CODEWITHSUNDEEP
javafile-iocharacter-encodingsavefilesystems
ChristofferAB
ChristofferAB

Reputation: 81

Can't use latin characters correctly when creating new files from within Java. Filenames get weird characters instead of the correct ones

Currently saving an int[] from hashmap in a file with the name of the key to the int[]. This exact key must be reachable from another program. Hence I can't switch name of the files to english only chars. But even though I use ISO_8859_1 as the charset for the filenames the files get all messed up in the file tree. The english letters are correct but not the special ones.

        /**
        * Save array to file
        */
        public void saveStatus(){
            try {
                for(String currentKey : hmap.keySet()) {
                    byte[] currentKeyByteArray = currentKey.getBytes();
                    String bytesString = new String(currentKeyByteArray, StandardCharsets.ISO_8859_1);
                    String fileLocation = "/var/tmp/" + bytesString + ".dat";
                    FileOutputStream saveFile = new FileOutputStream(fileLocation);
                    ObjectOutputStream out = new ObjectOutputStream(saveFile);
                    out.writeObject(hmap.get(currentKey));
                    out.close();
                    saveFile.close();
                    System.out.println("Saved file at " + fileLocation);
                }
            } catch (IOException e) {
                e.printStackTrace();
            }
        }

Could it have to do with how linux is encoding characters or is more likely to do with the Java code?

EDIT

I think the problem lies with the OS. Because when looking at text files with cat for example the problem is the same. However vim is able to decode the letters correctly. In that case I would have to perhaps change the language settings from the terminal?

Upvotes: 0

Views: 521

Answers (2)

sruetti
sruetti

Reputation: 533

The valid characters of a filename or path vary depending on the file system used. While it should be possible to just use a java string as filename (as long as it does not contain characters invalid in the given file system), there might be interoperability issues and bugs.

In other words, leave out all Charset-magic as @RealSkeptic recommends and it should work. But changing the environment might result in unexpected behavior.

Depending on your requirements, you might therefore want to encode the key to make sure it only uses a reduced character set. One variant of Base64 might work (assuming your file system is case sensitive!). You might even find a library (Apache Commons?) offering a function to reduce a string to characters safe for use in a file name.

Upvotes: 1

Chinmay jain
Chinmay jain

Reputation: 999

You have to change the charset in the getBytes function as well.

currentKey.getBytes(StandardCharsets.ISO_8859_1);

Also, why are you using StandardCharsets.ISO_8859_1? To accept a wider range of characters, use StandardCharsets.UTF_8.

Upvotes: 1

Related Questions