用不均匀的空格对空格分隔的列表进行排序

Question

I have a space delimited list that has an uneven amount of spaces in what would be the first column. I want to reverse sort this by the first number that appears after its string. I need to do this using bash commands.

Example:

Pontiac Firebird 19.0 6 250.0 100.0 3282. 15.0 71 US
Pontiac J2000 SE Hatchback 31.0 4 112.0 85.00 2575. 16.2 82 US
Oldsmobile Delta 88 Royale 12.0 8 350.0 160.0 4456. 13.5 72 US
Oldsmobile Omega 11.0 8 350.0 180.0 3664. 11.0 73 US
AMC Gremlin 20.0 6 232.0 100.0 2914. 16.0 75 US
AMC Gremlin 21.0 6 199.0 90.00 2648. 15.0 70 US
Pontiac Lemans V6 21.5 6 231.0 115.0 3245. 15.4 79 US

Would turn into:

Oldsmobile Omega 11.0 8 350.0 180.0 3664. 11.0 73 US
Oldsmobile Delta 88 Royale 12.0 8 350.0 160.0 4456. 13.5 72 US
Pontiac Firebird 19.0 6 250.0 100.0 3282. 15.0 71 US
AMC Gremlin 20.0 6 232.0 100.0 2914. 16.0 75 US
AMC Gremlin 21.0 6 199.0 90.00 2648. 15.0 70 US
Pontiac Lemans V6 21.5 6 231.0 115.0 3245. 15.4 79 US
Pontiac J2000 SE Hatchback 31.0 4 112.0 85.00 2575. 16.2 82 US

I've tried doing sort -nr to see what happens and it reverse sorts the list, but respective to it's alphabetized order. I want to sort based on all values.

The trick is that I must keep it space delimited. What's the best way to do this using bash?

Answer 1

I must keep it space delimited

You mean, the result has to be space delimited again, right? During processing, you can transform the input however you like.

Assuming you know a character that never appears in your file otherwise, delimit the value you want to sort with by that character using sed , then sort by that value, then remove the additional delimiters again.
Here we use a tab to delimit the key for sorting.

sed -E 's/ ([0-9]+\.[0-9]+) / \t\1\t /' | sort -t $'\t' -k2,2n | tr -d \\t

This is basically a Schwartzian transform .

Answer 2

here's a short ruby program:

ruby -e '
    puts IO.readlines(ARGV.shift, chomp: true)
        .map {|line|
            fields = line.split
            [fields[0..(fields.size - 9)].join(" ")] + fields[-8 .. -1]
        }
        .sort_by {|row| row[1]}
        .map {|row| row.join(" ")}
        .join("\n")
' file

Answer 3

I would use GNU AWK for this as follows, let file.txt content be

Pontiac Firebird 19.0 6 250.0 100.0 3282. 15.0 71 US
Pontiac J2000 SE Hatchback 31.0 4 112.0 85.00 2575. 16.2 82 US
Oldsmobile Delta 88 Royale 12.0 8 350.0 160.0 4456. 13.5 72 US
Oldsmobile Omega 11.0 8 350.0 180.0 3664. 11.0 73 US
AMC Gremlin 20.0 6 232.0 100.0 2914. 16.0 75 US
AMC Gremlin 21.0 6 199.0 90.00 2648. 15.0 70 US
Pontiac Lemans V6 21.5 6 231.0 115.0 3245. 15.4 79 US

then

awk 'BEGIN{FPAT="[0-9]*[.][0-9]*";PROCINFO["sorted_in"]="@ind_num_asc"}{arr[$1]=$0}END{for(i in arr){print arr[i]}}' file.txt

output

Oldsmobile Omega 11.0 8 350.0 180.0 3664. 11.0 73 US
Oldsmobile Delta 88 Royale 12.0 8 350.0 160.0 4456. 13.5 72 US
Pontiac Firebird 19.0 6 250.0 100.0 3282. 15.0 71 US
AMC Gremlin 20.0 6 232.0 100.0 2914. 16.0 75 US
AMC Gremlin 21.0 6 199.0 90.00 2648. 15.0 70 US
Pontiac Lemans V6 21.5 6 231.0 115.0 3245. 15.4 79 US
Pontiac J2000 SE Hatchback 31.0 4 112.0 85.00 2575. 16.2 82 US

Explanation: I inform GNU AWK that field is 0 or more digits followed by literal dot ( [.] ) followed by 0 or more digits (note: I assume that there will always be dot in first number and never dot in column with name) and that array traversal should be treat-indices-as-numbers-ascending which is one of Predefined Array Scanning Orders . For each line I add to array pair with key being first number ( $1 ) and value being whole line ( $0 ). After going through all lines I print values from array arr with order which observe selected array traversal.

(tested in gawk 4.2.1)