捕获有关文件和找到的正则表达式的信息

Question

I have a directory full of files that contain numbers I want to capture. I also want to know which numbers come from which particular files. Right now I'm just running grep, which outputs something like:

grep ./* -e 'expression'
./file1: expression numberA
./file1: expression numberB
./file1: expression numberA
./file2: expression numberC numberD
...

What I want is to extract a piece of the filenames (in this example, 1 for file1), and also all of the numbers that appear after my expression.

While I'd prefer to just do everything in bash, any solution is welcome.

EDIT: To be clear, I want as output the following:

file1:
numberA
numberB
file2:
numberC
numberD
...

I've also edited the earlier portion (./file1: expression numberA). Sorry for not being clear before.

Answer 1

Try this:

grep -e 'expression' * | perl -pe 's/^(.*?)(\d+)(:.*)$/$1$2$3 $2/'

This should produce all input lines. If the filename portion of the input line ends with a number, that number should be appended to the input line.

$1 , $2 and $3 are backreferences to the 3 subexpressions (those parts of the regular expression in parentheses).

The commandline switch -e instructs the Perl interpreter to execute the given expression. -p loops on the input and prints $_ .

However, since you also want all numbers from after your match, you probably need something more complex:

grep -e 'expression' * | perl -ne '
  chomp;
  ($a,$b) = split ":";
  $a =~ s/.*(\d+)$/$1/;
  $b =~ s/.*expression(.*)/$1/;
  $b =~ s/[^\d]+/ /g;
  print "$_ $a $b\n";
'

-n does the same as -p , only without implicitly printing $_ .

Edit: After reading your updated requirements I think you may be better off with an all-Perl solution.

#!/usr/bin/env perl

use strict;
use warnings;

foreach (@ARGV) {
  my $file = $_;
  open FILE, "<$file" or die "Can't open file $file.";
  my $first = 1;
  foreach (<FILE>) {
    if (m/expression(.*)/) {
      my $values = $1;
      if ($first) {
        print "$file:\n";
        $first = 0;
      }
      $values =~ s/(^ +| +$)//g;
      $values =~ s/ +/\n/g;
      print "$values\n";
    }
  }
  close FILE;
}

Answer 2

If you just want to see the file number and numbers you could use something like:

find . -exec sh -c "echo -n \;{}::;grep -e 'expression' {}" \; | perl -pe 's/^.*(\d+)::/File \1:\n/' | perl -pe 's/\D*(\d+)$/\1/'

Note: this would break if your expression contains ::number (used as a delimiter, can be changed). Prints last file name if no matches are found.

Would produce:

File 2:
878
File 3:
199
File 4:
123
234
9
0

Example file2:

foo 123
bar 123
expression 878
lorem ipsum

If you just want the number pairs (file number + number) then you could try:

grep ./* -e 'expression' | perl -pe 's/^.*?(\d+):.*?(\d+)$/\1 \2/'

Output:

2 878
3 199
4 123
4 234
4 9
4 0

As mentioned in my comment, your questions is somewhat unclear on exactly what you want. Providing some examples would be beneficial.

Answer 3

Thanks to the two who answered - with the information that both of you gave me, I was able to figure out a perfect solution:

grep -i expression ./* | perl -pe 's/.*(\d+).*:.*(\d+)/$1 $2/' | sort | uniq

This gives as output:

1 numberA
1 numberB
2 numberC numberD

Answer 4

I think grep is not needed in this case. Awk or perl needed to use to accumulate data, so they can look for the expression in the file. Here is an awk example:

awk '/expression/ {f[FILENAME]; for(i=2;i<=NF;++i) v[FILENAME,$i]} 
END {for(i in f) {print i":"; for(j in v) if(sub("^"i SUBSEP,"",j))print j}}' ./*

Output

file1:
numberB
numberA
file2:
numberD
numberC