public static void GetFirstCollection(
this RegulationList grammar,
out FIRSTCollection firstCollection, Dictionary <TreeNodeType, bool> nullableDict = null)
{
if (nullableDict == null)
{
GetNullableDict(grammar, out nullableDict);
}
FIRSTCollection firstList4Node;
grammar.GetFirstCollection4Node(out firstList4Node, nullableDict);
FIRSTCollection firstList4Regulation;
grammar.GetFirstCollection4Regulation(out firstList4Regulation,
nullableDict, firstList4Node);
firstCollection = new FIRSTCollection();
foreach (var item in firstList4Node)
{
firstCollection.TryInsert(item);
}
foreach (var item in firstList4Regulation)
{
firstCollection.TryInsert(item);
}
}
/// <summary>
/// 计算所有可能推导出null的结点。
/// </summary>
/// <param name="grammar"></param>
/// <returns></returns>
public static void GetNullableDict(
this RegulationList grammar, out Dictionary <TreeNodeType, bool> nullableDict)
{
nullableDict = new Dictionary <TreeNodeType, bool>();
// 初始化nullable
List <TreeNodeType> allNodeTypes = grammar.GetAllTreeNodeTypes();
foreach (var item in allNodeTypes)
{
nullableDict.Add(item, false);
}
// 迭代到不动点
bool changed = false;
do
{
changed = false;
foreach (var regulation in grammar)
{
// 如果RightPart的结点全部可为null，就说明此regulation.Left是可推导出"null"的。
if (Nullable(regulation.RightPart, 0, regulation.RightPart.Count(), nullableDict))
{
if (!nullableDict[regulation.Left])
{
nullableDict[regulation.Left] = true;
changed = true;
}
}
}
} while (changed);
}
/// <summary>
/// LR(0)的Closure操作。
/// 补全一个状态。
/// </summary>
/// <param name="list"></param>
/// <param name="state"></param>
/// <returns></returns>
static LR0State Closure(this RegulationList list, LR0State state)
{
Queue <LR0Item> queue = new Queue <LR0Item>();
foreach (var item in state)
{
queue.Enqueue(item);
}
while (queue.Count > 0)
{
LR0Item item = queue.Dequeue();
TreeNodeType node = item.GetNodeNext2Dot();
if (node == null)
{
continue;
}
foreach (var regulation in list)
{
if (regulation.Left == node)
{
var newItem = new LR0Item(regulation, 0);
if (state.TryInsert(newItem))
{
queue.Enqueue(newItem);
}
}
}
}
return(state);
}
/// <summary>
/// 计算文法的所有单个的结点的FIRST
/// </summary>
/// <param name="grammar"></param>
/// <param name="nullableDict"></param>
/// <returns></returns>
private static void GetFirstCollection4Node(
this RegulationList grammar,
out FIRSTCollection firstCollection,
Dictionary <TreeNodeType, bool> nullableDict = null)
{
// 初始化FIRST
firstCollection = new FIRSTCollection();
// 初始化非叶结点的FIRST
foreach (var item in grammar.GetAllTreeNodeNonLeaveTypes())
{
if (nullableDict[item])
{
firstCollection.TryInsert(new FIRST(item, TreeNodeType.NullNode));
}
else
{
firstCollection.TryInsert(new FIRST(item));
}
}
// 初始化叶结点的FIRST（叶结点的FIRST实际上已经完工）
foreach (var item in grammar.GetAllTreeNodeLeaveTypes())
{
firstCollection.TryInsert(new FIRST(item, item));
}
bool changed = false;
do
{
changed = false;
foreach (var regulation in grammar)
{
FIRST first = FindFirst(firstCollection, regulation.Left);
int rightPartCount = regulation.RightPart.Count();
for (int checkCount = 0; checkCount < rightPartCount; checkCount++)
{
// 如果前面checkCount个结点都可为null，
// 就说明RightPart[checkCount]的FIRST是此regulation.Left的FIRST的一部分。
if (Nullable(regulation.RightPart, 0, checkCount, nullableDict))
{
FIRST refFirst = FindFirst(firstCollection, regulation.RightNode(checkCount));
if (refFirst == null)
{
throw new Exception("algorithm error!");
}
if (refFirst != first)
{
foreach (var value in refFirst.Values)
{
if (value != TreeNodeType.NullNode)
{
changed = first.TryInsert(value) || changed;
}
}
}
}
}
}
} while (changed);
}
/// <summary>
/// 计算文法的FIRST和FOLLOW集
/// </summary>
/// <param name="grammar"></param>
/// <returns></returns>
public static FirstListAndFollowList GetFirstListAndFollowList(this RegulationList grammar)
{
Dictionary <TreeNodeType, bool> nullableDict = new Dictionary <TreeNodeType, bool>();
GetNullableDict(grammar, out nullableDict);
FIRSTCollection firstCollection4Node;
grammar.GetFirstCollection4Node(out firstCollection4Node, nullableDict);
FIRSTCollection firstCollection4Regulation;
grammar.GetFirstCollection4Regulation(out firstCollection4Regulation, nullableDict, firstCollection4Node);
FOLLOWCollection followCollection;
grammar.GetFollowCollection(out followCollection, nullableDict, firstCollection4Node);
var firstCollection = new FIRSTCollection();
foreach (var item in firstCollection4Node)
{
firstCollection.TryInsert(item);
}
foreach (var item in firstCollection4Regulation)
{
firstCollection.TryInsert(item);
}
return(new FirstListAndFollowList(firstCollection, followCollection));
}
public SyntaxTree Parse(TokenList tokenList)
{
LRParsingMap parsingMap = GetParsingMap();
RegulationList grammar = GetGrammar();
//TODO:这个convertor以后是可以配置的。
var tokenTypeConvertor = new TokenType2TreeNodeType();
var context = new ParsingContext(tokenList, grammar, parsingMap, tokenTypeConvertor);
while (context.CurrentTokenIndex < context.TokenList.Count + 1)
{
PrintParsingProgress(context);
TreeNodeType nodeType = context.CurrentNodeType();
int stateId = context.StateIdStack.Peek();
LRParsingAction action = parsingMap.GetAction(stateId, nodeType);
int currentTokenIndex = action.Execute(context);
context.CurrentTokenIndex = currentTokenIndex;
}
PrintLastState(context);
if (context.TreeStack.Count > 0)
{
return(context.TreeStack.Peek());
}
else
{
return(new SyntaxTree());
}
}
public ParsingContext(TokenList tokenList,
RegulationList grammar,
LRParsingMap parsingMap,
TokenType2TreeNodeType tokenTypeConvertor)
{
this.StateIdStack = new Stack <int>();
this.StateIdStack.Push(1);
this.TreeStack = new Stack <SyntaxTree>();
this.TokenList = tokenList;
this.Grammar = grammar;
this.ParsingMap = parsingMap;
this.TokenTypeConvertor = tokenTypeConvertor;
}
/// <summary>
/// LR(0)的Goto操作。
/// 将圆点移到所有LR(0)项中的符号<paramref name="x"/>之后。
/// </summary>
/// <param name="list"></param>
/// <param name="state"></param>
/// <param name="x">一个文法符号，终结点或非终结点。</param>
/// <returns></returns>
static LR0State Goto(this RegulationList list, LR0State state, TreeNodeType x)
{
LR0State toState = new LR0State();
foreach (var item in state)
{
TreeNodeType nextNode = item.GetNodeNext2Dot();
if (nextNode == x)
{
var newItem = new LR0Item(item.Regulation, item.DotPosition + 1);
toState.TryInsert(newItem);
}
}
return(Closure(list, toState));
}
/// <summary>
/// 计算文法的各个RightPart的FIRST集
/// </summary>
/// <param name="grammar"></param>
/// <param name="nullableDict"></param>
/// <param name="firstList4Node"></param>
/// <returns></returns>
private static void GetFirstCollection4Regulation(
this RegulationList grammar,
out FIRSTCollection firstCollection,
Dictionary <TreeNodeType, bool> nullableDict = null, FIRSTCollection firstList4Node = null)
{
// 初始化FIRST集
firstCollection = new FIRSTCollection();
foreach (var regulation in grammar)
{
firstCollection.TryInsert(new FIRST(regulation.RightPart));
}
bool changed = false;
do
{
changed = false;
foreach (var first in firstCollection)
{
int count = first.Target.Count();
for (int i = 0; i < count; i++)
{
// 如果前i个结点都可为null，就说明下一个结点（first.Target[i]）的FIRST是此RightPart的FIRST的一部分
if (Nullable(first.Target, 0, i, nullableDict))
{
// 找到下一个结点的FIRST
FIRST refFirst = FindFirst(firstList4Node, first.GetNode(i));
foreach (var value in refFirst.Values)
{
if (value != TreeNodeType.NullNode)
{
changed = first.TryInsert(value) || changed;
}
}
}
}
{
// 如果RightPart的全部结点都可为null，就说明此RightPart的FIRST包含"null"结点。
if (Nullable(first.Target, 0, first.Target.Count(), nullableDict))
{
changed = first.TryInsert(TreeNodeType.NullNode) || changed;
}
}
}
} while (changed);
}
/// <summary>
/// LR(1)的Goto操作。
/// 将圆点移到所有LR(1)项中的符号<paramref name="x"/>之后。
/// </summary>
/// <param name="list"></param>
/// <param name="state"></param>
/// <param name="x">一个文法符号，终结点或非终结点。</param>
/// <param name="firstList"></param>
/// <returns></returns>
static SmallerLR1State Goto(this RegulationList list, SmallerLR1State state, TreeNodeType x, Dictionary <TreeNodeType, bool> nullableDict, FIRSTCollection firstList = null)
{
var toState = new SmallerLR1State();
foreach (var group in state.GetGroups())
{
TreeNodeType nextNode = group.Item1.GetNodeNext2Dot();
if (nextNode == x)
{
toState.TryInsert(
new LR0Item(group.Item1.Regulation, group.Item1.DotPosition + 1),
group.Item2);
}
}
return(Closure(list, toState, nullableDict, firstList));
}
/// <summary>
/// LR(1)的Closure操作。
/// 补全一个状态。
/// </summary>
/// <param name="grammar"></param>
/// <param name="state"></param>
/// <returns></returns>
static SmallerLR1State Closure(this RegulationList grammar, SmallerLR1State state,
Dictionary <TreeNodeType, bool> nullableDict = null, FIRSTCollection firstCollection = null)
{
if (nullableDict == null)
{
grammar.GetNullableDict(out nullableDict);
}
if (firstCollection == null)
{
grammar.GetFirstCollection(out firstCollection, nullableDict);
}
Queue <LR1Item> queue = new Queue <LR1Item>();
foreach (var item in state)
{
queue.Enqueue(item);
}
while (queue.Count > 0)
{
LR1Item item = queue.Dequeue();
TreeNodeType node = item.GetNodeNext2Dot();
if (node == null || node.IsLeave)
{
continue;
}
List <TreeNodeType> betaZ = item.GetBetaZ();
FIRST first = grammar.GetFirst(firstCollection, nullableDict, betaZ);
List <Regulation> regulations = grammar.GetRegulations(node);
foreach (var regulation in regulations)
{
foreach (var value in first.Values)
{
LR1Item newItem = new LR1Item(regulation, 0, value);
if (state.TryInsert(newItem))
{
queue.Enqueue(newItem);
}
}
}
}
return(state);
}
/// <summary>
/// 根据文法和已经计算的所有结点的FIRST集，计算任意一个结点串的FIRST。
/// </summary>
/// <param name="grammar"></param>
/// <param name="firstList4Node"></param>
/// <param name="target"></param>
/// <returns></returns>
private static FIRST GetFirst(this RegulationList grammar, FIRSTCollection firstList4Node, Dictionary <TreeNodeType, bool> nullableDict, List <TreeNodeType> target)
{
// 初始化FIRST集
FIRST first = new FIRST(target);
bool changed = false;
do
{
changed = false;
int count = first.Target.Count();
for (int checkCount = 0; checkCount < count; checkCount++)
{
// 如果前i个结点都可为null，就说明下一个结点（first.Target[i]）的FIRST是此RightPart的FIRST的一部分
if (Nullable(first.Target, 0, checkCount, nullableDict))
{
// 找到下一个结点的FIRST
FIRST refFirst = FindFirst(firstList4Node, first.GetNode(checkCount));
foreach (var value in refFirst.Values)
{
if (value != TreeNodeType.NullNode)
{
changed = first.TryInsert(value) || changed;
}
}
}
}
{
// 如果RightPart的全部结点都可为null，就说明此RightPart的FIRST包含"null"结点。
if (Nullable(first.Target, 0, first.Target.Count(), nullableDict))
{
//throw new Exception("如果是在LR1分析器生成工具的设计时走到此处，说明代码有问题。");
if (!first.Values.Contains(TreeNodeType.NullNode))
{
changed = first.TryInsert(TreeNodeType.NullNode) || changed;
}
}
}
} while (changed);
return(first);
}
public static void GetFollowCollection(
this RegulationList grammar,
out FOLLOWCollection followCollection,
Dictionary <TreeNodeType, bool> nullableDict = null,
FIRSTCollection firstCollection = null)
{
if (nullableDict == null)
{
grammar.GetNullableDict(out nullableDict);
}
if (firstCollection == null)
{
grammar.GetFirstCollection(out firstCollection, nullableDict);
}
FIRSTCollection firstList4Node;
grammar.GetFirstCollection4Node(out firstList4Node, nullableDict);
FIRSTCollection firstList4Regulation;
grammar.GetFirstCollection4Regulation(out firstList4Regulation, nullableDict, firstList4Node);
grammar.DoGetFollowList(out followCollection, nullableDict, firstList4Node);
}
/// <summary>
/// 计算文法的FOLLOW集
/// </summary>
/// <param name="grammar"></param>
/// <param name="nullableDict"></param>
/// <param name="firstList4Node"></param>
/// <returns></returns>
private static void DoGetFollowList(
this RegulationList grammar,
out FOLLOWCollection followCollection,
Dictionary <TreeNodeType, bool> nullableDict, FIRSTCollection firstList4Node)
{
// 初始化Follow list
followCollection = new FOLLOWCollection();
foreach (var item in grammar.GetAllTreeNodeNonLeaveTypes())
{
followCollection.TryInsert(new FOLLOW(item));
}
// 迭代到不动点
bool changed = false;
do
{
changed = false;
foreach (var regulation in grammar)
{
int count = regulation.RightPart.Count();
for (int index = 0; index < count; index++)
{
// 准备为target添加follow元素
TreeNodeType target = regulation.RightNode(index);
if (target.IsLeave)
{
continue;
} // 叶结点没有follow
FOLLOW follow = FindFollow(followCollection, target); // 找到follow对象
for (int checkCount = 0; checkCount < count - (index + 1); checkCount++)
{
if (Nullable(regulation.RightPart, index + 1, checkCount, nullableDict))
{
// nullable之后的FIRST是target的follow的一部分
FIRST first = FindFirst(
firstList4Node, regulation.RightNode(index + 1 + checkCount));
foreach (var value in first.Values)
{
if (value != TreeNodeType.NullNode)
{
changed = follow.TryInsert(value) || changed;
}
}
}
}
{
// 如果target之后的全部结点都是nullable，说明此regulation.Left的folow也是target的一部分。
if (Nullable(regulation.RightPart, index + 1, count - (index + 1), nullableDict))
{
// 找到此regulation.Left的folow
FOLLOW refFollow = FindFollow(followCollection, regulation.Left);
if (refFollow != follow)
{
foreach (var item in refFollow.Values)
{
changed = follow.TryInsert(item) || changed;
}
}
}
}
}
}
} while (changed);
}
/// <summary>
/// 用SLR分析法计算分析表
/// </summary>
/// <param name="grammar"></param>
/// <returns></returns>
public static void GetSLRParsingMap(this RegulationList grammar,
out LRParsingMap map,
out LR0StateCollection stateCollection,
out LR0EdgeCollection edgeCollection, TextWriter writer)
{
// 给文法添加一个辅助的开始产生式 S' -> S $
// 如何添加一个外来的结点类型？用Enum是无法做到的。
var decoratedS = new TreeNodeType("__S2", "S'", "<S'>");
var decoratedEnd = TreeNodeType.endOfTokenListNode;
var decoratedRegulation = new Regulation(
decoratedS, grammar[0].Left, decoratedEnd);
var decoratedGrammar = new RegulationList(decoratedRegulation);
decoratedGrammar.AddRange(grammar);
// 初始化T为{ Closure(S' -> S $) }
var firstItem = new LR0Item(decoratedGrammar[0], 0);
var firstState = new LR0State(firstItem);
firstState = decoratedGrammar.Closure(firstState);
stateCollection = new LR0StateCollection(firstState);
edgeCollection = new LR0EdgeCollection(stateCollection);
var queue = new Queue <LR0State>();
queue.Enqueue(firstState);
int lastOutputLength = 0;
int stateListCount = 1;
int queueCount = 1;
while (queue.Count > 0)
{
LR0State fromState = queue.Dequeue(); queueCount--;
int itemIndex = 0;
int itemCount = fromState.Count();
foreach (var item in fromState)
{
{
TextWriter currentWriter = Console.Out;
if (Console.Out != writer)
{
Console.SetOut(writer);
}
for (int i = 0; i < lastOutputLength; i++)
{
Console.Write('\u0008');
}
string output = string.Format("Calculating SLR State List: {0} <-- {1}, working on {2}/{3} ...",
stateListCount, queueCount, 1 + itemIndex++, itemCount);
Console.Write(output);
lastOutputLength = output.Length;
Console.SetOut(currentWriter);
}
TreeNodeType x = item.GetNodeNext2Dot();
if (x == null || x == decoratedEnd)
{
continue;
}
LR0State toState = decoratedGrammar.Goto(fromState, x);
if (stateCollection.TryInsert(toState))
{
queue.Enqueue(toState);
stateListCount++;
queueCount++;
var edge = new LR0Edge(fromState, x, toState);
edgeCollection.TryInsert(edge);
}
else
{
int index = stateCollection.IndexOf(toState);
toState = stateCollection[index];
var edge = new LR0Edge(fromState, x, toState);
edgeCollection.TryInsert(edge);
}
}
}
{
TextWriter currentWriter = Console.Out;
if (Console.Out != writer)
{
Console.SetOut(writer);
}
Console.WriteLine();
Console.SetOut(currentWriter);
}
map = new LRParsingMap();
foreach (var edge in edgeCollection)
{
if (edge.X.IsLeave)
{
int stateId = edge.From.ParsingMapIndex + 1; //stateCollection.IndexOf(edge.From) + 1
int gotoStateId = edge.To.ParsingMapIndex + 1; //stateCollection.IndexOf(edge.To) + 1
map.SetAction(stateId, edge.X, new LR1ShiftInAction(gotoStateId));
}
else
{
int stateId = edge.From.ParsingMapIndex + 1; //stateCollection.IndexOf(edge.From) + 1
int gotoStateId = edge.To.ParsingMapIndex + 1; //stateCollection.IndexOf(edge.To) + 1
map.SetAction(stateId, edge.X, new LR1GotoAction(gotoStateId));
}
}
var endItem = new LR0Item(decoratedRegulation, 1);
FOLLOWCollection followCollection;
decoratedGrammar.GetFollowCollection(out followCollection);
foreach (var state in stateCollection)
{
if (state.Contains(endItem))
{
int stateId = state.ParsingMapIndex + 1;//stateCollection.IndexOf(state) + 1
map.SetAction(stateId, decoratedEnd, new LR1AceptAction());
}
foreach (var lr0Item in state)
{
if (lr0Item.GetNodeNext2Dot() == null)
{
FOLLOW follow = FindFollow(followCollection, lr0Item.Regulation.Left);
foreach (var value in follow.Values)
{
int stateId = state.ParsingMapIndex + 1;// stateCollection.IndexOf(state) + 1;
int reductionId = decoratedGrammar.IndexOf(lr0Item.Regulation);
var action = new LR1ReducitonAction(reductionId);
map.SetAction(stateId, value, action);
}
}
}
}
}
/// <summary>
/// 用LR(1)分析法计算分析表
/// </summary>
/// <param name="grammar"></param>
/// <returns></returns>
public static void GetLR1ParsingMap(this RegulationList grammar,
out LRParsingMap map,
out LR1StateCollection stateCollection,
out LR1EdgeCollection edgeCollection, TextWriter originalOut)
{
// 给文法添加一个辅助的开始产生式 S' -> S $
// 如何添加一个外来的结点类型？用Enum是无法做到的。
var decoratedS = new TreeNodeType("__S2", "S'", "<S'>");
var decoratedEnd = TreeNodeType.endOfTokenListNode;
var decoratedRegulation = new Regulation(
decoratedS, grammar[0].Left, decoratedEnd);
var decoratedGrammar = new RegulationList(decoratedRegulation);
decoratedGrammar.AddRange(grammar);
// 初始化T为{ Closure(S' -> S $) }
var firstItem = new LR1Item(decoratedGrammar[0], 0, decoratedEnd);
var firstState = new SmallerLR1State(firstItem);
Dictionary <TreeNodeType, bool> nullableDict;
decoratedGrammar.GetNullableDict(out nullableDict);
FIRSTCollection firstCollection;
decoratedGrammar.GetFirstCollection(out firstCollection, nullableDict);
firstState = decoratedGrammar.Closure(firstState, nullableDict, firstCollection);
stateCollection = new LR1StateCollection(firstState);
edgeCollection = new LR1EdgeCollection(stateCollection);
var queue = new Queue <SmallerLR1State>();
queue.Enqueue(firstState);
int lastOutputLength = 0;
int stateListCount = 1;
int queueCount = 1;
while (queue.Count > 0)
{
SmallerLR1State fromState = queue.Dequeue(); queueCount--;
int itemIndex = 0;
int groupCount = fromState.GroupCount;
foreach (var group in fromState.GetGroups())
{
{
TextWriter currentWriter = Console.Out;
if (Console.Out != originalOut)
{
Console.SetOut(originalOut);
}
for (int i = 0; i < lastOutputLength; i++)
{
Console.Write('\u0008');
}
string output = string.Format("Calculating LR(1) State List: {0} <-- {1}, working on group {2}/{3} ...",
stateListCount, queueCount, 1 + itemIndex++, groupCount);
Console.Write(output);
lastOutputLength = output.Length;
Console.SetOut(currentWriter);
}
TreeNodeType x = group.Item1.GetNodeNext2Dot();
if (x == null || x == decoratedEnd)
{
continue;
}
SmallerLR1State toState = decoratedGrammar.Goto(fromState, x, nullableDict, firstCollection);
if (stateCollection.TryInsert(toState))//融入组织之中吧
{
int index = stateCollection.IndexOf(toState);
toState = stateCollection[index];
queue.Enqueue(toState);
stateListCount++;
queueCount++;
var edge = new LR1Edge(fromState, x, toState);
edgeCollection.TryInsert(edge);
}
else
{
int index = stateCollection.IndexOf(toState);
toState = stateCollection[index];
var edge = new LR1Edge(fromState, x, toState);
edgeCollection.TryInsert(edge);
}
}
}
{
TextWriter currentWriter = Console.Out;
if (Console.Out != originalOut)
{
Console.SetOut(originalOut);
}
Console.WriteLine();
Console.SetOut(currentWriter);
}
map = new LRParsingMap();
foreach (var edge in edgeCollection)
{
if (edge.X.IsLeave)
{
int stateId = edge.From.ParsingMapIndex + 1; // stateCollection.IndexOf(edge.From) + 1;
int gotoId = edge.To.ParsingMapIndex + 1; //stateCollection.IndexOf(edge.To) + 1
map.SetAction(stateId, edge.X, new LR1ShiftInAction(gotoId));
}
else
{
int stateId = edge.From.ParsingMapIndex + 1; // stateCollection.IndexOf(edge.From) + 1;
int gotoId = edge.To.ParsingMapIndex + 1; //stateCollection.IndexOf(edge.To) + 1
map.SetAction(stateId, edge.X, new LR1GotoAction(gotoId));
}
}
// TODO: not implemented
var endItem = new LR1Item(decoratedRegulation, 1, decoratedEnd);
foreach (var state in stateCollection)
{
if (state.Contains(endItem))
{
int stateId = state.ParsingMapIndex + 1;// stateCollection.IndexOf(state) + 1;
map.SetAction(stateId, decoratedEnd, new LR1AceptAction());
}
foreach (var LR1Item in state)
{
if (LR1Item.GetNodeNext2Dot() == null)
{
int stateId = state.ParsingMapIndex + 1;// stateCollection.IndexOf(state) + 1;
map.SetAction(stateId, LR1Item.LookAheadNodeType,
new LR1ReducitonAction(decoratedGrammar.IndexOf(LR1Item.Regulation)));
}
}
}
}
