Starting with JDK 9, the Scanner class comes with a method that returns a stream of delimiter-separated tokens, Stream<String> tokens(). If we treat the text to search as the delimiter of Scanner and we count the entries of the Stream returned by tokens(), then we obtain the correct result:

public static long countOccurrences(
  Path path, String text, Charset ch) throws IOException {

  long count;

  try (Scanner scanner = new Scanner(path, ch)
      .useDelimiter(Pattern.quote(text))) {

    count = scanner.tokens().count() - 1;
  }

  return count;
}