help my program interpret /? as an argument 3

[darkreaper00]

I've got myself a program that parses another program's output and formats it in a way that is useful to our lab. The biggest hangup I have with the program is trying to get it to print my usage message when /? is used for an argument. Below is the first way I tried to do that, before I realized that / was not a metacharacter that gets escaped with a backslash (which, incidentally, will behave the way i want it to if i feed it '\?', which is something I don't understand but is irrelevant at the moment):

while ($ARGV[0] ne "") {
                        # code for arguments it does 
                        # know how to handle
     }
     elsif ($ARGV[0] =~ /^\/\?/) {
	     die($usage);
     }

since then I've tried:

($ARGV[0] =~ /^'/'\?/)
($ARGV[0] =~ /(^'/'\?)/)
($ARGV[0] =~ /^/+\?/)
($ARGV[0] =~ /^$e/)

, with $e being the string '/?'

and many others... I understand why putting / between // could be problematic, and I think my next route would be to try unicode names, but I'm inclined to think there must be an easier way to do this. Any thoughts?

Thanks,
Tony

 

[dchoulette]

($ARGV[0] =~ m:^/\?:)

You can use any character instead of / to surround your regular expression. You just have to add the 'm' (for match) operator which is implicit if you use / as delimitor.

 

[MikeLacey]

Tony,

this bit of code

foreach $arg (@ARGV){
print "$arg "
print "gotcha" if $arg =~ /\//;
print "\n";
}

seems to work ok

you have to "escape" the \ within the search pattern so that Perl understands that you're not terminating the search pattern

you could also use different search pattern delimiters, like this:

print "gotcha" if $arg =~ m#/#;
Mike
________________________________________________________________

"Experience is the comb that Nature gives us, after we are bald."

Is that a haiku?
I never could get the hang
of writing those things.

 

[darkreaper00]

Thanks for the prompt replies, but there're still some sticking points... namely, the problem of the phantom slash that you both solved creates a problem we'll call the furtive question mark. In both cases -- using the match operator as dchoulette wrote it, and with MikeLacey's standalone 'bit of code' -- perl gets very upset with /? as an argument, though they're all equally happy with anything else I've tried following the slash. Both programs behave almost normally when I feed them any of the following arguments:
/ /abc /;; /// /& /!

The first four all produce the correct output, while /& precedes the correct output with:

     [1] 225137
     >

(the number following [1] changes each time), and /! gives the correct output followed by

[1]  + Exit 255                      perl process-hits.pl /

from my program (named process-hits) and

[1]  + Done                          perl test /

from MikeLacey's.

When I feed /? to either one, however, the error message is the same:

perl: No match.

Again, any thoughts? (and again, thanks in advance..)

 

[danielhozac]

Maybe Perl itself "steals" those arguments and only sends parts of them to your script? That's sure what it seems like. //Daniel

 

[darkreaper00]

more interesting complications:

in the case of dchoulette's suggestion, the output is not a result of a successful match with

($ARGV[0] =~ m:^/\?:)

-- I also have the program die($usage) when it gets an argument that's not expected. (Funnily enough, /? is the one argument that fails before it gets to that point in the loop). Anyway, the /[.]* inputs match on MikeLacey's code (naturally, since all we're searching for is a /) but not dchoulette's, but /? does produce an error message from both.

Moreover, the "\?" which I thought was the correct input for what I coded initially is not -- it too dies on the unexpected argument premise.

I was thinking that I might just as well be rid of the line which solicits /? and have the program treat /? just as any other unexpected argument and display $usage by default (even though I still wanted to know how to fix it). But perl is making sure I pursue the problem, because removing that line doesn't fix the problem -- even when I'm not looking for a /?, perl gives me:

perl: No match.

It seems, then, perl is unhappy taking this as an argument... How, then, do I work around this? Surely someone sometime wanted to provide /? argument support to display a helpfile! Is there perhaps a defined way to do this -- maybe a variable which, when defined will respond to /? as an argument?

so much for a simple question, eh?

 

[justice41]

The problem is not with Perl but with the shell that you are using. I assume that you are using csh or tcsh. The problem is that '?' is a special character to csh so you have to escape it on the command line like

process-hits.pl /\?

it then gets passed into @ARGV as '/?'. ('?' is a single-character glob similar to '.' in perl regex). The '&' is also a special character, which is why you get strange behavior with it as well.

Perhaps /h would be a better choice for a help flag.

jaa

 

[dchoulette]

I agree with justice41.
The characters &!|;*? are interpreted by the shell before perl is even launch.

For example, with '/&' you run your perl script with last argument equal to '/' and in a background job. The

[1] 225137

means that your first (

[1]

) background job has process id 225137.

The shell tries to replace /? by the names of all the files in the root directory (/) whose name has only one letter (?).

When you want to use any 'strange' character (?;*! even $ if you do not want to expand a variable) in a shell command argument put the whole argument in single quotes.

process-hits.pl '/?'

 

[MikeLacey]

yep, this is all shell stuff

try the same commands on a DOS machine Mike
________________________________________________________________

"Experience is the comb that Nature gives us, after we are bald."

Is that a haiku?
I never could get the hang
of writing those things.

 

[darkreaper00]

Thank y'all for the help.

Tony