void UnifyingConflator::_reset()
{
if (_mergerFactory == 0)
{
_mergerFactory.reset(new MergerFactory());
// register the mark for review merger first so all reviews get tagged before another merger
// gets a chance.
_mergerFactory->registerCreator(new MarkForReviewMergerCreator());
_mergerFactory->registerDefaultCreators();
}
_e2m.clear();
_deleteAll(_matches);
_deleteAll(_mergers);
}
void _deleteAll(node<V>* n) {
if(n!=NULL) {
node<V>* c=n;
do {
node<V>* d=c;
c=c->next;
_deleteAll(d->child);
delete d;
} while(c!=n);
}
}
void UnifyingConflator::apply(shared_ptr<OsmMap>& map)
{
Timer timer;
_reset();
NamedOp(ConfigOptions().getUnifyPreOps().split(";", QString::SkipEmptyParts)).apply(map);
_stats.append(SingleStat("Apply Pre Ops Time (sec)", timer.getElapsedAndRestart()));
// will reproject if necessary.
MapReprojector::reprojectToPlanar(map);
_stats.append(SingleStat("Project to Planar Time (sec)", timer.getElapsedAndRestart()));
if (Log::getInstance().isDebugEnabled())
{
LOG_DEBUG("Writing debug map.");
OsmMapPtr debug(new OsmMap(map));
MapReprojector::reprojectToWgs84(debug);
OsmMapWriterFactory::write(debug, "tmp/debug.osm");
_stats.append(SingleStat("Write Debug Map Time (sec)", timer.getElapsedAndRestart()));
}
LOG_DEBUG("Creating matches...");
// find all the matches in this map
if (_matchThreshold.get())
{
//ScoreMatches logic seems to be the only one that needs to pass in the match threshold now when
//the optimize param is activated. Otherwise, we get the match threshold information from the
//config.
_matchFactory.createMatches(map, _matches, _bounds, _matchThreshold);
}
else
{
_matchFactory.createMatches(map, _matches, _bounds);
}
LOG_DEBUG("Match count: " << _matches.size());
LOG_DEBUG(SystemInfo::getMemoryUsageString());
double findMatchesTime = timer.getElapsedAndRestart();
_stats.append(SingleStat("Find Matches Time (sec)", findMatchesTime));
_stats.append(SingleStat("Number of Matches Found", _matches.size()));
_stats.append(SingleStat("Number of Matches Found per Second",
(double)_matches.size() / findMatchesTime));
vector<const Match*> allMatches = _matches;
// add review tags to all matches that have some review component
_addReviewTags(map, allMatches);
LOG_INFO("Pre-constraining match count: " << allMatches.size());
_stats.append(SingleStat("Number of Matches Before Whole Groups", _matches.size()));
// If there are groups of matches that should not be optimized, remove them before optimization.
MatchSetVector matchSets;
_removeWholeGroups(_matches, matchSets, map);
_stats.append(SingleStat("Number of Whole Groups", matchSets.size()));
// Globally optimize the set of matches to maximize the conflation score.
{
OptimalConstrainedMatches cm(map);
cm.addMatches(_matches.begin(), _matches.end());
cm.setTimeLimit(ConfigOptions(_settings).getUnifyOptimizerTimeLimit());
double cmStart = Time::getTime();
vector<const Match*> cmMatches = cm.calculateSubset();
LOG_INFO("CM took: " << Time::getTime() - cmStart << "s.");
LOG_INFO("CM Score: " << cm.getScore());
LOG_DEBUG(SystemInfo::getMemoryUsageString());
GreedyConstrainedMatches gm(map);
gm.addMatches(_matches.begin(), _matches.end());
double gmStart = Time::getTime();
vector<const Match*> gmMatches = gm.calculateSubset();
LOG_INFO("GM took: " << Time::getTime() - gmStart << "s.");
LOG_INFO("GM Score: " << gm.getScore());
if (gm.getScore() > cm.getScore())
{
_matches = gmMatches;
}
else
{
_matches = cmMatches;
}
}
double optimizeMatchesTime = timer.getElapsedAndRestart();
_stats.append(SingleStat("Optimize Matches Time (sec)", optimizeMatchesTime));
_stats.append(SingleStat("Number of Optimized Matches", _matches.size()));
_stats.append(SingleStat("Number of Matches Optimized per Second",
(double)allMatches.size() / optimizeMatchesTime));
LOG_DEBUG(SystemInfo::getMemoryUsageString());
// #warning validateConflictSubset is on, this is slow.
// _validateConflictSubset(map, _matches);
LOG_INFO("Post constraining match count: " << _matches.size());
{
// search the matches for groups (subgraphs) of matches. In other words, groups where all the
// matches are interrelated by element id
MatchGraph mg;
mg.addMatches(_matches.begin(), _matches.end());
vector< set<const Match*, MatchPtrComparator> > tmpMatchSets =
mg.findSubgraphs(map);
matchSets.insert(matchSets.end(), tmpMatchSets.begin(), tmpMatchSets.end());
LOG_DEBUG(SystemInfo::getMemoryUsageString());
}
LOG_DEBUG("Match sets count: " << matchSets.size());
LOG_DEBUG(SystemInfo::getMemoryUsageString());
/// @todo would it help to sort the matches so the biggest or best ones get merged first?
// convert all the match sets into mergers.
for (size_t i = 0; i < matchSets.size(); ++i)
{
_mergerFactory->createMergers(map, matchSets[i], _mergers);
}
LOG_DEBUG(SystemInfo::getMemoryUsageString());
// don't need the matches any more
_deleteAll(allMatches);
_matches.clear();
LOG_DEBUG(SystemInfo::getMemoryUsageString());
_mapElementIdsToMergers();
LOG_DEBUG(SystemInfo::getMemoryUsageString());
_stats.append(SingleStat("Create Mergers Time (sec)", timer.getElapsedAndRestart()));
vector< pair<ElementId, ElementId> > replaced;
for (size_t i = 0; i < _mergers.size(); ++i)
{
_mergers[i]->apply(map, replaced);
// update any mergers that reference the replaced values
_replaceElementIds(replaced);
replaced.clear();
if (Log::getInstance().getLevel() == Log::Debug)
{
cout << "Applying mergers: " << i + 1 << " / " << _mergers.size() << " \r" << flush;
}
}
if (Log::getInstance().getLevel() == Log::Debug)
{
cout << endl;
}
LOG_DEBUG(SystemInfo::getMemoryUsageString());
size_t mergerCount = _mergers.size();
// free up any used resources.
_reset();
LOG_DEBUG(SystemInfo::getMemoryUsageString());
double mergersTime = timer.getElapsedAndRestart();
_stats.append(SingleStat("Apply Mergers Time (sec)", mergersTime));
_stats.append(SingleStat("Mergers Applied per Second", (double)mergerCount / mergersTime));
NamedOp(ConfigOptions().getUnifyPostOps().split(";", QString::SkipEmptyParts)).apply(map);
_stats.append(SingleStat("Apply Post Ops Time (sec)", timer.getElapsedAndRestart()));
}
virtual ~FibonacciHeap() {
if(heap) {
_deleteAll(heap);
}
}
