910 lines
44 KiB
C++
910 lines
44 KiB
C++
/*
|
|
open source routing machine
|
|
Copyright (C) Dennis Luxen, others 2010
|
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
it under the terms of the GNU AFFERO General Public License as published by
|
|
the Free Software Foundation; either version 3 of the License, or
|
|
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 Affero General Public License
|
|
along with this program; if not, write to the Free Software
|
|
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
or see http://www.gnu.org/licenses/agpl.txt.
|
|
*/
|
|
|
|
#ifndef STATICRTREE_H_
|
|
#define STATICRTREE_H_
|
|
|
|
#include "MercatorUtil.h"
|
|
#include "TimingUtil.h"
|
|
#include "Coordinate.h"
|
|
#include "PhantomNodes.h"
|
|
#include "DeallocatingVector.h"
|
|
#include "HilbertValue.h"
|
|
#include "../typedefs.h"
|
|
|
|
#include <boost/assert.hpp>
|
|
#include <boost/bind.hpp>
|
|
#include <boost/foreach.hpp>
|
|
#include <boost/algorithm/minmax.hpp>
|
|
#include <boost/algorithm/minmax_element.hpp>
|
|
#include <boost/range/algorithm_ext/erase.hpp>
|
|
#include <boost/noncopyable.hpp>
|
|
#include <boost/thread.hpp>
|
|
|
|
#include <cassert>
|
|
#include <cfloat>
|
|
#include <climits>
|
|
|
|
#include <algorithm>
|
|
#include <fstream>
|
|
#include <queue>
|
|
#include <vector>
|
|
|
|
//tuning parameters
|
|
const static uint32_t RTREE_BRANCHING_FACTOR = 50;
|
|
const static uint32_t RTREE_LEAF_NODE_SIZE = 1170;
|
|
|
|
// Implements a static, i.e. packed, R-tree
|
|
|
|
static boost::thread_specific_ptr<std::ifstream> thread_local_rtree_stream;
|
|
|
|
template<class DataT>
|
|
class StaticRTree : boost::noncopyable {
|
|
private:
|
|
struct RectangleInt2D {
|
|
RectangleInt2D() :
|
|
min_lon(INT_MAX),
|
|
max_lon(INT_MIN),
|
|
min_lat(INT_MAX),
|
|
max_lat(INT_MIN) {}
|
|
|
|
int32_t min_lon, max_lon;
|
|
int32_t min_lat, max_lat;
|
|
|
|
inline void InitializeMBRectangle(
|
|
const DataT * objects,
|
|
const uint32_t element_count
|
|
) {
|
|
for(uint32_t i = 0; i < element_count; ++i) {
|
|
min_lon = std::min(
|
|
min_lon, std::min(objects[i].lon1, objects[i].lon2)
|
|
);
|
|
max_lon = std::max(
|
|
max_lon, std::max(objects[i].lon1, objects[i].lon2)
|
|
);
|
|
|
|
min_lat = std::min(
|
|
min_lat, std::min(objects[i].lat1, objects[i].lat2)
|
|
);
|
|
max_lat = std::max(
|
|
max_lat, std::max(objects[i].lat1, objects[i].lat2)
|
|
);
|
|
}
|
|
}
|
|
|
|
inline void AugmentMBRectangle(const RectangleInt2D & other) {
|
|
min_lon = std::min(min_lon, other.min_lon);
|
|
max_lon = std::max(max_lon, other.max_lon);
|
|
min_lat = std::min(min_lat, other.min_lat);
|
|
max_lat = std::max(max_lat, other.max_lat);
|
|
}
|
|
|
|
inline _Coordinate Centroid() const {
|
|
_Coordinate centroid;
|
|
//The coordinates of the midpoints are given by:
|
|
//x = (x1 + x2) /2 and y = (y1 + y2) /2.
|
|
centroid.lon = (min_lon + max_lon)/2;
|
|
centroid.lat = (min_lat + max_lat)/2;
|
|
return centroid;
|
|
}
|
|
|
|
inline bool Intersects(const RectangleInt2D & other) const {
|
|
_Coordinate upper_left (other.max_lat, other.min_lon);
|
|
_Coordinate upper_right(other.max_lat, other.max_lon);
|
|
_Coordinate lower_right(other.min_lat, other.max_lon);
|
|
_Coordinate lower_left (other.min_lat, other.min_lon);
|
|
|
|
return (
|
|
Contains(upper_left)
|
|
|| Contains(upper_right)
|
|
|| Contains(lower_right)
|
|
|| Contains(lower_left)
|
|
);
|
|
}
|
|
|
|
inline double GetMinDist(const _Coordinate & location) const {
|
|
bool is_contained = Contains(location);
|
|
if (is_contained) {
|
|
return 0.0;
|
|
}
|
|
|
|
double min_dist = DBL_MAX;
|
|
min_dist = std::min(
|
|
min_dist,
|
|
ApproximateDistance(
|
|
location.lat,
|
|
location.lon,
|
|
max_lat,
|
|
min_lon
|
|
)
|
|
);
|
|
min_dist = std::min(
|
|
min_dist,
|
|
ApproximateDistance(
|
|
location.lat,
|
|
location.lon,
|
|
max_lat,
|
|
max_lon
|
|
)
|
|
);
|
|
min_dist = std::min(
|
|
min_dist,
|
|
ApproximateDistance(
|
|
location.lat,
|
|
location.lon,
|
|
min_lat,
|
|
max_lon
|
|
)
|
|
);
|
|
min_dist = std::min(
|
|
min_dist,
|
|
ApproximateDistance(
|
|
location.lat,
|
|
location.lon,
|
|
min_lat,
|
|
min_lon
|
|
)
|
|
);
|
|
return min_dist;
|
|
}
|
|
|
|
inline double GetMinMaxDist(const _Coordinate & location) const {
|
|
double min_max_dist = DBL_MAX;
|
|
//Get minmax distance to each of the four sides
|
|
_Coordinate upper_left (max_lat, min_lon);
|
|
_Coordinate upper_right(max_lat, max_lon);
|
|
_Coordinate lower_right(min_lat, max_lon);
|
|
_Coordinate lower_left (min_lat, min_lon);
|
|
|
|
min_max_dist = std::min(
|
|
min_max_dist,
|
|
std::max(
|
|
ApproximateDistance(location, upper_left ),
|
|
ApproximateDistance(location, upper_right)
|
|
)
|
|
);
|
|
|
|
min_max_dist = std::min(
|
|
min_max_dist,
|
|
std::max(
|
|
ApproximateDistance(location, upper_right),
|
|
ApproximateDistance(location, lower_right)
|
|
)
|
|
);
|
|
|
|
min_max_dist = std::min(
|
|
min_max_dist,
|
|
std::max(
|
|
ApproximateDistance(location, lower_right),
|
|
ApproximateDistance(location, lower_left )
|
|
)
|
|
);
|
|
|
|
min_max_dist = std::min(
|
|
min_max_dist,
|
|
std::max(
|
|
ApproximateDistance(location, lower_left ),
|
|
ApproximateDistance(location, upper_left )
|
|
)
|
|
);
|
|
return min_max_dist;
|
|
}
|
|
|
|
inline bool Contains(const _Coordinate & location) const {
|
|
bool lats_contained =
|
|
(location.lat > min_lat) && (location.lat < max_lat);
|
|
bool lons_contained =
|
|
(location.lon > min_lon) && (location.lon < max_lon);
|
|
return lats_contained && lons_contained;
|
|
}
|
|
|
|
inline friend std::ostream & operator<< ( std::ostream & out, const RectangleInt2D & rect ) {
|
|
out << rect.min_lat/100000. << "," << rect.min_lon/100000. << " " << rect.max_lat/100000. << "," << rect.max_lon/100000.;
|
|
return out;
|
|
}
|
|
};
|
|
|
|
typedef RectangleInt2D RectangleT;
|
|
|
|
struct WrappedInputElement {
|
|
explicit WrappedInputElement(const uint32_t _array_index, const uint64_t _hilbert_value) :
|
|
m_array_index(_array_index), m_hilbert_value(_hilbert_value) {}
|
|
WrappedInputElement() : m_array_index(UINT_MAX), m_hilbert_value(0) {}
|
|
|
|
uint32_t m_array_index;
|
|
uint64_t m_hilbert_value;
|
|
|
|
inline bool operator<(const WrappedInputElement & other) const {
|
|
return m_hilbert_value < other.m_hilbert_value;
|
|
}
|
|
};
|
|
|
|
struct LeafNode {
|
|
LeafNode() : object_count(0) {}
|
|
uint32_t object_count;
|
|
DataT objects[RTREE_LEAF_NODE_SIZE];
|
|
};
|
|
|
|
struct TreeNode {
|
|
TreeNode() : child_count(0), child_is_on_disk(false) {}
|
|
RectangleT minimum_bounding_rectangle;
|
|
uint32_t child_count:31;
|
|
bool child_is_on_disk:1;
|
|
uint32_t children[RTREE_BRANCHING_FACTOR];
|
|
};
|
|
|
|
struct QueryCandidate {
|
|
explicit QueryCandidate(const uint32_t n_id, const double dist) : node_id(n_id), min_dist(dist)/*, minmax_dist(DBL_MAX)*/ {}
|
|
QueryCandidate() : node_id(UINT_MAX), min_dist(DBL_MAX)/*, minmax_dist(DBL_MAX)*/ {}
|
|
uint32_t node_id;
|
|
double min_dist;
|
|
// double minmax_dist;
|
|
inline bool operator<(const QueryCandidate & other) const {
|
|
return min_dist < other.min_dist;
|
|
}
|
|
};
|
|
|
|
std::vector<TreeNode> m_search_tree;
|
|
uint64_t m_element_count;
|
|
|
|
std::string m_leaf_node_filename;
|
|
public:
|
|
//Construct a pack R-Tree from the input-list with Kamel-Faloutsos algorithm [1]
|
|
explicit StaticRTree(std::vector<DataT> & input_data_vector, const char * tree_node_filename, const char * leaf_node_filename) :
|
|
m_leaf_node_filename(leaf_node_filename) {
|
|
m_element_count = input_data_vector.size();
|
|
//remove elements that are flagged to be ignored
|
|
// boost::remove_erase_if(input_data_vector, boost::bind(&DataT::isIgnored, _1 ));
|
|
|
|
INFO("constructing r-tree of " << m_element_count << " elements");
|
|
// INFO("sizeof(LeafNode)=" << sizeof(LeafNode));
|
|
// INFO("sizeof(TreeNode)=" << sizeof(TreeNode));
|
|
// INFO("sizeof(WrappedInputElement)=" << sizeof(WrappedInputElement));
|
|
double time1 = get_timestamp();
|
|
std::vector<WrappedInputElement> input_wrapper_vector(input_data_vector.size());
|
|
|
|
//generate auxiliary vector of hilbert-values
|
|
#pragma omp parallel for schedule(guided)
|
|
for(uint64_t element_counter = 0; element_counter < m_element_count; ++element_counter) {
|
|
//INFO("ID: " << input_data_vector[element_counter].id);
|
|
input_wrapper_vector[element_counter].m_array_index = element_counter;
|
|
//Get Hilbert-Value for centroid in mercartor projection
|
|
DataT & current_element = input_data_vector[element_counter];
|
|
_Coordinate current_centroid = current_element.Centroid();
|
|
current_centroid.lat = 100000*lat2y(current_centroid.lat/100000.);
|
|
|
|
uint64_t current_hilbert_value = HilbertCode::GetHilbertNumberForCoordinate(current_centroid);
|
|
input_wrapper_vector[element_counter].m_hilbert_value = current_hilbert_value;
|
|
|
|
}
|
|
//INFO("finished wrapper setup");
|
|
|
|
//open leaf file
|
|
std::ofstream leaf_node_file(leaf_node_filename, std::ios::binary);
|
|
leaf_node_file.write((char*) &m_element_count, sizeof(uint64_t));
|
|
|
|
//sort the hilbert-value representatives
|
|
std::sort(input_wrapper_vector.begin(), input_wrapper_vector.end());
|
|
// INFO("finished sorting");
|
|
std::vector<TreeNode> tree_nodes_in_level;
|
|
|
|
//pack M elements into leaf node and write to leaf file
|
|
uint64_t processed_objects_count = 0;
|
|
while(processed_objects_count < m_element_count) {
|
|
|
|
LeafNode current_leaf;
|
|
TreeNode current_node;
|
|
for(uint32_t current_element_index = 0; RTREE_LEAF_NODE_SIZE > current_element_index; ++current_element_index) {
|
|
if(m_element_count > (processed_objects_count + current_element_index)) {
|
|
// INFO("Checking element " << (processed_objects_count + current_element_index));
|
|
uint32_t index_of_next_object = input_wrapper_vector[processed_objects_count + current_element_index].m_array_index;
|
|
current_leaf.objects[current_element_index] = input_data_vector[index_of_next_object];
|
|
++current_leaf.object_count;
|
|
}
|
|
}
|
|
|
|
if(0 == processed_objects_count) {
|
|
for(uint32_t i = 0; i < current_leaf.object_count; ++i) {
|
|
//INFO("[" << i << "] id: " << current_leaf.objects[i].id << ", weight: " << current_leaf.objects[i].weight << ", " << current_leaf.objects[i].lat1/100000. << "," << current_leaf.objects[i].lon1/100000. << ";" << current_leaf.objects[i].lat2/100000. << "," << current_leaf.objects[i].lon2/100000.);
|
|
}
|
|
}
|
|
|
|
//generate tree node that resemble the objects in leaf and store it for next level
|
|
current_node.minimum_bounding_rectangle.InitializeMBRectangle(current_leaf.objects, current_leaf.object_count);
|
|
current_node.child_is_on_disk = true;
|
|
current_node.children[0] = tree_nodes_in_level.size();
|
|
tree_nodes_in_level.push_back(current_node);
|
|
|
|
//write leaf_node to leaf node file
|
|
leaf_node_file.write((char*)¤t_leaf, sizeof(current_leaf));
|
|
processed_objects_count += current_leaf.object_count;
|
|
}
|
|
|
|
// INFO("wrote " << processed_objects_count << " leaf objects");
|
|
|
|
//close leaf file
|
|
leaf_node_file.close();
|
|
|
|
uint32_t processing_level = 0;
|
|
while(1 < tree_nodes_in_level.size()) {
|
|
// INFO("processing " << (uint32_t)tree_nodes_in_level.size() << " tree nodes in level " << processing_level);
|
|
std::vector<TreeNode> tree_nodes_in_next_level;
|
|
uint32_t processed_tree_nodes_in_level = 0;
|
|
while(processed_tree_nodes_in_level < tree_nodes_in_level.size()) {
|
|
TreeNode parent_node;
|
|
//pack RTREE_BRANCHING_FACTOR elements into tree_nodes each
|
|
for(uint32_t current_child_node_index = 0; RTREE_BRANCHING_FACTOR > current_child_node_index; ++current_child_node_index) {
|
|
if(processed_tree_nodes_in_level < tree_nodes_in_level.size()) {
|
|
TreeNode & current_child_node = tree_nodes_in_level[processed_tree_nodes_in_level];
|
|
//add tree node to parent entry
|
|
parent_node.children[current_child_node_index] = m_search_tree.size();
|
|
m_search_tree.push_back(current_child_node);
|
|
//augment MBR of parent
|
|
parent_node.minimum_bounding_rectangle.AugmentMBRectangle(current_child_node.minimum_bounding_rectangle);
|
|
//increase counters
|
|
++parent_node.child_count;
|
|
++processed_tree_nodes_in_level;
|
|
}
|
|
}
|
|
tree_nodes_in_next_level.push_back(parent_node);
|
|
// INFO("processed: " << processed_tree_nodes_in_level << ", generating " << (uint32_t)tree_nodes_in_next_level.size() << " parents");
|
|
}
|
|
tree_nodes_in_level.swap(tree_nodes_in_next_level);
|
|
++processing_level;
|
|
}
|
|
BOOST_ASSERT_MSG(1 == tree_nodes_in_level.size(), "tree broken, more than one root node");
|
|
//last remaining entry is the root node;
|
|
// INFO("root node has " << (uint32_t)tree_nodes_in_level[0].child_count << " children");
|
|
//store root node
|
|
m_search_tree.push_back(tree_nodes_in_level[0]);
|
|
|
|
//reverse and renumber tree to have root at index 0
|
|
std::reverse(m_search_tree.begin(), m_search_tree.end());
|
|
#pragma omp parallel for schedule(guided)
|
|
for(uint32_t i = 0; i < m_search_tree.size(); ++i) {
|
|
TreeNode & current_tree_node = m_search_tree[i];
|
|
for(uint32_t j = 0; j < current_tree_node.child_count; ++j) {
|
|
const uint32_t old_id = current_tree_node.children[j];
|
|
const uint32_t new_id = m_search_tree.size() - old_id - 1;
|
|
current_tree_node.children[j] = new_id;
|
|
}
|
|
}
|
|
|
|
//open tree file
|
|
std::ofstream tree_node_file(tree_node_filename, std::ios::binary);
|
|
uint32_t size_of_tree = m_search_tree.size();
|
|
BOOST_ASSERT_MSG(0 < size_of_tree, "tree empty");
|
|
tree_node_file.write((char *)&size_of_tree, sizeof(uint32_t));
|
|
tree_node_file.write((char *)&m_search_tree[0], sizeof(TreeNode)*size_of_tree);
|
|
//close tree node file.
|
|
tree_node_file.close();
|
|
double time2 = get_timestamp();
|
|
// INFO("written " << processed_objects_count << " leafs in " << sizeof(LeafNode)*(1+(unsigned)std::ceil(processed_objects_count/RTREE_LEAF_NODE_SIZE) )+sizeof(uint64_t) << " bytes");
|
|
// INFO("written search tree of " << size_of_tree << " tree nodes in " << sizeof(TreeNode)*size_of_tree+sizeof(uint32_t) << " bytes");
|
|
INFO("finished r-tree construction in " << (time2-time1) << " seconds");
|
|
|
|
//todo: test queries
|
|
/* INFO("first MBR:" << m_search_tree[0].minimum_bounding_rectangle);
|
|
|
|
DataT result;
|
|
time1 = get_timestamp();
|
|
bool found_nearest = NearestNeighbor(_Coordinate(50.191085,8.466479), result);
|
|
time2 = get_timestamp();
|
|
INFO("found nearest element to (50.191085,8.466479): " << (found_nearest ? "yes" : "no") << " in " << (time2-time1) << "s at (" << result.lat1/100000. << "," << result.lon1/100000. << " " << result.lat2/100000. << "," << result.lon2/100000. << ")");
|
|
time1 = get_timestamp();
|
|
found_nearest = NearestNeighbor(_Coordinate(50.23979, 8.51882), result);
|
|
time2 = get_timestamp();
|
|
INFO("found nearest element to (50.23979, 8.51882): " << (found_nearest ? "yes" : "no") << " in " << (time2-time1) << "s at (" << result.lat1/100000. << "," << result.lon1/100000. << " " << result.lat2/100000. << "," << result.lon2/100000. << ")");
|
|
time1 = get_timestamp();
|
|
found_nearest = NearestNeighbor(_Coordinate(49.0316,2.6937), result);
|
|
time2 = get_timestamp();
|
|
INFO("found nearest element to (49.0316,2.6937): " << (found_nearest ? "yes" : "no") << " in " << (time2-time1) << "s at (" << result.lat1/100000. << "," << result.lon1/100000. << " " << result.lat2/100000. << "," << result.lon2/100000. << ")");
|
|
*/
|
|
}
|
|
|
|
//Read-only operation for queries
|
|
explicit StaticRTree(
|
|
const char * node_filename,
|
|
const char * leaf_filename
|
|
) : m_leaf_node_filename(leaf_filename) {
|
|
//INFO("Loading nodes: " << node_filename);
|
|
//INFO("opening leafs: " << leaf_filename);
|
|
//open tree node file and load into RAM.
|
|
std::ifstream tree_node_file(node_filename, std::ios::binary);
|
|
uint32_t tree_size = 0;
|
|
tree_node_file.read((char*)&tree_size, sizeof(uint32_t));
|
|
//INFO("reading " << tree_size << " tree nodes in " << (sizeof(TreeNode)*tree_size) << " bytes");
|
|
m_search_tree.resize(tree_size);
|
|
tree_node_file.read((char*)&m_search_tree[0], sizeof(TreeNode)*tree_size);
|
|
tree_node_file.close();
|
|
|
|
//open leaf node file and store thread specific pointer
|
|
std::ifstream leaf_node_file(leaf_filename, std::ios::binary);
|
|
leaf_node_file.read((char*)&m_element_count, sizeof(uint64_t));
|
|
leaf_node_file.close();
|
|
|
|
//INFO( tree_size << " nodes in search tree");
|
|
//INFO( m_element_count << " elements in leafs");
|
|
}
|
|
/*
|
|
inline void FindKNearestPhantomNodesForCoordinate(
|
|
const _Coordinate & location,
|
|
const unsigned zoom_level,
|
|
const unsigned candidate_count,
|
|
std::vector<std::pair<PhantomNode, double> > & result_vector
|
|
) const {
|
|
|
|
bool ignore_tiny_components = (zoom_level <= 14);
|
|
DataT nearest_edge;
|
|
|
|
uint32_t io_count = 0;
|
|
uint32_t explored_tree_nodes_count = 0;
|
|
INFO("searching for coordinate " << input_coordinate);
|
|
double min_dist = DBL_MAX;
|
|
double min_max_dist = DBL_MAX;
|
|
bool found_a_nearest_edge = false;
|
|
|
|
_Coordinate nearest, current_start_coordinate, current_end_coordinate;
|
|
|
|
//initialize queue with root element
|
|
std::priority_queue<QueryCandidate> traversal_queue;
|
|
traversal_queue.push(QueryCandidate(0, m_search_tree[0].minimum_bounding_rectangle.GetMinDist(input_coordinate)));
|
|
BOOST_ASSERT_MSG(FLT_EPSILON > (0. - traversal_queue.top().min_dist), "Root element in NN Search has min dist != 0.");
|
|
|
|
while(!traversal_queue.empty()) {
|
|
const QueryCandidate current_query_node = traversal_queue.top(); traversal_queue.pop();
|
|
|
|
++explored_tree_nodes_count;
|
|
bool prune_downward = (current_query_node.min_dist >= min_max_dist);
|
|
bool prune_upward = (current_query_node.min_dist >= min_dist);
|
|
if( !prune_downward && !prune_upward ) { //downward pruning
|
|
TreeNode & current_tree_node = m_search_tree[current_query_node.node_id];
|
|
if (current_tree_node.child_is_on_disk) {
|
|
LeafNode current_leaf_node;
|
|
LoadLeafFromDisk(current_tree_node.children[0], current_leaf_node);
|
|
++io_count;
|
|
for(uint32_t i = 0; i < current_leaf_node.object_count; ++i) {
|
|
DataT & current_edge = current_leaf_node.objects[i];
|
|
if(ignore_tiny_components && current_edge.belongsToTinyComponent) {
|
|
continue;
|
|
}
|
|
|
|
double current_ratio = 0.;
|
|
double current_perpendicular_distance = ComputePerpendicularDistance(
|
|
input_coordinate,
|
|
_Coordinate(current_edge.lat1, current_edge.lon1),
|
|
_Coordinate(current_edge.lat2, current_edge.lon2),
|
|
nearest,
|
|
¤t_ratio
|
|
);
|
|
|
|
if(
|
|
current_perpendicular_distance < min_dist
|
|
&& !DoubleEpsilonCompare(
|
|
current_perpendicular_distance,
|
|
min_dist
|
|
)
|
|
) { //found a new minimum
|
|
min_dist = current_perpendicular_distance;
|
|
result_phantom_node.edgeBasedNode = current_edge.id;
|
|
result_phantom_node.nodeBasedEdgeNameID = current_edge.nameID;
|
|
result_phantom_node.weight1 = current_edge.weight;
|
|
result_phantom_node.weight2 = INT_MAX;
|
|
result_phantom_node.location = nearest;
|
|
current_start_coordinate.lat = current_edge.lat1;
|
|
current_start_coordinate.lon = current_edge.lon1;
|
|
current_end_coordinate.lat = current_edge.lat2;
|
|
current_end_coordinate.lon = current_edge.lon2;
|
|
nearest_edge = current_edge;
|
|
found_a_nearest_edge = true;
|
|
} else if(
|
|
DoubleEpsilonCompare(current_perpendicular_distance, min_dist) &&
|
|
1 == abs(current_edge.id - result_phantom_node.edgeBasedNode )
|
|
&& CoordinatesAreEquivalent(
|
|
current_start_coordinate,
|
|
_Coordinate(
|
|
current_edge.lat1,
|
|
current_edge.lon1
|
|
),
|
|
_Coordinate(
|
|
current_edge.lat2,
|
|
current_edge.lon2
|
|
),
|
|
current_end_coordinate
|
|
)
|
|
) {
|
|
result_phantom_node.edgeBasedNode = std::min(current_edge.id, result_phantom_node.edgeBasedNode);
|
|
result_phantom_node.weight2 = current_edge.weight;
|
|
}
|
|
}
|
|
} else {
|
|
//traverse children, prune if global mindist is smaller than local one
|
|
for (uint32_t i = 0; i < current_tree_node.child_count; ++i) {
|
|
const int32_t child_id = current_tree_node.children[i];
|
|
TreeNode & child_tree_node = m_search_tree[child_id];
|
|
RectangleT & child_rectangle = child_tree_node.minimum_bounding_rectangle;
|
|
const double current_min_dist = child_rectangle.GetMinDist(input_coordinate);
|
|
const double current_min_max_dist = child_rectangle.GetMinMaxDist(input_coordinate);
|
|
if( current_min_max_dist < min_max_dist ) {
|
|
min_max_dist = current_min_max_dist;
|
|
}
|
|
if (current_min_dist > min_max_dist) {
|
|
continue;
|
|
}
|
|
if (current_min_dist > min_dist) { //upward pruning
|
|
continue;
|
|
}
|
|
traversal_queue.push(QueryCandidate(child_id, current_min_dist));
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
const double ratio = (found_a_nearest_edge ?
|
|
std::min(1., ApproximateDistance(_Coordinate(nearest_edge.lat1, nearest_edge.lon1),
|
|
result_phantom_node.location)/ApproximateDistance(_Coordinate(nearest_edge.lat1, nearest_edge.lon1), _Coordinate(nearest_edge.lat2, nearest_edge.lon2))
|
|
) : 0
|
|
);
|
|
result_phantom_node.weight1 *= ratio;
|
|
if(INT_MAX != result_phantom_node.weight2) {
|
|
result_phantom_node.weight2 *= (1.-ratio);
|
|
}
|
|
result_phantom_node.ratio = ratio;
|
|
|
|
//Hack to fix rounding errors and wandering via nodes.
|
|
if(std::abs(input_coordinate.lon - result_phantom_node.location.lon) == 1) {
|
|
result_phantom_node.location.lon = input_coordinate.lon;
|
|
}
|
|
if(std::abs(input_coordinate.lat - result_phantom_node.location.lat) == 1) {
|
|
result_phantom_node.location.lat = input_coordinate.lat;
|
|
}
|
|
|
|
INFO("mindist: " << min_dist << ", io's: " << io_count << ", nodes: " << explored_tree_nodes_count << ", loc: " << result_phantom_node.location << ", ratio: " << ratio << ", id: " << result_phantom_node.edgeBasedNode);
|
|
INFO("bidirected: " << (result_phantom_node.isBidirected() ? "yes" : "no") );
|
|
return found_a_nearest_edge;
|
|
|
|
}
|
|
|
|
*/
|
|
bool FindPhantomNodeForCoordinate(
|
|
const _Coordinate & input_coordinate,
|
|
PhantomNode & result_phantom_node,
|
|
const unsigned zoom_level
|
|
) {
|
|
|
|
bool ignore_tiny_components = (zoom_level <= 14);
|
|
DataT nearest_edge;
|
|
|
|
uint32_t io_count = 0;
|
|
uint32_t explored_tree_nodes_count = 0;
|
|
//INFO("searching for coordinate " << input_coordinate);
|
|
double min_dist = DBL_MAX;
|
|
double min_max_dist = DBL_MAX;
|
|
bool found_a_nearest_edge = false;
|
|
|
|
_Coordinate nearest, current_start_coordinate, current_end_coordinate;
|
|
|
|
//initialize queue with root element
|
|
std::priority_queue<QueryCandidate> traversal_queue;
|
|
double current_min_dist = m_search_tree[0].minimum_bounding_rectangle.GetMinDist(input_coordinate);
|
|
traversal_queue.push(
|
|
QueryCandidate(0, current_min_dist)
|
|
);
|
|
|
|
BOOST_ASSERT_MSG(
|
|
FLT_EPSILON > (0. - traversal_queue.top().min_dist),
|
|
"Root element in NN Search has min dist != 0."
|
|
);
|
|
|
|
while(!traversal_queue.empty()) {
|
|
const QueryCandidate current_query_node = traversal_queue.top(); traversal_queue.pop();
|
|
|
|
++explored_tree_nodes_count;
|
|
bool prune_downward = (current_query_node.min_dist >= min_max_dist);
|
|
bool prune_upward = (current_query_node.min_dist >= min_dist);
|
|
if( !prune_downward && !prune_upward ) { //downward pruning
|
|
TreeNode & current_tree_node = m_search_tree[current_query_node.node_id];
|
|
if (current_tree_node.child_is_on_disk) {
|
|
LeafNode current_leaf_node;
|
|
LoadLeafFromDisk(current_tree_node.children[0], current_leaf_node);
|
|
++io_count;
|
|
//INFO("checking " << current_leaf_node.object_count << " elements");
|
|
for(uint32_t i = 0; i < current_leaf_node.object_count; ++i) {
|
|
DataT & current_edge = current_leaf_node.objects[i];
|
|
if(ignore_tiny_components && current_edge.belongsToTinyComponent) {
|
|
continue;
|
|
}
|
|
if(current_edge.isIgnored()) {
|
|
continue;
|
|
}
|
|
|
|
double current_ratio = 0.;
|
|
double current_perpendicular_distance = ComputePerpendicularDistance(
|
|
input_coordinate,
|
|
_Coordinate(current_edge.lat1, current_edge.lon1),
|
|
_Coordinate(current_edge.lat2, current_edge.lon2),
|
|
nearest,
|
|
¤t_ratio
|
|
);
|
|
|
|
//INFO("[" << current_edge.id << "] (" << current_edge.lat1/100000. << "," << current_edge.lon1/100000. << ")==(" << current_edge.lat2/100000. << "," << current_edge.lon2/100000. << ") at distance " << current_perpendicular_distance << " min dist: " << min_dist
|
|
// << ", ratio " << current_ratio
|
|
// );
|
|
|
|
if(
|
|
current_perpendicular_distance < min_dist
|
|
&& !DoubleEpsilonCompare(
|
|
current_perpendicular_distance,
|
|
min_dist
|
|
)
|
|
) { //found a new minimum
|
|
min_dist = current_perpendicular_distance;
|
|
result_phantom_node.edgeBasedNode = current_edge.id;
|
|
result_phantom_node.nodeBasedEdgeNameID = current_edge.nameID;
|
|
result_phantom_node.weight1 = current_edge.weight;
|
|
result_phantom_node.weight2 = INT_MAX;
|
|
result_phantom_node.location = nearest;
|
|
current_start_coordinate.lat = current_edge.lat1;
|
|
current_start_coordinate.lon = current_edge.lon1;
|
|
current_end_coordinate.lat = current_edge.lat2;
|
|
current_end_coordinate.lon = current_edge.lon2;
|
|
nearest_edge = current_edge;
|
|
found_a_nearest_edge = true;
|
|
} else if(
|
|
DoubleEpsilonCompare(current_perpendicular_distance, min_dist) &&
|
|
1 == abs(current_edge.id - result_phantom_node.edgeBasedNode )
|
|
&& CoordinatesAreEquivalent(
|
|
current_start_coordinate,
|
|
_Coordinate(
|
|
current_edge.lat1,
|
|
current_edge.lon1
|
|
),
|
|
_Coordinate(
|
|
current_edge.lat2,
|
|
current_edge.lon2
|
|
),
|
|
current_end_coordinate
|
|
)
|
|
) {
|
|
BOOST_ASSERT_MSG(current_edge.id != result_phantom_node.edgeBasedNode, "IDs not different");
|
|
//INFO("found bidirected edge on nodes " << current_edge.id << " and " << result_phantom_node.edgeBasedNode);
|
|
result_phantom_node.weight2 = current_edge.weight;
|
|
if(current_edge.id < result_phantom_node.edgeBasedNode) {
|
|
result_phantom_node.edgeBasedNode = current_edge.id;
|
|
std::swap(result_phantom_node.weight1, result_phantom_node.weight2);
|
|
std::swap(current_end_coordinate, current_start_coordinate);
|
|
// INFO("case 2");
|
|
}
|
|
//INFO("w1: " << result_phantom_node.weight1 << ", w2: " << result_phantom_node.weight2);
|
|
}
|
|
}
|
|
} else {
|
|
//traverse children, prune if global mindist is smaller than local one
|
|
for (uint32_t i = 0; i < current_tree_node.child_count; ++i) {
|
|
const int32_t child_id = current_tree_node.children[i];
|
|
TreeNode & child_tree_node = m_search_tree[child_id];
|
|
RectangleT & child_rectangle = child_tree_node.minimum_bounding_rectangle;
|
|
const double current_min_dist = child_rectangle.GetMinDist(input_coordinate);
|
|
const double current_min_max_dist = child_rectangle.GetMinMaxDist(input_coordinate);
|
|
if( current_min_max_dist < min_max_dist ) {
|
|
min_max_dist = current_min_max_dist;
|
|
}
|
|
if (current_min_dist > min_max_dist) {
|
|
continue;
|
|
}
|
|
if (current_min_dist > min_dist) { //upward pruning
|
|
continue;
|
|
}
|
|
traversal_queue.push(QueryCandidate(child_id, current_min_dist));
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
const double ratio = (found_a_nearest_edge ?
|
|
std::min(1., ApproximateDistance(current_start_coordinate,
|
|
result_phantom_node.location)/ApproximateDistance(current_start_coordinate, current_end_coordinate)
|
|
) : 0
|
|
);
|
|
result_phantom_node.weight1 *= ratio;
|
|
if(INT_MAX != result_phantom_node.weight2) {
|
|
result_phantom_node.weight2 *= (1.-ratio);
|
|
}
|
|
result_phantom_node.ratio = ratio;
|
|
|
|
//Hack to fix rounding errors and wandering via nodes.
|
|
if(std::abs(input_coordinate.lon - result_phantom_node.location.lon) == 1) {
|
|
result_phantom_node.location.lon = input_coordinate.lon;
|
|
}
|
|
if(std::abs(input_coordinate.lat - result_phantom_node.location.lat) == 1) {
|
|
result_phantom_node.location.lat = input_coordinate.lat;
|
|
}
|
|
|
|
// INFO("start: (" << nearest_edge.lat1 << "," << nearest_edge.lon1 << "), end: (" << nearest_edge.lat2 << "," << nearest_edge.lon2 << ")" );
|
|
// INFO("mindist: " << min_dist << ", io's: " << io_count << ", nodes: " << explored_tree_nodes_count << ", loc: " << result_phantom_node.location << ", ratio: " << ratio << ", id: " << result_phantom_node.edgeBasedNode);
|
|
// INFO("weight1: " << result_phantom_node.weight1 << ", weight2: " << result_phantom_node.weight2);
|
|
// INFO("bidirected: " << (result_phantom_node.isBidirected() ? "yes" : "no") );
|
|
// INFO("NameID: " << result_phantom_node.nodeBasedEdgeNameID);
|
|
return found_a_nearest_edge;
|
|
|
|
}
|
|
/*
|
|
//Nearest-Neighbor query with the Roussopoulos et al. algorithm [2]
|
|
inline bool NearestNeighbor(const _Coordinate & input_coordinate, DataT & result_element) {
|
|
uint32_t io_count = 0;
|
|
uint32_t explored_tree_nodes_count = 0;
|
|
INFO("searching for coordinate " << input_coordinate);
|
|
double min_dist = DBL_MAX;
|
|
double min_max_dist = DBL_MAX;
|
|
bool found_return_value = false;
|
|
|
|
//initialize queue with root element
|
|
std::priority_queue<QueryCandidate> traversal_queue;
|
|
traversal_queue.push(QueryCandidate(0, m_search_tree[0].minimum_bounding_rectangle.GetMinDist(input_coordinate)));
|
|
BOOST_ASSERT_MSG(FLT_EPSILON > (0. - traversal_queue.top().min_dist), "Root element in NN Search has min dist != 0.");
|
|
|
|
while(!traversal_queue.empty()) {
|
|
const QueryCandidate current_query_node = traversal_queue.top(); traversal_queue.pop();
|
|
|
|
++explored_tree_nodes_count;
|
|
|
|
// INFO("popped node " << current_query_node.node_id << " at distance " << current_query_node.min_dist);
|
|
bool prune_downward = (current_query_node.min_dist >= min_max_dist);
|
|
bool prune_upward = (current_query_node.min_dist >= min_dist);
|
|
// INFO(" up prune: " << (prune_upward ? "y" : "n" ));
|
|
// INFO(" down prune: " << (prune_downward ? "y" : "n" ));
|
|
if( prune_downward || prune_upward ) { //downward pruning
|
|
// INFO(" pruned node " << current_query_node.node_id << " because " << current_query_node.min_dist << "<" << min_max_dist);
|
|
} else {
|
|
TreeNode & current_tree_node = m_search_tree[current_query_node.node_id];
|
|
if (current_tree_node.child_is_on_disk) {
|
|
// INFO(" Fetching child from disk for id: " << current_query_node.node_id);
|
|
LeafNode current_leaf_node;
|
|
LoadLeafFromDisk(current_tree_node.children[0], current_leaf_node);
|
|
++io_count;
|
|
double ratio = 0.;
|
|
_Coordinate nearest;
|
|
for(uint32_t i = 0; i < current_leaf_node.object_count; ++i) {
|
|
DataT & current_object = current_leaf_node.objects[i];
|
|
double current_perpendicular_distance = ComputePerpendicularDistance(
|
|
input_coordinate,
|
|
_Coordinate(current_object.lat1, current_object.lon1),
|
|
_Coordinate(current_object.lat2, current_object.lon2),
|
|
nearest,
|
|
&ratio
|
|
);
|
|
|
|
if(current_perpendicular_distance < min_dist && !DoubleEpsilonCompare(current_perpendicular_distance, min_dist)) { //found a new minimum
|
|
min_dist = current_perpendicular_distance;
|
|
result_element = current_object;
|
|
found_return_value = true;
|
|
}
|
|
}
|
|
} else {
|
|
//traverse children, prune if global mindist is smaller than local one
|
|
// INFO(" Checking " << current_tree_node.child_count << " children of node " << current_query_node.node_id);
|
|
for (uint32_t i = 0; i < current_tree_node.child_count; ++i) {
|
|
const int32_t child_id = current_tree_node.children[i];
|
|
TreeNode & child_tree_node = m_search_tree[child_id];
|
|
RectangleT & child_rectangle = child_tree_node.minimum_bounding_rectangle;
|
|
const double current_min_dist = child_rectangle.GetMinDist(input_coordinate);
|
|
const double current_min_max_dist = child_rectangle.GetMinMaxDist(input_coordinate);
|
|
if( current_min_max_dist < min_max_dist ) {
|
|
min_max_dist = current_min_max_dist;
|
|
}
|
|
if (current_min_dist > min_max_dist) {
|
|
continue;
|
|
}
|
|
if (current_min_dist > min_dist) { //upward pruning
|
|
continue;
|
|
}
|
|
// INFO(" pushing node " << child_id << " at distance " << current_min_dist);
|
|
traversal_queue.push(QueryCandidate(child_id, current_min_dist));
|
|
}
|
|
}
|
|
}
|
|
}
|
|
INFO("mindist: " << min_dist << ", io's: " << io_count << ", touched nodes: " << explored_tree_nodes_count);
|
|
return found_return_value;
|
|
}
|
|
*/
|
|
private:
|
|
inline void LoadLeafFromDisk(const uint32_t leaf_id, LeafNode& result_node) {
|
|
if(!thread_local_rtree_stream.get() || !thread_local_rtree_stream->is_open()) {
|
|
thread_local_rtree_stream.reset(
|
|
new std::ifstream(
|
|
m_leaf_node_filename.c_str(),
|
|
std::ios::in | std::ios::binary
|
|
)
|
|
);
|
|
}
|
|
if(!thread_local_rtree_stream->good()) {
|
|
thread_local_rtree_stream->clear(std::ios::goodbit);
|
|
DEBUG("Resetting stale filestream");
|
|
}
|
|
uint64_t seek_pos = sizeof(uint64_t) + leaf_id*sizeof(LeafNode);
|
|
thread_local_rtree_stream->seekg(seek_pos);
|
|
thread_local_rtree_stream->read((char *)&result_node, sizeof(LeafNode));
|
|
}
|
|
|
|
inline double ComputePerpendicularDistance(
|
|
const _Coordinate& inputPoint,
|
|
const _Coordinate& source,
|
|
const _Coordinate& target,
|
|
_Coordinate& nearest, double *r) const {
|
|
const double x = static_cast<double>(inputPoint.lat);
|
|
const double y = static_cast<double>(inputPoint.lon);
|
|
const double a = static_cast<double>(source.lat);
|
|
const double b = static_cast<double>(source.lon);
|
|
const double c = static_cast<double>(target.lat);
|
|
const double d = static_cast<double>(target.lon);
|
|
double p,q,mX,nY;
|
|
if(fabs(a-c) > FLT_EPSILON){
|
|
const double m = (d-b)/(c-a); // slope
|
|
// Projection of (x,y) on line joining (a,b) and (c,d)
|
|
p = ((x + (m*y)) + (m*m*a - m*b))/(1. + m*m);
|
|
q = b + m*(p - a);
|
|
} else {
|
|
p = c;
|
|
q = y;
|
|
}
|
|
nY = (d*p - c*q)/(a*d - b*c);
|
|
mX = (p - nY*a)/c;// These values are actually n/m+n and m/m+n , we need
|
|
// not calculate the explicit values of m an n as we
|
|
// are just interested in the ratio
|
|
if(std::isnan(mX)) {
|
|
*r = (target == inputPoint) ? 1. : 0.;
|
|
} else {
|
|
*r = mX;
|
|
}
|
|
if(*r<=0.){
|
|
nearest.lat = source.lat;
|
|
nearest.lon = source.lon;
|
|
return ((b - y)*(b - y) + (a - x)*(a - x));
|
|
// return std::sqrt(((b - y)*(b - y) + (a - x)*(a - x)));
|
|
} else if(*r >= 1.){
|
|
nearest.lat = target.lat;
|
|
nearest.lon = target.lon;
|
|
return ((d - y)*(d - y) + (c - x)*(c - x));
|
|
// return std::sqrt(((d - y)*(d - y) + (c - x)*(c - x)));
|
|
}
|
|
// point lies in between
|
|
nearest.lat = p;
|
|
nearest.lon = q;
|
|
// return std::sqrt((p-x)*(p-x) + (q-y)*(q-y));
|
|
return (p-x)*(p-x) + (q-y)*(q-y);
|
|
}
|
|
|
|
inline bool CoordinatesAreEquivalent(const _Coordinate & a, const _Coordinate & b, const _Coordinate & c, const _Coordinate & d) const {
|
|
return (a == b && c == d) || (a == c && b == d) || (a == d && b == c);
|
|
}
|
|
|
|
inline bool DoubleEpsilonCompare(const double d1, const double d2) const {
|
|
return (std::fabs(d1 - d2) < FLT_EPSILON);
|
|
}
|
|
|
|
};
|
|
|
|
//[1] "On Packing R-Trees"; I. Kamel, C. Faloutsos; 1993; DOI: 10.1145/170088.170403
|
|
//[2] "Nearest Neighbor Queries", N. Roussopulos et al; 1995; DOI: 10.1145/223784.223794
|
|
|
|
|
|
#endif /* STATICRTREE_H_ */
|