// This file is part of BULL, a program for phylogenetic simulations
// most of the code was written by Mark T. Holder.

//	This program is for internal use by the lab of Dr. Tandy Warnow only.
//	Do not redistribute the code.  It is distributed in the hope that it will be useful,
//	but WITHOUT ANY WARRANTY; without even the implied warranty of
//	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  
//
//	Some of the code is from publically available source by Paul Lewis, Ziheng Yang, 
//	John Huelsenbeck, David Swofford , and others  (as noted in the code).
//	In fact the main structure of the program was created by modifying Paul Lewis' 
//	basiccmdline.cpp from his NCL
//
//	This code was used in Mark's dissertation, some changes were made in order to 
//	get it to compile on gcc.  It is possible that this porting introduced bugs (very
//	little debugging has been done on UNIX platforms).	I would suggest checking 
//	the simulator by generating data on trees with short branches, etc.
 

// This file is part of BULL, a program for phylogenetic simulations
// most of the code was written by Mark T. Holder.

//	This program is for internal use by the lab of Dr. Tandy Warnow only.
//	Do not redistribute the code.  It is distributed in the hope that it will be useful,
//	but WITHOUT ANY WARRANTY; without even the implied warranty of
//	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  
//
//	Some of the code is from publically available source by Paul Lewis, Ziheng Yang, 
//	John Huelsenbeck, David Swofford , and others  (as noted in the code).
//	In fact the main structure of the program was created by modifying Paul Lewis' 
//	basiccmdline.cpp from his NCL
//
//	This code was used in Mark's dissertation, some changes were made in order to 
//	get it to compile on gcc.  It is possible that this porting introduced bugs (very
//	little debugging has been done on UNIX platforms).	I would suggest checking 
//	the simulator by generating data on trees with short branches, etc.
 

#include "nexus_defs.hpp"
#include "xbull.hpp"
#include "nexus_token.hpp"
#include "nexus.hpp"
#include "taxa_block.hpp"
#include "trees_block.hpp"

using namespace::std;
/**
 * @class	   TreesBlock
 * @file	   taxa_block.hpp
 * @file	   taxablock.cpp
 * @author	   Paul O. Lewis
 * @copyright  Copyright © 1999. All Rights Reserved.
 */

/**
 * @constructor
 *
 * Initializes id to "TREES" and ntrees and defaultTrees to 0.
 */
TreesBlock::TreesBlock( TaxaBlock& tb ) : ntrees(0), defaultTree(0), taxa(tb), NexusBlock()
{
   id = "TREES";
}

/**
 * @destructor
 *
 * Flushes translateList, rooted, and treeList.
 */
TreesBlock::~TreesBlock()
{
	translateList.erase( translateList.begin(), translateList.end() );
	rooted.erase( rooted.begin(), rooted.end() );
	treeName.erase( treeName.begin(), treeName.end() );
	treeDescription.erase( treeDescription.begin(), treeDescription.end() );
}

/**
 * @method Read [void:protected]
 * @param token [NexusToken&] the token used to read from in
 * @throws XBull
 *
 * This function provides the ability to read everything following the block name
 * (which is read by the Nexus object) to the end or endblock statement.
 * Characters are read from the input stream in. Overrides the
 * abstract virtual function in the base class.
 */
void TreesBlock::Read( NexusToken& token )
{
   isEmpty = false;
	token.GetNextToken(); // this should be the semicolon after the block name
	if ( !token.Equals(";") ) {
	  errormsg = "Expecting ';' after TREES block name, but found ";
	  errormsg += token.GetToken();
	  errormsg += " instead";
		throw XBull( errormsg, token);
   }

	for (;;)
	{
		token.GetNextToken();

		if ( token.Equals("TRANSLATE") ) {

		 int numEntries = taxa.GetNumTaxonLabels();
		 for ( int k = 0; k < numEntries; k++ ) {
			// create the Association

			// get the key
			token.GetNextToken();
			std::string skey = token.GetToken();

			// get the value
			token.GetNextToken();
			std::string sval = token.GetToken();

			// add the Association object to the translate list
			translateList[ skey ] = sval;

			// this should be a comma, unless we are at the last pair, in
			// which case it should be a semicolon
			token.GetNextToken();
			if ( k < numEntries-1 )
			{
			   if ( !token.Equals(",") ) {
						errormsg = "Expecting ',' to terminate each key/value pair in TRANSLATE command, but found ";
						errormsg +=token.GetToken();
						errormsg += " instead";
				  throw XBull( errormsg, token);
			   }
			}
			else
			{
			   if ( !token.Equals(";") ) {
						errormsg = "Expecting ';' to terminate the TRANSLATE command, but found ";
						errormsg +=token.GetToken();
						errormsg += " instead";
				  throw XBull( errormsg, token);
			   }
			}

		 }

		}
		else if ( token.Equals("TREE") ) {

			// this should be either an asterisk or a tree name
			token.GetNextToken();
			if ( token.Equals("*") ) {
				defaultTree = ntrees; // ntrees is not incremented until entire tree command has been read

				// this should be tree name
				token.GetNextToken();
			}

		 // save the tree name as the key
		 std::string skey = token.GetToken();

		 // this should be an equals sign
		 token.GetNextToken();
		 if ( !token.Equals("=") ) {
				errormsg = "Expecting '=' after tree name in TREE command, but found ";
			errormsg += token.GetToken();
			errormsg += " instead";
				throw XBull( errormsg, token);
		 }

		 // this should be either a tree description or a command comment specifying
		 // whether this tree is to be rooted ([&R]) or unrooted ([&U]).
		 token.SetLabileFlagBit( NexusToken::saveCommandComments );
		 token.SetLabileFlagBit( NexusToken::parentheticalToken );
		 token.GetNextToken();

		 std::string s = token.GetToken();
		 if ( s.size() < 2 ) {
				errormsg = "Expecting command comment or tree description in TREE command, but found ";
			errormsg += token.GetToken();
			errormsg += " instead";
				throw XBull( errormsg, token);
		 }

		 if ( s[0] == '&' ) {
			// command comment found
			if ( s[1] == 'R' || s[1] == 'r' )
			   rooted.push_back(true);
				else if ( s[1] == 'U' || s[1] == 'u' )
					rooted.push_back(false);
			else {
				errormsg = "[";
			   errormsg += token.GetToken();
			   errormsg += "] is not a valid command comment in a TREE command";
				throw XBull( errormsg, token);
			}

			// this should be only the tree description
			token.SetLabileFlagBit( NexusToken::parentheticalToken );
			token.GetNextToken();
		 }

		 std::string sval = token.GetToken();

		 // this should be a semicolon
		 token.GetNextToken();
		 if ( !token.Equals(";") ) {
				errormsg = "Expecting ';' to terminate the TREE command, but found ";
			errormsg += token.GetToken();
			errormsg += " instead";
			throw XBull( errormsg, token);
		 }

		 ntrees++;
		 treeName.push_back( skey );
		 treeDescription.push_back( sval );
		 if ( rooted.size() < ntrees )
			rooted.push_back(false);
		}
		else if ( token.Equals("END") ) {
			// get the semicolon following END
			token.GetNextToken();
		 if ( !token.Equals(";") ) {
				errormsg = "Expecting ';' to terminate the END command, but found ";
			errormsg += token.GetToken();
			errormsg += " instead";
				throw XBull( errormsg, token);
		 }
			break;
		}
		else if ( token.Equals("ENDBLOCK") ) {
			// get the semicolon following ENDBLOCK
			token.GetNextToken();
		 if ( !token.Equals(";") ) {
				errormsg = "Expecting ';' to terminate the ENDBLOCK command, but found ";
			errormsg += token.GetToken();
			errormsg += " instead";
				throw XBull( errormsg, token);
		 }
			break;
		}
		else {
		SkippingCommand( token.GetToken() );
		 do {
			token.GetNextToken();
		 } while ( !token.AtEOF() && !token.Equals(";") );
		 if ( token.AtEOF() ) {
				errormsg = "Unexpected end of file encountered";
				throw XBull( errormsg, token);
		 }
		}
	}
}

/**
 * @method Reset [void:protected]
 *
 * Flushes treeList, translateList and rooted, and sets ntrees to 0
 * in preparation for reading a new TREES block.
 */
void TreesBlock::Reset()
{
   isEmpty = true;
	treeName.erase( treeName.begin(), treeName.end() );
	treeDescription.erase( treeDescription.begin(), treeDescription.end() );
	translateList.erase( translateList.begin(), translateList.end() );
	rooted.erase( rooted.begin(), rooted.end() );
	ntrees = 0;
}

/**
 * @method GetNumDefaultTree [int:public]
 *
 * Returns the 0-offset index of the default tree, which will be 0 if there is
 * only one tree stored or no trees stored.	 If more than one tree is stored,
 * the default tree will be the one specifically indicated by the user
 * (using an asterisk in the data file) to be the default tree (or 0 if the
 * user failed to specify.
 */
int TreesBlock::GetNumDefaultTree()
{
	return defaultTree;
}

/**
 * @method GetNumTrees [int:public]
 *
 * Returns the number of trees stored in this TreesBlock object.
 */
int TreesBlock::GetNumTrees()
{
	return ntrees;
}

/**
 * @method GetTreeName [char*:public]
 * @param i [int] the index of the tree for which the name is to be returned
 *
 * Returns the name of the tree stored at position i in treeList.  Assumes
 * that i will be in the range 0...ntrees-1.
 */
std::string TreesBlock::GetTreeName( int i )
{
	assert( i >= 0 );
	assert( i < ntrees );
   return treeName[i];
//trash return treeList.GetKey(i);
}

static std::string& blanks_to_underscores( std::string& s )
{
   int len = s.length();
   for ( int k = 0; k < len; k++ ) {
	  if ( s[k] == ' ' )
		 s[k] = '_';
   }
   return s;
}

/**
 * @method GetTranslatedTreeDescription [std::string:public]
 * @param i [int] the index of the tree for which the description is to be returned
 *
 * Returns the description of the tree stored at position i in treeList.  Assumes
 * that i will be in the range 0...ntrees-1.  Node numbers will be translated to
 * names in the resulting tree description.	 Use GetTreeDescription if translation
 * is not desired.	When translating, blank spaces in names are converted to
 * underscores.
 */
std::string TreesBlock::GetTranslatedTreeDescription( int i )
{
	assert( i >= 0 );
	assert( i < ntrees );
   std::string s = treeDescription[i];
   std::string x;
   x += s[0];
   int slen = s.size();
   assert( slen > 1 );
   for ( int k = 1; k < slen; k++ )
   {
	  char prev;
	  if (isgraph(s[k-1]))
		prev = s[k-1];
	  char curr = s[k];
	  if ( isdigit(curr) && ( prev == '(' || prev == ',' ) ) {
		 std::string ns;
		 ns += curr;
		 for (;;) {
			curr = s[k+1];
			prev = s[k++];
			if ( isdigit(curr) )
			   ns += curr;
			else {
			   --k;
			   break;
			}
		 }
		 std::string nss = translateList[ns];
		 x += blanks_to_underscores( nss );
	  }
	  else
		 x += curr;
   }
   return x;
}

/**
 * @method GetTreeDescription [std::string:public]
 * @param i [int] the index of the tree for which the description is to be returned
 *
 * Returns the description of the tree stored at position i in treeList.  Assumes
 * that i will be in the range 0...ntrees-1.
 */
std::string TreesBlock::GetTreeDescription( int i )
{
	assert( i >= 0 );
	assert( i < ntrees );
   return treeDescription[i];
//trash return treeList.GetValue(i);
}

/**
 * @method IsDefaultTree [int:public]
 * @param i [int] the index of the tree in question
 *
 * Returns true if the ith tree (0-offset) is the default tree, 0 otherwise.
 * Assumes that i will be in the range 0...ntrees-1.
 */
int TreesBlock::IsDefaultTree( int i )
{
	assert( i >= 0 );
	assert( i < ntrees );
	if ( i == GetNumDefaultTree() )
		return 1;
	else
		return 0;
}

/**
 * @method IsRootedTree [int:public]
 * @param i [int] the index of the tree in question
 *
 * Returns true if the ith tree (0-offset) is rooted, 0 otherwise.
 * Assumes that i will be in the range 0...ntrees-1.
 */
int TreesBlock::IsRootedTree( int i )
{
	assert( i >= 0 );
	assert( i < ntrees );
	return (int)rooted[i];
}

/**
 * @method Report [void:public]
 * @param out [ostream&] the output stream to which to write the report
 *
 * This function outputs a brief report of the contents of this taxa block.
 * Overrides the abstract virtual function in the base class.
 */
void TreesBlock::Report(std::ostream& out )
{
	out << endl;
	out << id << " block contains ";
	if ( ntrees == 0 ) {
		out << "no trees" << endl;
	}
	else if ( ntrees == 1 )
		out << "one tree" << endl;
	else
		out << ntrees << " trees" << endl;

	if ( ntrees == 0 ) return;

	for ( int k = 0; k < ntrees; k++ ) {
		out << '\t' << (k+1) << '\t' << treeName[k];
		out << "\t(";
		if ( rooted[k] )
			out << "rooted";
		else
			out << "unrooted";
		if ( defaultTree == k )
			out << ",default tree)" << endl;
		else
			out << ')' << endl;
	}
}