Asp Forum - Escaped characters from the command-line

-j b-

5/5/2008 10:46:00 PM

I've run into a bit of an issue that I hope you can help with.
I am working on a parser that allows the user to provide a delimiter as
an argument to the script. However, I can not find a way to [elegantly]
handle escaped characters (such as \t and \n).
A simplified example to illustrate:

# Using
s = ARGV.shift
puts ['west', 'side', story'].join(s)

$ ./parse.rb '\t'
=> west\tside\tstory

Now, I can do something like:

s = ARGV.shift
sep = s =~ /^\\(.)/
when 't'
"\t"
when 'n'
"\n"
# ...
else
s
end
end

$ ./parse.rb '\t'
=> west side story

But I feel that is probably too brittle. I tried Shellwords, but that
didnt seem to help (although I could have been using it incorrectly).
Any suggestions are welcome.
Thanks.
--
Posted via http://www.ruby-....

4 Answers

Albert Schlef

5/6/2008 3:17:00 PM

-j b- wrote:
> I am working on a parser that allows the user to provide a delimiter as
> an argument to the script. However, I can not find a way to [elegantly]
> handle escaped characters (such as \t and \n).
> A simplified example to illustrate:
>
> # Using
> s = ARGV.shift
> puts ['west', 'side', story'].join(s)
>
> $ ./parse.rb '\t'
> => west\tside\tstory

It isn't a Ruby issue. It's a shell one. You need to find out how it's
possible, in your shell, to specify those special characters. Are you
using Linux? Bash?

$ echo one\ntwo
onentwo

$echo 'one\ntwo'
one\ntwo

$echo $'one\ntwo'
one
two

$echo 'one
two'
onw
two

$echo "one
two"
onw
two

So there are at least three ways to do \n in bash.
--
Posted via http://www.ruby-....

-j b-

5/6/2008 4:27:00 PM

Albert Schlef wrote:
> -j b- wrote:
>> I am working on a parser that allows the user to provide a delimiter as
>> an argument to the script. However, I can not find a way to [elegantly]
>> handle escaped characters (such as \t and \n).
>> A simplified example to illustrate:
>>
>> # Using
>> s = ARGV.shift
>> puts ['west', 'side', story'].join(s)
>>
>> $ ./parse.rb '\t'
>> => west\tside\tstory
>
> It isn't a Ruby issue. It's a shell one. You need to find out how it's
> possible, in your shell, to specify those special characters. Are you
> using Linux? Bash?

Yes. Yes. But the idea is that it shouldnt matter.

>
> So there are at least three ways to do \n in bash.

I know that it possible, from a user standpoint, to input a special
character in multiple ways. My concern is that I dont want to force that
restriction on the user. For example, the utility tr will accept '\t' as
a single character; I find this ultimately more useful than tr '^v[tab]'
' ' (or any of the above mentioned methods).
If it is a shell issue then I am forced to modify the arguments
accordingly. However, I'm still looking for a more elegant way then the
heavy case statement.
I've experimented with a gsub! solution but the way backslashes are
escaped in regular expressions (particularly when reinserting into a
string) has me blocked at the moment.

Thanks.

j
--
Posted via http://www.ruby-....

Albert Schlef

5/6/2008 7:54:00 PM

> I'm still looking for a more elegant way then the
> heavy case statement.
> I've experimented with a gsub! solution but the way backslashes are
> escaped in regular expressions (particularly when reinserting into a
> string) has me blocked at the moment.

All you're looking for is a cool one-liner? Here's one:

s.gsub!(/(\\[ntr])/) { eval %/"#$1"/ }

It replaces any \n and \t sequence by it's character. You can even
enhance the regexp to support \x<number> sequences as well:

s.gsub!(/(\\([ntr]|x\d+))/) { eval %/"#$1"/ }
--
Posted via http://www.ruby-....

Phillip Gawlowski

5/6/2008 8:04:00 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

- -j b- wrote:

|> It isn't a Ruby issue. It's a shell one. You need to find out how it's
|> possible, in your shell, to specify those special characters. Are you
|> using Linux? Bash?
|
| Yes. Yes. But the idea is that it shouldnt matter.

Have fun rewriting the *NIX shells, cmd.exe, and, possibly, PowerShell, too.

- --
Phillip Gawlowski
Twitter: twitter.com/cynicalryan
Blog: http://justarubyist.bl...

Use statement labels that mean something.
~ - The Elements of Programming Style (Kernighan & Plaugher)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail....

iEYEARECAAYFAkgguakACgkQbtAgaoJTgL/PPwCgplzxuF3E608P8jtSCWukpA4o
zlgAn1oQVkLOnSLkmMtwrhMWCGuNro0V
=iHkS
-----END PGP SIGNATURE-----

comp.lang.ruby

Escaped characters from the command-line

-j b-

Albert Schlef

-j b-

Albert Schlef

Phillip Gawlowski

x Login to ForumsZone