public void VerifyTokenTrieCombine()
{
byte[] hello = Encoding.UTF8.GetBytes("hello");
byte[] helloBang = Encoding.UTF8.GetBytes("hello!");
byte[] hi = Encoding.UTF8.GetBytes("hi");
byte[] there = Encoding.UTF8.GetBytes("there!");
TokenTrie t = new TokenTrie();
t.AddToken(hello);
t.AddToken(helloBang);
TokenTrie t2 = new TokenTrie();
t.AddToken(hi);
t.AddToken(there);
TokenTrie combined = new TokenTrie();
combined.Append(t);
combined.Append(t2);
byte[] source1 = Encoding.UTF8.GetBytes("hello there");
byte[] source2 = Encoding.UTF8.GetBytes("hello! there");
byte[] source3 = Encoding.UTF8.GetBytes("hi there");
byte[] source4 = Encoding.UTF8.GetBytes("there!");
int token;
int pos = 0;
Assert.True(t.GetOperation(source1, source1.Length, ref pos, out token));
Assert.Equal(0, token);
pos = 0;
Assert.True(t.GetOperation(source2, source2.Length, ref pos, out token));
Assert.Equal(1, token);
pos = 0;
Assert.True(t.GetOperation(source3, source3.Length, ref pos, out token));
Assert.Equal(2, token);
pos = 0;
Assert.True(t.GetOperation(source4, source4.Length, ref pos, out token));
Assert.Equal(3, token);
}
public void VerifySimpleTrieAtBegin()
{
byte[] hello = Encoding.UTF8.GetBytes("hello");
byte[] helloBang = Encoding.UTF8.GetBytes("hello!");
byte[] hi = Encoding.UTF8.GetBytes("hi");
TokenTrie t = new TokenTrie();
t.AddToken(hello);
t.AddToken(helloBang);
t.AddToken(hi);
byte[] source1 = Encoding.UTF8.GetBytes("hello there");
byte[] source2 = Encoding.UTF8.GetBytes("hello1 there");
byte[] source3 = Encoding.UTF8.GetBytes("hello! there");
byte[] source4 = Encoding.UTF8.GetBytes("hi there");
byte[] source5 = Encoding.UTF8.GetBytes("hi");
byte[] source6 = Encoding.UTF8.GetBytes("he");
int token;
int pos = 0;
Assert.True(t.GetOperation(source1, source1.Length, ref pos, out token));
Assert.Equal(0, token);
pos = 0;
Assert.True(t.GetOperation(source2, source2.Length, ref pos, out token));
Assert.Equal(0, token);
pos = 0;
Assert.True(t.GetOperation(source3, source3.Length, ref pos, out token));
Assert.Equal(1, token);
pos = 0;
Assert.True(t.GetOperation(source4, source4.Length, ref pos, out token));
Assert.Equal(2, token);
pos = 0;
Assert.True(t.GetOperation(source5, source5.Length, ref pos, out token));
Assert.Equal(2, token);
pos = 0;
Assert.False(t.GetOperation(source6, source6.Length, ref pos, out token));
Assert.Equal(-1, token);
}
static GlobbingPatternMatcher()
{
Trie = new TokenTrie();
byte[] anyNumberOfPathParts = Encoding.UTF8.GetBytes("**/");
byte[] onePathPart = Encoding.UTF8.GetBytes("*");
byte[] wildcard = Encoding.UTF8.GetBytes("?");
byte[] openCharSet = Encoding.UTF8.GetBytes("[");
byte[] closeCharSet = Encoding.UTF8.GetBytes("]");
byte[] separatorChar = Encoding.UTF8.GetBytes("/");
byte[] separatorChar2 = Encoding.UTF8.GetBytes("\\");
Trie.AddToken(anyNumberOfPathParts);
Trie.AddToken(onePathPart);
Trie.AddToken(wildcard);
Trie.AddToken(openCharSet);
Trie.AddToken(closeCharSet);
Trie.AddToken(separatorChar);
Trie.AddToken(separatorChar2);
}
private static ITokenTrie GetSymbols(IProcessorState processor)
{
ITokenTrie tokens;
if (!TokenCache.TryGetValue(processor.Encoding, out tokens))
{
TokenTrie trie = new TokenTrie();
//Logic
trie.AddToken(processor.Encoding.GetBytes("AND"));
trie.AddToken(processor.Encoding.GetBytes("OR"));
trie.AddToken(processor.Encoding.GetBytes("!"));
trie.AddToken(processor.Encoding.GetBytes(">"));
trie.AddToken(processor.Encoding.GetBytes(">="));
trie.AddToken(processor.Encoding.GetBytes("<"));
trie.AddToken(processor.Encoding.GetBytes("<="));
trie.AddToken(processor.Encoding.GetBytes("=="));
trie.AddToken(processor.Encoding.GetBytes("!="));
//Braces
trie.AddToken(processor.Encoding.GetBytes("("));
trie.AddToken(processor.Encoding.GetBytes(")"));
//Whitespace
trie.AddToken(processor.Encoding.GetBytes(" "));
trie.AddToken(processor.Encoding.GetBytes("\t"));
//EOLs
trie.AddToken(processor.Encoding.GetBytes("\r\n"));
trie.AddToken(processor.Encoding.GetBytes("\n"));
trie.AddToken(processor.Encoding.GetBytes("\r"));
// quotes
trie.AddToken(processor.Encoding.GetBytes("'"));
// variable start
trie.AddToken(processor.Encoding.GetBytes("$("));
TokenCache[processor.Encoding] = tokens = trie;
}
return(tokens);
}
public IOperation GetOperation(Encoding encoding, IProcessorState processorState)
{
TokenTrie trie = new TokenTrie();
List <IToken> tokens = new List <IToken>(TokenTypeModulus * LongestTokenVariantListSize);
for (int i = 0; i < tokens.Capacity; i++)
{
tokens.Add(null);
}
AddTokensOfTypeToTokenListAndTrie(trie, tokens, Tokens.IfTokens, IfTokenBaseIndex, encoding);
AddTokensOfTypeToTokenListAndTrie(trie, tokens, Tokens.ElseTokens, ElseTokenBaseIndex, encoding);
AddTokensOfTypeToTokenListAndTrie(trie, tokens, Tokens.ElseIfTokens, ElseIfTokenBaseIndex, encoding);
AddTokensOfTypeToTokenListAndTrie(trie, tokens, Tokens.EndIfTokens, EndTokenBaseIndex, encoding);
AddTokensOfTypeToTokenListAndTrie(trie, tokens, Tokens.ActionableIfTokens, IfTokenActionableBaseIndex, encoding);
AddTokensOfTypeToTokenListAndTrie(trie, tokens, Tokens.ActionableElseTokens, ElseTokenActionableBaseIndex, encoding);
AddTokensOfTypeToTokenListAndTrie(trie, tokens, Tokens.ActionableElseIfTokens, ElseIfTokenActionableBaseIndex, encoding);
return(new Impl(this, tokens, trie, _id, _initialState));
}
public void VerifyTokenTrieLookArounds(string original, int checkPosition, int expectedPosition, bool success, string after, string value, string before)
{
byte[] data = Encoding.UTF8.GetBytes(original);
TokenTrie t = new TokenTrie();
TokenConfig builder = (value ?? "").TokenConfigBuilder();
if (!string.IsNullOrEmpty(after))
{
builder = builder.OnlyIfAfter(after);
}
if (!string.IsNullOrEmpty(before))
{
builder = builder.OnlyIfBefore(before);
}
t.AddToken(builder.ToToken(Encoding.UTF8));
int pos = checkPosition;
Assert.Equal(success, t.GetOperation(data, data.Length, ref pos, out int token));
Assert.Equal(expectedPosition, pos);
}
public static bool Evaluate(IProcessorState processor, ref int bufferLength, ref int currentBufferPosition, out bool faulted)
{
faulted = false;
TokenTrie trie = new TokenTrie();
//Logic
trie.AddToken(processor.Encoding.GetBytes("&&"), 0);
trie.AddToken(processor.Encoding.GetBytes("||"), 1);
trie.AddToken(processor.Encoding.GetBytes("^"), 2);
trie.AddToken(processor.Encoding.GetBytes("!"), 3);
trie.AddToken(processor.Encoding.GetBytes(">"), 4);
trie.AddToken(processor.Encoding.GetBytes(">="), 5);
trie.AddToken(processor.Encoding.GetBytes("<"), 6);
trie.AddToken(processor.Encoding.GetBytes("<="), 7);
trie.AddToken(processor.Encoding.GetBytes("=="), 8);
trie.AddToken(processor.Encoding.GetBytes("="), 9);
trie.AddToken(processor.Encoding.GetBytes("!="), 10);
//Bitwise
trie.AddToken(processor.Encoding.GetBytes("&"), 11);
trie.AddToken(processor.Encoding.GetBytes("|"), 12);
trie.AddToken(processor.Encoding.GetBytes("<<"), 13);
trie.AddToken(processor.Encoding.GetBytes(">>"), 14);
//Braces
trie.AddToken(processor.Encoding.GetBytes("("), 15);
trie.AddToken(processor.Encoding.GetBytes(")"), 16);
//Whitespace
trie.AddToken(processor.Encoding.GetBytes(" "), 17);
trie.AddToken(processor.Encoding.GetBytes("\t"), 18);
//EOLs
trie.AddToken(processor.Encoding.GetBytes("\r\n"), 19);
trie.AddToken(processor.Encoding.GetBytes("\n"), 20);
trie.AddToken(processor.Encoding.GetBytes("\r"), 21);
// quotes
trie.AddToken(processor.Encoding.GetBytes("\""), 22);
trie.AddToken(processor.Encoding.GetBytes("'"), 23);
//Tokens
trie.Append(processor.EncodingConfig.Variables);
//Run forward to EOL and collect args
TokenFamily currentTokenFamily;
List <byte> currentTokenBytes = new List <byte>();
List <TokenRef> tokens = new List <TokenRef>();
if (!trie.GetOperation(processor.CurrentBuffer, bufferLength, ref currentBufferPosition, out int token))
{
currentTokenFamily = TokenFamily.Literal;
currentTokenBytes.Add(processor.CurrentBuffer[currentBufferPosition++]);
}
else if (token > ReservedTokenMaxIndex)
{
currentTokenFamily = TokenFamily.Reference | (TokenFamily)token;
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
else
{
currentTokenFamily = (TokenFamily)token;
if (currentTokenFamily != TokenFamily.WindowsEOL && currentTokenFamily != TokenFamily.LegacyMacEOL && currentTokenFamily != TokenFamily.UnixEOL)
{
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
else
{
return(EvaluateCondition(tokens, processor.EncodingConfig.VariableValues));
}
}
int braceDepth = 0;
if (tokens[0].Family == TokenFamily.OpenBrace)
{
++braceDepth;
}
bool first = true;
QuotedRegionKind inQuoteType = QuotedRegionKind.None;
while ((first || braceDepth > 0) && bufferLength > 0)
{
int targetLen = Math.Min(bufferLength, trie.MaxLength);
for (; currentBufferPosition < bufferLength - targetLen + 1;)
{
int oldBufferPos = currentBufferPosition;
if (trie.GetOperation(processor.CurrentBuffer, bufferLength, ref currentBufferPosition, out token))
{
if (braceDepth == 0)
{
switch (tokens[tokens.Count - 1].Family)
{
case TokenFamily.Whitespace:
case TokenFamily.Tab:
case TokenFamily.CloseBrace:
case TokenFamily.WindowsEOL:
case TokenFamily.UnixEOL:
case TokenFamily.LegacyMacEOL:
TokenFamily thisFamily = (TokenFamily)token;
if (thisFamily == TokenFamily.WindowsEOL || thisFamily == TokenFamily.UnixEOL || thisFamily == TokenFamily.LegacyMacEOL)
{
currentBufferPosition = oldBufferPos;
}
break;
default:
currentBufferPosition = oldBufferPos;
first = false;
break;
}
if (!first)
{
break;
}
}
// We matched an item, so whatever this is, it's not a literal.
// if the current token is a literal, end it.
if (currentTokenFamily == TokenFamily.Literal)
{
string literal = processor.Encoding.GetString(currentTokenBytes.ToArray());
tokens.Add(new TokenRef
{
Family = TokenFamily.Literal,
Literal = literal
});
currentTokenBytes.Clear();
}
TokenFamily foundTokenFamily = (TokenFamily)token;
if (foundTokenFamily == TokenFamily.QuotedLiteral || foundTokenFamily == TokenFamily.SingleQuotedLiteral)
{
QuotedRegionKind incomingQuoteKind;
switch (foundTokenFamily)
{
case TokenFamily.QuotedLiteral:
incomingQuoteKind = QuotedRegionKind.DoubleQuoteRegion;
break;
case TokenFamily.SingleQuotedLiteral:
incomingQuoteKind = QuotedRegionKind.SingleQuoteRegion;
break;
default:
incomingQuoteKind = QuotedRegionKind.None;
break;
}
if (inQuoteType == QuotedRegionKind.None)
{
// starting quote found
currentTokenBytes.AddRange(trie.Tokens[token].Value);
inQuoteType = incomingQuoteKind;
}
else if (incomingQuoteKind == inQuoteType)
{
// end quote found
currentTokenBytes.AddRange(trie.Tokens[token].Value);
tokens.Add(new TokenRef
{
Family = TokenFamily.Literal,
Literal = processor.Encoding.GetString(currentTokenBytes.ToArray())
});
currentTokenBytes.Clear();
inQuoteType = QuotedRegionKind.None;
}
else
{
// this is a different quote type. Treat it like a non-match, just add the token to the currentTokenBytes
currentTokenBytes.AddRange(trie.Tokens[token].Value);
}
}
else if (inQuoteType != QuotedRegionKind.None)
{
// we're inside a quoted literal, the token found by the trie should not be processed, just included with the literal
currentTokenBytes.AddRange(trie.Tokens[token].Value);
}
else if (token > ReservedTokenMaxIndex)
{
currentTokenFamily = TokenFamily.Reference | (TokenFamily)token;
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
else
{
//If we have a normal token...
currentTokenFamily = (TokenFamily)token;
if (currentTokenFamily != TokenFamily.WindowsEOL && currentTokenFamily != TokenFamily.LegacyMacEOL && currentTokenFamily != TokenFamily.UnixEOL)
{
switch (currentTokenFamily)
{
case TokenFamily.OpenBrace:
++braceDepth;
break;
case TokenFamily.CloseBrace:
--braceDepth;
break;
}
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
else
{
return(EvaluateCondition(tokens, processor.EncodingConfig.VariableValues));
}
}
}
else if (inQuoteType != QuotedRegionKind.None)
{
// we're in a quoted literal but did not match a token at the current position.
// so just add the current byte to the currentTokenBytes
currentTokenBytes.Add(processor.CurrentBuffer[currentBufferPosition++]);
}
else if (braceDepth > 0)
{
currentTokenFamily = TokenFamily.Literal;
currentTokenBytes.Add(processor.CurrentBuffer[currentBufferPosition++]);
}
else
{
first = false;
break;
}
}
processor.AdvanceBuffer(currentBufferPosition);
currentBufferPosition = processor.CurrentBufferPosition;
bufferLength = processor.CurrentBufferLength;
}
#if DEBUG
Debug.Assert(
inQuoteType == QuotedRegionKind.None,
$"Malformed predicate due to unmatched quotes. InitialBuffer = {processor.Encoding.GetString(processor.CurrentBuffer)} currentTokenFamily = {currentTokenFamily} | TokenFamily.QuotedLiteral = {TokenFamily.QuotedLiteral} | TokenFamily.SingleQuotedLiteral = {TokenFamily.SingleQuotedLiteral}");
#endif
return(EvaluateCondition(tokens, processor.EncodingConfig.VariableValues));
}
public static bool CppStyleEvaluator(IProcessorState processor, ref int bufferLength, ref int currentBufferPosition)
{
TokenTrie trie = new TokenTrie();
//Logic
trie.AddToken(processor.Encoding.GetBytes("&&"), 0);
trie.AddToken(processor.Encoding.GetBytes("||"), 1);
trie.AddToken(processor.Encoding.GetBytes("^"), 2);
trie.AddToken(processor.Encoding.GetBytes("!"), 3);
trie.AddToken(processor.Encoding.GetBytes(">"), 4);
trie.AddToken(processor.Encoding.GetBytes(">="), 5);
trie.AddToken(processor.Encoding.GetBytes("<"), 6);
trie.AddToken(processor.Encoding.GetBytes("<="), 7);
trie.AddToken(processor.Encoding.GetBytes("=="), 8);
trie.AddToken(processor.Encoding.GetBytes("="), 9);
trie.AddToken(processor.Encoding.GetBytes("!="), 10);
//Bitwise
trie.AddToken(processor.Encoding.GetBytes("&"), 11);
trie.AddToken(processor.Encoding.GetBytes("|"), 12);
trie.AddToken(processor.Encoding.GetBytes("<<"), 13);
trie.AddToken(processor.Encoding.GetBytes(">>"), 14);
//Braces
trie.AddToken(processor.Encoding.GetBytes("("), 15);
trie.AddToken(processor.Encoding.GetBytes(")"), 16);
//Whitespace
trie.AddToken(processor.Encoding.GetBytes(" "), 17);
trie.AddToken(processor.Encoding.GetBytes("\t"), 18);
//EOLs
trie.AddToken(processor.Encoding.GetBytes("\r\n"), 19);
trie.AddToken(processor.Encoding.GetBytes("\n"), 20);
trie.AddToken(processor.Encoding.GetBytes("\r"), 21);
//Tokens
trie.Append(processor.EncodingConfig.Variables);
//Run forward to EOL and collect args
TokenFamily currentTokenFamily;
List <byte> currentTokenBytes = new List <byte>();
List <TokenRef> tokens = new List <TokenRef>();
int token;
if (!trie.GetOperation(processor.CurrentBuffer, bufferLength, ref currentBufferPosition, out token))
{
currentTokenFamily = TokenFamily.Literal;
currentTokenBytes.Add(processor.CurrentBuffer[currentBufferPosition++]);
}
else if (token > ReservedTokenMaxIndex)
{
currentTokenFamily = TokenFamily.Reference | (TokenFamily)token;
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
else
{
currentTokenFamily = (TokenFamily)token;
if (currentTokenFamily != TokenFamily.WindowsEOL && currentTokenFamily != TokenFamily.LegacyMacEOL && currentTokenFamily != TokenFamily.UnixEOL)
{
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
else
{
return(EvaluateCondition(tokens, processor.EncodingConfig.VariableValues));
}
}
int braceDepth = 0;
if (tokens[0].Family == TokenFamily.OpenBrace)
{
++braceDepth;
}
bool first = true;
while ((first || braceDepth > 0) && bufferLength > 0)
{
int targetLen = Math.Min(bufferLength, trie.MaxLength);
for (; currentBufferPosition < bufferLength - targetLen + 1;)
{
int oldBufferPos = currentBufferPosition;
if (trie.GetOperation(processor.CurrentBuffer, bufferLength, ref currentBufferPosition, out token))
{
if (braceDepth == 0)
{
switch (tokens[tokens.Count - 1].Family)
{
case TokenFamily.Whitespace:
case TokenFamily.Tab:
case TokenFamily.CloseBrace:
case TokenFamily.WindowsEOL:
case TokenFamily.UnixEOL:
case TokenFamily.LegacyMacEOL:
TokenFamily thisFamily = (TokenFamily)token;
if (thisFamily == TokenFamily.WindowsEOL || thisFamily == TokenFamily.UnixEOL || thisFamily == TokenFamily.LegacyMacEOL)
{
currentBufferPosition = oldBufferPos;
}
break;
default:
currentBufferPosition = oldBufferPos;
first = false;
break;
}
if (!first)
{
break;
}
}
//We matched an item, so whatever this is, it's not a literal, end the current literal if that's
// what we currently have
if (currentTokenFamily == TokenFamily.Literal)
{
string literal = processor.Encoding.GetString(currentTokenBytes.ToArray());
tokens.Add(new TokenRef
{
Family = TokenFamily.Literal,
Literal = literal
});
currentTokenBytes.Clear();
}
//If we have a token from the args...
if (token > ReservedTokenMaxIndex)
{
if (currentTokenFamily == TokenFamily.Literal)
{
TokenRef previous = tokens[tokens.Count - 1];
previous.Literal += processor.Encoding.GetString(currentTokenBytes.ToArray());
currentTokenBytes = processor.Encoding.GetBytes(previous.Literal).ToList();
tokens.RemoveAt(tokens.Count - 1);
}
else
{
currentTokenFamily = TokenFamily.Reference | (TokenFamily)token;
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
}
//If we have a normal token...
else
{
currentTokenFamily = (TokenFamily)token;
if (currentTokenFamily != TokenFamily.WindowsEOL && currentTokenFamily != TokenFamily.LegacyMacEOL && currentTokenFamily != TokenFamily.UnixEOL)
{
if (currentTokenFamily == TokenFamily.OpenBrace)
{
++braceDepth;
}
else if (currentTokenFamily == TokenFamily.CloseBrace)
{
--braceDepth;
}
tokens.Add(new TokenRef
{
Family = currentTokenFamily
});
}
else
{
return(EvaluateCondition(tokens, processor.EncodingConfig.VariableValues));
}
}
}
else if (braceDepth > 0)
{
currentTokenFamily = TokenFamily.Literal;
currentTokenBytes.Add(processor.CurrentBuffer[currentBufferPosition++]);
}
else
{
first = false;
break;
}
}
processor.AdvanceBuffer(currentBufferPosition);
currentBufferPosition = processor.CurrentBufferPosition;
bufferLength = processor.CurrentBufferLength;
}
return(EvaluateCondition(tokens, processor.EncodingConfig.VariableValues));
}
public Impl(InlineMarkupConditional definition, IReadOnlyList <byte[]> tokens, TokenTrie structureTrie, TokenTrie closeConditionTrie, TokenTrie scanBackTrie, MarkupTokenMapping mapping, string id)
{
_definition = definition;
Id = id;
Tokens = tokens;
_mapping = mapping;
_structureTrie = structureTrie;
_scanBackTrie = scanBackTrie;
_closeConditionTrie = closeConditionTrie;
}
