Delete()返回 false，即使 file.vis()、 file.canRead()、 file.canWrite()、 file.canExecute()都返回 true

小开

There is no reason you should not be able to delete this file. I would look to see who has a hold on this file. In unix/linux, you can use the lsof utility to check which process has a lock on the file. In windows, you can use process explorer.

for lsof, it's as simple as saying:

lsof /path/and/name/of/the/file

for process explorer you can use the find menu and enter the file name to show you the handle which will point you to the process locking the file.

here is some code that does what I think you need to do:

FileOutputStream to;


try {
String file = "/tmp/will_delete.txt";
to = new FileOutputStream(file );
to.write(new String("blah blah").getBytes());
to.flush();
to.close();
File f = new File(file);
System.out.print(f.delete());
} catch (FileNotFoundException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}

It works fine on OS X. I haven't tested it on windows but I suspect it should work on Windows too. I will also admit seeing some unexpected behavior on Windows w.r.t. file handling.

小开

最佳答案

It was pretty odd the trick that worked. The thing is when I have previously read the content of the file, I used BufferedReader. After reading, I closed the buffer.

Meanwhile I switched and now I'm reading the content using FileInputStream. Also after finishing reading I close the stream. And now it's working.

The problem is I don't have the explanation for this.

I don't know BufferedReader and FileOutputStream to be incompatible.

小开

As Jon Skeet commented, you should close your file in the finally {...} block, to ensure that it's always closed. And, instead of swallowing the exceptions with the e.printStackTrace, simply don't catch and add the exception to the method signature. If you can't for any reason, at least do this:

catch(IOException ex) {
throw new RuntimeException("Error processing file XYZ", ex);
}

Now, question number #2:

What if you do this:

...
to.close();
System.out.println("Please delete the file and press <enter> afterwards!");
System.in.read();
...

Would you be able to delete the file?

Also, files are flushed when they're closed. I use IOUtils.closeQuietly(...), so I use the flush method to ensure that the contents of the file are there before I try to close it (IOUtils.closeQuietly doesn't throw exceptions). Something like this:

...
try {
...
to.flush();
} catch(IOException ex) {
throw new CannotProcessFileException("whatever", ex);
} finally {
IOUtils.closeQuietly(to);
}

So I know that the contents of the file are in there. As it usually matters to me that the contents of the file are written and not if the file could be closed or not, it really doesn't matter if the file was closed or not. In your case, as it matters, I would recommend closing the file yourself and treating any exceptions according.

小开

Hopefully this will help. I came across similar problem where i couldn't delete my file after my java code made a copy of the content to the other folder. After extensive googling, i explicitly declared every single file operation related variables and called the close() method of each file operation object, and set them to NULL. Then, there is a function called System.gc(), which will clear up the file i/o mapping (i'm not sure, i just tell what is given on the web sites).

Here is my example code:

public void start() {
File f = new File(this.archivePath + "\\" + this.currentFile.getName());
this.Copy(this.currentFile, f);


if(!this.currentFile.canWrite()){
System.out.println("Write protected file " +
this.currentFile.getAbsolutePath());


return;
}




boolean ok = this.currentFile.delete();
if(ok == false){
System.out.println("Failed to remove " + this.currentFile.getAbsolutePath());
return;
}
}


private void Copy(File source, File dest) throws IOException {
FileInputStream fin;
FileOutputStream fout;
FileChannel cin = null, cout = null;
try {
fin = new FileInputStream(source);
cin = fin.getChannel();
fout = new FileOutputStream(dest);
cout = fout.getChannel();


long size = cin.size();
MappedByteBuffer buf = cin.map(FileChannel.MapMode.READ_ONLY, 0, size);


cout.write(buf);
buf.clear();
buf = null;


cin.close();
cin = null;


fin.close();
fin = null;


cout.close();
cout = null;


fout.close();
fout = null;


System.gc();


} catch (Exception e){
this.message = e.getMessage();
e.printStackTrace();
}
}

小开

Another bug in Java. I seldom find them, only my second in my 10 year career. This is my solution, as others have mentioned. I have nether used System.gc(). But here, in my case, it is absolutely crucial. Weird? YES!

finally
{
try
{
in.close();
in = null;
out.flush();
out.close();
out = null;
System.gc();
}
catch (IOException e)
{
logger.error(e.getMessage());
e.printStackTrace();
}
}

小开

There was a problem once in ruby where files in windows needed an "fsync" to actually be able to turn around and re-read the file after writing it and closing it. Maybe this is a similar manifestation (and if so, I think a windows bug, really).

小开

I tried this simple thing and it seems to be working.

file.setWritable(true);
file.delete();

It works for me.

If this does not work try to run your Java application with sudo if on linux and as administrator when on windows. Just to make sure Java has rights to change the file properties.

小开

None of the solutions listed here worked in my situation. My solution was to use a while loop, attempting to delete the file, with a 5 second (configurable) limit for safety.

File f = new File("/path/to/file");


int limit = 20; //Only try for 5 seconds, for safety
while(!f.delete() && limit > 0){
synchronized(this){
try {
this.wait(250); //Wait for 250 milliseconds
} catch (InterruptedException e) {
e.printStackTrace();
}
}
limit--;
}

Using the above loop worked without having to do any manual garbage collecting or setting the stream to null, etc.

小开

If you are working in Eclipse IDE, that could mean that you haven't close the file in the previous launch of the application. When I had the same error message at trying to delete a file, that was the reason. It seems, Eclipse IDE doesn't close all files after termination of an application.

小开

The problem could be that the file is still seen as opened and locked by a program; or maybe it is a component from your program that it had been opened in, so you have to ensure you use the dispose() method to solve that problem. i.e. JFrame frame; .... frame.dispose();

小开

You have to close all of the streams or use try-with-resource block

static public String head(File file) throws FileNotFoundException, UnsupportedEncodingException, IOException
{
final String readLine;
try (FileInputStream fis = new FileInputStream(file);
InputStreamReader isr = new InputStreamReader(fis, "UTF-8");
LineNumberReader lnr = new LineNumberReader(isr))
{
readLine = lnr.readLine();
}
return readLine;
}

小开

the answer is when you load the file, you need apply the "close" method, in any line of code, works to me

小开

if file.delete() is sending false then in most of the cases your Bufferedreader handle will not be closed. Just close and it seems to work for me normally.

小开

Before trying to delete/rename any file, you must ensure that all the readers or writers (for ex: BufferedReader/InputStreamReader/BufferedWriter) are properly closed.

When you try to read/write your data from/to a file, the file is held by the process and not released until the program execution completes. If you want to perform the delete/rename operations before the program ends, then you must use the close() method that comes with the java.io.* classes.

小开

I had the same problem on Windows. I used to read the file in scala line by line with

Source.fromFile(path).getLines()

Now I read it as a whole with

import org.apache.commons.io.FileUtils._


// encoding is null for platform default
val content=readFileToString(new File(path),null.asInstanceOf[String])

which closes the file properly after reading and now

new File(path).delete

works.

小开

FOR Eclipse/NetBeans

Restart your IDE and run your code again this is only trick work for me after one hour long struggle.

Here is my code:

File file = new File("file-path");
if(file.exists()){
if(file.delete()){
System.out.println("Delete");
}
else{


System.out.println("not delete");
}
}

Output:

Delete

小开

Another corner case that this could happen: if you read/write a JAR file through a URL and later try to delete the same file within the same JVM session.

File f = new File("/tmp/foo.jar");
URL j = f.toURI().toURL();


URL u = new URL("jar:" + j + "!/META-INF/MANIFEST.MF");
URLConnection c = u.openConnection();


// open a Jar entry in auto-closing manner
try (InputStream i = c.getInputStream()) {


// just read some stuff; for demonstration purposes only
byte[] first16 = new byte[16];
i.read(first16);
System.out.println(new String(first16));
}


// ...


// i is now closed, so we should be good to delete the jar; but...
System.out.println(f.delete());     // says false!

Reason is that the internal JAR file handling logic of Java, tends to cache JarFile entries:

// inner class of `JarURLConnection` that wraps the actual stream returned by `getInputStream()`


class JarURLInputStream extends FilterInputStream {
JarURLInputStream(InputStream var2) {
super(var2);
}


public void close() throws IOException {
try {
super.close();
} finally {


// if `getUseCaches()` is set, `jarFile` won't get closed!


if (!JarURLConnection.this.getUseCaches()) {
JarURLConnection.this.jarFile.close();
}
}
}
}

And each JarFile (rather, the underlying ZipFile structure) would hold a handle to the file, right from the time of construction up until close() is invoked:

public ZipFile(File file, int mode, Charset charset) throws IOException {
// ...


jzfile = open(name, mode, file.lastModified(), usemmap);


// ...
}


// ...


private static native long open(String name, int mode, long lastModified,
boolean usemmap) throws IOException;

There's a good explanation on this NetBeans issue.

Apparently there are two ways to "fix" this:

You can disable the JAR file caching - for the current URLConnection, or for all future URLConnections (globally) in the current JVM session:

URL u = new URL("jar:" + j + "!/META-INF/MANIFEST.MF");
URLConnection c = u.openConnection();


// for only c
c.setUseCaches(false);


// globally; for some reason this method is not static,
// so we still need to access it through a URLConnection instance :(
c.setDefaultUseCaches(false);

[HACK WARNING!] You can manually purge the JarFile from the cache when you are done with it. The cache manager sun.net.www.protocol.jar.JarFileFactory is package-private, but some reflection magic can get the job done for you:

class JarBridge {


static void closeJar(URL url) throws Exception {


// JarFileFactory jarFactory = JarFileFactory.getInstance();
Class<?> jarFactoryClazz = Class.forName("sun.net.www.protocol.jar.JarFileFactory");
Method getInstance = jarFactoryClazz.getMethod("getInstance");
getInstance.setAccessible(true);
Object jarFactory = getInstance.invoke(jarFactoryClazz);


// JarFile jarFile = jarFactory.get(url);
Method get = jarFactoryClazz.getMethod("get", URL.class);
get.setAccessible(true);
Object jarFile = get.invoke(jarFactory, url);


// jarFactory.close(jarFile);
Method close = jarFactoryClazz.getMethod("close", JarFile.class);
close.setAccessible(true);
//noinspection JavaReflectionInvocation
close.invoke(jarFactory, jarFile);


// jarFile.close();
((JarFile) jarFile).close();
}
}


// and in your code:


// i is now closed, so we should be good to delete the jar
JarBridge.closeJar(j);
System.out.println(f.delete());     // says true, phew.

Please note: All this is based on Java 8 codebase (1.8.0_144); they may not work with other / later versions.