Class UTF16Input

java/com/google/re2j/MachineInput.java:174–266

  // |pos| and |width| are in Java "char" units.
  private static class UTF16Input extends MachineInput {
    final CharSequence str;
    final int start;
    final int end;

    public UTF16Input(CharSequence str, int start, int end) {
      this.str = str;
      this.start = start;
      this.end = end;
    }

    @Override
    int step(int pos) {
      pos += start;
      if (pos < end) {
        int rune = Character.codePointAt(str, pos);
        return rune << 3 | Character.charCount(rune);
      } else {
        return EOF;
      }
    }

    @Override
    boolean canCheckPrefix() {
      return true;
    }

    @Override
    int index(RE2 re2, int pos) {
      pos += start;
      int i = indexOf(str, re2.prefix, pos);
      return i < 0 ? i : i - pos;
    }

    @Override
    int context(int pos) {
      pos += start;
      int r1 = pos > 0 && pos <= str.length() ? Character.codePointBefore(str, pos) : -1;
      int r2 = pos < str.length() ? Character.codePointAt(str, pos) : -1;
      return Utils.emptyOpContext(r1, r2);
    }

    @Override
    int endPos() {
      return end;
    }

    private int indexOf(CharSequence hayStack, String needle, int pos) {
      if (hayStack instanceof String) {
        return ((String) hayStack).indexOf(needle, pos);
      }
      if (hayStack instanceof StringBuilder) {
        return ((StringBuilder) hayStack).indexOf(needle, pos);
      }
      return indexOfFallback(hayStack, needle, pos);
    }

    // Modified version of {@link String#indexOf(String) that allows a CharSequence.