//------------------------------------------------------------------------------
char Tokeniser::GetNextChar ()
{
int ch;
#ifdef __MWERKS__
if (putBuffer != '\0')
{
ch = putBuffer;
putBuffer = '\0';
}
else
ch = in.get();
#else
if (putBackChar == '\0')
{
ch = in.get();
}
else
{
ch = putBackChar;
}
putBackChar = '\0';
#endif
int failed = in.bad();
if( failed )
throw XTokeniser ( "Unknown error reading data file (check to make sure file exists)" );
//std::cout << "[" << (char)ch << "]" << std::endl;
if( ch == 13 || ch == 10 )
{
fileline++;
filecol = 1L;
if( ch == 13 && (int)in.peek() == 10 )
ch = in.get();
atEOL = true;
}
else if( ch == EOF )
{
atEOF = true;
}
else
{
filecol++;
atEOL = false;
}
filepos = in.tellg();
if (atEOF )
return '\0';
else if (atEOL )
return '\n';
else
return (char)ch;
}
//------------------------------------------------------------------------------
// Uses a simple pushdown automaton to read Newick-style trees
bool TreeReader::Read (TreePtr t)
{
// States of pushdown automaton that reads trees
enum statetype
{
GETNAME,
GETINTERNODE,
NEXTMOVE,
DOSIBLING,
FINISHCHILDREN,
ACCEPTED,
CLEANUP,
QUIT
} state;
std::stack< NodePtr, std::vector<NodePtr> > stk;
Tokeniser::tokentype token;
tree = t;
tree->MakeRoot();
token = parser.GetNextToken ();
if (token == Tokeniser::EMPTY)
return false;
// Parse the tree description
state = GETNAME;
while ((state != QUIT) && (state != ACCEPTED))
{
switch (state)
{
case GETNAME:
switch (token)
{
case Tokeniser::STRING:
case Tokeniser::NUMBER:
LabelLeaf (parser.GetToken());
token = parser.GetNextToken ();
state = GETINTERNODE;
break;
case Tokeniser::LPAR:
state = NEXTMOVE;
break;
default:
errormsg = "Syntax error [GETNAME]: expecting a \"(\" or leaf name, got \"";
errormsg += parser.GetToken();
errormsg += "\" instead";
state = QUIT;
break;
}
break;
case GETINTERNODE:
switch (token)
{
case Tokeniser::COLON:
case Tokeniser::COMMA:
case Tokeniser::RPAR:
state = NEXTMOVE;
break;
default:
errormsg = "Syntax error [GETINTERNODE]: expecting one of \":,)\", got ";
errormsg += parser.GetToken();
errormsg += " instead";
state = QUIT;
break;
}
break;
case NEXTMOVE:
switch (token)
{
case Tokeniser::COLON:
if (LabelEdge ())
token = parser.GetNextToken ();
else
state = QUIT;
break;
// The next node encountered will be a sibling
// of Curnode and a descendant of the node on
// the top of the node stack.
case Tokeniser::COMMA:
if (stk.empty())
{
errormsg = "Tree description unbalanced, this \")\" has no matching \"(\"";
state = QUIT;
}
else
{
tree->MakeSibling ();
//token = parser.GetNextToken ();
token = GetTaxonName();
state = GETNAME;
}
break;
// The next node will be a child of CurNode, hence
// we create the node and push CurNode onto the
// node stack.
case Tokeniser::LPAR:
stk.push (tree->GetCurNode());
tree->MakeChild();
//token = parser.GetNextToken ();
token = GetTaxonName();
state = GETNAME;
break;
// We've finished ready the descendants of the node
// at the top of the node stack so pop it off.
case Tokeniser::RPAR:
if (stk.empty())
{
errormsg = "Tree description unbalanced (an extra \")\")";
state = QUIT;
}
else
{
NodePtr q = stk.top();
q->AddWeight(tree->GetCurNode()->GetWeight());
tree->SetCurNode (q);
stk.pop ();
token = parser.GetNextToken ();
state = FINISHCHILDREN;
}
break;
// We should have finished the tree
case Tokeniser::SEMICOLON:
if (stk.empty())
{
state = ACCEPTED;
}
else
{
errormsg = "Tree description ended prematurely (stack not empty)";
state = QUIT;
}
break;
default:
errormsg = "Syntax error [NEXTMOVE]: expecting one of \":,();\", got ";
errormsg += parser.GetToken();
errormsg += " instead";
state = QUIT;
break;
}
break;
case FINISHCHILDREN:
switch (token)
{
case Tokeniser::STRING:
case Tokeniser::NUMBER:
LabelInternalNode (parser.GetToken());
token = parser.GetNextToken ();
break;
case Tokeniser::COLON:
if (LabelEdge ())
token = parser.GetNextToken ();
else
state = QUIT;
break;
// We've completed traversing the descendants of the
// node at the top of the stack, so pop it off.
case Tokeniser::RPAR:
if (stk.empty())
{
errormsg = "Tree description unbalanced, this \")\" has no matching \"(\"";
state = QUIT;
}
else
{
NodePtr q = stk.top();
q->AddWeight(tree->GetCurNode()->GetWeight());
tree->SetCurNode (q);
stk.pop ();
token = parser.GetNextToken ();
}
break;
// The node at the top of the stack still has some
// descendants.
case Tokeniser::COMMA:
if (stk.empty())
{
errormsg = "Tree description unbalanced, missing a \"(\"";
state = QUIT;
}
else
{
tree->MakeSibling ();
//token = parser.GetNextToken ();
token = GetTaxonName();
state = GETNAME;
}
break;
case Tokeniser::SEMICOLON:
state = NEXTMOVE;
break;
default:
if (stk.empty())
{
errormsg = "Tree description unbalanced";
state = QUIT;
}
else
{
errormsg = "Syntax error [FINISHCHILDREN]: expecting one of \":,();\" or internal label, got ";
errormsg += parser.GetToken();
errormsg += " instead";
}
state = QUIT;
break;
}
break;
}
}
// Handle errors
if (state == QUIT)
{
// Clean memory here...
// ... then throw exception
throw XTokeniser (errormsg, parser.GetFilePosition(),
parser.GetFileLine (), parser.GetFileColumn());
}
else
{
tree->GetRoot()->SetWeight(tree->GetNumLeaves());
doAdjust ();
}
return true;
}
