/* ScummVM - Graphic Adventure Engine * * ScummVM is the legal property of its developers, whose names * are too numerous to list here. Please refer to the COPYRIGHT * file distributed with this source distribution. * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation, either version 3 of the License, or * (at your option) any later version. * * This program 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. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program. If not, see . * */ /* This is a lightly modified version of MicroPather, from github.com/leethomason/MicroPather. Modifications were made to fit with ScummVM coding style and APIs. The original copyright message is: ------- Copyright (c) 2000-2009 Lee Thomason (www.grinninglizard.com) Grinning Lizard Utilities. This software is provided 'as-is', without any express or implied warranty. In no event will the authors be held liable for any damages arising from the use of this software. Permission is granted to anyone to use this software for any purpose, including commercial applications, and to alter it and redistribute it freely, subject to the following restrictions: 1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required. 2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software. 3. This notice may not be removed or altered from any source distribution. */ #define MPASSERT assert //#define TETRAEDGE_MICROPATHER_DEBUG //#define TETRAEDGE_MICROPATHER_DEBUG_PATH //#define TETRAEDGE_MICROPATHER_DEBUG_PATH_DEEP //#define TETRAEDGE_MICROPATHER_TRACK_COLLISION //#define TETRAEDGE_MICROPATHER_DEBUG_CACHING //#define TETRAEDGE_MICROPATHER_STRESS //#ifdef TETRAEDGE_MICROPATHER_DEBUG_CACHING //#include "../grinliz/gldebug.h" //#endif #include "micropather.h" using namespace Tetraedge::micropather; class OpenQueue { public: OpenQueue( Graph* _graph ) { graph = _graph; sentinel = (PathNode*) sentinelMem; sentinel->InitSentinel(); #ifdef TETRAEDGE_MICROPATHER_DEBUG sentinel->CheckList(); #endif } ~OpenQueue() {} void Push( PathNode* pNode ); PathNode* Pop(); void Update( PathNode* pNode ); bool Empty() { return sentinel->next == sentinel; } private: OpenQueue( const OpenQueue& ); // undefined and unsupported void operator=( const OpenQueue& ); PathNode* sentinel; int sentinelMem[ ( sizeof( PathNode ) + sizeof( int ) ) / sizeof( int ) ]; Graph* graph; // for debugging }; void OpenQueue::Push( PathNode* pNode ) { MPASSERT( pNode->inOpen == 0 ); MPASSERT( pNode->inClosed == 0 ); #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH_DEEP debug( "Open Push: " ); graph->PrintStateInfo( pNode->state ); debug( " total=%.1f\n", pNode->totalCost ); #endif // Add sorted. Lowest to highest cost path. Note that the sentinel has // a value of FLT_MAX, so it should always be sorted in. MPASSERT( pNode->totalCost < FLT_MAX ); PathNode* iter = sentinel->next; while ( true ) { if ( pNode->totalCost < iter->totalCost ) { iter->AddBefore( pNode ); pNode->inOpen = 1; break; } iter = iter->next; } MPASSERT( pNode->inOpen ); // make sure this was actually added. #ifdef TETRAEDGE_MICROPATHER_DEBUG sentinel->CheckList(); #endif } PathNode* OpenQueue::Pop() { MPASSERT( sentinel->next != sentinel ); PathNode* pNode = sentinel->next; pNode->Unlink(); #ifdef TETRAEDGE_MICROPATHER_DEBUG sentinel->CheckList(); #endif MPASSERT( pNode->inClosed == 0 ); MPASSERT( pNode->inOpen == 1 ); pNode->inOpen = 0; #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH_DEEP debug( "Open Pop: " ); graph->PrintStateInfo( pNode->state ); debug( " total=%.1f\n", pNode->totalCost ); #endif return pNode; } void OpenQueue::Update( PathNode* pNode ) { #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH_DEEP debug( "Open Update: " ); graph->PrintStateInfo( pNode->state ); debug( " total=%.1f\n", pNode->totalCost ); #endif MPASSERT( pNode->inOpen ); // If the node now cost less than the one before it, // move it to the front of the list. if ( pNode->prev != sentinel && pNode->totalCost < pNode->prev->totalCost ) { pNode->Unlink(); sentinel->next->AddBefore( pNode ); } // If the node is too high, move to the right. if ( pNode->totalCost > pNode->next->totalCost ) { PathNode* it = pNode->next; pNode->Unlink(); while ( pNode->totalCost > it->totalCost ) it = it->next; it->AddBefore( pNode ); #ifdef TETRAEDGE_MICROPATHER_DEBUG sentinel->CheckList(); #endif } } class ClosedSet { public: ClosedSet( Graph* _graph ) { this->graph = _graph; } ~ClosedSet() {} void Add( PathNode* pNode ) { #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH_DEEP debug( "Closed add: " ); graph->PrintStateInfo( pNode->state ); debug( " total=%.1f\n", pNode->totalCost ); #endif #ifdef TETRAEDGE_MICROPATHER_DEBUG MPASSERT( pNode->inClosed == 0 ); MPASSERT( pNode->inOpen == 0 ); #endif pNode->inClosed = 1; } void Remove( PathNode* pNode ) { #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH_DEEP debug( "Closed remove: " ); graph->PrintStateInfo( pNode->state ); debug( " total=%.1f\n", pNode->totalCost ); #endif MPASSERT( pNode->inClosed == 1 ); MPASSERT( pNode->inOpen == 0 ); pNode->inClosed = 0; } private: ClosedSet( const ClosedSet& ); void operator=( const ClosedSet& ); Graph* graph; }; PathNodePool::PathNodePool( unsigned _allocate, unsigned _typicalAdjacent ) : firstBlock( 0 ), blocks( 0 ), #ifdef TETRAEDGE_MICROPATHER_STRESS allocate( 32 ), #else allocate( _allocate ), #endif nAllocated( 0 ), nAvailable( 0 ) { freeMemSentinel.InitSentinel(); cacheCap = allocate * _typicalAdjacent; cacheSize = 0; cache = (NodeCost*)malloc(cacheCap * sizeof(NodeCost)); // Want the behavior that if the actual number of states is specified, the cache // will be at least that big. hashShift = 3; // 8 (only useful for stress testing) #ifndef TETRAEDGE_MICROPATHER_STRESS while( HashSize() < allocate ) ++hashShift; #endif hashTable = (PathNode**)calloc( HashSize(), sizeof(PathNode*) ); blocks = firstBlock = NewBlock(); //debug( "HashSize=%d allocate=%d\n", HashSize(), allocate ); totalCollide = 0; } PathNodePool::~PathNodePool() { Clear(); free( firstBlock ); free( cache ); free( hashTable ); #ifdef TETRAEDGE_MICROPATHER_TRACK_COLLISION debug( "Total collide=%d HashSize=%d HashShift=%d\n", totalCollide, HashSize(), hashShift ); #endif } bool PathNodePool::PushCache( const NodeCost* nodes, int nNodes, int* start ) { *start = -1; if ( nNodes+cacheSize <= cacheCap ) { for( int i=0; i= 0 && start < cacheCap ); MPASSERT( nNodes > 0 ); MPASSERT( start + nNodes <= cacheCap ); memcpy( nodes, &cache[start], sizeof(NodeCost)*nNodes ); } void PathNodePool::Clear() { #ifdef TETRAEDGE_MICROPATHER_TRACK_COLLISION // Collision tracking code. int collide=0; for( unsigned i=0; ichild[0] || hashTable[i]->child[1]) ) ++collide; } //debug( "PathNodePool %d/%d collision=%d %.1f%%\n", nAllocated, HashSize(), collide, 100.0f*(float)collide/(float)HashSize() ); totalCollide += collide; #endif Block* b = blocks; while( b ) { Block* temp = b->nextBlock; if ( b != firstBlock ) { free( b ); } b = temp; } blocks = firstBlock; // Don't delete the first block (we always need at least that much memory.) // Set up for new allocations (but don't do work we don't need to. Reset/Clear can be called frequently.) if ( nAllocated > 0 ) { freeMemSentinel.next = &freeMemSentinel; freeMemSentinel.prev = &freeMemSentinel; memset( hashTable, 0, sizeof(PathNode*)*HashSize() ); for( unsigned i=0; ipathNode[i] ); } } nAvailable = allocate; nAllocated = 0; cacheSize = 0; } PathNodePool::Block* PathNodePool::NewBlock() { Block* block = (Block*) calloc( 1, sizeof(Block) + sizeof(PathNode)*(allocate-1) ); block->nextBlock = 0; nAvailable += allocate; for( unsigned i=0; ipathNode[i] ); } return block; } unsigned PathNodePool::Hash( void* voidval ) { /* Spent quite some time on this, and the result isn't quite satifactory. The input set is the size of a void*, and is generally (x,y) pairs or memory pointers. FNV resulting in about 45k collisions in a (large) test and some other approaches about the same. Simple folding reduces collisions to about 38k - big improvement. However, that may be an artifact of the (x,y) pairs being well distributed. And for either the x,y case or the pointer case, there are probably very poor hash table sizes that cause "overlaps" and grouping. (An x,y encoding with a hashShift of 8 is begging for trouble.) The best tested results are simple folding, but that seems to beg for a pathelogical case. FNV-1a was the next best choice, without obvious pathelogical holes. Finally settled on h%HashMask(). Simple, but doesn't have the obvious collision cases of folding. */ /* // Time: 567 // FNV-1a // http://isthe.com/chongo/tech/comp/fnv/ // public domain. MP_UPTR val = (MP_UPTR)(voidval); const unsigned char *p = (unsigned char *)(&val); uint h = 2166136261; for( size_t i=0; i>hashShift) ^ (h>>(hashShift*2)) ^ (h>>(hashShift*3)) ) & HashMask(); */ /* // Time: 526 MP_UPTR h = (MP_UPTR)(voidval); return ( h ^ (h>>hashShift) ^ (h>>(hashShift*2)) ^ (h>>(hashShift*3)) ) & HashMask(); */ // Time: 512 // The HashMask() is used as the divisor. h%1024 has lots of common // repetitions, but h%1023 will move things out more. MP_UPTR h = (MP_UPTR)(voidval); return h % HashMask(); } PathNode* PathNodePool::Alloc() { if ( freeMemSentinel.next == &freeMemSentinel ) { MPASSERT( nAvailable == 0 ); Block* b = NewBlock(); b->nextBlock = blocks; blocks = b; MPASSERT( freeMemSentinel.next != &freeMemSentinel ); } PathNode* pathNode = freeMemSentinel.next; pathNode->Unlink(); ++nAllocated; MPASSERT( nAvailable > 0 ); --nAvailable; return pathNode; } void PathNodePool::AddPathNode( unsigned key, PathNode* root ) { if ( hashTable[key] ) { PathNode* p = hashTable[key]; while( true ) { int dir = (root->state < p->state) ? 0 : 1; if ( p->child[dir] ) { p = p->child[dir]; } else { p->child[dir] = root; break; } } } else { hashTable[key] = root; } } PathNode* PathNodePool::FetchPathNode( void* state ) { unsigned key = Hash( state ); PathNode* root = hashTable[key]; while( root ) { if ( root->state == state ) { break; } root = ( state < root->state ) ? root->child[0] : root->child[1]; } MPASSERT( root ); return root; } PathNode* PathNodePool::GetPathNode( unsigned frame, void* _state, float _costFromStart, float _estToGoal, PathNode* _parent ) { unsigned key = Hash( _state ); PathNode* root = hashTable[key]; while( root ) { if ( root->state == _state ) { if ( root->frame == frame ) // This is the correct state and correct frame. break; // Correct state, wrong frame. root->Init( frame, _state, _costFromStart, _estToGoal, _parent ); break; } root = ( _state < root->state ) ? root->child[0] : root->child[1]; } if ( !root ) { // allocate new one root = Alloc(); root->Clear(); root->Init( frame, _state, _costFromStart, _estToGoal, _parent ); AddPathNode( key, root ); } return root; } void PathNode::Init( unsigned _frame, void* _state, float _costFromStart, float _estToGoal, PathNode* _parent ) { state = _state; costFromStart = _costFromStart; estToGoal = _estToGoal; CalcTotalCost(); parent = _parent; frame = _frame; inOpen = 0; inClosed = 0; } void PathNode::Clear() { memset( this, 0, sizeof( PathNode ) ); numAdjacent = -1; cacheIndex = -1; } MicroPather::MicroPather( Graph* _graph, unsigned allocate, unsigned typicalAdjacent, bool cache ) : pathNodePool( allocate, typicalAdjacent ), graph( _graph ), frame( 0 ) { MPASSERT( allocate ); MPASSERT( typicalAdjacent ); pathCache = 0; if ( cache ) { pathCache = new PathCache( allocate*4 ); // untuned arbitrary constant } } MicroPather::~MicroPather() { delete pathCache; } void MicroPather::Reset() { pathNodePool.Clear(); if ( pathCache ) { pathCache->Reset(); } frame = 0; } void MicroPather::GoalReached( PathNode* node, void* start, void* end, Common::Array< void* > *_path ) { Common::Array< void* >& path = *_path; path.clear(); // We have reached the goal. // How long is the path? Used to allocate the vector which is returned. int count = 1; PathNode* it = node; while( it->parent ) { ++count; it = it->parent; } // Now that the path has a known length, allocate // and fill the vector that will be returned. if ( count < 3 ) { // Handle the short, special case. path.resize(2); path[0] = start; path[1] = end; } else { path.resize(count); path[0] = start; path[count-1] = end; count-=2; it = node->parent; while ( it->parent ) { path[count] = it->state; it = it->parent; --count; } } if ( pathCache ) { costVec.clear(); PathNode* pn0 = pathNodePool.FetchPathNode( path[0] ); PathNode* pn1 = 0; for( unsigned i=0; iAdd( path, costVec ); } #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH debug( "Path: " ); int counter=0; #endif for ( unsigned k=0; kPrintStateInfo( path[k] ); debug( " " ); ++counter; if ( counter == 8 ) { debug( "\n" ); counter = 0; } #endif } #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH debug( "Cost=%.1f Checksum %d\n", node->costFromStart, checksum ); #endif } void MicroPather::GetNodeNeighbors( PathNode* node, Common::Array< NodeCost >* pNodeCost ) { if ( node->numAdjacent == 0 ) { // it has no neighbors. pNodeCost->resize( 0 ); } else if ( node->cacheIndex < 0 ) { // Not in the cache. Either the first time or just didn't fit. We don't know // the number of neighbors and need to call back to the client. stateCostVec.resize( 0 ); graph->AdjacentCost( node->state, &stateCostVec ); #ifdef TETRAEDGE_MICROPATHER_DEBUG { // If this assert fires, you have passed a state // as its own neighbor state. This is impossible -- // bad things will happen. for ( unsigned i=0; istate ); } #endif pNodeCost->resize( stateCostVec.size() ); node->numAdjacent = stateCostVec.size(); if ( node->numAdjacent > 0 ) { // Now convert to pathNodes. // Note that the microsoft std library is actually pretty slow. // Move things to temp vars to help. const unsigned stateCostVecSize = stateCostVec.size(); const StateCost* stateCostVecPtr = &stateCostVec[0]; NodeCost* pNodeCostPtr = &(*pNodeCost)[0]; for( unsigned i=0; isize() > 0 && pathNodePool.PushCache( pNodeCostPtr, pNodeCost->size(), &start ) ) { node->cacheIndex = start; } } } else { // In the cache! pNodeCost->resize( node->numAdjacent ); NodeCost* pNodeCostPtr = &(*pNodeCost)[0]; pathNodePool.GetCache( node->cacheIndex, node->numAdjacent, pNodeCostPtr ); // A node is uninitialized (even if memory is allocated) if it is from a previous frame. // Check for that, and Init() as necessary. for( int i=0; inumAdjacent; ++i ) { PathNode* pNode = pNodeCostPtr[i].node; if ( pNode->frame != frame ) { pNode->Init( frame, pNode->state, FLT_MAX, FLT_MAX, 0 ); } } } } #ifdef TETRAEDGE_MICROPATHER_DEBUG void MicroPather::DumpStats() { int hashTableEntries = 0; for( int i=0; i* stateVec ) { stateVec->clear(); pathNodePool.AllStates( frame, stateVec ); } void PathNodePool::AllStates( unsigned frame, Common::Array< void* >* stateVec ) { for ( Block* b=blocks; b; b=b->nextBlock ) { for( unsigned i=0; ipathNode[i].frame == frame ) stateVec->push_back( b->pathNode[i].state ); } } } PathCache::PathCache( int _allocated ) { mem = new Item[_allocated]; memset( mem, 0, sizeof(*mem)*_allocated ); allocated = _allocated; nItems = 0; hit = 0; miss = 0; } PathCache::~PathCache() { delete [] mem; } void PathCache::Reset() { if ( nItems ) { memset( mem, 0, sizeof(*mem)*allocated ); nItems = 0; hit = 0; miss = 0; } } void PathCache::Add( const Common::Array< void* >& path, const Common::Array< float >& cost ) { if ( nItems + (int)path.size() > allocated*3/4 ) { return; } for( unsigned i=0; ib->c->d // Huge memory saving to only store 3 paths to 'd' // Can put more in cache with also adding path to b, c, & d // But uses much more memory. Experiment with this commented // in and out and how to set. void* end = path[path.size()-1]; Item item = { path[i], end, path[i+1], cost[i] }; AddItem( item ); } } void PathCache::AddNoSolution( void* end, void* states[], int count ) { if ( count + nItems > allocated*3/4 ) { return; } for( int i=0; i* path, float* totalCost ) { const Item* item = Find( start, end ); if ( item ) { if ( item->cost == FLT_MAX ) { ++hit; return MicroPather::NO_SOLUTION; } path->clear(); path->push_back( start ); *totalCost = 0; for ( ;start != end; start=item->next, item=Find(start, end) ) { MPASSERT( item ); *totalCost += item->cost; path->push_back( item->next ); } ++hit; return MicroPather::SOLVED; } ++miss; return MicroPather::NOT_CACHED; } void PathCache::AddItem( const Item& item ) { MPASSERT( allocated ); int index = item.Hash() % allocated; while( true ) { if ( mem[index].Empty() ) { mem[index] = item; ++nItems; #ifdef TETRAEDGE_MICROPATHER_DEBUG_CACHING GLOUTPUT(( "Add: start=%x next=%x end=%x\n", item.start, item.next, item.end )); #endif break; } else if ( mem[index].KeyEqual( item ) ) { MPASSERT( (mem[index].next && item.next) || (mem[index].next==0 && item.next == 0) ); // do nothing; in cache break; } ++index; if ( index == allocated ) index = 0; } } const PathCache::Item* PathCache::Find( void* start, void* end ) { MPASSERT( allocated ); Item fake = { start, end, 0, 0 }; int index = fake.Hash() % allocated; while( true ) { if ( mem[index].Empty() ) { return 0; } if ( mem[index].KeyEqual( fake )) { return mem + index; } ++index; if ( index == allocated ) index = 0; } } void MicroPather::GetCacheData( CacheData* data ) { if (data) { data->reset(); } if ( pathCache ) { data->nBytesAllocated = pathCache->AllocatedBytes(); data->nBytesUsed = pathCache->UsedBytes(); data->memoryFraction = (float)( (double)data->nBytesUsed / (double)data->nBytesAllocated ); data->hit = pathCache->hit; data->miss = pathCache->miss; if ( data->hit + data->miss ) { data->hitFraction = (float)( (double)(data->hit) / (double)(data->hit + data->miss) ); } else { data->hitFraction = 0; } } } int MicroPather::Solve( void* startNode, void* endNode, Common::Array< void* >* path, float* cost ) { // Important to clear() in case the caller doesn't check the return code. There // can easily be a left over path from a previous call. path->clear(); #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH debug( "Path: " ); graph->PrintStateInfo( startNode ); debug( " --> " ); graph->PrintStateInfo( endNode ); debug( " min cost=%f\n", graph->LeastCostEstimate( startNode, endNode ) ); #endif *cost = 0.0f; if ( startNode == endNode ) return START_END_SAME; if ( pathCache ) { int cacheResult = pathCache->Solve( startNode, endNode, path, cost ); if ( cacheResult == SOLVED || cacheResult == NO_SOLUTION ) { #ifdef TETRAEDGE_MICROPATHER_DEBUG_CACHING GLOUTPUT(( "PathCache hit. result=%s\n", cacheResult == SOLVED ? "solved" : "no_solution" )); #endif return cacheResult; } #ifdef TETRAEDGE_MICROPATHER_DEBUG_CACHING GLOUTPUT(( "PathCache miss\n" )); #endif } ++frame; OpenQueue open( graph ); ClosedSet closed( graph ); PathNode* newPathNode = pathNodePool.GetPathNode( frame, startNode, 0, graph->LeastCostEstimate( startNode, endNode ), 0 ); open.Push( newPathNode ); stateCostVec.resize(0); nodeCostVec.resize(0); while ( !open.Empty() ) { PathNode* node = open.Pop(); if ( node->state == endNode ) { GoalReached( node, startNode, endNode, path ); *cost = node->costFromStart; #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH DumpStats(); #endif return SOLVED; } else { closed.Add( node ); // We have not reached the goal - add the neighbors. GetNodeNeighbors( node, &nodeCostVec ); for( int i=0; inumAdjacent; ++i ) { // Not actually a neighbor, but useful. Filter out infinite cost. if ( nodeCostVec[i].cost == FLT_MAX ) { continue; } PathNode* child = nodeCostVec[i].node; float newCost = node->costFromStart + nodeCostVec[i].cost; PathNode* inOpen = child->inOpen ? child : 0; PathNode* inClosed = child->inClosed ? child : 0; PathNode* inEither = (PathNode*)( ((MP_UPTR)inOpen) | ((MP_UPTR)inClosed) ); MPASSERT( inEither != node ); MPASSERT( !( inOpen && inClosed ) ); if ( inEither ) { if ( newCost < child->costFromStart ) { child->parent = node; child->costFromStart = newCost; child->estToGoal = graph->LeastCostEstimate( child->state, endNode ); child->CalcTotalCost(); if ( inOpen ) { open.Update( child ); } } } else { child->parent = node; child->costFromStart = newCost; child->estToGoal = graph->LeastCostEstimate( child->state, endNode ), child->CalcTotalCost(); MPASSERT( !child->inOpen && !child->inClosed ); open.Push( child ); } } } } #ifdef TETRAEDGE_MICROPATHER_DEBUG_PATH DumpStats(); #endif if ( pathCache ) { // Could add a bunch more with a little tracking. pathCache->AddNoSolution( endNode, &startNode, 1 ); } return NO_SOLUTION; } int MicroPather::SolveForNearStates( void* startState, Common::Array< StateCost >* near, float maxCost ) { /* https://en.wikipedia.org/wiki/Dijkstra%27s_algorithm 1 function Dijkstra(Graph, source): 2 for each vertex v in Graph: // Initializations 3 dist[v] := infinity // Unknown distance function from source to v 4 previous[v] := undefined // Previous node in optimal path from source 5 dist[source] := 0 // Distance from source to source 6 Q := the set of all nodes in Graph // All nodes in the graph are unoptimized - thus are in Q 7 while Q is not empty: // The main loop 8 u := vertex in Q with smallest dist[] 9 if dist[u] = infinity: 10 break // all remaining vertices are inaccessible from source 11 remove u from Q 12 for each neighbor v of u: // where v has not yet been removed from Q. 13 alt := dist[u] + dist_between(u, v) 14 if alt < dist[v]: // Relax (u,v,a) 15 dist[v] := alt 16 previous[v] := u 17 return dist[] */ ++frame; OpenQueue open( graph ); // nodes to look at ClosedSet closed( graph ); nodeCostVec.resize(0); stateCostVec.resize(0); PathNode closedSentinel; closedSentinel.Clear(); closedSentinel.Init( frame, 0, FLT_MAX, FLT_MAX, 0 ); closedSentinel.next = closedSentinel.prev = &closedSentinel; PathNode* newPathNode = pathNodePool.GetPathNode( frame, startState, 0, 0, 0 ); open.Push( newPathNode ); while ( !open.Empty() ) { PathNode* node = open.Pop(); // smallest dist closed.Add( node ); // add to the things we've looked at closedSentinel.AddBefore( node ); if ( node->totalCost > maxCost ) continue; // Too far away to ever get here. GetNodeNeighbors( node, &nodeCostVec ); for( int i=0; inumAdjacent; ++i ) { MPASSERT( node->costFromStart < FLT_MAX ); float newCost = node->costFromStart + nodeCostVec[i].cost; PathNode* inOpen = nodeCostVec[i].node->inOpen ? nodeCostVec[i].node : 0; PathNode* inClosed = nodeCostVec[i].node->inClosed ? nodeCostVec[i].node : 0; MPASSERT( !( inOpen && inClosed ) ); PathNode* inEither = inOpen ? inOpen : inClosed; MPASSERT( inEither != node ); if ( inEither && inEither->costFromStart <= newCost ) { continue; // Do nothing. This path is not better than existing. } // Groovy. We have new information or improved information. PathNode* child = nodeCostVec[i].node; MPASSERT( child->state != newPathNode->state ); // should never re-process the parent. child->parent = node; child->costFromStart = newCost; child->estToGoal = 0; child->totalCost = child->costFromStart; if ( inOpen ) { open.Update( inOpen ); } else if ( !inClosed ) { open.Push( child ); } } } near->clear(); for( PathNode* pNode=closedSentinel.next; pNode != &closedSentinel; pNode=pNode->next ) { if ( pNode->totalCost <= maxCost ) { StateCost sc; sc.cost = pNode->totalCost; sc.state = pNode->state; near->push_back( sc ); } } #ifdef TETRAEDGE_MICROPATHER_DEBUG for( unsigned i=0; isize(); ++i ) { for( unsigned k=i+1; ksize(); ++k ) { MPASSERT( (*near)[i].state != (*near)[k].state ); } } #endif return SOLVED; }