private CharsetGrammar()
{
var provider = new UnicodeCharSetProvider();
var mapper = new UnicodeUtf16Mapper(false, false);
var rx = RegexLexer.CreateRx(mapper);
var rxWhitespace = new RxAccept <char>(RxOfSymbol <char> .Extract(rx, RegexLexer.SymWhitespace), SymWhitespace, 0);
var rxCharset = new RxAccept <char>(RxOfSymbol <char> .Extract(rx, RegexLexer.SymCharset), SymCharset, 0);
var rxRegexCharset = new RxAccept <char>(RxOfSymbol <char> .Extract(rx, RegexLexer.SymRegexCharset), SymRegexCharset, 0);
var rxUnion = new RxAccept <char>(RegexMatchSet.FromChars('|', '+').ToInvariant(mapper, provider, true), SymUnion, 0);
var rxSubtract = new RxAccept <char>(RegexMatchSet.FromChars('-').ToInvariant(mapper, provider, true), SymSubtract, 0);
var rxIntersect = new RxAccept <char>(RegexMatchSet.FromChars('&').ToInvariant(mapper, provider, true), SymIntersect, 0);
var rxDifference = new RxAccept <char>(RegexMatchSet.FromChars('^').ToInvariant(mapper, provider, true), SymDifference, 0);
var rxNegate = new RxAccept <char>(RegexMatchSet.FromChars('~').ToInvariant(mapper, provider, true), SymNegate, 0);
var rxParensOpen = new RxAccept <char>(RegexMatchSet.FromChars('(').ToInvariant(mapper, provider, true), SymParensOpen, 0);
var rxParensClose = new RxAccept <char>(RegexMatchSet.FromChars(')').ToInvariant(mapper, provider, true), SymParensClose, 0);
var alpha = new AlphabetBuilder <char>(
new RxAlternation <char>(rxWhitespace,
new RxAlternation <char>(rxCharset,
new RxAlternation <char>(rxRegexCharset,
new RxAlternation <char>(rxUnion,
new RxAlternation <char>(rxSubtract,
new RxAlternation <char>(rxIntersect,
new RxAlternation <char>(rxDifference,
new RxAlternation <char>(rxNegate,
new RxAlternation <char>(rxParensOpen, rxParensClose))))))))),
Utf16Chars.EOF,
Utf16Chars.ValidBmp);
var nfa = NfaBuilder <LetterId> .Build(alpha.Expression);
var dfa = DfaBuilder <LetterId> .Build(nfa, LetterId.Eof);
if (dfa.StartState.Id != default(Id <DfaState <LetterId> >))
{
throw new InvalidOperationException($"Internal error: Unexpected DFA start state {dfa.StartState.Id}");
}
this.stateMachine = DfaStateMachineEmitter.CreateExpression(dfa, AlphabetMapperEmitter <char> .CreateExpression(alpha)).Compile();
this.table = new LalrTableGenerator(new GrammarBuilder(-2, -1, SymExpression)
{
{ SymUnionExpression, SymExpression, SymUnion, SymNegateExpression },
{ SymExpression, SymUnionExpression },
{ SymSubtractExpression, SymExpression, SymSubtract, SymNegateExpression },
{ SymExpression, SymSubtractExpression },
{ SymIntersectExpression, SymExpression, SymIntersect, SymNegateExpression },
{ SymExpression, SymIntersectExpression },
{ SymDifferenceExpression, SymExpression, SymDifference, SymNegateExpression },
{ SymExpression, SymDifferenceExpression },
{ SymExpression, SymNegateExpression },
{ SymNegateExpression, SymNegate, SymValueExpression },
{ SymNegateExpression, SymValueExpression },
{ SymValueExpression, SymParensOpen, SymExpression, SymParensClose },
{ SymValueExpression, SymCharset },
{ SymValueExpression, SymRegexCharset }
})
.ComputeTable();
}
public static Expression <Action <TParser, ParserContextBase <TAstNode, TInput, TPosition>, SymbolId, Capture <TInput>, TPosition> > CreateExpression <TParser, TAstNode, TInput, TPosition>(LalrTable table, IParserFragmentEmitter <TParser, TAstNode, TInput, TPosition> fragmentEmitter, Func <SymbolId, string> resolver = null)
{
var paramParser = Expression.Parameter(typeof(TParser), "parser");
var paramContext = Expression.Parameter(typeof(ParserContextBase <TAstNode, TInput, TPosition>), "context");
var paramTokenSymbolId = Expression.Parameter(typeof(SymbolId), "tokenSymbolId");
var paramTokenValue = Expression.Parameter(typeof(Capture <TInput>), "tokenValue");
var paramPosition = Expression.Parameter(typeof(TPosition), "position");
var varInitialState = Expression.Variable(typeof(ParserState <TAstNode>), "initialState");
var exprCurrentState = Expression.Field(paramContext, Reflect <ParserContext <TAstNode, TInput, TPosition> > .GetField(c => c.currentState));
var lblBreak = Expression.Label("Break");
var lblContinue = Expression.Label("Continue");
var ctor_ParserState = Reflect.GetConstructor(() => new ParserState <TAstNode>(default(TAstNode), default(int), default(ParserState <TAstNode>)));
var prop_ParserState_State = Reflect <ParserState <TAstNode> > .GetProperty(s => s.State);
var prop_ParserState_Node = Reflect <ParserState <TAstNode> > .GetProperty(s => s.Node);
var prop_ParserState_Parent = Reflect <ParserState <TAstNode> > .GetProperty(s => s.Parent);
var meth_Accept = Reflect <ParserContext <TAstNode, TInput, TPosition> > .GetMethod(c => c.Accept(default(TAstNode)));
var meth_SyntaxError = Reflect <ParserContext <TAstNode, TInput, TPosition> > .GetMethod(c => c.SyntaxError(default(SymbolId), default(Capture <TInput>), default(TPosition), default(IEnumerable <SymbolId>)));
Expression DoShift(SymbolId symbolId, ShiftAction action)
{
return(Expression.Block(typeof(void),
Expression.Assign(
exprCurrentState,
Expression.New(
ctor_ParserState,
fragmentEmitter.CreateTerminal(symbolId, paramParser, paramTokenValue, paramPosition),
Expression.Constant(action.NewState),
exprCurrentState)),
Expression.Break(lblBreak)));
}
Expression DoReduce(ReduceSingleAction action)
{
var varCurrentState = Expression.Variable(typeof(ParserState <TAstNode>), "currentState");
var varsNodes = action.ProductionRule.RuleSymbolIds.Select(s => Expression.Variable(typeof(TAstNode), s.ToString(resolver))).ToArray();
var body = new List <Expression>(varsNodes.Length * 2 + 3);
body.Add(Expression.Assign(
varCurrentState,
exprCurrentState));
for (var i = varsNodes.Length - 1; i >= 0; i--)
{
body.Add(Expression.Assign(
varsNodes[i],
Expression.Property(
varCurrentState,
prop_ParserState_Node)));
body.Add(Expression.Assign(
varCurrentState,
Expression.Property(
varCurrentState,
prop_ParserState_Parent)));
}
body.Add(Expression.Assign(
exprCurrentState,
Expression.New(
ctor_ParserState,
fragmentEmitter.CreateNonterminal(action.ProductionRule, paramParser, varsNodes),
Expression.Switch(
Expression.Property(
varCurrentState,
prop_ParserState_State),
Expression.Throw(
Expression.New(ctor_InvalidOperationException), typeof(int)),
table.Action
.Where(a => (a.Value.Type == ActionType.Goto) && (a.Key.Value == action.ProductionRule.ProductionSymbolId))
.Select(a => Expression.SwitchCase(
Expression.Constant(((GotoAction)a.Value).NewState),
Expression.Constant(a.Key.State)
)).ToArray()),
varCurrentState)));
body.Add(Expression.Continue(lblContinue));
return(Expression.Block(typeof(void),
varsNodes.Append(varCurrentState),
body));
}
Expression DoAccept()
{
return(Expression.Block(typeof(void),
Expression.Call(
paramContext,
meth_Accept,
Expression.Property(
exprCurrentState,
prop_ParserState_Node)),
Expression.Assign(
exprCurrentState,
Expression.Property(
exprCurrentState,
prop_ParserState_Parent)),
Expression.Break(lblBreak)));
}
Expression DoThrow()
{
throw new InvalidOperationException("Internal error: unexpected action type");
}
Expression DoSyntaxError()
{
var varSimulatedState = Expression.Variable(typeof(ParserState <TAstNode>), "simulatedState");
var lblSimulationBreak = Expression.Label(typeof(bool), "simulationBreak");
Expression DoSimulateReduce(ProductionRule rule)
{
Expression exprNewSimulatedState = Expression.Assign(
varSimulatedState,
Expression.New(
ctor_ParserState,
Expression.Default(typeof(TAstNode)),
Expression.Switch(
Expression.Property(
varSimulatedState,
prop_ParserState_State),
Expression.Break(
lblSimulationBreak,
Expression.Constant(false),
typeof(int)),
table.Action
.Where(a => (a.Key.Value == rule.ProductionSymbolId) && (a.Value.Type == ActionType.Goto))
.GroupBy(a => ((GotoAction)a.Value).NewState)
.Select(g => Expression.SwitchCase(
Expression.Constant(g.Key),
g.Select(a => Expression.Constant(a.Key.State)))).ToArray()),
varSimulatedState));
return(rule.RuleSymbolIds.Count == 0
? exprNewSimulatedState
: Expression.Block(
Expression.Assign(varSimulatedState,
rule.RuleSymbolIds.Aggregate <SymbolId, Expression>(varSimulatedState, (expression, id) => Expression.Property(expression, prop_ParserState_Parent))),
exprNewSimulatedState));
}
var varExpectedTokens = Expression.Variable(typeof(List <SymbolId>), "expectedTokens");
var terminalSymbolIds = table.Action
.Where(a => (a.Value.Type == ActionType.Accept) || (a.Value.Type == ActionType.Shift))
.Select(a => a.Key.Value)
.Distinct()
.OrderBy(id => id.ToInt32())
.ToArray();
var body = new List <Expression>(terminalSymbolIds.Length + 2);
body.Add(Expression.Assign(
varExpectedTokens,
Expression.New(
ctor_ListOfSymbolId,
Expression.Constant(terminalSymbolIds.Length))));
foreach (var symbolId in terminalSymbolIds)
{
body.Add(Expression.IfThen(
Expression.Block(typeof(bool), new[] { varSimulatedState },
Expression.Assign(
varSimulatedState,
exprCurrentState),
Expression.Loop(
Expression.Switch(typeof(void),
Expression.Property(
varSimulatedState,
prop_ParserState_State),
Expression.Break(lblSimulationBreak,
Expression.Constant(false)),
null,
table.Action
.Where(a => (a.Key.Value == symbolId) && (a.Value.Type == ActionType.Reduce))
.GroupBy(a => ((ReduceSingleAction)a.Value).ProductionRule)
.Select(g => Expression.SwitchCase(
DoSimulateReduce(g.Key),
g.Select(a => Expression.Constant(a.Key.State))))
.Append(Expression.SwitchCase(
Expression.Break(lblSimulationBreak,
Expression.Constant(true)),
table.Action
.Where(a => (a.Key.Value == symbolId) && ((a.Value.Type == ActionType.Accept) || (a.Value.Type == ActionType.Shift)))
.Select(a => Expression.Constant(a.Key.State))
)).ToArray()), lblSimulationBreak)),
Expression.Call(
varExpectedTokens,
meth_ListOfSymbolId_Add,
Expression.New(
ctor_SymbolId_Int32,
Expression.Constant(symbolId.ToInt32())))));
}
body.Add(Expression.Convert(
varExpectedTokens,
typeof(IEnumerable <SymbolId>)));
var exprExpectedSymbolIds = Expression.Block(typeof(IEnumerable <SymbolId>), new[] { varExpectedTokens }, body);
var exprCustomSyntaxError = fragmentEmitter.SyntaxError(paramParser, paramTokenSymbolId, paramTokenValue, paramPosition, exprExpectedSymbolIds);
if (exprCustomSyntaxError?.Type == typeof(bool))
{
exprCustomSyntaxError = Expression.IfThen(
exprCustomSyntaxError,
Expression.Continue(lblContinue));
}
return(Expression.Block(typeof(void),
Expression.Assign(
exprCurrentState,
varInitialState),
exprCustomSyntaxError ?? Expression.Call(
paramContext,
meth_SyntaxError,
paramTokenSymbolId,
paramTokenValue,
paramPosition,
exprExpectedSymbolIds),
Expression.Break(lblBreak)));
}
return(Expression.Lambda <Action <TParser, ParserContextBase <TAstNode, TInput, TPosition>, SymbolId, Capture <TInput>, TPosition> >(
Expression.Block(typeof(void), new[] { varInitialState },
Expression.Assign(
varInitialState,
exprCurrentState),
Expression.Loop(
Expression.Block(typeof(void),
Expression.Switch(
Expression.Property(
exprCurrentState,
prop_ParserState_State),
table.Action
.Where(a => (a.Value.Type == ActionType.Shift) || (a.Value.Type == ActionType.Reduce) || (a.Value.Type == ActionType.Accept))
.GroupBy(a => a.Key.State)
.Select(g => Expression.SwitchCase(
Expression.Switch(
Expression.Call(
paramTokenSymbolId,
meth_SymbolId_ToInt32),
g.Select(a => Expression.SwitchCase(
a.Value.Type == ActionType.Shift
? DoShift(a.Key.Value, (ShiftAction)a.Value)
: a.Value.Type == ActionType.Reduce
? DoReduce((ReduceSingleAction)a.Value)
: a.Value.Type == ActionType.Accept
? DoAccept()
: DoThrow(),
Expression.Constant(a.Key.Value.ToInt32()))).ToArray()),
Expression.Constant(g.Key))).ToArray()),
DoSyntaxError()
), lblBreak, lblContinue)),
paramParser,
paramContext,
paramTokenSymbolId,
paramTokenValue,
paramPosition));
}
protected ParserBase(LalrTable table, ParserContextBase <TAstNode, TInput, TPosition> context)
{
this.table = table;
this.context = context;
}