1

Just would like to know if there is a more elegant and maintainable approach for this:

private String replaceSpecialChars(String fileName) {
    if (fileName.length() < 1) return null;

    if (fileName.contains("Ü")) {
        fileName = fileName.replace("Ü", "Ue");
    }

    if (fileName.contains("Ä")) {
        fileName = fileName.replace("Ä", "Ae");
    }

    if (fileName.contains("Ö")) {
        fileName = fileName.replace("Ö", "Oe");
    }

    if (fileName.contains("ü")) {
        fileName = fileName.replace("ü", "ue");
    }

    ...

    return fileName;
}

I'm restricted to Java 6.

jzibell
  • 29
  • 2

2 Answers2

4

Before you go any further on this, note that what you're doing is effectively impossible. For example, the 'ascii-fication' of 'Ö' in swedish is 'O' and not 'Oe'. There is no way to know if a word is swedish or german; after all, swedes sometimes move to germany, for example. If you open a german phonebook and you see a Mrs. Sjögren, and you asciify that to Sjoegren, you messed it up.

If you want to run 'case and asciification insensitive comparisons', well, first you have to answer a few questions. Is muller equal to mueller equal to müller? That rabbit hole goes quite deep.

The general solution is trigrams or other generalized text search tools such as provided by postgres. Alternatively, opt out of this mechanism and store this stuff in unicode, and be clear that to find Ms. Sjögren, you're going to have search for "Sjögren" for the same reason that to find Mr. Johnson, you're not going to if you try to search for Jahnson.

Note that most filesystems allow unicode filenames; there is no need to try to replace a Ü.

This also goes some way as to explain why there are no ready libraries available for this seemingly common job; the job is, in fact, impossible.

You can simplify this code by using a Map<String, String> with replacements if you must. I advise against it for the above reasons. Or, just.. keep it as is, but ditch the contains. This code is needlessly slow and lengthy.

There is no difference between:

if (fileName.contains("x")) fileName = fileName.replace("x", "y");

and just fileName = fileName.replace("x", "y"); except that the former is strictly slower (replace does not make a new string and returns itself, if you ask it to replace a string that it does not contain. The former will search twice, the latter only once, and either one will make no new strings unless actual string replacing needs to be done.

You can then chain it:

if (fileName.isEmpty()) return null;
return fileName
    .replace("Ü", "Ue")
    .replace("Ä", "Ae")
    ...
    ;

But, as I said, you probably don't want to do that, unless you want an aggravated person on the line at some point in the future complaining that you bungled up the asciification of their surname.

rzwitserloot
  • 85,357
  • 5
  • 51
  • 72
  • That's really a useful hint, thank you :-). Fortunately, the filenames are not really relevant, users can select any file name in a 3rd party software, so no chance to force any conventions. Then we have a pair of files (meta data and tif) that belong together and can only be associated via name and name in the meta data xml. – jzibell Aug 27 '20 at 05:49
0

You can remove unnecessary if statements an use a chain of String.replace methods. Your code might look something like this:

private static String replaceSpecialChars(String fileName) {
    if (fileName == null)
        return null;
    else
        return fileName
                .replace("Ü", "Ue")
                .replace("Ä", "Ae")
                .replace("Ö", "Oe")
                .replace("ü", "ue");
}

public static void main(String[] args) {
    System.out.println(replaceSpecialChars("ABc"));       // ABc
    System.out.println(replaceSpecialChars("ÜÄÖü"));      // UeAeOeue
    System.out.println(replaceSpecialChars("").length()); // 0
    System.out.println(replaceSpecialChars(null));        // null
}