Infinite Recursion in Linux file system
I抦 working on some code that will recursively search through a directory and enumerate a list of files. However I抦 having trouble with infinite loops. When my code reaches a link (shortcut) in the file system it will register as a directory and start from there over and over again until memory is full. Anyone have any thoughts or recommendations?
Here抯 a sample of the code:
publicstaticvoid enumerateFiles(String parentDir)
{
File searchDir =new File(parentDir);
File[] list = searchDir.listFiles();
if(list !=null)
{
for(int i=0;i<list.length;i++)
{
if(list[i].isDirectory())
enumerateFiles(list[i].toString());
else
superList.file.add(list[i]);// public static ArrayList
}
}
}
Thanks in advance.>
[1338 byte] By [
sisslacka] at [2007-10-3 4:28:02]
> use getCanonicalFile() or getCanonicalPath(), they
> resolve symlinks
Well the problem I had there was that it would still resolve to a directory and still loop infinitely.
But thanks to your post, I played around with the File class a little more and I noticed if you compare getCanonicalFile() or getCanonicalPath() to getAbsoluteFile() or getAbsolutePath() they are only equivalent when it is a normal file and they are different when it is a symlink.
Here is some output from a test app I made:
file.exists() = true
file.getAbsolutePath() = /sys/block/hdc/device/driver/1.0
file.getCanonicalPath() = /sys/devices/pci0000:00/0000:00:1f.1/ide1/1.0
file.getName() = 1.0
file.getParent() = /sys/block/hdc/device/driver
file.getPath() = /sys/block/hdc/device/driver/1.0
file.getAbsoluteFile() = /sys/block/hdc/device/driver/1.0
file.getCanonicalFile() = /sys/devices/pci0000:00/0000:00:1f.1/ide1/1.0
file.getClass() = class java.io.File
file.getParentFile() = /sys/block/hdc/device/driver
file.isAbsolute() = true
file.isDirectory() = true
file.isFile() = false
file.isHidden() = false
file.toURI() = file:/sys/block/hdc/device/driver/1.0/
file.toURL() = file:/sys/block/hdc/device/driver/1.0/
Since my app my have to enumerate all the files in / , I basically just coded the logic to throw out the symlinks and only pass the true files back into enumerateFiles(). Now I'm only enumerating the real files and not getting stuck in infinite loops.
