Partial string match -- with a twist.

smw6181 has asked for the wisdom of the Perl Monks concerning the following question:

Hello and thanks for taking the time to look at my issue.

I have a script that I'm working on that handles setup of new servers, via user input. The user enters server information which is then stored in a hash. The hash keys all start with 'node'.

The issue is that the hash entries are stored as node0-9wwn0, etc.

Sample:

Key: node0  Value: servername0
Key: node0wwn0 Value: servername0wwn0
...
[download]

What I need is to pull JUST the node key (which is variable based on how many servers are entered...could be node0-node199).

If I regex match on node0-9 inside the loop it pulls the last hash entry (node0wwn0 in this case), and all I need is 'node0' and it's corresponding key. After several days of searching and getting nowhere with substr and regexes I decided to throw myself on the mercy of the Monks.

foreach my $key (sort keys %$ref) {
 if ($key ne "Server") {
 print $key .":" . " " . $ref->{$key} . "\n";
 } 
 

### Search the hash for cluster nodes.  
### If they exist, perform operations
### Regex looks for the node.* keys, which indicates a cluster node.
### If there's no cluster, there won't be any node.* keys.

 
if ($key =~ m/node.*$/) {
print "We have a cluster!  Here are the entries: " .$key. ":" . $ref->
+{$key}. "\n";
 } 
if ($key =~ m/node[0-9]/) {
 print $key . ":" . $ref->{$key}. "\n";
  }
 }
[download]

As you can see from above regex, that matches anything on that line with node0-9, when all I want to pull out of the hash is node0-9...my problem is that the key is variable, as I said, so I can't hard code it. Can someone point me in the right direction? Thank you much!

Comment on Partial string match -- with a twist. Select or Download Code

Replies are listed 'Best First'.
Re: Partial string match -- with a twist. by SuicideJunkie (Vicar) on Jul 20, 2012 at 13:28 UTC
See: Capture groups Basically, just slap some brackets around your `node[0-9]`, and that portion of the string will be saved into the `$1` variable for you to use later.	[reply] [d/l] [select]
Re: Partial string match -- with a twist. by Anonymous Monk on Jul 20, 2012 at 13:27 UTC
What you want to match is: `/(node[0-9])/` If you match on `.` that means "everything else" and that's exactly what you got.	[reply]
Re^2: Partial string match -- with a twist. by smw6181 (Initiate) on Jul 20, 2012 at 13:55 UTC
Revised code: `if ($key =~ m/(node[0-9]*)/) { if ($key = $1) { print "You entered " .$key. ":" . $ref->{$key}. "\n"; } }` [download] That did the trick. Amazing how something so simple can be so easily missed. Thank you both very much.	[reply] [d/l]
Re^3: Partial string match -- with a twist. by aitap (Curate) on Jul 20, 2012 at 14:16 UTC
This form of matching is more common: `if ($key =~ m/(node[0-9]*)/) { print "You entered " .$1. ":" . $ref->{$1}. "\n"; }` [download] Sorry if my advice was wrong.	[reply] [d/l]
Re^4: Partial string match -- with a twist. by smw6181 (Initiate) on Jul 20, 2012 at 14:52 UTC
Re^4: Partial string match -- with a twist. by smw6181 (Initiate) on Jul 20, 2012 at 15:03 UTC
Re: Partial string match -- with a twist. by AnomalousMonk (Archbishop) on Jul 20, 2012 at 15:13 UTC
What I need is to pull JUST the node key (which is variable based on how many servers are entered...could be node0-node199). [emphases added] From the quote from the OP and from the examples therein, I get the notion that the pattern is best described as "the literal 'node' at the start of a string, immediately followed by 1 to 3 (and no more) decimal digits". The regex `/(node[0-9])/` matches `'node'` anywhere,* followed by any number (including zero) of digits. IMHO, a better (because more specific) regex would be `m{ \A (node \d{1,3}) (?! \d) }xms` (which has not been tested).	[reply] [d/l] [select]
Re: Partial string match -- with a twist. by Anonymous Monk on Jul 20, 2012 at 18:24 UTC
A potentially better version would have been: `/^(node[0-9])/` ... notice underlined character ... ... which would force the string `node` to be considered only if it was anchored to the left-margin of the string, not anywhere in the string. It is also sometimes necessary to specify that a regular expression should not be "greedy." Normally, a regex will seek the longest* substring that matches the pattern ("greedy") when what you actually want is the shortest one.	[reply]