterWriteLock();
try
{
Encoding cachedEncoding;
if (_encodings.TryGetValue(codepage, out cachedEncoding))
return cachedEncoding;
_encodings.Add(codepage, result);
}
finally
{
_cacheLock.ExitWriteLock();
}
}
finally
{
_cacheLock.ExitUpgradeableReadLock();
}
return result;
}
public static Encoding GetEncoding(int codepage)
{
if ((codepage < 0) || (codepage > 0xffff))
{
throw new ArgumentOutOfRangeException("codepage", Environment.GetResourceString("ArgumentOutOfRange_Range", new object[] { 0, 0xffff }));
}
Encoding unicode = null;
if (encodings != null)
{
unicode = (Encoding) encodings[codepage];
}
if (unicode == null)
{
lock (InternalSyncObject)
{
if (encodings == null)
{
encodings = new Hashtable();
}
unicode = (Encoding) encodings[codepage];
if (unicode != null)
{
return unicode;
}
switch (codepage)
{
case 0:
unicode = Default;
break;
case 1:
case 2:
case 3:
case 0x2a:
throw new ArgumentException(Environment.GetResourceString("Argument_CodepageNotSupported", new object[] { codepage }), "codepage");
case 0x4b0:
unicode = Unicode;
break;
case 0x4b1:
unicode = BigEndianUnicode;
break;
case 0x6faf:
unicode = Latin1;
break;
case 0xfde9:
unicode = UTF8;
break;
case 0x4e4:
unicode = new SBCSCodePageEncoding(codepage);
break;
case 0x4e9f:
unicode = ASCII;
break;
default:
unicode = GetEncodingCodePage(codepage);
if (unicode == null)
{
unicode = GetEncodingRare(codepage);
}
break;
}
encodings.Add(codepage, unicode);
}
}
return unicode;
}
[System.Security.SecurityCritical] // auto-generated
private static Encoding GetEncodingRare(int codepage)
{
Contract.Assert(codepage != 0 && codepage != 1200 && codepage != 1201 && codepage != 65001,
"[Encoding.GetEncodingRare]This code page (" + codepage + ") isn't supported by GetEncodingRare!");
Encoding result;
switch (codepage)
{
case CodePageUTF7: // 65000
result = UTF7;
break;
case CodePageUTF32: // 12000
result = UTF32;
break;
case CodePageUTF32BE: // 12001
result = new UTF32Encoding(true, true);
break;
case ISCIIAssemese:
case ISCIIBengali:
case ISCIIDevanagari:
case ISCIIGujarathi:
case ISCIIKannada:
case ISCIIMalayalam:
case ISCIIOriya:
case ISCIIPanjabi:
case ISCIITamil:
case ISCIITelugu:
result = new ISCIIEncoding(codepage);
break;
// GB2312-80 uses same code page for 20936 and mac 10008
case CodePageMacGB2312:
// case CodePageGB2312:
// result = new DBCSCodePageEncoding(codepage, EUCCN);
result = new DBCSCodePageEncoding(CodePageMacGB2312, CodePageGB2312);
break;
// Mac Korean 10003 and 20949 are the same
case CodePageMacKorean:
result = new DBCSCodePageEncoding(CodePageMacKorean, CodePageDLLKorean);
break;
// GB18030 Code Pages
case GB18030:
result = new GB18030Encoding();
break;
// ISO2022 Code Pages
case ISOKorean:
// case ISOSimplifiedCN
case ChineseHZ:
case ISO2022JP: // JIS JP, full-width Katakana mode (no half-width Katakana)
case ISO2022JPESC: // JIS JP, esc sequence to do Katakana.
case ISO2022JPSISO: // JIS JP with Shift In/ Shift Out Katakana support
result = new ISO2022Encoding(codepage);
break;
// Duplicate EUC-CN (51936) just calls a base code page 936,
// so does ISOSimplifiedCN (50227), which's gotta be broken
case DuplicateEUCCN:
case ISOSimplifiedCN:
result = new DBCSCodePageEncoding(codepage, EUCCN); // Just maps to 936
break;
case EUCJP:
result = new EUCJPEncoding();
break;
case EUCKR:
result = new DBCSCodePageEncoding(codepage, CodePageDLLKorean); // Maps to 20949
break;
case ENC50229:
throw new NotSupportedException(Environment.GetResourceString("NotSupported_CodePage50229"));
case ISO_8859_8I:
result = new SBCSCodePageEncoding(codepage, ISO_8859_8_Visual); // Hebrew maps to a different code page
break;
default:
// Not found, already tried codepage table code pages in GetEncoding()
throw new NotSupportedException(
Environment.GetResourceString("NotSupported_NoCodepageData", codepage));
}
return result;
}
private static Encoding GetEncodingRare(int codepage)
{
Encoding result = null;
switch (codepage)
{
case ISCIIAssemese:
case ISCIIBengali:
case ISCIIDevanagari:
case ISCIIGujarathi:
case ISCIIKannada:
case ISCIIMalayalam:
case ISCIIOriya:
case ISCIIPanjabi:
case ISCIITamil:
case ISCIITelugu:
result = new ISCIIEncoding(codepage);
break;
// GB2312-80 uses same code page for 20936 and mac 10008
case CodePageMacGB2312:
// case CodePageGB2312:
// result = new DBCSCodePageEncoding(codepage, EUCCN);
result = new DBCSCodePageEncoding(CodePageMacGB2312, CodePageGB2312);
break;
// Mac Korean 10003 and 20949 are the same
case CodePageMacKorean:
result = new DBCSCodePageEncoding(CodePageMacKorean, CodePageDLLKorean);
break;
// GB18030 Code Pages
case GB18030:
result = new GB18030Encoding();
break;
// ISO2022 Code Pages
case ISOKorean:
// case ISOSimplifiedCN
case ChineseHZ:
case ISO2022JP: // JIS JP, full-width Katakana mode (no half-width Katakana)
case ISO2022JPESC: // JIS JP, esc sequence to do Katakana.
case ISO2022JPSISO: // JIS JP with Shift In/ Shift Out Katakana support
result = new ISO2022Encoding(codepage);
break;
// Duplicate EUC-CN (51936) just calls a base code page 936,
// so does ISOSimplifiedCN (50227), which's gotta be broken
case DuplicateEUCCN:
case ISOSimplifiedCN:
result = new DBCSCodePageEncoding(codepage, EUCCN); // Just maps to 936
break;
case EUCJP:
result = new EUCJPEncoding();
break;
case EUCKR:
result = new DBCSCodePageEncoding(codepage, CodePageDLLKorean); // Maps to 20949
break;
case ISO_8859_8I:
result = new SBCSCodePageEncoding(codepage, ISO_8859_8_Visual); // Hebrew maps to a different code page
break;
}
return result;
}
[System.Security.SecurityCritical] // auto-generated
private static Encoding CreateDefaultEncoding()
{
Encoding enc;
#if FEATURE_CODEPAGES_FILE
int codePage = Win32Native.GetACP();
// For US English, we can save some startup working set by not calling
// GetEncoding(int codePage) since JITting GetEncoding will force us to load
// all the Encoding classes for ASCII, UTF7 & UTF8, & UnicodeEncoding.
if (codePage == 1252)
enc = new SBCSCodePageEncoding(codePage);
else
enc = GetEncoding(codePage);
#else // FEATURE_CODEPAGES_FILE
// For silverlight we use UTF8 since ANSI isn't available
enc = UTF8;
#endif //FEATURE_CODEPAGES_FILE
return (enc);
}
[System.Security.SecuritySafeCritical] // auto-generated
#endif
public static Encoding GetEncoding(int codepage)
{
Encoding result = EncodingProvider.GetEncodingFromProvider(codepage);
if (result != null)
return result;
//
// NOTE: If you add a new encoding that can be get by codepage, be sure to
// add the corresponding item in EncodingTable.
// Otherwise, the code below will throw exception when trying to call
// EncodingTable.GetDataItem().
//
if (codepage < 0 || codepage > 65535) {
throw new ArgumentOutOfRangeException(
"codepage", Environment.GetResourceString("ArgumentOutOfRange_Range",
0, 65535));
}
Contract.EndContractBlock();
// Our Encoding
// See if we have a hash table with our encoding in it already.
if (encodings != null) {
result = (Encoding)encodings[codepage];
}
if (result == null)
{
// Don't conflict with ourselves
lock (InternalSyncObject)
{
// Need a new hash table
// in case another thread beat us to creating the Dictionary
if (encodings == null) {
encodings = new Hashtable();
}
// Double check that we don't have one in the table (in case another thread beat us here)
if ((result = (Encoding)encodings[codepage]) != null)
return result;
// Special case the commonly used Encoding classes here, then call
// GetEncodingRare to avoid loading classes like MLangCodePageEncoding
// and ASCIIEncoding. ASP.NET uses UTF-8 & ISO-8859-1.
switch (codepage)
{
case CodePageDefault: // 0, default code page
result = Encoding.Default;
break;
case CodePageUnicode: // 1200, Unicode
result = Unicode;
break;
case CodePageBigEndian: // 1201, big endian unicode
result = BigEndianUnicode;
break;
#if FEATURE_CODEPAGES_FILE
case CodePageWindows1252: // 1252, Windows
result = new SBCSCodePageEncoding(codepage);
break;
#else
#if FEATURE_UTF7
// on desktop, UTF7 is handled by GetEncodingRare.
// On Coreclr, we handle this directly without bringing GetEncodingRare, so that we get real UTF-7 encoding.
case CodePageUTF7: // 65000, UTF7
result = UTF7;
break;
#endif
#if FEATURE_UTF32
case CodePageUTF32: // 12000
result = UTF32;
break;
case CodePageUTF32BE: // 12001
result = new UTF32Encoding(true, true);
break;
#endif
#endif
case CodePageUTF8: // 65001, UTF8
result = UTF8;
break;
// These are (hopefully) not very common, but also shouldn't slow us down much and make default
// case able to handle more code pages by calling GetEncodingCodePage
case CodePageNoOEM: // 1
case CodePageNoMac: // 2
case CodePageNoThread: // 3
case CodePageNoSymbol: // 42
// Win32 also allows the following special code page values. We won't allow them except in the
// CP_ACP case.
// #define CP_ACP 0 // default to ANSI code page
// #define CP_OEMCP 1 // default to OEM code page
// #define CP_MACCP 2 // default to MAC code page
// #define CP_THREAD_ACP 3 // current thread's ANSI code page
// #define CP_SYMBOL 42 // SYMBOL translations
throw new ArgumentException(Environment.GetResourceString(
"Argument_CodepageNotSupported", codepage), "codepage");
#if FEATURE_ASCII
// Have to do ASCII and Latin 1 first so they don't get loaded as code pages
case CodePageASCII: // 20127
result = ASCII;
break;
#endif
#if FEATURE_LATIN1
case ISO_8859_1: // 28591
result = Latin1;
break;
#endif
default:
{
#if FEATURE_CODEPAGES_FILE
// 1st assume its a code page.
result = GetEncodingCodePage(codepage);
if (result == null)
result = GetEncodingRare(codepage);
break;
#else
// Is it a valid code page?
if (EncodingTable.GetCodePageDataItem(codepage) == null)
{
throw new NotSupportedException(
Environment.GetResourceString("NotSupported_NoCodepageData", codepage));
}
result = UTF8;
break;
#endif // FEATURE_CODEPAGES_FILE
}
}
encodings.Add(codepage, result);
}
}
return result;
}
public override Encoding GetEncoding(int codepage)
{
if (codepage < 0 || codepage > 65535)
{
return(null);
}
if (codepage == 0)
{
// Retrieve the system default non-unicode code page if possible, or return null,
// giving the rest of the EncodingProviders a chance to return a default.
int systemDefaultCodePage = SystemDefaultCodePage;
return(systemDefaultCodePage != 0 ?
GetEncoding(systemDefaultCodePage) :
null);
}
Encoding result = null;
_cacheLock.EnterUpgradeableReadLock();
try
{
if (_encodings.TryGetValue(codepage, out result))
{
return(result);
}
int i = BaseCodePageEncoding.GetCodePageByteSize(codepage);
if (i == 1)
{
result = new SBCSCodePageEncoding(codepage);
}
else if (i == 2)
{
result = new DBCSCodePageEncoding(codepage);
}
else
{
result = GetEncodingRare(codepage);
if (result == null)
{
return(null);
}
}
_cacheLock.EnterWriteLock();
try
{
Encoding cachedEncoding;
if (_encodings.TryGetValue(codepage, out cachedEncoding))
{
return(cachedEncoding);
}
_encodings.Add(codepage, result);
}
finally
{
_cacheLock.ExitWriteLock();
}
}
finally
{
_cacheLock.ExitUpgradeableReadLock();
}
return(result);
}
private static Encoding GetEncodingRare(int codepage)
{
Encoding result = null;
switch (codepage)
{
case ISCIIAssemese:
case ISCIIBengali:
case ISCIIDevanagari:
case ISCIIGujarathi:
case ISCIIKannada:
case ISCIIMalayalam:
case ISCIIOriya:
case ISCIIPanjabi:
case ISCIITamil:
case ISCIITelugu:
result = new ISCIIEncoding(codepage);
break;
// GB2312-80 uses same code page for 20936 and mac 10008
case CodePageMacGB2312:
// case CodePageGB2312:
// result = new DBCSCodePageEncoding(codepage, EUCCN);
result = new DBCSCodePageEncoding(CodePageMacGB2312, CodePageGB2312);
break;
// Mac Korean 10003 and 20949 are the same
case CodePageMacKorean:
result = new DBCSCodePageEncoding(CodePageMacKorean, CodePageDLLKorean);
break;
// GB18030 Code Pages
case GB18030:
result = new GB18030Encoding();
break;
// ISO2022 Code Pages
case ISOKorean:
// case ISOSimplifiedCN
case ChineseHZ:
case ISO2022JP: // JIS JP, full-width Katakana mode (no half-width Katakana)
case ISO2022JPESC: // JIS JP, esc sequence to do Katakana.
case ISO2022JPSISO: // JIS JP with Shift In/ Shift Out Katakana support
result = new ISO2022Encoding(codepage);
break;
// Duplicate EUC-CN (51936) just calls a base code page 936,
// so does ISOSimplifiedCN (50227), which's gotta be broken
case DuplicateEUCCN:
case ISOSimplifiedCN:
result = new DBCSCodePageEncoding(codepage, EUCCN); // Just maps to 936
break;
case EUCJP:
result = new EUCJPEncoding();
break;
case EUCKR:
result = new DBCSCodePageEncoding(codepage, CodePageDLLKorean); // Maps to 20949
break;
case ISO_8859_8I:
result = new SBCSCodePageEncoding(codepage, ISO_8859_8_Visual); // Hebrew maps to a different code page
break;
}
return(result);
}
public static Encoding GetEncoding(int codepage)
{
Encoding result = EncodingProvider.GetEncodingFromProvider(codepage);
if (result != null)
return result;
//
// NOTE: If you add a new encoding that can be get by codepage, be sure to
// add the corresponding item in EncodingTable.
// Otherwise, the code below will throw exception when trying to call
// EncodingTable.GetDataItem().
//
if (codepage < 0 || codepage > 65535) {
throw new ArgumentOutOfRangeException(
nameof(codepage), Environment.GetResourceString("ArgumentOutOfRange_Range",
0, 65535));
}
Contract.EndContractBlock();
// Our Encoding
// See if the encoding is cached in a static field.
switch (codepage)
{
case CodePageDefault: return Default; // 0
case CodePageUnicode: return Unicode; // 1200
case CodePageBigEndian: return BigEndianUnicode; // 1201
case CodePageUTF32: return UTF32; // 12000
case CodePageUTF32BE: return BigEndianUTF32; // 12001
case CodePageUTF7: return UTF7; // 65000
case CodePageUTF8: return UTF8; // 65001
case CodePageASCII: return ASCII; // 20127
case ISO_8859_1: return Latin1; // 28591
// We don't allow the following special code page values that Win32 allows.
case CodePageNoOEM: // 1 CP_OEMCP
case CodePageNoMac: // 2 CP_MACCP
case CodePageNoThread: // 3 CP_THREAD_ACP
case CodePageNoSymbol: // 42 CP_SYMBOL
throw new ArgumentException(Environment.GetResourceString(
"Argument_CodepageNotSupported", codepage), nameof(codepage));
}
#if FEATURE_CODEPAGES_FILE
object key = codepage; // Box once
// See if we have a hash table with our encoding in it already.
if (encodings != null) {
result = (Encoding)encodings[key];
}
if (result == null)
{
// Don't conflict with ourselves
lock (InternalSyncObject)
{
// Need a new hash table
// in case another thread beat us to creating the Dictionary
if (encodings == null) {
encodings = new Hashtable();
}
// Double check that we don't have one in the table (in case another thread beat us here)
if ((result = (Encoding)encodings[key]) != null)
return result;
if (codepage == CodePageWindows1252)
{
result = new SBCSCodePageEncoding(codepage);
}
else
{
result = GetEncodingCodePage(codepage) ?? GetEncodingRare(codepage);
}
Debug.Assert(result != null, "result != null");
encodings.Add(key, result);
}
}
return result;
#else
// Is it a valid code page?
if (EncodingTable.GetCodePageDataItem(codepage) == null)
{
throw new NotSupportedException(
Environment.GetResourceString("NotSupported_NoCodepageData", codepage));
}
return UTF8;
#endif // FEATURE_CODEPAGES_FILE
}
private static Encoding CreateDefaultEncoding()
{
// defaultEncoding should be null if we get here, but we can't
// assert that in case another thread beat us to the initialization
Encoding enc;
#if FEATURE_CODEPAGES_FILE
int codePage = Win32Native.GetACP();
// For US English, we can save some startup working set by not calling
// GetEncoding(int codePage) since JITting GetEncoding will force us to load
// all the Encoding classes for ASCII, UTF7 & UTF8, & UnicodeEncoding.
if (codePage == 1252)
enc = new SBCSCodePageEncoding(codePage);
else
enc = GetEncoding(codePage);
#else // FEATURE_CODEPAGES_FILE
// For silverlight we use UTF8 since ANSI isn't available
enc = UTF8;
#endif // FEATURE_CODEPAGES_FILE
// This method should only ever return one Encoding instance
return Interlocked.CompareExchange(ref defaultEncoding, enc, null) ?? enc;
}
private static Encoding CreateDefaultEncoding()
{
Encoding enc;
int codePage = Win32Native.GetACP();
// For US English, we can save some startup working set by not calling
// GetEncoding(int codePage) since JITting GetEncoding will force us to load
// all the Encoding classes for ASCII, UTF7 & UTF8, & UnicodeEncoding.
if (codePage == 1252)
enc = new SBCSCodePageEncoding(codePage);
else
enc = GetEncoding(codePage);
return (enc);
}
