Changeset 12283 in josm for trunk/src/org/openstreetmap/josm/data/validation/tests/SimilarNamedWays.java

Timestamp:

2017-05-31T03:12:36+02:00 (7 years ago)

Author:

Don-vip

Message:

fix #14858 - "Similarly named ways" test: detect accent and case variations for strings of same length

File:

• trunk/src/org/openstreetmap/josm/data/validation/tests/SimilarNamedWays.java (modified) (4 diffs)

trunk/src/org/openstreetmap/josm/data/validation/tests/SimilarNamedWays.java

@@ -7 +7 @@
 
 import java.awt.geom.Point2D;
+import java.text.Normalizer;
 import java.util.ArrayList;
 import java.util.Arrays;
@@ -34 +35 @@
 
     protected static final int SIMILAR_NAMED = 701;
+
+    private static final Pattern REMOVE_DIACRITICS = Pattern.compile("\\p{InCombiningDiacriticalMarks}+");
 
     /** All ways, grouped by cells */
@@ -202 +205 @@
         boolean similar = distance > 0 && distance <= 2;
 
+        // check if only the case differs, so we don't consider large distance as different strings
+        if (distance > 2 && name.length() == name2.length()) {
+            similar = deAccent(name).equalsIgnoreCase(deAccent(name2));
+        }
+
         // try all rules
         for (NormalizeRule rule : rules) {
@@ -216 +224 @@
     }
 
+    /**
+     * Removes diacritics (accents) from string.
+     * @param str string
+     * @return {@code str} without any diacritic (accent)
+     * @since 12283
+     */
+    public static String deAccent(String str) {
+        // https://stackoverflow.com/a/1215117/2257172
+        return REMOVE_DIACRITICS.matcher(Normalizer.normalize(str, Normalizer.Form.NFD)).replaceAll("");
+    }
+
     @FunctionalInterface
     public interface NormalizeRule {