This might not be a suitable question for the groovy users, but I am using it
in a groovy script, so thought it might solicit some useful feedback and
alternate ideas.
In Java's NIO there are a couple of ways to walk the tree of files in order to
return a file count. There must be lots of ways! I don't know Groovy has a
better built in method - but I could not see it.
This one should be the fastest, using concurrent processes. However, if the set
of files changes during the count it errors. I have no idea how to catch and
ignore that error.
long fileCountV2(Path dir) {
return Files.walk(dir)
.parallel()
.filter(p -> !p.toFile().isDirectory())
.count();
}
This one works regardless of any file changes, and uses a Stream for low memory
overhead. But it could do with concurrency to speed it up, but I really cannot
figure how to GPARS this for performance. This script did once fail with "too
many files open" error on Linux (and a directory with 200,000+ files). I can't
determine how - no files are actually opened!
long fileCount(Path dir)
{
long i = 0
try {
Files.newDirectoryStream(dir).each {i++}
} catch (e) {
println (e)
}
return i
}
Merlin Beedell
