Hi,
I tried to echo some Non-English message in my build files.
For example:
<echo message="some Chinese characters here!"/>
which replaces all Chinese character with a "?":
[echo] ??????
I'm aware that it is PrintStream converts the Strings to bytes with the
system default character encoding (ISO-8859-1).
Well, I decided to write my own logger instead of using the
ant's DefaultLogger. The code comes here:
// ========================
// DefaultLogger.java
// ========================
package com.alibaba.toolkit.ant.loggers;
import org.apache.tools.ant.BuildEvent;
import java.io.PrintStream;
public class DefaultLogger extends org.apache.tools.ant.DefaultLogger {
public void setErrorPrintStream(PrintStream err) {
this.err = CharsetPrintStream.createWrapper(err);
}
public void setOutputPrintStream(PrintStream out) {
this.out = CharsetPrintStream.createWrapper(out);
}
}
// ========================
// CharsetPrintStream.java
// ========================
package com.alibaba.toolkit.ant.loggers;
import java.io.OutputStream;
import java.io.OutputStreamWriter;
import java.io.PrintStream;
import java.io.UnsupportedEncodingException;
import org.apache.tools.ant.BuildEvent;
public class CharsetPrintStream extends PrintStream {
private final static String SYSTEM_CHARSET = new
OutputStreamWriter(System.out).getEncoding();
private String charset = SYSTEM_CHARSET;
public static CharsetPrintStream createWrapper(OutputStream stream) {
if (stream == null || stream instanceof CharsetPrintStream) {
return (CharsetPrintStream) stream;
}
try {
return new CharsetPrintStream(stream,
System.getProperty("ant.logger.charset"));
} catch (UnsupportedEncodingException e) {
try {
return new CharsetPrintStream(stream, SYSTEM_CHARSET);
} catch (UnsupportedEncodingException e2) {
return null;// should not occur
}
}
}
public CharsetPrintStream(OutputStream out, String charset)
throws UnsupportedEncodingException {
this(out, charset, false);
}
public CharsetPrintStream(OutputStream out, String charset, boolean
autoFlush)
throws UnsupportedEncodingException {
super(out, autoFlush);
if (charset != null && !charset.trim().equals("")) {
this.charset = new OutputStreamWriter(out, charset).getEncoding();
}
}
public void print(char c) {
super.print(correctCharset(String.valueOf(c)));
}
public void print(char[] s) {
super.print(correctCharset(String.valueOf(s)));
}
public void print(String s) {
super.print(correctCharset(String.valueOf(s)));
}
public void print(Object obj) {
super.print(correctCharset(String.valueOf(obj)));
}
private String correctCharset(String s) {
if (s == null) {
return null;
}
try {
if (SYSTEM_CHARSET.equals(charset)) {
return s;
} else {
return new String(s.getBytes(charset), SYSTEM_CHARSET);
}
} catch (UnsupportedEncodingException e) {
return "should not occur!";
}
}
}
Now the echo task do the work:
# export ant.logger.charset=GBK
# ant -logger com.alibaba.toolkit.ant.loggers.DefaultLogger
But there's another problem! When I tried to describe my ant target in
Chinese, like this:
<target name="test" description="Chinese Description">
...
</target>
# ant -projecthelp
The console output is:
Buildfile: build.xml
Main targets:
test ???????
Default target: test
That is because these messages are output by calling System.out.println,
and in turn captured and forwared by ant. The problem is that the
System.out/System.err converts the Non-English characters to "?" BEFORE
forward these messages to logger handler!
It's a big problem to me!
Why not use PrintWriter instead of PrintStream? It's more natural to
Unicode and Java.
--
Michael Zhou <[EMAIL PROTECTED]>
--
To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
