Charset mixed up in Dnd / File List / URI

[mtp]

Hello,

i work on linux/gnome (jdk1.4.2_10 to compile, jdk 1.5.0_07 to run), and
when i drag and drop a file whose name is ("données.xls") on my java
application i accept the Drop with DataFlavor:

java.awt.datatransfer.DataFlavor[mimetype=application/x-java-serialized-object;representationclass=java.lang.String]

the drop data is "file:///home/user/.../donn%E9es.xls\r\n"

I drop the "\r\n", and i try to decode the %xx :

URI uri = new URI(filename);
String decodedPath = uri.getPath();

but the decoded path is: "/home/user/.../donn�es.xls"

So i guess it's a Charset problem but what can i do to solve it?

In short:
- "ls" and gnome explorer shows "données.xls"
- the drop data is "donn%E9es.xls"
- the decoded path is "donn�es.xls"

Does anyone know how to solve this?

TIA

 

[Soren Kuula]

Hello,

the drop data is "file:///home/user/.../donn%E9es.xls\r\n"

I drop the "\r\n", and i try to decode the %xx :

URI uri = new URI(filename);
String decodedPath = uri.getPath();

try

String decodedPath = URLDecoder.decode(filename, "ISO-8859-1");

instead. My guess at ISO-8859 is inzzzztinct (from seeing the %E9).

but the decoded path is: "/home/user/.../donn�es.xls"

So i guess it's a Charset problem but what can i do to solve

Unless there is a way to query the Java drag and drop stuff about the
encoding used, my suggesting is to kick that penguin out of your
computer. Drag and drop has always been a pain in Linux, and character
encoding issues are not taken too seriously there either.

Søren

 

[Oliver Wong]

Hello,

i work on linux/gnome (jdk1.4.2_10 to compile, jdk 1.5.0_07 to run), and
when i drag and drop a file whose name is ("données.xls") on my java
application i accept the Drop with DataFlavor:

java.awt.datatransfer.DataFlavor[mimetype=application/x-java-serialized-object;representationclass=java.lang.String]

the drop data is "file:///home/user/.../donn%E9es.xls\r\n"

I drop the "\r\n", and i try to decode the %xx :

URI uri = new URI(filename);
String decodedPath = uri.getPath();

but the decoded path is: "/home/user/.../donn�es.xls"

So i guess it's a Charset problem but what can i do to solve it?

In short:
- "ls" and gnome explorer shows "données.xls"
- the drop data is "donn%E9es.xls"
- the decoded path is "donn�es.xls"

Does anyone know how to solve this?

In your "incorrect" decoded path, what's the unicode value of the
incorrect character? E9 is indeed the correct unicode value for the
"lowercase latin e with acute accent":
http://www.eki.ee/letter/chardata.cgi?ucode=00E9

- Oliver

 

[mtp]

try

String decodedPath = URLDecoder.decode(filename, "ISO-8859-1");

instead. My guess at ISO-8859 is inzzzztinct (from seeing the %E9).

you were right. My only idea about where the encoding come from is the
file.encoding system property:

String fileEncodingCharsetName = System.getProperty("file.encoding");
String decodedUrl = URLDecoder.decode(s, fileEncodingCharsetName);
URL url = new URL(decodedUrl);
File f = new File(url.getPath());

Unless there is a way to query the Java drag and drop stuff about the
encoding used, my suggesting is to kick that penguin out of your
computer. Drag and drop has always been a pain in Linux, and character
encoding issues are not taken too seriously there either.

true, but i can't drop it ...

 

[mtp]

In your "incorrect" decoded path, what's the unicode value of the
incorrect character? E9 is indeed the correct unicode value for the
"lowercase latin e with acute accent":
http://www.eki.ee/letter/chardata.cgi?ucode=00E9

it's also true for ISO-8859-1 and ISO-8859-15:

found in charsets: 8859-1 (E9); 8859-10 (E9); 8859-13 (E9); 8859-14
(E9); 8859-15 (E9); ...

but the page explains it:
UTF-8 (c3, a9) é
^^^^^^

which i checked :

bsh % print(URLDecoder.decode("donn%E9es", "UTF-8"));
donn�es

bsh % print(URLDecoder.decode("donn%E9es", "ISO-8859-1"));
données

bsh % print(URLEncoder.encode("données", "UTF-8"));
donn%C3%A9es
^^^^^^