Different OS, same file, different results

[Alex Armenteros]

I'm developing a XML file parser and I'm having problems with characters like Ñ or Á

On windows, it reads the file correctly.

On CentOS 5, the special characters are read wrong

//My class that will do the needed operations
DocumentParser par = new DocumentParser();
Reader inr = new InputStreamReader(new FileInputStream("/route/to/file.xml"), "latin1");
			BufferedReader in = new BufferedReader(inr);
			String s = "";
			String res = "";
//I know this is not the best method but I don't care
			while ((s = in.readLine()) != null) {

				res += new String(s.getBytes(), "latin1");
			}

			par.parseDocument(res);

I've tried changing the encoding with ultraedit and no result.

Thanks in advance

[Jeff Verdegan]

How are you determining that they're wrong? If you're looking at them in a text editor or console window, there's a decent chance that you've got the right character, but your viewer isn't displaying it correctly, either because it's not using the right encoding or because the font doesn't include a glyph for that character. Here's one way to know for sure what you're getting:

// verify what we are writing out
char ch = ...
int codePoint = ch;
System.out.println(codePoint);
// now write the character out


// verify what we are reading in
char ch = ...; // read the character
int codePoint = ch;
System.out.println(codePoint)

Presumably you're writing and reading lines at a time in a loop, but the same thing still applies. Pick a couple of troublesome characters from the Strings, and print out their int values before writing and after reading. If you're not seeing the same values, there's a problem with your code, so post an SSCCE that shows what you're doing and your expected and actual output. If you do see the same values, then the problem is just with the viewer you're using.

[Alex Armenteros]

Yeah I forgot that after parsing the xml file I insert values in an Oracle DB

With the process running on Windows Ñ and accents are show correctly in the DB,
running on CentOS don't.

[Jeff Verdegan]

Everything I said still applies. You need to look at the int values of the characters to determine whether the data is getting corrupted or if it's just your viewer that's not displaying them correctly.

[Alex Armenteros]

As you aid mr Verdegan, the bytes are different.


Windows

> 209

CentOS

< 195
< 145

This is a file reading issue

because if I do this:

String res = "Ñ";
			// while ((s = in.readLine()) != null) {

			// res += new String(s.getBytes(), "latin1");
			// }
			for (int i = 0; i < res.length(); i++) {
				char ch = res.charAt(i);
				int k = ch;

				System.out.println(k);
			}

On both cases 209 is printed.

[Alex Armenteros]

Finally I got it to work by doing the most absurd and dirty way I could think of...

File f = new File(args[1]);
			FileInputStream fis = new FileInputStream(f);
			int size = 0;
			while (fis.read() != -1) {

				size++;
			}
			byte[] ba = new byte[size];
			int b = 0;
			fis.close();
			fis = new FileInputStream(f);
			int i = 0;

			while ((b = fis.read()) != -1) {
				ba[i] = (byte) b;
				i++;
			}

			String res = new String(ba, "latin1");