Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

head/tail split for path

0 views
Skip to first unread message

Trans

unread,
Jul 26, 2008, 3:54:43 PM7/26/08
to
I need to split a path by head/*tail.

Ex.

File.head_tail_split('home/foo/bar') #=> [ 'home', 'foo/bar' ]

Sure, I can write a clumsy loop like the following:

def File.head_tail_split(fname)
s = fname
t = []
h = nil
until s == '.'
t << h
s, h = *split(s)
end
return h, File.join(*t.compact)
end

But I'm betting there's a better way. Or maybe there's already an easy
way I'm overlooking?

T.

David A. Black

unread,
Jul 26, 2008, 5:04:29 PM7/26/08
to
Hi --

Check out the Pathname feature:

require 'pathname'

p = Pathname.new("a/b/c")

You get p.basename and p.dirname. Both return Pathname objects, but
they're quite string-like and easily converted.


David

--
Rails training from David A. Black and Ruby Power and Light:
Intro to Ruby on Rails July 21-24 Edison, NJ
* Advancing With Rails August 18-21 Edison, NJ
* Co-taught by D.A. Black and Erik Kastner
See http://www.rubypal.com for details and updates!

Thomas Sawyer

unread,
Jul 26, 2008, 5:44:57 PM7/26/08
to
Hi David,

David A. Black wrote:
>
> Check out the Pathname feature:
>
> require 'pathname'
>
> p = Pathname.new("a/b/c")
>
> You get p.basename and p.dirname. Both return Pathname objects, but
> they're quite string-like and easily converted.

p = Pathname.new("a/b/c")

p.dirname #=> "a/b"
p.basename #=> "c"

But I need "a" and "b/c".

T.
--
Posted via http://www.ruby-forum.com/.

David A. Black

unread,
Jul 26, 2008, 5:59:01 PM7/26/08
to
Hi --

On Sun, 27 Jul 2008, Thomas Sawyer wrote:

> Hi David,
>
> David A. Black wrote:
>>
>> Check out the Pathname feature:
>>
>> require 'pathname'
>>
>> p = Pathname.new("a/b/c")
>>
>> You get p.basename and p.dirname. Both return Pathname objects, but
>> they're quite string-like and easily converted.
>
> p = Pathname.new("a/b/c")
>
> p.dirname #=> "a/b"
> p.basename #=> "c"
>
> But I need "a" and "b/c".

Whoops. Well, you could do:

require 'enumerator'
[path.to_enum(:ascend).to_a[1], path.basename]

or something like:

path.scan(/([^\/]+)\/(.*)/) # with the String path

Pathname#cleanpath might come in handy if you're rolling your own. So
might File::Separator.

Bill Walton

unread,
Jul 27, 2008, 2:13:55 PM7/27/08
to
Hi Trans,

Trans wrote:

> I need to split a path by head/*tail.
>
> Ex.
>
> File.head_tail_split('home/foo/bar') #=> [ 'home', 'foo/bar' ]
>

You could do like this.

def head_tail_split(fname)
components = fname.split('/')
[components.shirt, components.join('/')]
end

irb(main):001:0> fname = 'home/foo/bar'
=> "home/foo/bar"
irb(main):002:0> components = fname.split('/')
=> ["home", "foo", "bar"]
irb(main):003:0> [components.shift, components.join('/')]
=> ["home", "foo/bar"]

HTH,
Bill

Phlip

unread,
Jul 27, 2008, 2:20:18 PM7/27/08
to
Bill Walton wrote:

> You could do like this.
>
> def head_tail_split(fname)
> components = fname.split('/')
> [components.shirt, components.join('/')]
> end
>
> irb(main):001:0> fname = 'home/foo/bar'
> => "home/foo/bar"
> irb(main):002:0> components = fname.split('/')
> => ["home", "foo", "bar"]
> irb(main):003:0> [components.shift, components.join('/')]
> => ["home", "foo/bar"]

That would be safer if fname were a Pathname.

But then, would split only split off the last path part? Or the first?

If the last, how could we roll many splits up, then pop the first?

--
Phlip

Thomas Sawyer

unread,
Jul 27, 2008, 2:47:33 PM7/27/08
to

That's basically were I ended up too, but using David's File::Separator
suggestion.

However you made me think it would be helpful for Pathname to have:

class Pathname
def to_a
to_s.split(File::Separator) # better definition ?
end
end

The funny thing is that reminds me of a rewrite of Pathname I did a
while back that used an internal array instead of a string to store the
path. It was ~20% faster than the current lib. But alas, no one cared :(

ara.t.howard

unread,
Jul 27, 2008, 11:27:20 PM7/27/08
to

cfp:~ > cat a.rb
path = File.join 'a', 'b', 'c'

head, tail = path.split( File::SEPARATOR, 2 )

p :head => head, :tail => tail

cfp:~ > ruby a.rb
{:tail=>"b/c", :head=>"a"}


a @ http://codeforpeople.com/
--
we can deny everything, except that we have the possibility of being
better. simply reflect on that.
h.h. the 14th dalai lama


Bill Walton

unread,
Jul 27, 2008, 11:29:06 PM7/27/08
to
Hi Phlip,
Phlip wrote:

> That would be safer if fname were a Pathname.

Not sure what you mean here by 'safer'. Say more?

Best regards,
Bill

Martin DeMello

unread,
Jul 28, 2008, 1:48:26 AM7/28/08
to
On Sun, Jul 27, 2008 at 11:47 AM, Thomas Sawyer <tran...@gmail.com> wrote:
>> irb(main):002:0> components = fname.split('/')
>> => ["home", "foo", "bar"]
>> irb(main):003:0> [components.shift, components.join('/')]
>> => ["home", "foo/bar"]
>
> That's basically were I ended up too, but using David's File::Separator
> suggestion.

Don't forget the second argument to split - it'd be handy here.

first, rest = fname.split(File::Separator, 2)

though you'd probably have to thrown in an
if fname.index(File::Separator) == 0 then fname = fname[1..-1]
first

martin

Phlip

unread,
Jul 28, 2008, 7:34:38 AM7/28/08
to
Bill Walton wrote:

>> That would be safer if fname were a Pathname.
>
> Not sure what you mean here by 'safer'. Say more?

At work we just finished a rewrite of a system that generously manipulated
folders, paths, and files. The old system originally used only strings, and
string surgery, to manipulate paths. (The old system was also very shabby and
patched up; it started as a one-shot script with no structure, etc.)

In the new system we follow a simple rule: If it's a filename, relative path, or
absolute path of any kind, it's a Pathname. This allows us to stay within the
Pathname feature set, and manipulate paths without any string surgery. The
result is much more typesafe.

In theory, Pathnames would be safer if you needed to support \ path delimiters,
and if you needed to support paths with embedded \ or / characters. We are very
good string surgeons, so we never had those problems.

--
Phlip

David A. Black

unread,
Jul 28, 2008, 7:35:02 AM7/28/08
to
Hi --

On Mon, 28 Jul 2008, ara.t.howard wrote:

>
> On Jul 26, 2008, at 1:54 PM, Trans wrote:
>
>> I need to split a path by head/*tail.
>>
>> Ex.
>>
>> File.head_tail_split('home/foo/bar') #=> [ 'home', 'foo/bar' ]
>>
>> Sure, I can write a clumsy loop like the following:
>>
>> def File.head_tail_split(fname)
>> s = fname
>> t = []
>> h = nil
>> until s == '.'
>> t << h
>> s, h = *split(s)
>> end
>> return h, File.join(*t.compact)
>> end
>>
>> But I'm betting there's a better way. Or maybe there's already an easy
>> way I'm overlooking
>
>
>
> cfp:~ > cat a.rb
> path = File.join 'a', 'b', 'c'
>
> head, tail = path.split( File::SEPARATOR, 2 )
>
> p :head => head, :tail => tail
>
>
>
> cfp:~ > ruby a.rb
> {:tail=>"b/c", :head=>"a"}

Ironically, I tried that first and discarded it because I had
misunderstood Tom's question and thought he wanted dirname/basename,
and then didn't resurrect it when I finally understood :-) The only
tweak I might make would be to use Pathname#clean (or roll one's own)
in case there are consecutive separators.


David

--
Rails training from David A. Black and Ruby Power and Light:

ara.t.howard

unread,
Jul 28, 2008, 11:27:51 AM7/28/08
to

On Jul 28, 2008, at 5:39 AM, Phlip wrote:

> In the new system we follow a simple rule: If it's a filename,
> relative path, or absolute path of any kind, it's a Pathname. This
> allows us to stay within the Pathname feature set, and manipulate
> paths without any string surgery. The result is much more typesafe.
>
> In theory, Pathnames would be safer if you needed to support \ path
> delimiters, and if you needed to support paths with embedded \ or /
> characters. We are very good string surgeons, so we never had those
> problems.

i've had the same experience. this only issue is

cfp:~ > ruby -r pathname -e' Pathname.new("does-not-exist").realpath '
/opt/local/lib/ruby/1.8/pathname.rb:420:in `lstat': No such file or
directory - /Users/ahoward/does-not-exist (Errno::ENOENT)
from /opt/local/lib/ruby/1.8/pathname.rb:420:in `realpath_rec'
from /opt/local/lib/ruby/1.8/pathname.rb:453:in `realpath'
from -e:1


which is to say that certain pathname operations blow up when files do
not exist so code must work with that fact. also, pathnames cannot be
given to many functions that expect strings so you end up with a lot
of 'pathname.to_s' lines in the code. that said, using pathname will
result in better code every time because it deals with the common
errors people make slinging strings as path components.

cheers.

0 new messages