Implementing python's os.walk


#1

Hello,

I seem to be having some difficulty creating a version of python’s
os.walk() for ruby, and I was hoping for some pointers.

As some background, python’s os.walk() [1] is a generator function.
It is passed the top of a directory tree and it returns the following
for each subdirectory that it encounters:
. the working directory
. an Array of subdirectories
. an Array of non-directory files

Here is some truncated output for my use case:

import os
repo = ‘/usr/local/nfsen/profiles-data/live/lax1er1’
for root, dirs, files in os.walk(repo):
… if len(files) == 288:
… print root

/usr/local/nfsen/profiles-data/live/lax1er1/2008/11/11

/usr/local/nfsen/profiles-data/live/lax1er1/2008/12/13

Essentially, when there are exactly 288 files in a subdirectory, I
want to print or otherwise do something with the working directory.

Here is my attempt at simply translating this library function to ruby:

#! /usr/bin/env ruby

require ‘pp’

dirs = [ ‘/usr/local/nfsen/profiles-data/live/lax1er1’ ]

def find_dirs(top)
dirs = []
nondirs = []
Dir.entries(top).each do |f|
next if f =~ /(.$|..$)/
full_path = [top, f].join(’/’)
if File.directory?(full_path)
dirs.push(full_path)
else
nondirs.push(full_path)
end
end

yield top, dirs, nondirs

dirs.each do |d|
if File.directory?(d)
for o in find_dirs(d) { |a,b,c| puts “#{a} #{b} #{c}”}
yield o
end
end
end
end

find_dirs(dirs[0]) do |top,dirs,nondirs|
if nondirs.length == 288
puts “#{top}”
end
end

There are some things that I know are wrong or missing, but that is
due to trying to get something to run at all without throwing an
exception.

The part that I think is totally wrong is:

for o in find_dirs(d) { |a,b,c| puts “#{a} #{b} #{c}”}

It’s really only in there currently to keep the from getting
‘LocalJumpError: no block given.’ Unfortunately, I have no idea what
the correct way to deal with this would be.

The missing part would be including the directory contents in addition
to the working directory and the Array of subdirectories.

So, I guess my main questions would be: What do I need to do to get
this sort of a generator to work? Do I need to wrap this up in a
‘class’ or is a ‘def’ sufficient? What should the block look like,
and where should it be in the code?

Thanks for reading,

Brad

[1]
http://svn.python.org/view/python/branches/release25-maint/Lib/os.py?rev=62757&view=auto


#2

On Tue, 16 Dec 2008, Brad V. wrote:

Hello,

I seem to be having some difficulty creating a version of python’s os.walk()
for ruby, and I was hoping for some pointers.

As some background, python’s os.walk() [1] is a generator function. It is

You can have generators in Ruby
ri Generator
furnishes you with the details…

for root, dirs, files in os.walk(repo):
… if len(files) == 288:
… print root

/usr/local/nfsen/profiles-data/live/lax1er1/2008/11/11

/usr/local/nfsen/profiles-data/live/lax1er1/2008/12/13

Essentially, when there are exactly 288 files in a subdirectory, I want to
print or otherwise do something with the working directory.

OK.

dirs = []
nondirs = []
Dir.entries(top).each do |f|
next if f =~ /(.$|..$)/

or maybe
next if f =~ /^..?$/
or
next if f =~ /^.{1,2}$/

full_path = [top, f].join(’/’)

 full_path = File.join(tmp,f)  # separator agnostic

if File.directory?(full_path)
dirs.push(full_path)
else
nondirs.push(full_path)
end
end

yield top, dirs, nondirs

yielding to a proc with arity 3…

dirs.each do |d|
if File.directory?(d)
for o in find_dirs(d) { |a,b,c| puts “#{a} #{b} #{c}”}
yield o

yielding to a proc with arity 1

That may be one problem

There are some things that I know are wrong or missing, but that is due to
trying to get something to run at all without throwing an exception.

ri Find is short enough to quote:

------------------------------------------------------------ Class: Find
The +Find+ module supports the top-down traversal of a set of file
paths.

 For example, to total the size of all files under your home
 directory, ignoring anything in a "dot" directory (e.g.
 $HOME/.ssh):

   require 'find'

   total_size = 0

   Find.find(ENV["HOME"]) do |path|
     if FileTest.directory?(path)
       if File.basename(path)[0] == ?.
         Find.prune       # Don't look any further into this 

directory.
else
next
end
else
total_size += FileTest.size(path)
end
end


Instance methods:

 find, prune

That will do most of the lifting for you…
[…]
Hugh


#3

Brad V. wrote:

As some background, python’s os.walk() [1] is a generator function.
It is passed the top of a directory tree and it returns the following
for each subdirectory that it encounters:
. the working directory
. an Array of subdirectories
. an Array of non-directory files

The normal ‘ruby way’ to do this would be as an object which yields
each of these things in turn, rather than returning them.

In many cases you can use it directly like this. If you want to turn it
into a generator you can wrap it using generator.rb; or more cleanly in
ruby 1.9, turn it into an Enumerator, which has this functionality built
in.

class Foo
def all_dirs
yield “dir1”
yield “dir2”
yield “dir3”
end
end

foo = Foo.new

Normal style

foo.all_dirs { |x| p x }

Generator style (ruby 1.9, uses Fiber)

g = foo.to_enum(:all_dirs)
3.times { p g.next }

Generator style (ruby 1.8, beware uses callcc)

require ‘generator’
require ‘enumerator’
g = Generator.new(foo.to_enum(:all_dirs))
3.times { p g.next }


#4

2008/12/16 Brad V. removed_email_address@domain.invalid:

Here is some truncated output for my use case:

Essentially, when there are exactly 288 files in a subdirectory, I want to
print or otherwise do something with the working directory.

So, I guess my main questions would be: What do I need to do to get this
sort of a generator to work? Do I need to wrap this up in a ‘class’ or is a
‘def’ sufficient? What should the block look like, and where should it be
in the code?

You have a recursive algorithm here but you want each call of the
method invoke the same block. This can be achieved by forwarding
the block with the &b notation:

def find_dirs(top, &b)

enter recursion

find_dirs(d, &b)
end

The way you did it, every invocation yields to the caller’s block
which is only the correct one for the first caller; all others yield
to the block in their parent calling find_dirs.

You might as well be able to create a totally different solution using
Find:

require ‘find’

roots.each do |root|
dir_count = Hash.new 0

Find.find root do |file|
d, f = File.split file
next if /\A.{1,2}\z/ =~ f
dir_count[d] += 1 if File.file? file
end

dir_count.each do |dir, cnt|
puts root if cnt == 288
end
end

Kind regards

robert


#5

Many thanks for the suggestions.

I wasn’t previously aware of Find, I have been able to get it to
provide all of the information that I need.

Thanks again!

Brad


#6

Brad V. wrote:

Hello,

I seem to be having some difficulty creating a version of python’s
os.walk() for ruby, and I was hoping for some pointers.

You should really know how to use Ruby’s stdlib named “find”.
http://www.ruby-doc.org/stdlib/libdoc/find/rdoc/index.html

Below is another twisted example in Ruby 1.9. It can be used much like
python’s
generators:

irb(main):060:0> for root, dirs, files in os_walk("/dev/disk") do
irb(main):061:1* puts root
irb(main):062:1> end

require ‘find’
require ‘pathname’

def os_walk(dir)
ret = Fiber.new do
Pathname(dir).find do |ent|
next unless ent.directory?
dirs, files = ent.children.partition {|i| i.directory? }
Fiber.yield ent, dirs, files
end
raise
end
def ret.each
loop { yield resume } rescue self
end
ret
end