Special Array method about Unix pathes

Une_BSSSSvS_ue · September 19, 2007, 6:03pm

here is the goal:

i’ve an Array of Unix pathes :
forbidden = ["/Volumes/U3 System",
“/Volumes/EMTEC KEY/emtec_dl”,
“/Volumes/EMTEC KEY/emtec_dl/Win98SE new driver.exe”,
“/Volumes/EMTEC KEY/emtec_dl/v233r001”
]

i want to forbid access to those pathes BUT since, for example,
“/Volumes/EMTEC KEY/emtec_dl/Win98SE new driver.exe” « belongs »
to the path (it’s below this path) “/Volumes/EMTEC KEY/emtec_dl”
it is unusefull to have it in the Array.

in the case above i’d like to return :

forbidden = ["/Volumes/U3 System",
“/Volumes/EMTEC KEY/emtec_dl”
]

what i’ve done :

class Array
def includePath?(path)
self.each { | x |
return true if path.startsWith?(x)
}
return false
end
def getContainers!
self.delete_if { | x | ( self - [ x ] ).includePath?( x ) }
end
end

class String
def startsWith?(dir)
self.index(dir) === 0
end
end

then forbidden.getContainers! gives me the expected result.

what do you think about this solution ?
could you propose a better one ?

also, because english isn’t my mother tongue, are my methods naming
correct ?

Une_BSSSSvS_ue · September 19, 2007, 6:32pm

Une Bévue wrote:

“/Volumes/EMTEC KEY/emtec_dl/Win98SE new driver.exe” « belongs »

end
what do you think about this solution ?
could you propose a better one ?

also, because english isn’t my mother tongue, are my methods naming
correct ?

Array doesn’t seem the proper class to have a includePath? method,
otherwise the names look reasonable. If the method is in Array, then a
reasonable name might be includesStart?

Une_BSSSSvS_ue · September 19, 2007, 6:38pm

Charles D Hixson wrote:

 self.each { | x |
   return true if path.startsWith?(x)
 }
 return false

self.any? {|x| path.starts_with?(x) }

HTH,
Sebastian

Une_BSSSSvS_ue · September 19, 2007, 7:06pm

Sebastian H. [email protected] wrote:

self.any? {|x| path.starts_with?(x) }

fine thanks, i didn’t know the #any?

Une_BSSSSvS_ue · September 19, 2007, 7:07pm

Charles D Hixson [email protected] wrote:

Array doesn’t seem the proper class to have a includePath? method,
otherwise the names look reasonable.

i’ll lokk if there is something more related to Unix pathes…

If the method is in Array, then a
reasonable name might be includesStart?

OK, thanks!

Une_BSSSSvS_ue · September 19, 2007, 8:26pm

Alle mercoledì 19 settembre 2007, Une Bévue ha scritto:

“/Volumes/EMTEC KEY/emtec_dl/Win98SE new driver.exe” « belongs »

end
what do you think about this solution ?
could you propose a better one ?

also, because english isn’t my mother tongue, are my methods naming
correct ?

This is my alternative solution:

def select_toplevel_paths arg
a = arg.dup.sort_by{|i| i.size}
a.inject([]) do |res, i|
res.any?{|j| i.index(j) == 0 } ? res : (res << i)
end
end

It doesn’t change the contents of the array and it’s not a method of
class
Array, because, in my opinion, it’s a too specialized method to be put
in a
core class.

Regarding your code, I’d say names are correct, as far as English is
concerned. I’d like to give you a copule of suggestions:

use underscores instead of camel-case (include_path? instead of
includePath,
for example), because it’s the more common style in the ruby community
use start_with? instead of starts_with?. In the ruby standard library,
there
are several cases in which both methods exist, with the ones ending in s
being
obsolete (for example, File.exists? and File.exist?) For consistency, I
think
it’s better to follow the same convention.

Stefano

Une_BSSSSvS_ue · September 19, 2007, 8:44pm

Hi,

At Thu, 20 Sep 2007 03:24:58 +0900,
Stefano C. wrote in [ruby-talk:269890]:

def select_toplevel_paths arg
a = arg.dup.sort_by{|i| i.size}
a.inject([]) do |res, i|
res.any?{|j| i.index(j) == 0 } ? res : (res << i)

i.rindex(j, 0) is faster than i.index(j) == 0 when j matches in
the middle.

use start_with? instead of starts_with?. In the ruby standard library, there
are several cases in which both methods exist, with the ones ending in s being
obsolete (for example, File.exists? and File.exist?) For consistency, I think
it’s better to follow the same convention.

And ruby 1.9 has String#start_with? already in fact.

Une_BSSSSvS_ue · September 19, 2007, 9:31pm

Stefano C. [email protected] wrote:

This is my alternative solution:

def select_toplevel_paths arg
a = arg.dup.sort_by{|i| i.size}
a.inject([]) do |res, i|
res.any?{|j| i.index(j) == 0 } ? res : (res << i)
end
end

clever solution, thanks a lot !

however with the following (part of a yaml file) :

forbiddenList:

/Volumes/EMTEC KEY/emtec_dl
/Volumes/EMTEC KEY/emtec_dl/v233r001/Setup.exe
/Volumes/U3 System
/Volumes/U3 System/Launchpad.zip

your select_toplevel_paths don’t remove /Volumes/EMTEC
KEY/emtec_dl/v233r001/Setup.exe and /Volumes/U3 System/Launchpad.zip

It doesn’t change the contents of the array and it’s not a method of class
Array, because, in my opinion, it’s a too specialized method to be put in a
core class.

U’re right, i was embarassed with that point.

Regarding your code, I’d say names are correct, as far as English is
concerned. I’d like to give you a copule of suggestions:

use underscores instead of camel-case (include_path? instead of includePath,
for example), because it’s the more common style in the ruby community

use start_with? instead of starts_with?. In the ruby standard library, there
are several cases in which both methods exist, with the ones ending in s being
obsolete (for example, File.exists? and File.exist?) For consistency, I think
it’s better to follow the same convention.

OK, done !

Une_BSSSSvS_ue · September 19, 2007, 8:46pm

2007/9/19, Sebastian H. [email protected]:

Charles D Hixson wrote:

self.each { | x |
return true if path.startsWith?(x)
}
return false

self.any? {|x| path.starts_with?(x) }

You can even leave the “self.” out.

I can think of several approaches to solve this:

order the array and do a second iteration where each item is
checked against the last seen prefix; the idea is that every string
comes before the strings that contain it as prefix.
insert names into a tree and output only prefixes during a tree walk.

Bévue, the only issue I see with your method naming is that in Ruby
methods typically use under_scores and not CamelCase as in Java.

Kind regards

robert

Une_BSSSSvS_ue · September 19, 2007, 9:33pm

Robert K. [email protected] wrote:

You can even leave the “self.” out.

OK !

I can think of several approaches to solve this:

order the array and do a second iteration where each item is
checked against the last seen prefix; the idea is that every string
comes before the strings that contain it as prefix.

fine, the Array is allready ordered elsewhere, i thought putting sort!
in this method ))

insert names into a tree and output only prefixes during a tree walk.

ok, i see what you mean, may be better not to do that when output rather
than when building the three having a means to compare three deepness.

Bévue, the only issue I see with your method naming is that in Ruby
methods typically use under_scores and not CamelCase as in Java.

Yes i’ve corrected that, it’s an habit coming from java, as the getter
“getContainers!”…

Une_BSSSSvS_ue · September 19, 2007, 9:44pm

Alle mercoledì 19 settembre 2007, Une Bévue ha scritto:

forbiddenList:

/Volumes/EMTEC KEY/emtec_dl

/Volumes/EMTEC KEY/emtec_dl/v233r001/Setup.exe

/Volumes/U3 System

/Volumes/U3 System/Launchpad.zip

your select_toplevel_paths don’t remove /Volumes/EMTEC
KEY/emtec_dl/v233r001/Setup.exe and /Volumes/U3 System/Launchpad.zip

It works for me:

irb: 001> def select_toplevel_path arg
irb: 002+> a = arg.dup.sort_by{|i| i.size}
irb: 003+> a.inject([]) do |res, i|
irb: 004*> res.any?{|j| i[0, j.size] == j} ? res : (res << i)
irb: 005+> end
irb: 006+> end
nil
irb: 007> forbidden = ["/Volumes/EMTEC KEY/emtec_dl",
irb: 008*> “/Volumes/EMTEC KEY/emtec_dl/v233r001/Setup.exe”,
irb: 009*> “/Volumes/U3 System”,
irb: 010*> “/Volumes/U3 System/Launchpad.zip”
irb: 011+> ]
["/Volumes/EMTEC KEY/emtec_dl", “/Volumes/EMTEC
KEY/emtec_dl/v233r001/Setup.exe”, “/Volumes/U3 System”, “/Volumes/U3
System/Launchpad.zip”]
irb: 012> select_toplevel_path forbidden
["/Volumes/U3 System", “/Volumes/EMTEC KEY/emtec_dl”]

Notice that this doesn’t change the forbidden array, but creates a new
one.

Stefano

Une_BSSSSvS_ue · September 19, 2007, 10:11pm

Une Bévue [email protected] wrote:

KEY/emtec_dl/v233r001/Setup.exe and /Volumes/U3 System/Launchpad.zip
sorry, in fact that works wheb used as :

forbidden = [
“/Volumes/U3 System”,
“/Volumes/EMTEC KEY/emtec_dl”,
“/Volumes/EMTEC KEY/emtec_dl/Win98SE new driver.exe”,
“/Volumes/EMTEC KEY/emtec_dl/v233r001/Setup.exe”,
“/Volumes/EMTEC KEY/emtec_dl/v233r001”,
“/Volumes/U3 System/LaunchU3.exe”,
“/Volumes/U3 System/Launchpad.zip”
]

p forbidden
forbidden = select_toplevel_paths( forbidden )
p forbidden

but not when used in a class :

class SyncDataSource < OSX::NSObject

def initialize

[…]

    @forbiddenList = [ ]

[…]

end

[…]

def addForbidden( notification )

[…]

    p @forbiddenList
    @forbiddenList = select_toplevel_paths( @forbiddenList )
    p @forbiddenList

[…]

end

end

i don’t understand why for the time being…

Une_BSSSSvS_ue · September 19, 2007, 10:55pm

On 19.09.2007 21:26, Une Bévue wrote:

Robert K. [email protected] wrote:

I can think of several approaches to solve this:

order the array and do a second iteration where each item is
checked against the last seen prefix; the idea is that every string
comes before the strings that contain it as prefix.

fine, the Array is allready ordered elsewhere, i thought putting sort!
in this method ))

If it’s sorted anyway that’s probably the approach I’d take.

insert names into a tree and output only prefixes during a tree walk.

ok, i see what you mean, may be better not to do that when output rather
than when building the three having a means to compare three deepness.

I am not sure what you mean here. You have to build the tree first
before you can decide which branches you can omit. With “output” I did
not necessarily mean output to screen, but rather output from the
method. Maybe I should have chosen a different wording. I am sorry if
that caused confusion.

Bévue, the only issue I see with your method naming is that in Ruby
methods typically use under_scores and not CamelCase as in Java.

Yes i’ve corrected that, it’s an habit coming from java, as the getter
“getContainers!”…

Different languages, different conventions.

Kind regards

robert

Une_BSSSSvS_ue · September 20, 2007, 5:02pm

 self.delete_if { | x | ( self - [ x ] ).includePath?( x ) }

end
end

In contrast to everyone else’s points about your naming, I’m not happy
with your
implementation.

/home/gareth/files/my_file
is not a child of
/home/gareth/fil

Gareth

Une_BSSSSvS_ue · September 20, 2007, 7:46am

Robert K. [email protected] wrote:

ok, i see what you mean, may be better not to do that when output rather
than when building the three having a means to compare three deepness.

I am not sure what you mean here. You have to build the tree first
before you can decide which branches you can omit. With “output” I did
not necessarily mean output to screen, but rather output from the
method. Maybe I should have chosen a different wording. I am sorry if
that caused confusion.

i’ve found a simpler approach, in fact, this list of pathes is entered
by the user using a File/folder chooser window, then instead of walking
a three or something else, xhen the user enter a new path right now i
check only if this path is allready taken into account by :

files.each { |f|
f_to_s = f.to_s
@forbiddenList << f_to_s unless @forbiddenList.include_path?( f_to_s )
}

with :
files the File(s)/Folder(s) entered by the user ;
f.to_s converting NSString to Ruby String (NS stands for NextStep) ;
Array#include_path?( a_path ) being :
class Array
def include_path?(path)
self.any? { | x | a_path.rindex( x, 0 ) }
end
end

the only thing i dislike is in that pathes have nothing to do with
Array,i must rexrite this part.

Bévue, the only issue I see with your method naming is that in Ruby
methods typically use under_scores and not CamelCase as in Java.

Yes i’ve corrected that, it’s an habit coming from java, as the getter
“getContainers!”…

Different languages, different conventions.

yes right, here i’m embarassed because my code isn’t “pure” Ruby, it’s
RubyCocoa (a bridge between Cocoa/ObjC on Mac OS X) and in RubyCocoa the
Cocoa methods like :

NSOpenPanel#beginSheetForDirectory:file:types:modalForWindow:modalDelega
te:didEndSelector:contextInfo:

are translated, on the Ruby side, in :

beginSheetForDirectory_file_types_modalForWindow_modalDelegate_didEndSel
ector:contextInfo( … )

then “_” replaces “:” and still camelized…