在文本文件行中提取部分重复的模式

Question

Given a text file of the form:

firstword<number1>,<string1>:<number2>,<string2>:<number3>,<string>:<number4>...
firstword<number1>,<string1>:<number2>,<string2>:<number3>,<string>:<number4>...
firstword<number1>,<string1>:<number2>,<string2>:<number3>,<string>:<number4>...
...

where each line can differ from each other, and can have any number of string:number pairs. "firstword" is always the same. The contents of the strings and numbers can change, eg numbers could be "12345", string could be "abc" (without the quotes).

In addition, a line can have multiple times the same string (how many times is unknown and different per line), each with a different associated number. For example:

firstword123,abc:123,cde:234,abc:345,def:456

If one now wants to only extract the first word and number (in this case firstword123), as well as all string:number pairs in a line for a specific string, how can one do this? In the above example, if one choses for the string the value "abc", then the extracted line should look like:

firstword123,abc:123,abc:345

I am looking for a solution which works with Bash (and possibly other commands).

Answer 1

you can use perl for this

#!/usr/bin/perl
my $first='firstword123';
my $str='abc';

while (<DATA>) {
    next if not /^$first/;
    print "$first";
    print ",$_" for ($_ =~ /$str:\d+/g);
}

__DATA__
firstword123,abc:123,cde:234,abc:345,def:456

out:

firstword123,abc:123,abc:345

Answer 2

Not a one-liner, but an all-bash solution. If you need faster code we can write something in awk or perl ...

$: cat keyscan
#! /bin/env bash

key="$1"
while read line
do start=${line//,*/}
   line=${line#$start}
   line=${line#,}
   while [[ -n "$line" ]]
   do case "$line" in
      $key:[0-9]*) lead="${line//,*/}"
                   start="$start,$lead"
                   line="${line#$lead}"
                   line="${line#,}"  ;;
              *,*) line="${line#*,}" ;;
                *) line='' ;;
      esac
   done
   printf "$start\n"
done

$: cat data
firstword123,abc:123,cde:234,abc:345,def:456

$: ./keyscan abc < data
firstword123,abc:123,abc:345

$: ./keyscan def < data
firstword123,def:456

$: ./keyscan cde < data
firstword123,cde:234

It will not be fast because it has a processing loop on every line of input, but it works on the sample line of data you gave.