Ticket #16310: Addresses.java.patch

src/org/openstreetmap/josm/data/validation/tests/Addresses.java

@@ -13 +13 @@
 import java.util.Map;
 import java.util.Map.Entry;
 import java.util.Set;
+import java.util.stream.Collectors;
+import java.util.stream.Stream;
 
 import org.openstreetmap.josm.data.coor.EastNorth;
+import org.openstreetmap.josm.data.coor.LatLon;
 import org.openstreetmap.josm.data.osm.Node;
 import org.openstreetmap.josm.data.osm.OsmPrimitive;
 import org.openstreetmap.josm.data.osm.Relation;
@@ -47 +50 @@
     protected static final String ADDR_NEIGHBOURHOOD = "addr:neighbourhood";
     protected static final String ADDR_PLACE         = "addr:place";
     protected static final String ADDR_STREET        = "addr:street";
+    protected static final String ADDR_CITY          = "addr:city";
+    protected static final String ADDR_UNIT          = "addr:unit";
+    protected static final String ADDR_FLATS         = "addr:flats";
+    protected static final String ADDR_HOUSE_NAME    = "addr:housename";
+    protected static final String ADDR_POSTCODE      = "addr:postcode";
     protected static final String ASSOCIATED_STREET  = "associatedStreet";
     // CHECKSTYLE.ON: SingleSpaceSeparator
 
+    private Map<String, Collection<OsmPrimitive>> addresses = null;
+    private Set<String> ignored_addresses = null;
+
     /**
      * Constructor
      */
@@ -102 +113 @@
         }
     }
 
+    private boolean isPOI(OsmPrimitive p) {
+        return p.hasKey("shop", "amenity", "tourism", "leisure", "emergency", "craft", "name");
+    }
+
+    private boolean hasAddress(OsmPrimitive p) {
+        return (p.hasKey(ADDR_HOUSE_NUMBER) && p.hasKey(ADDR_STREET, ADDR_PLACE));
+    }
+
+    /**
+     * adds the OsmPrimitive to the address map if it complies to the restrictions
+     * @param p OsmPrimitive that has an address
+     */
+    private void collectAddress(OsmPrimitive p) {
+        if (isPOI(p)) {
+            return;
+        }
+        String simplified_address = getSimplifiedAddress(p);
+        if (!ignored_addresses.contains(simplified_address)) {
+            if (addresses.containsKey(simplified_address)) {
+                addresses.get(simplified_address).add(p);
+            } else {
+                ArrayList<OsmPrimitive> objects = new ArrayList<>();
+                objects.add(p);
+                addresses.put(simplified_address, objects);
+            }
+        }
+    }
+
+    protected void initAddressMap(OsmPrimitive primitive) {
+        addresses = new HashMap<>();
+        ignored_addresses = new HashSet<>();
+        Collection<OsmPrimitive> primitives = primitive.getDataSet().getPrimitives(p -> !p.isDeleted());
+        String simplified_address;
+        for (OsmPrimitive p : primitives) {
+            if (p.hasKey(ADDR_UNIT, ADDR_FLATS) && p instanceof Node) {
+                for (OsmPrimitive r : p.getReferrers()) {
+                    if (hasAddress(r)) {
+                        // ignore addresses of buildings that are connected to addr:unit nodes
+                        // it's quite reasonable that there are more buildings with this address
+                        simplified_address = getSimplifiedAddress(r);
+                        if (!ignored_addresses.contains(simplified_address)) {
+                            ignored_addresses.add(simplified_address);
+                        } else if (addresses.containsKey(simplified_address)) {
+                                addresses.remove(simplified_address);
+                        }
+                    }
+                }
+            }
+            if (hasAddress(p)) {
+                collectAddress(p);
+            }
+        }
+    }
+
     @Override
+    public void endTest() {
+        addresses = null;
+        ignored_addresses = null;
+        super.endTest();
+    }
+
+    protected void checkForDuplicate(OsmPrimitive p) {
+        if (this.addresses == null) {
+            initAddressMap(p);
+        }
+        if (!isPOI(p) && hasAddress(p)) {
+            String simplified_address = getSimplifiedAddress(p);
+            if (ignored_addresses.contains(simplified_address)) {
+                return;
+            }
+            if (addresses.containsKey(simplified_address)) {
+                for (OsmPrimitive p2 : addresses.get(simplified_address)) {
+                    if (p == p2) {
+                        continue;
+                    }
+                    Severity severity_level = Severity.WARNING;
+                    List<OsmPrimitive> primitives = new ArrayList<>(2);
+                    primitives.add(p);
+                    primitives.add(p2);
+                    String city1 = p.get(ADDR_CITY);
+                    String city2 = p2.get(ADDR_CITY);
+                    double distance = getDistance(p, p2);
+                    if (city1 != null && city2 != null) {
+                        if (city1.equals(city2)) {
+                            if (!p.hasKey(ADDR_POSTCODE) || !p2.hasKey(ADDR_POSTCODE) || p.get(ADDR_POSTCODE).equals(p2.get(ADDR_POSTCODE))) {
+                                severity_level = Severity.WARNING;
+                            } else {
+                                // address including city identical but postcode differs
+                                // most likely perfectly fine
+                                severity_level = Severity.OTHER;
+                            }
+                        } else {
+                            // address differs only by city - notify if very close, otherwise ignore
+                            if (distance < 200.0) {
+                                severity_level = Severity.OTHER;
+                            } else {
+                                continue;
+                            }
+                        }
+
+                    } else {
+                        // at least one address has no city specified
+                        if (p.hasKey(ADDR_POSTCODE) && p2.hasKey(ADDR_POSTCODE) && p.get(ADDR_POSTCODE).equals(p2.get(ADDR_POSTCODE))) {
+                            // address including postcode identical
+                            severity_level = Severity.WARNING;
+                        } else {
+                            // city/postcode unclear - warn if very close, otherwise only notify
+                            // TODO: get city from surrounding boundaries?
+                            if (distance < 200.0) {
+                                severity_level = Severity.WARNING;
+                            } else {
+                                severity_level = Severity.OTHER;
+                            }
+                        }
+                    }
+                    errors.add(TestError.builder(this, severity_level, DUPLICATE_HOUSE_NUMBER)
+                            .message(tr("Duplicate house numbers"), marktr("''{0}'' ({1}m)"), simplified_address, (int) distance)
+                            .primitives(primitives).build());
+                }
+                addresses.get(simplified_address).remove(p); // otherwise we would get every warning two times
+            }
+        }
+    }
+
+    private String getSimplifiedAddress(OsmPrimitive p) {
+        String simplified_street_name = p.hasKey(ADDR_STREET) ? p.get(ADDR_STREET) : p.get(ADDR_PLACE);
+        // ignore whitespaces and dashes in street name, so that "Mozart-Gasse", "Mozart Gasse" and "Mozartgasse" are all seen as equal
+        simplified_street_name = simplified_street_name.toUpperCase().replaceAll("[ -]", "");
+        String simplified_address = Stream.of(
+                simplified_street_name,
+                p.get(ADDR_HOUSE_NUMBER),
+                p.get(ADDR_HOUSE_NAME),
+                p.get(ADDR_UNIT),
+                p.get(ADDR_FLATS))
+            .map(s -> (s == null ? "" : s))
+            .collect(Collectors.joining(" "));
+        simplified_address = simplified_address.trim().toUpperCase();
+        return simplified_address;
+    }
+
+    @Override
     public void visit(Node n) {
         checkHouseNumbersWithoutStreet(n);
+        checkForDuplicate(n);
     }
 
     @Override
     public void visit(Way w) {
         checkHouseNumbersWithoutStreet(w);
+        checkForDuplicate(w);
     }
 
     @Override
     public void visit(Relation r) {
         checkHouseNumbersWithoutStreet(r);
+        checkForDuplicate(r);
         if (r.hasTag("type", ASSOCIATED_STREET)) {
             // Used to count occurences of each house number in order to find duplicates
             Map<String, List<OsmPrimitive>> map = new HashMap<>();
@@ -185 +339 @@
         }
     }
 
+    /**
+     * returns rough distance between two OsmPrimitives
+     * @param a primitive a
+     * @param b primitive b
+     * @return distance of center of bounding boxes in meters
+     */
+    private double getDistance(OsmPrimitive a, OsmPrimitive b) {
+        LatLon center_a = a.getBBox().getCenter();
+        LatLon center_b = b.getBBox().getCenter();
+        return (center_a.greatCircleDistance(center_b));
+    }
+
     protected void checkDistance(OsmPrimitive house, Collection<Way> street) {
         EastNorth centroid;
         if (house instanceof Node) {