Changeset 15613 in josm

Timestamp:

2019-12-25T19:04:23+01:00 (5 years ago)

Author:

Don-vip

Message:

fix #18449 - exclude lowercase schwa character (ə) from "unusual unicode character" check, as it is used in azerbaidjani alphabet

Location:

trunk

Files:

• src/org/openstreetmap/josm/data/validation/tests/TagChecker.java (modified) (2 diffs)
• test/unit/org/openstreetmap/josm/data/validation/tests/TagCheckerTest.java (modified) (1 diff)

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

@@ -425 +425 @@
 
     static boolean containsUnusualUnicodeCharacter(String key, String value) {
-        return value != null && value.chars().anyMatch(c -> isUnusualUnicodeBlock(key, UnicodeBlock.of(c)));
+        return value != null && value.chars().anyMatch(c -> isUnusualUnicodeBlock(key, c));
     }
 
@@ -431 +431 @@
      * Detects highly suspicious Unicode characters that have been seen in OSM database.
      * @param key tag key
-     * @param b Unicode block of the current character
+     * @param c current character code point
      * @return {@code true} if the current unicode block is very unusual for the given key
      */
-    private static boolean isUnusualUnicodeBlock(String key, UnicodeBlock b) {
-        return isUnusualPhoneticUse(key, b) || isUnusualBmpUse(b) || isUnusualSmpUse(b);
-    }
-
-    private static boolean isUnusualPhoneticUse(String key, UnicodeBlock b) {
-        return (b == UnicodeBlock.IPA_EXTENSIONS                        // U+0250..U+02AF
+    private static boolean isUnusualUnicodeBlock(String key, int c) {
+        UnicodeBlock b = UnicodeBlock.of(c);
+        return isUnusualPhoneticUse(key, b, c) || isUnusualBmpUse(b) || isUnusualSmpUse(b);
+    }
+
+    private static boolean isUnusualPhoneticUse(String key, UnicodeBlock b, int c) {
+        return c != 0x0259                                              // U+0259 is used as a standard character in azerbaidjani
+            && (b == UnicodeBlock.IPA_EXTENSIONS                        // U+0250..U+02AF
              || b == UnicodeBlock.PHONETIC_EXTENSIONS                   // U+1D00..U+1D7F
              || b == UnicodeBlock.PHONETIC_EXTENSIONS_SUPPLEMENT)       // U+1D80..U+1DBF

trunk/test/unit/org/openstreetmap/josm/data/validation/tests/TagCheckerTest.java

@@ -315 +315 @@
         assertFalse(TagChecker.containsUnusualUnicodeCharacter("name", "Tallinn — Narva"));
     }
+
+    /**
+     * Non-regression test for <a href="https://josm.openstreetmap.de/ticket/18449">Bug #18449</a>.
+     */
+    @Test
+    public void testTicket18449() {
+        assertFalse(TagChecker.containsUnusualUnicodeCharacter("name", "Hökumət Evi"));
+    }
 }