// finds cycles
void dfsFindCycles(ROMol &mol,int atomIdx,int inBondIdx,
std::vector<AtomColors> &colors,
const UINT_VECT &ranks,
INT_VECT &atomOrders,
VECT_INT_VECT &atomRingClosures,
const boost::dynamic_bitset<> *bondsInPlay,
const std::vector<std::string> *bondSymbols
){
Atom *atom = mol.getAtomWithIdx(atomIdx);
atomOrders.push_back(atomIdx);
colors[atomIdx] = GREY_NODE;
// ---------------------
//
// Build the list of possible destinations from here
//
// ---------------------
std::vector< PossibleType > possibles;
possibles.resize(0);
ROMol::OBOND_ITER_PAIR bondsPair = mol.getAtomBonds(atom);
possibles.reserve(bondsPair.second-bondsPair.first);
while(bondsPair.first != bondsPair.second){
BOND_SPTR theBond = mol[*(bondsPair.first)];
bondsPair.first++;
if(bondsInPlay && !(*bondsInPlay)[theBond->getIdx()]) continue;
if(inBondIdx<0 || theBond->getIdx() != static_cast<unsigned int>(inBondIdx)){
int otherIdx = theBond->getOtherAtomIdx(atomIdx);
long rank=ranks[otherIdx];
// ---------------------
//
// things are a bit more complicated if we are sitting on a
// ring atom. we would like to traverse first to the
// ring-closure atoms, then to atoms outside the ring first,
// then to atoms in the ring that haven't already been visited
// (non-ring-closure atoms).
//
// Here's how the black magic works:
// - non-ring atom neighbors have their original ranks
// - ring atom neighbors have this added to their ranks:
// (MAX_BONDTYPE - bondOrder)*MAX_NATOMS*MAX_NATOMS
// - ring-closure neighbors lose a factor of:
// (MAX_BONDTYPE+1)*MAX_NATOMS*MAX_NATOMS
//
// This tactic biases us to traverse to non-ring neighbors first,
// original ordering if bond orders are all equal... crafty, neh?
//
// ---------------------
if( colors[otherIdx] == GREY_NODE ) {
rank -= static_cast<int>(MAX_BONDTYPE+1) *
MAX_NATOMS*MAX_NATOMS;
if(!bondSymbols){
rank += static_cast<int>(MAX_BONDTYPE - theBond->getBondType()) *
MAX_NATOMS;
} else {
const std::string &symb=(*bondSymbols)[theBond->getIdx()];
boost::uint32_t hsh=gboost::hash_range(symb.begin(),symb.end());
rank += (hsh%MAX_NATOMS) * MAX_NATOMS;
}
} else if( theBond->getOwningMol().getRingInfo()->numBondRings(theBond->getIdx()) ){
if(!bondSymbols){
rank += static_cast<int>(MAX_BONDTYPE - theBond->getBondType()) *
MAX_NATOMS*MAX_NATOMS;
} else {
const std::string &symb=(*bondSymbols)[theBond->getIdx()];
boost::uint32_t hsh=gboost::hash_range(symb.begin(),symb.end());
rank += (hsh%MAX_NATOMS)*MAX_NATOMS*MAX_NATOMS;
}
}
//std::cerr<<"aIdx: "<< atomIdx <<" p: "<<otherIdx<<" Rank: "<<ranks[otherIdx] <<" "<<colors[otherIdx]<<" "<<theBond->getBondType()<<" "<<rank<<std::endl;
possibles.push_back(PossibleType(rank,otherIdx,theBond.get()));
}
}
// ---------------------
//
// Sort on ranks
//
// ---------------------
std::sort(possibles.begin(),possibles.end(),_possibleCompare());
// ---------------------
//
// Now work the children
//
// ---------------------
for(std::vector<PossibleType>::iterator possiblesIt=possibles.begin();
possiblesIt!=possibles.end();
possiblesIt++){
int possibleIdx = possiblesIt->get<1>();
Bond *bond = possiblesIt->get<2>();
Atom *otherAtom=mol.getAtomWithIdx(possibleIdx);
switch(colors[possibleIdx]){
case WHITE_NODE:
// -----
// we haven't seen this node at all before, traverse
// -----
dfsFindCycles(mol,possibleIdx,bond->getIdx(),colors,
ranks,atomOrders,
atomRingClosures,
bondsInPlay,bondSymbols);
break;
case GREY_NODE:
// -----
// we've seen this, but haven't finished it (we're finishing a ring)
// -----
atomRingClosures[possibleIdx].push_back(bond->getIdx());
atomRingClosures[atomIdx].push_back(bond->getIdx());
break;
default:
// -----
// this node has been finished. don't do anything.
// -----
break;
}
}
colors[atomIdx] = BLACK_NODE;
}
void canonicalDFSTraversal(ROMol &mol,int atomIdx,int inBondIdx,
std::vector<AtomColors> &colors,
VECT_INT_VECT &cycles,
INT_VECT &ranks,
INT_VECT &cyclesAvailable,
MolStack &molStack,
INT_VECT &atomOrders,
INT_VECT &bondVisitOrders,
VECT_INT_VECT &atomRingClosures,
std::vector<INT_LIST> &atomTraversalBondOrder,
const boost::dynamic_bitset<> *bondsInPlay,
const std::vector<std::string> *bondSymbols
){
PRECONDITION(colors.size()>=mol.getNumAtoms(),"vector too small");
PRECONDITION(ranks.size()>=mol.getNumAtoms(),"vector too small");
PRECONDITION(atomOrders.size()>=mol.getNumAtoms(),"vector too small");
PRECONDITION(bondVisitOrders.size()>=mol.getNumBonds(),"vector too small");
PRECONDITION(atomRingClosures.size()>=mol.getNumAtoms(),"vector too small");
PRECONDITION(atomTraversalBondOrder.size()>=mol.getNumAtoms(),"vector too small");
PRECONDITION(!bondsInPlay || bondsInPlay->size()>=mol.getNumBonds(),"bondsInPlay too small");
PRECONDITION(!bondSymbols || bondSymbols->size()>=mol.getNumBonds(),"bondSymbols too small");
int nAttached=0;
Atom *atom = mol.getAtomWithIdx(atomIdx);
INT_LIST directTravList,cycleEndList;
molStack.push_back(MolStackElem(atom));
atomOrders[atom->getIdx()] = molStack.size();
colors[atomIdx] = GREY_NODE;
// ---------------------
//
// Build the list of possible destinations from here
//
// ---------------------
std::vector< PossibleType > possibles;
possibles.resize(0);
ROMol::OBOND_ITER_PAIR bondsPair = mol.getAtomBonds(atom);
possibles.reserve(bondsPair.second-bondsPair.first);
while(bondsPair.first != bondsPair.second){
BOND_SPTR theBond = mol[*(bondsPair.first)];
bondsPair.first++;
if(bondsInPlay && !(*bondsInPlay)[theBond->getIdx()]) continue;
if(inBondIdx<0 || theBond->getIdx() != static_cast<unsigned int>(inBondIdx)){
int otherIdx = theBond->getOtherAtomIdx(atomIdx);
long rank=ranks[otherIdx];
// ---------------------
//
// things are a bit more complicated if we are sitting on a
// ring atom we would like to traverse first to the
// ring-closure atoms, then to atoms outside the ring first,
// then to atoms in the ring that haven't already been visited
// (non-ring-closure atoms).
//
// Here's how the black magic works:
// - non-ring atom neighbors have their original ranks
// - ring atom neighbors have this added to their ranks:
// (Bond::OTHER - bondOrder)*MAX_NATOMS*MAX_NATOMS
// - ring-closure neighbors lose a factor of:
// (Bond::OTHER+1)*MAX_NATOMS*MAX_NATOMS
//
// This tactic biases us to traverse to non-ring neighbors first,
// original ordering if bond orders are all equal... crafty, neh?
//
// ---------------------
if( colors[otherIdx] == GREY_NODE ) {
rank -= static_cast<int>(Bond::OTHER+1) *
MAX_NATOMS*MAX_NATOMS;
if(!bondSymbols){
rank += static_cast<int>(Bond::OTHER - theBond->getBondType()) *
MAX_NATOMS;
} else {
const std::string &symb=(*bondSymbols)[theBond->getIdx()];
boost::uint32_t hsh=gboost::hash_range(symb.begin(),symb.end());
rank += (hsh%MAX_NATOMS) * MAX_NATOMS;
}
} else if( theBond->getOwningMol().getRingInfo()->numBondRings(theBond->getIdx()) ){
if(!bondSymbols){
rank += static_cast<int>(Bond::OTHER - theBond->getBondType()) *
MAX_NATOMS*MAX_NATOMS;
} else {
const std::string &symb=(*bondSymbols)[theBond->getIdx()];
boost::uint32_t hsh=gboost::hash_range(symb.begin(),symb.end());
rank += (hsh%MAX_NATOMS)*MAX_NATOMS*MAX_NATOMS;
}
}
possibles.push_back(PossibleType(rank,otherIdx,theBond.get()));
}
}
// ---------------------
//
// Sort on ranks
//
// ---------------------
std::sort(possibles.begin(),possibles.end(),_possibleComp);
// ---------------------
//
// Now work the children
//
// ---------------------
std::vector<MolStack> subStacks;
for(std::vector<PossibleType>::iterator possiblesIt=possibles.begin();
possiblesIt!=possibles.end();
possiblesIt++){
MolStack subStack;
#if 0
int possibleIdx = possiblesIt->second.first;
Bond *bond = possiblesIt->second.second;
#endif
int possibleIdx = possiblesIt->get<1>();
Bond *bond = possiblesIt->get<2>();
Atom *otherAtom=mol.getAtomWithIdx(possibleIdx);
unsigned int lowestRingIdx;
INT_VECT::const_iterator cAIt;
switch(colors[possibleIdx]){
case WHITE_NODE:
// -----
// we haven't seen this node at all before
// -----
// it might have some residual data from earlier calls, clean that up:
if(otherAtom->hasProp("_TraversalBondIndexOrder")){
otherAtom->clearProp("_TraversalBondIndexOrder");
}
directTravList.push_back(bond->getIdx());
subStack.push_back(MolStackElem(bond,atomIdx));
canonicalDFSTraversal(mol,possibleIdx,bond->getIdx(),colors,
cycles,ranks,cyclesAvailable,subStack,
atomOrders,bondVisitOrders,atomRingClosures,atomTraversalBondOrder,
bondsInPlay,bondSymbols);
subStacks.push_back(subStack);
nAttached += 1;
break;
case GREY_NODE:
// -----
// we've seen this, but haven't finished it (we're finishing a ring)
// -----
cycleEndList.push_back(bond->getIdx());
cAIt=std::find(cyclesAvailable.begin(),
cyclesAvailable.end(),1);
if(cAIt==cyclesAvailable.end()){
throw ValueErrorException("Too many rings open at once. SMILES cannot be generated.");
}
lowestRingIdx = cAIt-cyclesAvailable.begin();
cyclesAvailable[lowestRingIdx] = 0;
cycles[possibleIdx].push_back(lowestRingIdx);
++lowestRingIdx;
bond->setProp("_TraversalRingClosureBond",lowestRingIdx);
molStack.push_back(MolStackElem(bond,
atom->getIdx()));
molStack.push_back(MolStackElem(lowestRingIdx));
// we need to add this bond (which closes the ring) to the traversal list for the
// other atom as well:
atomTraversalBondOrder[otherAtom->getIdx()].push_back(bond->getIdx());
atomRingClosures[otherAtom->getIdx()].push_back(bond->getIdx());
break;
default:
// -----
// this node has been finished. don't do anything.
// -----
break;
}
}
INT_VECT &ringClosures=atomRingClosures[atom->getIdx()];
CHECK_INVARIANT(ringClosures.size()==cycles[atomIdx].size(),
"ring closure mismatch");
for(unsigned int i=0;i<ringClosures.size();i++){
int ringIdx=cycles[atomIdx][i];
ringIdx += 1;
molStack.push_back(MolStackElem(ringIdx));
}
cycles[atomIdx].resize(0);
MolStack::const_iterator ciMS;
for(int i=0;i<nAttached;i++){
if(i<nAttached-1){
int branchIdx=0;
if(subStacks[i].begin()->type==MOL_STACK_ATOM){
branchIdx=subStacks[i].begin()->obj.atom->getIdx();
} else if(subStacks[i].begin()->type==MOL_STACK_BOND){
branchIdx=-1*subStacks[i].begin()->obj.bond->getIdx();
} else {
ASSERT_INVARIANT(0,"branch started with something other than an atom or bond");
}
molStack.push_back(MolStackElem("(",branchIdx));
for(ciMS=subStacks[i].begin();ciMS!=subStacks[i].end();ciMS++){
molStack.push_back(*ciMS);
switch(ciMS->type){
case MOL_STACK_ATOM:
atomOrders[ciMS->obj.atom->getIdx()] = molStack.size();
break;
case MOL_STACK_BOND:
bondVisitOrders[ciMS->obj.bond->getIdx()] = molStack.size();
break;
default:
break;
}
}
molStack.push_back(MolStackElem(")",branchIdx));
} else {
for(ciMS=subStacks[i].begin();ciMS!=subStacks[i].end();ciMS++){
molStack.push_back(*ciMS);
switch(ciMS->type){
case MOL_STACK_ATOM:
atomOrders[ciMS->obj.atom->getIdx()] = molStack.size();
break;
case MOL_STACK_BOND:
bondVisitOrders[ciMS->obj.bond->getIdx()] = molStack.size();
break;
default:
break;
}
}
}
}
//std::cerr<<"*****>>>>>> Traversal results for atom: "<<atom->getIdx()<<"> ";
INT_LIST travList;
// first push on the incoming bond:
if(inBondIdx >= 0){
//std::cerr<<" "<<inBondIdx;
travList.push_back(inBondIdx);
}
// ... ring closures that end here:
for(INT_LIST_CI ilci=cycleEndList.begin();ilci!=cycleEndList.end();++ilci){
//std::cerr<<" ["<<*ilci<<"]";
travList.push_back(*ilci);
}
// ... ring closures that start here:
// if(atom->hasProp("_TraversalBondIndexOrder")){
// INT_LIST indirectTravList;
// atom->getProp("_TraversalBondIndexOrder",indirectTravList);
// for(INT_LIST_CI ilci=indirectTravList.begin();ilci!=indirectTravList.end();++ilci){
// //std::cerr<<" ("<<*ilci<<")";
// travList.push_back(*ilci);
// }
// }
BOOST_FOREACH(int ili,atomTraversalBondOrder[atom->getIdx()]){
travList.push_back(ili);
}