sorting help urgent

[viseshu]

I have an input file attached!!

I need the output to be sorted by the line number along with the record number as well.
(both line and record number are specified in the file).
for the lines where field number is not given, they shud be sorted along the line sorting.

[Peter Godron · Accepted solution]

Hi,
my solution would be delimiter replacement:
#!/usr/bin/sh
sed "s/- line/*/" data > data2
sed "s/,field/*/" data2 > data3
sort -t'*' -k 2,2n -k 3,3n data3 > data4
sed "s/*/- line/" data4 > data5
sed "s/*/,field/" data5 > data6
cat data6

If you input file is called data it substitues the "- line" with a "*" and repeats with ",field".
Then a sort is carried out to sort first on line, then on field.
Then the placeholders are replaced with their initial values.
You need to use a unique delimiter ! Base din your inputfile I used "*", you may need to use something else.

[James R. Ferguson]

Hi Viseshu:

Here's a simple solution:

# cat ./filter
#!/usr/bin/perl
use strict;
use warnings;
my @list;
my $line;
my $field;
while (<>) {
$line = m/line\s(\d+)/ ? $1 : 0;
$field = m/field\s(\d+)/ ? $1 : 0;
push @list, sprintf "%6d%6d%s", $line, $field, $_;
}
@list = sort @list;
for $line (@list) {
print substr( $line, 12 );
}
1;

...Run this passing your input file to be sorted:

# ./filter file

Regards!

...JRF...

[Sandman!]

Try the awk construct below...

awk -F"-" '{
gsub("[\",]"," ",$NF)
split($NF,z," ")
ml < z[2] ? ml=z[2] : ml
mr < z[4] ? mr=z[4] : mr
z[4] ? x[z[2]""z[4]]=$0 : x[z[2]""0]=$0
} END {
for (i=0;i for (j=0;j if (x[i""j])
print x[i""j]
}' file

[Hemmetter]

Hi viseshu,

What about:
$sort -k9,9n -k10,10n data

rgds
HGH

[Dennis Handly]

A possible problem with Hemmetter's solution is that there are two formats for the messages. If there are any fewer words in the message, it will fail. A possibly safer sort command is:
$ sort -t- -k2,2 283372.txt

Unfortunately it does an ascii sort, not numeric. But Peter has solved that problem.

In Peter's solution, you can combine the first two seds and the last two:
#!/usr/bin/sh
sed -e 's/- line/*/' -e 's/,field/*/' data |
sort -t'*' -k2,2n -k3,3n |
sed -e 's/*/- line/"' -e 's/*/,field/'

[James R. Ferguson]

Hi:

I would respectively ask, "Didn't you find the other solutions with merit? Did you try any of them?

...JRF...