// Where stringToEscape is intended to be URI data, but not an entire URI. // This method will escape any character that is not an unreserved character, including percent signs. public static string EscapeDataString(string stringToEscape) => UriHelper.EscapeString(stringToEscape, checkExistingEscaped: false, UriHelper.UnreservedTable);
//
// Resolves into either baseUri or relativeUri according to conditions OR if not possible it uses newUriString
// to return combined URI strings from both Uris
// otherwise if e != null on output the operation has failed
//
internal static Uri?ResolveHelper(Uri baseUri, Uri?relativeUri, ref string?newUriString, ref bool userEscaped)
{
Debug.Assert(!baseUri.IsNotAbsoluteUri && !baseUri.UserDrivenParsing, "Uri::ResolveHelper()|baseUri is not Absolute or is controlled by User Parser.");
string relativeStr;
if ((object?)relativeUri != null)
{
if (relativeUri.IsAbsoluteUri)
{
return(relativeUri);
}
relativeStr = relativeUri.OriginalString;
userEscaped = relativeUri.UserEscaped;
}
else
{
relativeStr = string.Empty;
}
// Here we can assert that passed "relativeUri" is indeed a relative one
if (relativeStr.Length > 0 && (UriHelper.IsLWS(relativeStr[0]) || UriHelper.IsLWS(relativeStr[relativeStr.Length - 1])))
{
relativeStr = relativeStr.Trim(UriHelper.s_WSchars);
}
if (relativeStr.Length == 0)
{
newUriString = baseUri.GetParts(UriComponents.AbsoluteUri,
baseUri.UserEscaped ? UriFormat.UriEscaped : UriFormat.SafeUnescaped);
return(null);
}
// Check for a simple fragment in relative part
if (relativeStr[0] == '#' && !baseUri.IsImplicitFile && baseUri.Syntax !.InFact(UriSyntaxFlags.MayHaveFragment))
{
newUriString = baseUri.GetParts(UriComponents.AbsoluteUri & ~UriComponents.Fragment,
UriFormat.UriEscaped) + relativeStr;
return(null);
}
// Check for a simple query in relative part
if (relativeStr[0] == '?' && !baseUri.IsImplicitFile && baseUri.Syntax !.InFact(UriSyntaxFlags.MayHaveQuery))
{
newUriString = baseUri.GetParts(UriComponents.AbsoluteUri & ~UriComponents.Query & ~UriComponents.Fragment,
UriFormat.UriEscaped) + relativeStr;
return(null);
}
// Check on the DOS path in the relative Uri (a special case)
if (relativeStr.Length >= 3 &&
(relativeStr[1] == ':' || relativeStr[1] == '|') &&
UriHelper.IsAsciiLetter(relativeStr[0]) &&
(relativeStr[2] == '\\' || relativeStr[2] == '/'))
{
if (baseUri.IsImplicitFile)
{
// It could have file:/// prepended to the result but we want to keep it as *Implicit* File Uri
newUriString = relativeStr;
return(null);
}
else if (baseUri.Syntax !.InFact(UriSyntaxFlags.AllowDOSPath))
{
// The scheme is not changed just the path gets replaced
string prefix;
if (baseUri.InFact(Flags.AuthorityFound))
{
prefix = baseUri.Syntax.InFact(UriSyntaxFlags.PathIsRooted) ? ":///" : "://";
}
else
{
prefix = baseUri.Syntax.InFact(UriSyntaxFlags.PathIsRooted) ? ":/" : ":";
}
newUriString = baseUri.Scheme + prefix + relativeStr;
return(null);
}
// If we are here then input like "http://host/path/" + "C:\x" will produce the result http://host/path/c:/x
}
GetCombinedString(baseUri, relativeStr, userEscaped, ref newUriString);
if (ReferenceEquals(newUriString, baseUri._string))
{
return(baseUri);
}
return(null);
}
//
// Need to check for invalid utf sequences that may not have given any chars.
// We got the unescaped chars, we then re-encode them and match off the bytes
// to get the invalid sequence bytes that we just copy off
//
internal static unsafe void MatchUTF8Sequence(char *pDest, char[] dest, ref int destOffset, Span <char> unescapedChars,
int charCount, byte[] bytes, int byteCount, bool isQuery, bool iriParsing)
{
Span <byte> maxUtf8EncodedSpan = stackalloc byte[4];
int count = 0;
fixed(char *unescapedCharsPtr = unescapedChars)
{
for (int j = 0; j < charCount; ++j)
{
bool isHighSurr = char.IsHighSurrogate(unescapedCharsPtr[j]);
Span <byte> encodedBytes = maxUtf8EncodedSpan;
int bytesWritten = Encoding.UTF8.GetBytes(unescapedChars.Slice(j, isHighSurr ? 2 : 1), encodedBytes);
encodedBytes = encodedBytes.Slice(0, bytesWritten);
// we have to keep unicode chars outside Iri range escaped
bool inIriRange = false;
if (iriParsing)
{
if (!isHighSurr)
{
inIriRange = IriHelper.CheckIriUnicodeRange(unescapedChars[j], isQuery);
}
else
{
bool surrPair = false;
inIriRange = IriHelper.CheckIriUnicodeRange(unescapedChars[j], unescapedChars[j + 1],
ref surrPair, isQuery);
}
}
while (true)
{
// Escape any invalid bytes that were before this character
while (bytes[count] != encodedBytes[0])
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
EscapeAsciiChar((char)bytes[count++], dest, ref destOffset);
}
// check if all bytes match
bool allBytesMatch = true;
int k = 0;
for (; k < encodedBytes.Length; ++k)
{
if (bytes[count + k] != encodedBytes[k])
{
allBytesMatch = false;
break;
}
}
if (allBytesMatch)
{
count += encodedBytes.Length;
if (iriParsing)
{
if (!inIriRange)
{
// need to keep chars not allowed as escaped
for (int l = 0; l < encodedBytes.Length; ++l)
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
EscapeAsciiChar((char)encodedBytes[l], dest, ref destOffset);
}
}
else if (!UriHelper.IsBidiControlCharacter(unescapedCharsPtr[j]) || !UriParser.DontKeepUnicodeBidiFormattingCharacters)
{
//copy chars
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = unescapedCharsPtr[j];
if (isHighSurr)
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = unescapedCharsPtr[j + 1];
}
}
}
else
{
//copy chars
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = unescapedCharsPtr[j];
if (isHighSurr)
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = unescapedCharsPtr[j + 1];
}
}
break; // break out of while (true) since we've matched this char bytes
}
else
{
// copy bytes till place where bytes don't match
for (int l = 0; l < k; ++l)
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
EscapeAsciiChar((char)bytes[count++], dest, ref destOffset);
}
}
}
if (isHighSurr)
{
j++;
}
}
}
// Include any trailing invalid sequences
while (count < byteCount)
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
EscapeAsciiChar((char)bytes[count++], dest, ref destOffset);
}
}
internal unsafe static string UnicodeEquivalent(char *hostname, int start, int end, ref bool allAscii, ref bool atLeastOneValidIdn)
{
IdnMapping map = new IdnMapping();
// hostname already validated
allAscii = true;
atLeastOneValidIdn = false;
string idn = null;
if (end <= start)
{
return(idn);
}
string unescapedHostname = UriHelper.StripBidiControlCharacter(hostname, start, (end - start));
string unicodeEqvlHost = null;
int curPos = 0;
int newPos = 0;
int length = unescapedHostname.Length;
bool asciiLabel = true;
bool foundAce = false;
bool checkedAce = false;
bool foundDot = false;
// We run a loop where for every label
// a) if label is ascii and no ace then we lowercase it
// b) if label is ascii and ace and not valid idn then just lowercase it
// c) if label is ascii and ace and is valid idn then get its unicode eqvl
// d) if label is unicode then clean it by running it through idnmapping
do
{
asciiLabel = true;
foundAce = false;
checkedAce = false;
foundDot = false;
//find the dot or hit the end
newPos = curPos;
while (newPos < length)
{
char c = unescapedHostname[newPos];
if (!checkedAce)
{
checkedAce = true;
if ((newPos + 3 < length) && (c == 'x') && IsIdnAce(unescapedHostname, newPos))
{
foundAce = true;
}
}
if (asciiLabel && (c > '\x7F'))
{
asciiLabel = false;
allAscii = false;
}
if ((c == '.') || (c == '\u3002') || //IDEOGRAPHIC FULL STOP
(c == '\uFF0E') || //FULLWIDTH FULL STOP
(c == '\uFF61')) //HALFWIDTH IDEOGRAPHIC FULL STOP
{
foundDot = true;
break;
}
++newPos;
}
if (!asciiLabel)
{
string asciiForm = unescapedHostname.Substring(curPos, newPos - curPos);
try
{
asciiForm = map.GetAscii(asciiForm);
}
catch (ArgumentException)
{
throw new UriFormatException(SR.net_uri_BadUnicodeHostForIdn);
}
unicodeEqvlHost += map.GetUnicode(asciiForm);
if (foundDot)
{
unicodeEqvlHost += ".";
}
}
else
{
bool aceValid = false;
if (foundAce)
{
// check ace validity
try
{
unicodeEqvlHost += map.GetUnicode(unescapedHostname.Substring(curPos, newPos - curPos));
if (foundDot)
{
unicodeEqvlHost += ".";
}
aceValid = true;
atLeastOneValidIdn = true;
}
catch (ArgumentException)
{
// not valid ace so treat it as a normal ascii label
}
}
if (!aceValid)
{
// for invalid aces we just lowercase the label
unicodeEqvlHost += unescapedHostname.Substring(curPos, newPos - curPos).ToLowerInvariant();
if (foundDot)
{
unicodeEqvlHost += ".";
}
}
}
curPos = newPos + (foundDot ? 1 : 0);
} while (curPos < length);
return(unicodeEqvlHost);
}
public static unsafe int UnescapePercentEncodedUTF8Sequence(char *input, int length, ref ValueStringBuilder dest, bool isQuery, bool iriParsing)
{
// The following assertions rely on the input not mutating mid-operation, as is the case currently since callers are working with strings
// If we start accepting input such as spans, this method must be audited to ensure no buffer overruns/infinite loops could occur
// As an optimization, this method should only be called after the first character is known to be a part of a non-ascii UTF8 sequence
Debug.Assert(length >= 3);
Debug.Assert(input[0] == '%');
Debug.Assert(UriHelper.DecodeHexChars(input[1], input[2]) != Uri.c_DummyChar);
Debug.Assert(UriHelper.DecodeHexChars(input[1], input[2]) >= 128);
uint fourByteBuffer = 0;
int bytesLeftInBuffer = 0;
int totalCharsConsumed = 0;
int charsToCopy = 0;
int bytesConsumed = 0;
RefillBuffer:
int i = totalCharsConsumed + (bytesLeftInBuffer * 3);
ReadByteFromInput:
if ((uint)(length - i) <= 2 || input[i] != '%')
{
goto NoMoreOrInvalidInput;
}
uint value = input[i + 1];
if ((uint)((value - 'A') & ~0x20) <= ('F' - 'A'))
{
value = (value | 0x20) - 'a' + 10;
}
else if ((value - '8') <= ('9' - '8'))
{
value -= '0';
}
else
{
goto NoMoreOrInvalidInput; // First character wasn't hex or was <= 7F (Ascii)
}
uint second = (uint)input[i + 2] - '0';
if (second <= 9)
{
// second is already [0, 9]
}
else if ((uint)((second - ('A' - '0')) & ~0x20) <= ('F' - 'A'))
{
second = ((second + '0') | 0x20) - 'a' + 10;
}
else
{
goto NoMoreOrInvalidInput; // Second character wasn't Hex
}
value = (value << 4) | second;
Debug.Assert(value >= 128);
// Rotate the buffer and overwrite the last byte
if (BitConverter.IsLittleEndian)
{
fourByteBuffer = (fourByteBuffer >> 8) | (value << 24);
}
else
{
fourByteBuffer = (fourByteBuffer << 8) | value;
}
if (++bytesLeftInBuffer != 4)
{
i += 3;
goto ReadByteFromInput;
}
DecodeRune:
Debug.Assert(totalCharsConsumed % 3 == 0);
Debug.Assert(bytesLeftInBuffer == 2 || bytesLeftInBuffer == 3 || bytesLeftInBuffer == 4);
Debug.Assert((fourByteBuffer & (BitConverter.IsLittleEndian ? 0x00000080 : 0x80000000)) != 0);
Debug.Assert((fourByteBuffer & (BitConverter.IsLittleEndian ? 0x00008000 : 0x00800000)) != 0);
Debug.Assert(bytesLeftInBuffer < 3 || (fourByteBuffer & (BitConverter.IsLittleEndian ? 0x00800000 : 0x00008000)) != 0);
Debug.Assert(bytesLeftInBuffer < 4 || (fourByteBuffer & (BitConverter.IsLittleEndian ? 0x80000000 : 0x00000080)) != 0);
uint temp = fourByteBuffer; // make a copy so that the *copy* (not the original) is marked address-taken
if (Rune.DecodeFromUtf8(new ReadOnlySpan <byte>(&temp, bytesLeftInBuffer), out Rune rune, out bytesConsumed) == OperationStatus.Done)
{
Debug.Assert(bytesConsumed >= 2, $"Rune.DecodeFromUtf8 consumed {bytesConsumed} bytes, likely indicating input was modified concurrently during UnescapePercentEncodedUTF8Sequence's execution");
if (!iriParsing || IriHelper.CheckIriUnicodeRange((uint)rune.Value, isQuery))
{
if (charsToCopy != 0)
{
dest.Append(input + totalCharsConsumed - charsToCopy, charsToCopy);
charsToCopy = 0;
}
dest.Append(rune);
goto AfterDecodeRune;
}
}
//
// IRI normalization for strings containing characters that are not allowed or
// escaped characters that should be unescaped in the context of the specified Uri component.
//
internal static unsafe string EscapeUnescapeIri(char *pInput, int start, int end, UriComponents component)
{
int size = end - start;
ValueStringBuilder dest = new ValueStringBuilder(size);
byte[]? bytes = null;
int next = start;
char ch;
for (; next < end; ++next)
{
if ((ch = pInput[next]) == '%')
{
if (next + 2 < end)
{
ch = UriHelper.EscapedAscii(pInput[next + 1], pInput[next + 2]);
// Do not unescape a reserved char
if (ch == Uri.c_DummyChar || ch == '%' || CheckIsReserved(ch, component) || UriHelper.IsNotSafeForUnescape(ch))
{
// keep as is
dest.Append(pInput[next++]);
dest.Append(pInput[next++]);
dest.Append(pInput[next]);
continue;
}
else if (ch <= '\x7F')
{
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
//ASCII
dest.Append(ch);
next += 2;
continue;
}
else
{
// possibly utf8 encoded sequence of unicode
// check if safe to unescape according to Iri rules
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
int startSeq = next;
int byteCount = 1;
// lazy initialization of max size, will reuse the array for next sequences
if ((object?)bytes == null)
{
bytes = new byte[end - next];
}
bytes[0] = (byte)ch;
next += 3;
while (next < end)
{
// Check on exit criterion
if ((ch = pInput[next]) != '%' || next + 2 >= end)
{
break;
}
// already made sure we have 3 characters in str
ch = UriHelper.EscapedAscii(pInput[next + 1], pInput[next + 2]);
//invalid hex sequence ?
if (ch == Uri.c_DummyChar)
{
break;
}
// character is not part of a UTF-8 sequence ?
else if (ch < '\x80')
{
break;
}
else
{
//a UTF-8 sequence
bytes[byteCount++] = (byte)ch;
next += 3;
}
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
}
next--; // for loop will increment
// Using encoder with no replacement fall-back will skip all invalid UTF-8 sequences.
Encoding noFallbackCharUTF8 = Encoding.GetEncoding(
Encoding.UTF8.CodePage,
new EncoderReplacementFallback(""),
new DecoderReplacementFallback(""));
char[] unescapedChars = new char[bytes.Length];
int charCount = noFallbackCharUTF8.GetChars(bytes, 0, byteCount, unescapedChars, 0);
if (charCount != 0)
{
// If invalid sequences were present in the original escaped string, we need to
// copy the escaped versions of those sequences.
// Decoded Unicode values will be kept only when they are allowed by the URI/IRI RFC
// rules.
UriHelper.MatchUTF8Sequence(ref dest, unescapedChars, charCount, bytes,
byteCount, component == UriComponents.Query, true);
}
else
{
// copy escaped sequence as is
for (int i = startSeq; i <= next; ++i)
{
dest.Append(pInput[i]);
}
}
}
}
else
{
dest.Append(pInput[next]);
}
}
else if (ch > '\x7f')
{
// unicode
bool escape;
bool surrogatePair = false;
char ch2 = '\0';
if ((char.IsHighSurrogate(ch)) && (next + 1 < end))
{
ch2 = pInput[next + 1];
escape = !CheckIriUnicodeRange(ch, ch2, ref surrogatePair, component == UriComponents.Query);
}
else
{
escape = !CheckIriUnicodeRange(ch, component == UriComponents.Query);
}
if (escape)
{
Span <byte> encodedBytes = stackalloc byte[4];
Rune rune;
if (surrogatePair)
{
rune = new Rune(ch, ch2);
}
else if (!Rune.TryCreate(ch, out rune))
{
rune = Rune.ReplacementChar;
}
int bytesWritten = rune.EncodeToUtf8(encodedBytes);
encodedBytes = encodedBytes.Slice(0, bytesWritten);
foreach (byte b in encodedBytes)
{
UriHelper.EscapeAsciiChar(b, ref dest);
}
}
else
{
dest.Append(ch);
if (surrogatePair)
{
dest.Append(ch2);
}
}
if (surrogatePair)
{
next++;
}
}
else
{
// just copy the character
dest.Append(pInput[next]);
}
}
string result = dest.ToString();
return(result);
}
//
// Check reserved chars according to RFC 3987 in a specific component
//
internal static bool CheckIsReserved(char ch, UriComponents component)
{
if ((component != UriComponents.Scheme) &&
(component != UriComponents.UserInfo) &&
(component != UriComponents.Host) &&
(component != UriComponents.Port) &&
(component != UriComponents.Path) &&
(component != UriComponents.Query) &&
(component != UriComponents.Fragment)
)
{
return((component == (UriComponents)0) ? UriHelper.IsGenDelim(ch) : false);
}
else if (UriParser.DontEnableStrictRFC3986ReservedCharacterSets)
{
// Since we aren't enabling strict RFC 3986 reserved sets, we stick with the old behavior
// (for app-compat) which was a broken mix of RFCs 2396 and 3986.
switch (component)
{
case UriComponents.UserInfo:
if (ch == '/' || ch == '?' || ch == '#' || ch == '[' || ch == ']' || ch == '@')
{
return(true);
}
break;
case UriComponents.Host:
if (ch == ':' || ch == '/' || ch == '?' || ch == '#' || ch == '[' || ch == ']' || ch == '@')
{
return(true);
}
break;
case UriComponents.Path:
if (ch == '/' || ch == '?' || ch == '#' || ch == '[' || ch == ']')
{
return(true);
}
break;
case UriComponents.Query:
if (ch == '#' || ch == '[' || ch == ']')
{
return(true);
}
break;
case UriComponents.Fragment:
if (ch == '#' || ch == '[' || ch == ']')
{
return(true);
}
break;
default:
break;
}
return(false);
}
else
{
return(UriHelper.RFC3986ReservedMarks.IndexOf(ch) >= 0);
}
}
//
// IRI normalization for strings containing characters that are not allowed or
// escaped characters that should be unescaped in the context of the specified Uri component.
//
internal static unsafe string EscapeUnescapeIri(char *pInput, int start, int end, UriComponents component)
{
char[] dest = new char[end - start];
byte[] bytes = null;
// Pin the array to do pointer accesses
GCHandle destHandle = GCHandle.Alloc(dest, GCHandleType.Pinned);
char * pDest = (char *)destHandle.AddrOfPinnedObject();
const int percentEncodingLen = 3; // Escaped UTF-8 will take 3 chars: %AB.
const int bufferCapacityIncrease = 30 * percentEncodingLen;
int bufferRemaining = 0;
int next = start;
int destOffset = 0;
char ch;
bool escape = false;
bool surrogatePair = false;
for (; next < end; ++next)
{
escape = false;
surrogatePair = false;
if ((ch = pInput[next]) == '%')
{
if (next + 2 < end)
{
ch = UriHelper.EscapedAscii(pInput[next + 1], pInput[next + 2]);
// Do not unescape a reserved char
if (ch == Uri.c_DummyChar || ch == '%' || CheckIsReserved(ch, component) || UriHelper.IsNotSafeForUnescape(ch))
{
// keep as is
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next++];
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next++];
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next];
continue;
}
else if (ch <= '\x7F')
{
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
//ASCII
pDest[destOffset++] = ch;
next += 2;
continue;
}
else
{
// possibly utf8 encoded sequence of unicode
// check if safe to unescape according to Iri rules
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
int startSeq = next;
int byteCount = 1;
// lazy initialization of max size, will reuse the array for next sequences
if ((object)bytes == null)
{
bytes = new byte[end - next];
}
bytes[0] = (byte)ch;
next += 3;
while (next < end)
{
// Check on exit criterion
if ((ch = pInput[next]) != '%' || next + 2 >= end)
{
break;
}
// already made sure we have 3 characters in str
ch = UriHelper.EscapedAscii(pInput[next + 1], pInput[next + 2]);
//invalid hex sequence ?
if (ch == Uri.c_DummyChar)
{
break;
}
// character is not part of a UTF-8 sequence ?
else if (ch < '\x80')
{
break;
}
else
{
//a UTF-8 sequence
bytes[byteCount++] = (byte)ch;
next += 3;
}
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
}
next--; // for loop will increment
// Using encoder with no replacement fall-back will skip all invalid UTF-8 sequences.
Encoding noFallbackCharUTF8 = Encoding.GetEncoding(
Encoding.UTF8.CodePage,
new EncoderReplacementFallback(""),
new DecoderReplacementFallback(""));
char[] unescapedChars = new char[bytes.Length];
int charCount = noFallbackCharUTF8.GetChars(bytes, 0, byteCount, unescapedChars, 0);
if (charCount != 0)
{
// If invalid sequences were present in the original escaped string, we need to
// copy the escaped versions of those sequences.
// Decoded Unicode values will be kept only when they are allowed by the URI/IRI RFC
// rules.
UriHelper.MatchUTF8Sequence(pDest, dest, ref destOffset, unescapedChars, charCount, bytes,
byteCount, component == UriComponents.Query, true);
}
else
{
// copy escaped sequence as is
for (int i = startSeq; i <= next; ++i)
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[i];
}
}
}
}
else
{
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next];
}
}
else if (ch > '\x7f')
{
// unicode
char ch2;
if ((char.IsHighSurrogate(ch)) && (next + 1 < end))
{
ch2 = pInput[next + 1];
escape = !CheckIriUnicodeRange(ch, ch2, ref surrogatePair, component == UriComponents.Query);
if (!escape)
{
// copy the two chars
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next++];
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next];
}
}
else
{
if (CheckIriUnicodeRange(ch, component == UriComponents.Query))
{
if (!UriHelper.IsBidiControlCharacter(ch) || !UriParser.DontKeepUnicodeBidiFormattingCharacters)
{
// copy it
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next];
}
}
else
{
// escape it
escape = true;
}
}
}
else
{
// just copy the character
Debug.Assert(dest.Length > destOffset, "Destination length exceeded destination offset.");
pDest[destOffset++] = pInput[next];
}
if (escape)
{
const int MaxNumberOfBytesEncoded = 4;
if (bufferRemaining < MaxNumberOfBytesEncoded * percentEncodingLen)
{
int newBufferLength = 0;
checked
{
// may need more memory since we didn't anticipate escaping
newBufferLength = dest.Length + bufferCapacityIncrease;
bufferRemaining += bufferCapacityIncrease;
}
char[] newDest = new char[newBufferLength];
fixed(char *pNewDest = newDest)
{
Buffer.MemoryCopy((byte *)pDest, (byte *)pNewDest, newBufferLength * sizeof(char), destOffset * sizeof(char));
}
if (destHandle.IsAllocated)
{
destHandle.Free();
}
dest = newDest;
// re-pin new dest[] array
destHandle = GCHandle.Alloc(dest, GCHandleType.Pinned);
pDest = (char *)destHandle.AddrOfPinnedObject();
}
byte[] encodedBytes = new byte[MaxNumberOfBytesEncoded];
fixed(byte *pEncodedBytes = &encodedBytes[0])
{
int encodedBytesCount = Encoding.UTF8.GetBytes(pInput + next, surrogatePair ? 2 : 1, pEncodedBytes, MaxNumberOfBytesEncoded);
Debug.Assert(encodedBytesCount <= MaxNumberOfBytesEncoded, "UTF8 encoder should not exceed specified byteCount");
bufferRemaining -= encodedBytesCount * percentEncodingLen;
for (int count = 0; count < encodedBytesCount; ++count)
{
UriHelper.EscapeAsciiChar((char)encodedBytes[count], dest, ref destOffset);
}
}
}
}
if (destHandle.IsAllocated)
{
destHandle.Free();
}
Debug.Assert(destOffset <= dest.Length, "Destination length met or exceeded destination offset.");
return(new string(dest, 0, destOffset));
}
//
// IRI normalization for strings containing characters that are not allowed or
// escaped characters that should be unescaped in the context of the specified Uri component.
//
internal static unsafe string EscapeUnescapeIri(char *pInput, int start, int end, UriComponents component)
{
int size = end - start;
ValueStringBuilder dest = new ValueStringBuilder(size);
byte[]? bytes = null;
const int percentEncodingLen = 3; // Escaped UTF-8 will take 3 chars: %AB.
int bufferRemaining = 0;
int next = start;
char ch;
bool escape = false;
bool surrogatePair = false;
for (; next < end; ++next)
{
escape = false;
surrogatePair = false;
if ((ch = pInput[next]) == '%')
{
if (next + 2 < end)
{
ch = UriHelper.EscapedAscii(pInput[next + 1], pInput[next + 2]);
// Do not unescape a reserved char
if (ch == Uri.c_DummyChar || ch == '%' || CheckIsReserved(ch, component) || UriHelper.IsNotSafeForUnescape(ch))
{
// keep as is
dest.Append(pInput[next++]);
dest.Append(pInput[next++]);
dest.Append(pInput[next]);
continue;
}
else if (ch <= '\x7F')
{
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
//ASCII
dest.Append(ch);
next += 2;
continue;
}
else
{
// possibly utf8 encoded sequence of unicode
// check if safe to unescape according to Iri rules
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
int startSeq = next;
int byteCount = 1;
// lazy initialization of max size, will reuse the array for next sequences
if ((object?)bytes == null)
{
bytes = new byte[end - next];
}
bytes[0] = (byte)ch;
next += 3;
while (next < end)
{
// Check on exit criterion
if ((ch = pInput[next]) != '%' || next + 2 >= end)
{
break;
}
// already made sure we have 3 characters in str
ch = UriHelper.EscapedAscii(pInput[next + 1], pInput[next + 2]);
//invalid hex sequence ?
if (ch == Uri.c_DummyChar)
{
break;
}
// character is not part of a UTF-8 sequence ?
else if (ch < '\x80')
{
break;
}
else
{
//a UTF-8 sequence
bytes[byteCount++] = (byte)ch;
next += 3;
}
Debug.Assert(ch < 0xFF, "Expecting ASCII character.");
}
next--; // for loop will increment
// Using encoder with no replacement fall-back will skip all invalid UTF-8 sequences.
Encoding noFallbackCharUTF8 = Encoding.GetEncoding(
Encoding.UTF8.CodePage,
new EncoderReplacementFallback(""),
new DecoderReplacementFallback(""));
char[] unescapedChars = new char[bytes.Length];
int charCount = noFallbackCharUTF8.GetChars(bytes, 0, byteCount, unescapedChars, 0);
if (charCount != 0)
{
// If invalid sequences were present in the original escaped string, we need to
// copy the escaped versions of those sequences.
// Decoded Unicode values will be kept only when they are allowed by the URI/IRI RFC
// rules.
UriHelper.MatchUTF8Sequence(ref dest, unescapedChars, charCount, bytes,
byteCount, component == UriComponents.Query, true);
}
else
{
// copy escaped sequence as is
for (int i = startSeq; i <= next; ++i)
{
dest.Append(pInput[i]);
}
}
}
}
else
{
dest.Append(pInput[next]);
}
}
else if (ch > '\x7f')
{
// unicode
char ch2;
if ((char.IsHighSurrogate(ch)) && (next + 1 < end))
{
ch2 = pInput[next + 1];
escape = !CheckIriUnicodeRange(ch, ch2, ref surrogatePair, component == UriComponents.Query);
if (!escape)
{
// copy the two chars
dest.Append(pInput[next++]);
dest.Append(pInput[next]);
}
}
else
{
if (CheckIriUnicodeRange(ch, component == UriComponents.Query))
{
// copy it
dest.Append(pInput[next]);
}
else
{
// escape it
escape = true;
}
}
}
else
{
// just copy the character
dest.Append(pInput[next]);
}
if (escape)
{
const int MaxNumberOfBytesEncoded = 4;
byte[] encodedBytes = new byte[MaxNumberOfBytesEncoded];
fixed(byte *pEncodedBytes = &encodedBytes[0])
{
int encodedBytesCount = Encoding.UTF8.GetBytes(pInput + next, surrogatePair ? 2 : 1, pEncodedBytes, MaxNumberOfBytesEncoded);
Debug.Assert(encodedBytesCount <= MaxNumberOfBytesEncoded, "UTF8 encoder should not exceed specified byteCount");
bufferRemaining -= encodedBytesCount * percentEncodingLen;
for (int count = 0; count < encodedBytesCount; ++count)
{
UriHelper.EscapeAsciiChar((char)encodedBytes[count], ref dest);
}
}
}
}
string result = dest.ToString();
return(result);
}
//
// Parse
//
// Convert this IPv6 address into a sequence of 8 16-bit numbers
//
// Inputs:
// <member> Name
// The validated IPv6 address
//
// Outputs:
// <member> numbers
// Array filled in with the numbers in the IPv6 groups
//
// <member> PrefixLength
// Set to the number after the prefix separator (/) if found
//
// Assumes:
// <Name> has been validated and contains only hex digits in groups of
// 16-bit numbers, the characters ':' and '/', and a possible IPv4
// address
//
// Returns:
// true if this is a loopback, false otherwise. There is no failure indication as the sting must be a valid one.
//
// Throws:
// Nothing
//
unsafe internal static bool Parse(string address, ushort *numbers, int start, ref string scopeId)
{
int number = 0;
int index = 0;
int compressorIndex = -1;
bool numberIsValid = true;
//This used to be a class instance member but have not been used so far
int PrefixLength = 0;
if (address[start] == '[')
{
++start;
}
for (int i = start; i < address.Length && address[i] != ']';)
{
switch (address[i])
{
case '%':
if (numberIsValid)
{
numbers[index++] = (ushort)number;
numberIsValid = false;
}
start = i;
for (++i; address[i] != ']' && address[i] != '/'; ++i)
{
;
}
scopeId = address.Substring(start, i - start);
// ignore prefix if any
for (; address[i] != ']'; ++i)
{
;
}
break;
case ':':
numbers[index++] = (ushort)number;
number = 0;
++i;
if (address[i] == ':')
{
compressorIndex = index;
++i;
}
else if ((compressorIndex < 0) && (index < 6))
{
//
// no point checking for IPv4 address if we don't
// have a compressor or we haven't seen 6 16-bit
// numbers yet
//
break;
}
//
// check to see if the upcoming number is really an IPv4
// address. If it is, convert it to 2 ushort numbers
//
for (int j = i; (address[j] != ']') &&
(address[j] != ':') &&
(address[j] != '%') &&
(address[j] != '/') &&
(j < i + 4); ++j)
{
if (address[j] == '.')
{
//
// we have an IPv4 address. Find the end of it:
// we know that since we have a valid IPv6
// address, the only things that will terminate
// the IPv4 address are the prefix delimiter '/'
// or the end-of-string (which we conveniently
// delimited with ']')
//
while ((address[j] != ']') && (address[j] != '/') && (address[j] != '%'))
{
++j;
}
number = IPv4AddressHelper.ParseHostNumber(address, i, j);
numbers[index++] = (ushort)(number >> 16);
numbers[index++] = (ushort)number;
i = j;
//
// set this to avoid adding another number to
// the array if there's a prefix
//
number = 0;
numberIsValid = false;
break;
}
}
break;
case '/':
if (numberIsValid)
{
numbers[index++] = (ushort)number;
numberIsValid = false;
}
//
// since we have a valid IPv6 address string, the prefix
// length is the last token in the string
//
for (++i; address[i] != ']'; ++i)
{
PrefixLength = PrefixLength * 10 + (address[i] - '0');
}
break;
default:
number = number * 16 + UriHelper.FromHex(address[i++]);
break;
}
}
//
// add number to the array if its not the prefix length or part of
// an IPv4 address that's already been handled
//
if (numberIsValid)
{
numbers[index++] = (ushort)number;
}
//
// if we had a compressor sequence ("::") then we need to expand the
// numbers array
//
if (compressorIndex > 0)
{
int toIndex = NumberOfLabels - 1;
int fromIndex = index - 1;
for (int i = index - compressorIndex; i > 0; --i)
{
numbers[toIndex--] = numbers[fromIndex];
numbers[fromIndex--] = 0;
}
}
//
// is the address loopback? Loopback is defined as one of:
//
// 0:0:0:0:0:0:0:1
// 0:0:0:0:0:0:127.0.0.1 == 0:0:0:0:0:0:7F00:0001
// 0:0:0:0:0:FFFF:127.0.0.1 == 0:0:0:0:0:FFFF:7F00:0001
//
return(((numbers[0] == 0) &&
(numbers[1] == 0) &&
(numbers[2] == 0) &&
(numbers[3] == 0) &&
(numbers[4] == 0)) &&
(((numbers[5] == 0) &&
(numbers[6] == 0) &&
(numbers[7] == 1)) ||
(((numbers[6] == 0x7F00) &&
(numbers[7] == 0x0001)) &&
((numbers[5] == 0) ||
(numbers[5] == 0xFFFF)))));
}
//
// InternalIsValid
//
// Determine whether a name is a valid IPv6 address. Rules are:
//
// * 8 groups of 16-bit hex numbers, separated by ':'
// * a *single* run of zeros can be compressed using the symbol '::'
// * an optional string of a ScopeID delimited by '%'
// * an optional (last) 1 or 2 character prefix length field delimited by '/'
// * the last 32 bits in an address can be represented as an IPv4 address
//
// Inputs:
// <argument> name
// Domain name field of a URI to check for pattern match with
// IPv6 address
// validateStrictAddress: if set to true, it expects strict ipv6 address. Otherwise it expects
// part of the string in ipv6 format.
//
// Outputs:
// Nothing
//
// Assumes:
// the correct name is terminated by ']' character
//
// Returns:
// true if <name> has IPv6 format/ipv6 address based on validateStrictAddress, else false
//
// Throws:
// Nothing
//
// Remarks: MUST NOT be used unless all input indexes are verified and trusted.
// start must be next to '[' position, or error is reported
unsafe private static bool InternalIsValid(char *name, int start, ref int end, bool validateStrictAddress)
{
int sequenceCount = 0;
int sequenceLength = 0;
bool haveCompressor = false;
bool haveIPv4Address = false;
bool havePrefix = false;
bool expectingNumber = true;
int lastSequence = 1;
int i;
for (i = start; i < end; ++i)
{
if (havePrefix ? (name[i] >= '0' && name[i] <= '9') : UriHelper.IsHexDigit(name[i]))
{
++sequenceLength;
expectingNumber = false;
}
else
{
if (sequenceLength > 4)
{
return(false);
}
if (sequenceLength != 0)
{
++sequenceCount;
lastSequence = i - sequenceLength;
}
switch (name[i])
{
case '%':
while (true)
{
//accept anything in scopeID
if (++i == end)
{
// no closing ']', fail
return(false);
}
if (name[i] == ']')
{
goto case ']';
}
else if (name[i] == '/')
{
goto case '/';
}
}
case ']':
start = i;
i = end;
//this will make i = end+1
continue;
case ':':
if ((i > 0) && (name[i - 1] == ':'))
{
if (haveCompressor)
{
//
// can only have one per IPv6 address
//
return(false);
}
haveCompressor = true;
expectingNumber = false;
}
else
{
expectingNumber = true;
}
break;
case '/':
if (validateStrictAddress)
{
return(false);
}
if ((sequenceCount == 0) || havePrefix)
{
return(false);
}
havePrefix = true;
expectingNumber = true;
break;
case '.':
if (haveIPv4Address)
{
return(false);
}
i = end;
if (!IPv4AddressHelper.IsValid(name, lastSequence, ref i, true, false, false))
{
return(false);
}
// ipv4 address takes 2 slots in ipv6 address, one was just counted meeting the '.'
++sequenceCount;
haveIPv4Address = true;
--i; // it will be incremented back on the next loop
break;
default:
return(false);
}
sequenceLength = 0;
}
}
//
// if the last token was a prefix, check number of digits
//
if (havePrefix && ((sequenceLength < 1) || (sequenceLength > 2)))
{
return(false);
}
//
// these sequence counts are -1 because it is implied in end-of-sequence
//
int expectedSequenceCount = 8 + (havePrefix ? 1 : 0);
if (!expectingNumber && (sequenceLength <= 4) && (haveCompressor ? (sequenceCount < expectedSequenceCount) : (sequenceCount == expectedSequenceCount)))
{
if (i == end + 1)
{
// ']' was found
end = start + 1;
return(true);
}
return(false);
}
return(false);
}