I don't know how well it would scale, but I'd build a hash from the first file ($objects{'object1'}=23.12;), then scan the second file. If the object in the second file has a defined hash, then print out the hashkey and both values..

Untested code... and I'm assuming a space delim. as per your examples..

open(FILE, "file1");
while(<FILE>){
  if(/(\S+)\s+(\S+)/){
    $hash{$1} = $2;
  }
}
close FILE;
open(FILE, "file2");
while(<FILE>){
  if(/(\S+)\s+(\S+)/){
    print "$1: $hash{$1} $2\n" if defined $hash{$1};
  }
}
[download]

Yup, it's really just that simple (well, maybe exists rather than defined; but that's a minor nit). If your files are really, really big you probably want to use something like Berkeley_DB or one of the other DBM modules rather than reading everything into memory, but that's just an implementation detail; the basic algorithm remains the same.

or, just slightly different:

open(FILE1, "file1");
open(FILE2, "file2");
while(<FILE1>){
    /(\S*)\s+(\S*)/;
    $hash{$1}=$2;
}
while(<FILE2>){
    /(\S*)\s+(\S*)/;
    print "$1 $hash{$1} $2\n" if $hash{$1}
}
[download]



~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"one who asks a question is a fool for five minutes; one who does not ask a question remains a fool forever."

mk at perl dot org dot br

This is even less memory efficient, but I couldn't resist turning your problem into a golfed one-liner. I'm sure someone else will squeeze a few extra characters out of it:

perl -ane "push@{$h{$F[0]}},$_;END{while(($k,$v)=each%h){print@{$v}if@
+{$v}>1}}" dat1.txt dat2.txt
[download]

-a = autosplit into @F. -n means wrap the -e code in a while(<>){.....} loop. So as this one-liner iterates over the two (or more) files, it pushes each line into an anonymous array held in a hash where the keys are the "objectN" (the first element of @F).

After the first implicit while. loop (-n) finishes, the END{} block is executed. Here we test each hash element to see if its anonymous array holds more than one element. If it does, print the array. We're taking advantage of the fact that each array element still contains the \n newline from the original file's line endings, and that's why "print @ARRAY." results in one element per print-out line.

I hope my description of this solution helps, but you can also brush up on perlrun for more details. There are a couple of caveats with this one-liner. First, both files are slurped into a hash in their entirety. Second, the output is in no particular order.

based on my original code! (:
perl -ne '/(\S*)\s+(\S*)/;(!$h{$1})?$h{$1}=$2:print "$1 $h{$1} $2\n";' file1 file2


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"one who asks a question is a fool for five minutes; one who does not ask a question remains a fool forever."

mk at perl dot org dot br

That one formats nicely. It does have a quirk with regards to always printing the "value" of the first find next to the current find for each of multiple repeats. That's a mouthful, let me demonstrate with a contrived data set:

file1.....

test1 abc
test2 def
test3 ghi
test4 jkl

file2.....
test1 ghi
test3 jkl
test3 mno
[download]

And the output.....

test1 abc ghi
test3 ghi jkl
test3 ghi mno
[download]

As you can see, test3's "ghi" (the first sequence found) gets repeated for each 'test3' found. Not that there's anything wrong with that. ;)

If you use the -a switch, you will shave off a few more keystrokes from your solution though, and that's got to be worth something!

perl -ane '($a,$b)=@F;!$h{$a}?$h{$a}=$b:print"$a $h{$a} $b\n"' file1 f
+ile2
[download]

I do like your solution since it preserves order and formats nicely. Good job.




Dave

Hi,

Updated: removed code cause the first code example may fit better, but replacing defined in favor of exists

If using some kind of db file or solution, I would use first the DBM one, because the BerkeleyDB, will need to have installed the libraries, instead DBM comes with standard perl.

Regards,