public static cSet <cLexem> cm_FirstTerminals(List <cLexem> a_listLexem)
{
cSet <cLexem> _tempSet = new cSet <cLexem>();
if (a_listLexem.Count > 0)
{
int _pos = 0;
cLexem _currLexem = a_listLexem[_pos];
do
{
_tempSet.Remove(cLexem.cc_EpsilonLexem);
_currLexem = a_listLexem[_pos];
_tempSet.UnionWith(_currLexem.cp_FirstCache);
_pos++;
} while (_tempSet.Contains(cLexem.cc_EpsilonLexem) && (_pos < a_listLexem.Count));
}
cSet <cLexem> _retSet = new cSet <cLexem>();
foreach (cLexem _lexem in _tempSet)
{
if (_lexem.cp_Type != eLexType.NonTerminal)
{
_retSet.Add(_lexem);
}
}
return(_retSet);
}
public object Clone()
{
cSet <T> _retSet = new cSet <T>();
_retSet.UnionWith(this);
return(_retSet);
}
private cLexem(string a_name, eLexType a_type)
{
cf_firstCache = null;
cf_EpsilonCount = 0;
cf_Name = a_name;
cf_type = a_type;
if (a_name == cc_Epsilon)
{
cf_type = eLexType.Epsilon;
}
}
private cSet <cLexem> cm_getAllPossibleGotoLexems(cSet <cConfiguration> a_set)
{
cSet <cLexem> _retSet = new cSet <cLexem>();
foreach (cConfiguration _configuration in a_set)
{
if (_configuration.cf_Production.cp_RightPart.Count > _configuration.cf_Position)
{
_retSet.Add(_configuration.cf_Production.cp_RightPart[_configuration.cf_Position]);
}
}
_retSet.Remove(cLexem.cc_EpsilonLexem);
return(_retSet);
}
public static cSet <cConfiguration> cm_Goto(cSet <cConfiguration> a_listConfiguration, cLexem a_lexem)
{
cSet <cConfiguration> _validConfigurations = new cSet <cConfiguration>();
foreach (cConfiguration _configuration in a_listConfiguration)
{
if (_configuration.cf_Production.cp_RightPart.Count > _configuration.cf_Position)
{
if (_configuration.cf_Production.cp_RightPart[_configuration.cf_Position] == a_lexem)
{
cConfiguration _newConfiguration = cConfiguration.cm_GetConfiguration(_configuration.cf_Production, _configuration.cf_Position + 1, _configuration.cf_Terminal);
_validConfigurations.Add(_newConfiguration);
}
}
}
return(cm_Closure(_validConfigurations));
}
private cSet <cLexem> cm_First(List <cLexem> a_listLexem)
{
cSet <cLexem> _retSet = new cSet <cLexem>();
if (a_listLexem.Count > 0)
{
int _pos = a_listLexem.Count - 1;
cLexem _currLexem = a_listLexem[_pos];
do
{
_retSet.Remove(cLexem.cc_EpsilonLexem);
_currLexem = a_listLexem[_pos];
_retSet.UnionWith(_currLexem.cp_FirstCache);
_pos--;
} while (_retSet.Contains(cLexem.cc_EpsilonLexem) && (_pos >= 0));
}
return(_retSet);
}
public static cSet <cConfiguration> cm_Closure(cSet <cConfiguration> a_listConfiguration)
{
bool _added = false;
if (a_listConfiguration != null)
{
cSet <cConfiguration> _retList = a_listConfiguration.Clone() as cSet <cConfiguration>;
cSet <cConfiguration> _addRetList = new cSet <cConfiguration>();
do
{
_addRetList.Clear();
foreach (cConfiguration _configuration in _retList)
{
if (_configuration.cf_Production.cp_RightPart.Count > _configuration.cf_Position)
{
cLexem _lexem = _configuration.cf_Production.cp_RightPart[_configuration.cf_Position];
if (_lexem.cp_Type == eLexType.NonTerminal)
{
List <cLexem> _listTerminals = new List <cLexem>();
_listTerminals.AddRange(_configuration.cf_Production.cp_RightPart.GetRange(_configuration.cf_Position + 1, _configuration.cf_Production.cp_RightPart.Count - _configuration.cf_Position - 1));
_listTerminals.Add(_configuration.cf_Terminal);
cSet <cLexem> _firstTerminals = cm_FirstTerminals(_listTerminals);
foreach (cProduction _production in _lexem.cp_ListProducts)
{
foreach (cLexem _terminal in _firstTerminals)
{
cConfiguration _newConfiguration = cConfiguration.cm_GetConfiguration(_production, 0, _terminal);
_addRetList.Add(_newConfiguration);
}
}
}
}
}
_added = _retList.AddRange(_addRetList);
} while (_added == true);
return(_retList);
}
else
{
return(new cSet <cConfiguration>());
}
}
private bool cm_addItemGroup(List <cSet <cConfiguration> > a_items, cSet <cConfiguration> a_item, cLexem a_lexem, cSet <cConfiguration> a_set, Dictionary <cSet <cConfiguration>, Dictionary <cLexem, cSet <cConfiguration> > > a_goto)
{
bool _notSubset = true;
foreach (cSet <cConfiguration> _set in a_items)
{
if (_set.IsSupersetOf(a_set))
{
_notSubset = false;
cm_addToGoto(a_goto, a_item, a_lexem, _set);
break;
}
}
if (_notSubset)
{
a_items.Add(a_set);
cm_addToGoto(a_goto, a_item, a_lexem, a_set);
}
return(_notSubset);
}
List <cLexem> cm_getLeadLexemsInternal(List <cLexem> a_watchedLexems)
{
List <cLexem> _retLst = new List <cLexem>();
if (a_watchedLexems.Contains(this))
{
//if (cf_type == eLexType.Terminal || cf_type == eLexType.NonTerminal)
throw new cNotLL1Exception(this, null, String.Empty);
}
else
{
a_watchedLexems.Add(this);
switch (cf_type)
{
case eLexType.NonTerminal:
foreach (KeyValuePair <cLexem, List <cLexem> > _kvp in cf_listProducts)
{
if (_kvp.Key.cp_Type != eLexType.Epsilon)
{
cm_fillNonTerminalLeadingLexList(a_watchedLexems, _kvp.Value, 0, _retLst);
}
else
{
_retLst.Add(_kvp.Key);
}
}
if (cf_leadLexems == null)
{
cf_leadLexems = new cSet <cLexem>();
cf_leadLexems.AddRange(_retLst);
}
break;
case eLexType.Terminal:
case eLexType.Epsilon:
_retLst.Add(this);
break;
}
}
return(_retLst);
}
private KeyValuePair <List <cSet <cConfiguration> >, Dictionary <cSet <cConfiguration>, Dictionary <cLexem, cSet <cConfiguration> > > > cm_Items(cLexem a_root)
{
List <cSet <cConfiguration> > _retList = new List <cSet <cConfiguration> >();
cConfiguration _rootConfiguration = cConfiguration.cm_GetConfiguration(a_root.cp_ListProducts[0], 0, cLexem.cc_StopLexem);
cSet <cConfiguration> _rootClosure = cConfiguration.cm_Closure(new cSet <cConfiguration>()
{
_rootConfiguration
});
Dictionary <cSet <cConfiguration>, Dictionary <cLexem, cSet <cConfiguration> > > _gotoResults = new Dictionary <cSet <cConfiguration>, Dictionary <cLexem, cSet <cConfiguration> > >();
_retList.Add(_rootClosure);
bool _added = false;
int _lastChecked = 0;
do
{
_added = false;
for (; _lastChecked < _retList.Count; _lastChecked++)
{
cSet <cConfiguration> _currentItem = _retList[_lastChecked];
cSet <cLexem> _currentChecked = cm_getAllPossibleGotoLexems(_currentItem);
foreach (cLexem _lexem in _currentChecked)
{
cSet <cConfiguration> _possibleItem = cConfiguration.cm_Goto(_currentItem, _lexem);
if (cm_addItemGroup(_retList, _currentItem, _lexem, _possibleItem, _gotoResults))
{
_added = true;
}
}
}
} while (_added);
return(new KeyValuePair <List <cSet <cConfiguration> >, Dictionary <cSet <cConfiguration>, Dictionary <cLexem, cSet <cConfiguration> > > >(_retList, _gotoResults));
}
//private static cSet<cLexem> cm_FirstInList(List<cLexem> a_listLexem)
//{
// cSet<cLexem> _retSet = new cSet<cLexem>();
// if (a_listLexem.Count > 0)
// {
// int _pos = 0;
// cLexem _currLexem = a_listLexem[_pos];
// do
// {
// _retSet.Remove(cLexem.cc_EpsilonLexem);
// _currLexem = a_listLexem[_pos++];
// _currLexem.cm_First(_retSet);
// } while (_retSet.Contains(cLexem.cc_EpsilonLexem) && (_pos < a_listLexem.Count));
// }
// return _retSet;
//}
//public cSet<cLexem> cm_First(cSet<cLexem> a_set)
//{
// if (cf_firstCache == null)
// {
// cSet<cLexem> _retSet = a_set ?? new cSet<cLexem>();
// switch (cf_type)
// {
// case (eLexType.Epsilon):
// case (eLexType.Stop):
// case (eLexType.Terminal):
// _retSet.Add(this);
// break;
// case (eLexType.NonTerminal):
// if (cf_EpsilonCount > 0)
// {
// _retSet.Add(cc_EpsilonLexem);
// }
// bool _added = false;
// do
// {
// _added = false;
// foreach (cProduction _production in cf_listProducts)
// {
// if (_production.cp_RightPart.Count > 0)
// {
// int _pos = 0;
// cLexem _currLexem = _production.cp_RightPart[_pos];
// do
// {
// _retSet.Remove(cLexem.cc_EpsilonLexem);
// _currLexem = _production.cp_RightPart[_pos++];
// if (_currLexem != this)
// {
// _currLexem.cm_First(_retSet);
// }
// else
// {
// break;
// //if (_retSet.Contains(cLexem.cc_EpsilonLexem))
// //{
// // // int _countBefore=_retSet.Count;
// // // for
// //}
// //else
// //{
// // break;
// //}
// }
// } while (_retSet.Contains(cLexem.cc_EpsilonLexem) && (_pos < _production.cp_RightPart.Count));
// }
// }
// } while (_added);
// break;
// default:
// throw new Exception("Передан некорректный тип в cm_First: " + cf_type.ToString());
// }
// cf_firstCache = _retSet.Clone() as cSet<cLexem>;
// return _retSet;
// }
// else
// {
// if (a_set == null)
// {
// return cf_firstCache.Clone() as cSet<cLexem>;
// }
// else
// {
// a_set.AddRange(cf_firstCache);
// return a_set;
// }
// }
//}
public static void cm_CacheFirst()
{
Dictionary <cLexem, cSet <cLexem> > _firstPerLexem = new Dictionary <cLexem, cSet <cLexem> >();
foreach (cLexem _lexem in cf_LexemDic.Values)
{
_firstPerLexem.Add(_lexem, new cSet <cLexem>());
}
_firstPerLexem.Add(cLexem.cc_EpsilonLexem, new cSet <cLexem>());
_firstPerLexem.Add(cLexem.cc_StopLexem, new cSet <cLexem>());
bool _added = false;
do
{
_added = false;
foreach (cLexem _lexem in _firstPerLexem.Keys)
{
switch (_lexem.cf_type)
{
case eLexType.Epsilon:
case eLexType.Stop:
case eLexType.Terminal:
_added |= _firstPerLexem[_lexem].Add(_lexem);
break;
case eLexType.NonTerminal:
foreach (cProduction _production in _lexem.cf_listProducts)
{
if (_production.cp_RightPart.Count > 0)
{
int _pos = 0;
cLexem _currLexem = _production.cp_RightPart[_pos];
bool _hasEpsilon = false;
do
{
_currLexem = _production.cp_RightPart[_pos];
_hasEpsilon = false;
cSet <cLexem> _currentSet = _firstPerLexem[_currLexem];
_hasEpsilon |= _currentSet.Contains(cLexem.cc_EpsilonLexem);
if (_hasEpsilon)
{
_currentSet = _currentSet.Clone() as cSet <cLexem>;
_currentSet.Remove(cLexem.cc_EpsilonLexem);
}
_added |= _firstPerLexem[_lexem].AddRange(_currentSet);
_pos++;
} while (_hasEpsilon && (_pos < _production.cp_RightPart.Count));
if (_hasEpsilon)
{
_added |= _firstPerLexem[_lexem].Add(cLexem.cc_EpsilonLexem);
}
}
}
break;
case eLexType.Action:
break;
default:
throw new Exception("Некорректный тип в cm_CacheFirst: " + _lexem.cf_type.ToString());
}
}
} while (_added);
foreach (cLexem _lexem in _firstPerLexem.Keys)
{
_lexem.cf_firstCache = _firstPerLexem[_lexem];
}
}
public void cm_Init(cParser a_parser)
{
cf_valid = false;
cf_root = a_parser.cp_Root;
// Заполнение множества FOLLOW
cf_follow = new Dictionary <cLexem, cSet <cLexem> >();
foreach (cLexem _nonTerminal in cp_Lexems)
{
if (_nonTerminal.cp_Type == eLexType.NonTerminal)
{
cf_follow[_nonTerminal] = new cSet <cLexem>();
}
}
// 2
cSet <cLexem> _first = new cSet <cLexem>();
foreach (cLexem _nonTerminal in cp_Lexems)
{
if (_nonTerminal.cp_Type == eLexType.NonTerminal)
{
foreach (List <cLexem> _lstLex in _nonTerminal.cp_ListProducts.Values)
{
int _count = _lstLex.Count;
_first.Clear();
for (int i = _lstLex.Count - 1; i >= 0; i--)
{
cLexem _lex = _lstLex[i];
switch (_lex.cp_Type)
{
case eLexType.NonTerminal:
cf_follow[_lex].AddRange(_first);
if (!_lex.cm_IsLeadingLexem(cLexem.cc_EpsilonLexem))
{
_first.Clear();
}
_first.AddRange(_lex.cp_LeadLexems);
break;
case eLexType.Terminal:
_first.Clear();
_first.Add(_lex);
break;
case eLexType.Epsilon:
_first.Clear();
break;
}
}
}
}
}
// 3
bool _added = true;
while (_added)
{
_added = false;
foreach (cLexem _nonTerminal in cp_Lexems)
{
if (_nonTerminal.cp_Type == eLexType.NonTerminal)
{
foreach (List <cLexem> _lstLex in _nonTerminal.cp_ListProducts.Values)
{
int _count = _lstLex.Count;
_first.Clear();
bool _break = false;
for (int i = _lstLex.Count - 1; i >= 0; i--)
{
cLexem _lex = _lstLex[i];
switch (_lex.cp_Type)
{
case eLexType.NonTerminal:
cf_follow[_lex].AddRange(_first);
if (!_lex.cm_IsLeadingLexem(cLexem.cc_EpsilonLexem))
{
_break = true;
}
//else
_added = cf_follow[_lex].AddRange(cf_follow[_nonTerminal]) | _added;
break;
case eLexType.Action:
break;
default:
_break = true;
break;
}
if (_break)
{
break;
}
}
}
}
}
}
cf_valid = true;
}
DataTable cm_fillDataTable(cProductInfo[] a_data)
{
cm_checkProducts(a_data);
Dictionary <cLexem, int> _dicJump = new Dictionary <cLexem, int>();
for (int k = 0; k < a_data.Length; k++)
{
if (!_dicJump.ContainsKey(a_data[k].cp_Root.Key))
{
_dicJump.Add(a_data[k].cp_Root.Key, a_data[k].cp_Root.Value);
}
}
DataTable _retDT = new DataTable("table");
_retDT.Columns.Add("i", typeof(int));
_retDT.Columns.Add("terminals", typeof(string[]) /*typeof(object)*/);
_retDT.Columns.Add("jump", typeof(int));
_retDT.Columns.Add("accept", typeof(bool));
_retDT.Columns.Add("stack", typeof(bool));
_retDT.Columns.Add("return", typeof(bool));
_retDT.Columns.Add("error", typeof(bool));
_retDT.Columns.Add("action", typeof(string));
int _count = a_data.Length;
bool _warning = false;
for (int k = 0; k < _count; k++)
{
cProductInfo _info = a_data[k];
// Проверка LL(1)
if (k == 0 || a_data[k - 1].cp_Root.Key != _info.cp_Root.Key)
{
_warning = _info.cp_Root.Key.cm_HasEpsilonProduct();
}
foreach (cLexem _lex in _info.cp_ArrDirLexems)
{
if (cf_follow[_info.cp_Root.Key].ContainsKey(_lex))
{
if (_warning)
{
throw new Exception("Невозможно однозначно определить переход из-за " + cLexem.cc_Epsilon + "-порождающей продукции в " + _info.cp_Root.Key + ".");
}
if (_lex.cp_Type != eLexType.Epsilon)
{
_warning = true;
}
}
else if (_lex.cp_Type == eLexType.Epsilon)
{
if (_warning)
{
throw new Exception("Невозможно однозначно определить переход из-за " + cLexem.cc_Epsilon + "-порождающей продукции в " + _info.cp_Root.Key + ".");
}
_warning = true;
}
}
int _i = _info.cp_Root.Value;
cSet <string> _terminals = new cSet <string>();
foreach (cLexem _lex in _info.cp_ArrDirLexems)
{
if (_lex.cp_Type == eLexType.Terminal)
{
_terminals.Add(_lex.cf_Name);
}
}
int _jump = _info.cp_ArrLexems[0].Value;
bool _accept = false;
bool _stack = false;
bool _return = false;
bool _error = false;
if (k == _count - 1 || a_data[k + 1].cp_Root.Key != _info.cp_Root.Key)
{
_error = true;
}
string _action = "";
//_terminals.Sort();
_retDT.Rows.Add(_i, _terminals.ToArray(), _jump, _accept, _stack, _return, _error, _action);
KeyValuePair <cLexem, int>[] _arrLexems = _info.cp_ArrLexems;
int _lexCount = _arrLexems.Length;
for (int j = 0; j < _lexCount; j++)
{
cLexem _lexem = _arrLexems[j].Key;
_i = _arrLexems[j].Value;
// ERROR
_error = true;
// TERMINALS
switch (_lexem.cp_Type)
{
case eLexType.Epsilon:
_terminals = new cSet <string>();
//foreach (cLexem _lex in _info.cp_ArrDirLexems)
// _terminals.Add(_lex.cf_Name);
break;
case eLexType.Terminal:
_terminals = new cSet <string>();
_terminals.Add(_lexem.cf_Name);
break;
case eLexType.NonTerminal:
_terminals = new cSet <string>();
foreach (cLexem _lex in _lexem.cp_LeadLexems)
{
if (_lex.cp_Type == eLexType.Epsilon)
{
if (j == _lexCount - 1)
{
_retDT.Rows.Add(_i + 1, new string[0], 0, false, false, true, true, "");
}
_error = false;
}
else
{
_terminals.Add(_lex.cf_Name);
}
}
break;
case eLexType.Action:
_terminals = new cSet <string>();
break;
}
// JUMP
if (_lexem.cp_Type == eLexType.NonTerminal)
{
_jump = _dicJump[_lexem];
}
else
{
// Не последний терминал или действие
//if (j < _lexCount - 1)
_jump = _i + 1;
// Последний терминал или действие
//else
// _jump = 0;
}
// ACCEPT
_accept = _lexem.cp_Type == eLexType.Terminal;
// STACK
// Не последний нетерминал правой части
_stack = (_lexem.cp_Type == eLexType.NonTerminal && j < _lexCount - 1);
// RETURN
// Крайний правый терминал или действие
_return = ((_lexem.cp_Type != eLexType.NonTerminal) && j == _lexCount - 1);
// ACTION
_action = "";
if (_lexem.cp_Type == eLexType.Action)
{
_action = _lexem.cf_Name;
}
//_terminals.Sort();
_retDT.Rows.Add(_i, _terminals.ToArray(), _jump, _accept, _stack, _return, _error, _action);
}
}
// Сортировка строк
DataTable _oldDT = _retDT;
_retDT = _oldDT.Clone();
foreach (DataRow _dr in _oldDT.Select(null, "i"))
{
_retDT.Rows.Add(_dr.ItemArray);
}
return(_retDT);
}
private bool cm_addToGoto(Dictionary <cSet <cConfiguration>, Dictionary <cLexem, cSet <cConfiguration> > > a_goto, cSet <cConfiguration> a_item, cLexem a_lexem, cSet <cConfiguration> a_gotoItem)
{
bool _retVal;
if (!a_goto.ContainsKey(a_item))
{
a_goto.Add(a_item, new Dictionary <cLexem, cSet <cConfiguration> >());
}
if (!a_goto[a_item].ContainsKey(a_lexem))
{
a_goto[a_item].Add(a_lexem, a_gotoItem);
_retVal = true;
}
else
{
_retVal = false;
throw new Exception("Попытка заменить значение в дереве Goto при построении списка пунктов");
}
return(_retVal);
}
private Dictionary <cLexem, cSet <cLexem> > cm_Follow()
{
Dictionary <cLexem, cSet <cLexem> > _retDic = new Dictionary <cLexem, cSet <cLexem> >();
foreach (cLexem _nonTerminal in cp_Lexems)
{
if (_nonTerminal.cp_Type == eLexType.NonTerminal)
{
_retDic[_nonTerminal] = new cSet <cLexem>();
}
}
_retDic[cf_root].Add(cLexem.cc_StopLexem);
// 2
foreach (cLexem _nonTerminal in _retDic.Keys)
{
foreach (cProduction _production in _nonTerminal.cp_ListProducts)
{
int _count = _production.cp_RightPart.Count - 1;
List <cLexem> _revListProduct = new List <cLexem>();
for (int i = _count; i >= 0; i--)
{
cLexem _lex = _production.cp_RightPart[i];
switch (_lex.cp_Type)
{
case eLexType.NonTerminal:
cSet <cLexem> _first = cm_First(_revListProduct);
_first.Remove(cLexem.cc_EpsilonLexem);
_retDic[_lex].AddRange(_first);
break;
default:
break;
}
_revListProduct.Add(_lex);
}
}
}
// 3
bool _added = true;
while (_added)
{
_added = false;
foreach (cLexem _nonTerminal in _retDic.Keys)
{
foreach (cProduction _production in _nonTerminal.cp_ListProducts)
{
int _count = _production.cp_RightPart.Count - 1;
List <cLexem> _revListProduct = new List <cLexem>();
bool _break = false; //?
for (int i = _count; i >= 0; i--)
{
cLexem _lex = _production.cp_RightPart[i];
switch (_lex.cp_Type)
{
case eLexType.NonTerminal:
cSet <cLexem> _first = cm_First(_revListProduct);
if (_first.Contains(cLexem.cc_EpsilonLexem) | (i == _count))
{
//_production.cp_Root.cm_First(_first);
//_first.UnionWith(_production.cp_Root.cp_FirstCache);
_first.Remove(cLexem.cc_EpsilonLexem);
_first.UnionWith(_retDic[_nonTerminal]);
}
_added |= _retDic[_lex].AddRange(_first);
break;
case eLexType.Action:
break;
default:
_break = true;
break;
}
_revListProduct.Add(_lex);
if (_break)
{
break;
}
}
}
}
}
return(_retDic);
}
