Skip literal prefix index hits inside surrogate pairs (fixes #207)

java/com/google/re2j/MachineInput.java

@@ -200,9 +200,18 @@ boolean canCheckPrefix() {
 
     @Override
     int index(RE2 re2, int pos) {
-      pos += start;
-      int i = indexOf(str, re2.prefix, pos);
-      return i < 0 ? i : i - pos;
+      int p = pos + start;
+      while (true) {
+        int i = indexOf(str, re2.prefix, p);
+        if (i < 0) {
+          return -1;
+        }
+        if (i > 0 && Character.isLowSurrogate(str.charAt(i)) && Character.isHighSurrogate(str.charAt(i - 1))) {
+          p = i + 1;
+          continue;
+        }
+        return i - (pos + start);
+      }
     }
 
     @Override

javatests/com/google/re2j/MatcherTest.java

@@ -524,4 +524,23 @@ public void testPatternLongestMatch() {
       assertEquals("aaa bbb", text.substring(matcher.start(), matcher.end()));
     }
   }
+
+  @Test
+  public void testSurrogatePairInteriorNoMatch() {
+    // A pattern whose literal prefix is a lone low surrogate should not match
+    // inside a valid surrogate pair (issue #207).
+    assertFalse(Pattern.compile("\uDC21").matcher("a\uD801\uDC21b").find());
+    assertFalse(Pattern.compile("\uDC21|\uDC22").matcher("a\uD801\uDC21b").find());
+    assertFalse(Pattern.compile("[\uD800-\uDFFF]").matcher("a\uD801\uDC21b").find());
+
+    // Search must skip the interior hit and resume to find an isolated low surrogate:
+    Matcher m = Pattern.compile("\uDC21").matcher("\uD801\uDC21\uDC21");
+    assertTrue(m.find());
+    assertEquals(2, m.start());
+    assertEquals(3, m.end());
+
+    // StringBuilder CharSequence input should also behave identically:
+    Matcher mSb = Pattern.compile("\uDC21").matcher(new StringBuilder("a\uD801\uDC21b"));
+    assertFalse(mSb.find());
+  }
 }