/*
* Copyright (C) 2007 Steve Ratcliffe
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
* 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.
*
*
* Author: Steve Ratcliffe
* Create date: Jan 5, 2008
*/
package uk.me.parabola.imgfmt.app.net;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import uk.me.parabola.imgfmt.Utils;
import uk.me.parabola.imgfmt.app.BufferedImgFileWriter;
import uk.me.parabola.imgfmt.app.ImgFile;
import uk.me.parabola.imgfmt.app.ImgFileWriter;
import uk.me.parabola.imgfmt.app.Label;
import uk.me.parabola.imgfmt.app.lbl.City;
import uk.me.parabola.imgfmt.app.srt.IntegerSortKey;
import uk.me.parabola.imgfmt.app.srt.MultiSortKey;
import uk.me.parabola.imgfmt.app.srt.Sort;
import uk.me.parabola.imgfmt.app.srt.SortKey;
import uk.me.parabola.imgfmt.fs.ImgChannel;
/**
* The NET file. This consists of information about roads. It is not clear
* what this file brings on its own (without NOD) but may allow some better
* searching, street addresses etc.
*
* @author Steve Ratcliffe
*/
public class NETFile
extends ImgFile
{
private final NETHeader netHeader =
new NETHeader
();
private List<RoadDef
> roads
;
private Sort sort
;
public NETFile
(ImgChannel chan
) {
setHeader
(netHeader
);
setWriter
(new BufferedImgFileWriter
(chan
));
position
(NETHeader.
HEADER_LEN);
}
/**
* Write out NET1.
* @param numCities The number of cities in the map. Needed for the size of the written fields.
* @param numZips The number of zips in the map. Needed for the size of the written fields.
*/
public void write
(int numCities,
int numZips
) {
// Write out the actual file body.
ImgFileWriter writer = netHeader.
makeRoadWriter(getWriter
());
try {
for (RoadDef rd : roads
)
rd.
writeNet1(writer, numCities, numZips
);
} finally {
Utils.
closeFile(writer
);
}
}
/**
* Final writing out of net sections.
*
* We patch the NET offsets into the RGN file and create the sorted roads section.
*
* @param rgn The region file, this has to be patched with the calculated net offsets.
*/
public void writePost
(ImgFileWriter rgn
) {
for (RoadDef rd : roads
)
rd.
writeRgnOffsets(rgn
);
ImgFileWriter writer = netHeader.
makeSortedRoadWriter(getWriter
());
try {
List<LabeledRoadDef
> labeledRoadDefs = sortRoads
();
for (LabeledRoadDef labeledRoadDef : labeledRoadDefs
)
labeledRoadDef.
roadDef.
putSortedRoadEntry(writer, labeledRoadDef.
label);
} finally {
Utils.
closeFile(writer
);
}
getHeader
().
writeHeader(getWriter
());
}
/**
* Sort the roads by name and remove duplicates.
*
* We want a list of roads such that every entry in the list is a different road. Since in osm
* roads are frequently chopped into small pieces we have to remove the duplicates.
* This doesn't have to be perfect, it needs to be useful when searching for roads.
*
* So we have a separate entry if the road is in a different city. This would probably be enough
* except that associating streets with cities is not always very good in OSM. So I also create an
* extra entry for each subdivision. Finally there a search for disconnected roads within the subdivision
* with the same name.
*
* Performance note: The previous implementation was very, very slow when there were a large number
* of roads with the same name. Although this was an unusual situation, when it happened it appears
* that mkgmap has hung. This implementation takes a fraction of a second even for large numbers of
* same named roads.
*
* @return A sorted list of road labels that identify all the different roads.
*/
private List<LabeledRoadDef
> sortRoads
() {
List<SortKey<LabeledRoadDef
>> sortKeys =
new ArrayList<SortKey<LabeledRoadDef
>>(roads.
size());
Map<String,
byte[]> cache =
new HashMap<String,
byte[]>();
for (RoadDef rd : roads
) {
Label[] labels = rd.
getLabels();
for (int i =
0; i
< labels.
length && labels
[i
] !=
null; ++i
) {
Label label = labels
[i
];
if (label.
getLength() ==
0)
continue;
// Sort by name, city, region/country and subdivision number.
LabeledRoadDef lrd =
new LabeledRoadDef
(label, rd
);
SortKey<LabeledRoadDef
> nameKey = sort.
createSortKey(lrd, label.
getText(),
0, cache
);
// If there is a city add it to the sort.
City city = rd.
getCity();
SortKey<LabeledRoadDef
> cityKey
;
if (city
!=
null) {
int region = city.
getRegionNumber();
int country = city.
getCountryNumber();
cityKey = sort.
createSortKey(null, city.
getName(),
(region
& 0xffff
) << 16 |
(country
& 0xffff
), cache
);
} else {
cityKey = sort.
createSortKey(null,
"",
0, cache
);
}
SortKey<LabeledRoadDef
> sortKey =
new MultiSortKey
<LabeledRoadDef
>(nameKey, cityKey,
new IntegerSortKey
<LabeledRoadDef
>(null, rd.
getStartSubdivNumber(),
0));
sortKeys.
add(sortKey
);
}
}
Collections.
sort(sortKeys
);
List<LabeledRoadDef
> out =
new ArrayList<LabeledRoadDef
>(sortKeys.
size());
String lastName =
null;
City lastCity =
null;
List<LabeledRoadDef
> dupes =
new ArrayList<LabeledRoadDef
>();
// Since they are sorted we can easily remove the duplicates.
// The duplicates are saved to the dupes list.
for (SortKey<LabeledRoadDef
> key : sortKeys
) {
LabeledRoadDef lrd = key.
getObject();
String name = lrd.
label.
getText();
RoadDef road = lrd.
roadDef;
City city = road.
getCity();
if (road.
hasHouseNumbers() ||
!name.
equals(lastName
) || city
!= lastCity
) {
// process any previously collected duplicate road names and reset.
addDisconnected
(dupes, out
);
dupes =
new ArrayList<LabeledRoadDef
>();
lastName = name
;
lastCity = city
;
}
dupes.
add(lrd
);
}
// Finish off the final set of duplicates.
addDisconnected
(dupes, out
);
return out
;
}
/**
* Take a set of roads with the same name/city etc and find sets of roads that do not
* connect with each other. One of the members of each set is added to the road list.
*
* @param in A list of duplicate roads.
* @param out The list of sorted roads. Any new road is added to this.
*/
private void addDisconnected
(List<LabeledRoadDef
> in,
List<LabeledRoadDef
> out
) {
// switch out to different routines depending on the input size. A normal number of
// roads with the same name in the same city is a few tens.
if (in.
size() > 200) {
addDisconnectedLarge
(in, out
);
} else {
addDisconnectedSmall
(in, out
);
}
}
/**
* Split the input set of roads into disconnected groups and output one member from each group.
*
* This is done in an accurate manner which is slow for large numbers (eg thousands) of items in the
* input.
*
* @param in Input set of roads with the same name.
* @param out List to add the discovered groups.
*/
private void addDisconnectedSmall
(List<LabeledRoadDef
> in,
List<LabeledRoadDef
> out
) {
// Each road starts out with a different group number
int[] groups =
new int[in.
size()];
for (int i =
0; i
< groups.
length; i++
)
groups
[i
] = i
;
// Go through pairs of roads, any that are connected we mark with the same (lowest) group number.
boolean done
;
do {
done =
true;
for (int current =
0; current
< groups.
length; current++
) {
RoadDef first = in.
get(current
).
roadDef;
for (int i = current
; i
< groups.
length; i++
) {
// If the groups are already the same, then no need to test
if (groups
[current
] == groups
[i
])
continue;
if (first.
connectedTo(in.
get(i
).
roadDef)) {
groups
[current
] = groups
[i
] =
Math.
min(groups
[current
], groups
[i
]);
done =
false;
}
}
}
} while (!done
);
// Output the first road in each group
int last = -
1;
for (int i =
0; i
< groups.
length; i++
) {
if (groups
[i
] > last
) {
LabeledRoadDef lrd = in.
get(i
);
out.
add(lrd
);
last = groups
[i
];
}
}
}
/**
* Split the input set of roads into disconnected groups and output one member from each group.
*
* This is an modified algorithm for large numbers in the input set (eg thousands).
* First sort into groups by subdivision and then call {@link #addDisconnectedSmall} on each
* one. Since roads in the same subdivision are near each other this finds most connected roads, but
* since there is a maximum number of roads in a subdivision, the test can be done very quickly.
* You will get a few extra duplicate entries in the index.
*
* In normal cases this routine gives almost the same results as {@link #addDisconnectedSmall}.
*
* @param in Input set of roads with the same name.
* @param out List to add the discovered groups.
*/
private void addDisconnectedLarge
(List<LabeledRoadDef
> in,
List<LabeledRoadDef
> out
) {
Collections.
sort(in,
new Comparator<LabeledRoadDef
>() {
public int compare
(LabeledRoadDef o1, LabeledRoadDef o2
) {
Integer i1 = o1.
roadDef.
getStartSubdivNumber();
Integer i2 = o2.
roadDef.
getStartSubdivNumber();
return i1.
compareTo(i2
);
}
});
int lastDiv =
0;
List<LabeledRoadDef
> dupes =
new ArrayList<LabeledRoadDef
>();
for (LabeledRoadDef lrd : in
) {
int sd = lrd.
roadDef.
getStartSubdivNumber();
if (sd
!= lastDiv
) {
addDisconnectedSmall
(dupes, out
);
dupes =
new ArrayList<LabeledRoadDef
>();
lastDiv = sd
;
}
dupes.
add(lrd
);
}
addDisconnectedSmall
(dupes, out
);
}
public void setNetwork
(List<RoadDef
> roads
) {
this.
roads = roads
;
}
public void setSort
(Sort sort
) {
this.
sort = sort
;
}
/**
* A road can have several names. Keep an association between a road def
* and one of its names.
*/
class LabeledRoadDef
{
private final Label label
;
private final RoadDef roadDef
;
LabeledRoadDef
(Label label, RoadDef roadDef
) {
this.
label = label
;
this.
roadDef = roadDef
;
}
}
}