Although i tried multiple option i couldn't find a way to get the rigt ouput.
Say i have the following data
cat file.txt
[1] C request
[1:1]
[1:4]
[1:2]
[1:3]
[1] C response
[2:3]
[2] C request
[2:1]
[2:2]
[2] C response
The output should look like
[1] C request
[1:1]
[1:2]
[1:3]
[1:4]
[1] C response
[2] C request
[2:1]
[2:2]
[2:3]
[2] C response
I've tried
sort -t: -k1,1n -k2,2n or sort -t: +1.1 -1.3
but no luck.
any idea?
Cheers.
---------- Post updated 06-06-12 at 09:09 AM ---------- Previous update was 06-05-12 at 06:37 PM ----------
Sorry - i'll remember and use it next time.
Do you have any idea on my request? Maybe i need to use awk + sort?
Thx in advance!
Hi
$ sort -n file.txt | sed -n '/request/{p;s/.*//;x;p;d;}; /response/{p;n;h;T}; /request/!{1h;1!H;}'
[1] C request
[1:1]
[1:2]
[1:3]
[1:4]
[1] C response
[2] C request
[2:1]
[2:2]
[2:3]
[2] C response
Guru.
Thanks a lot!! If you don't mind, can you please explain the sed command?
Many thanks!!
# awk '/req||!~res/{while($0!~/res/){a[x++]=$0;getline}};{a[l]=$0;for(cc in a)if(a[cc]~/req/){print a[cc];a[cc]="";};
for(i=0;i<x;i++){split(a,s,":");sub("]","",s[2]);ll=s[1];f=0;s[2]=s[2];if(s[2]==1&&!f){print a;f=1;xx=s[2]+1;}
if(s[2]==xx&&a){print s[1]":"s[2]"]";xx++;f=1;}if(a&&f!=1)n[xxx++]=s[2]}for(lll=1;lll<20;lll++){
for(i=0;i<xxx;i++)if(n==(ll+lll))print ll":"n"]"}print a[l];;;x=xxx=0;xx=""}' unsorted_file
[1] C request
[1:1]
[1:2]
[1:3]
[1:4]
[1] C response
[2] C request
[2:1]
[2:2]
[2:3]
[2] C response
regards
Hi all,
Can anyone help me understand the above SED command?
Thanks!!
drl
June 10, 2012, 7:54am
6
Hi.
Here is an alternate solution. The utility msort has an interesting sort-field key called a hybrid. Essentially, it breaks apart a key-field into sub-fields. It then sorts the sub-fields as character unless the field is numeric, in which case it sorts as a numeric. The header and trailer lines in this data file have no second field, so a pair of perl scripts adds the field to request and response lines, and removes the second field after the hybrid sort. So for the perl add action, the lines would become:
[1:0] C request
[1:32000] C response
ready for input into msort.
#!/usr/bin/env bash
# @(#) s2 Demonstrate hybrid sort, msort.
# http://freecode.com/projects/msort
pe() { for _i;do printf "%s" "$_i";done; printf "\n"; }
pl() { pe;pe "-----" ;pe "$*"; }
db() { ( printf " db, ";for _i;do printf "%s" "$_i";done;printf "\n" ) >&2 ; }
db() { : ; }
C=$HOME/bin/context && [ -f $C ] && $C msort perl
FILE=${1-data1}
pl " Input data file $FILE, and expected output:"
head -n 11 $FILE expected-output.txt
pl " Helper perl scripts:"
head p*
pl " Results:"
./p1 $FILE |
tee f1 |
msort -q -l -n 1,1 -c hybrid |
tee f2 |
./p2
exit 0
producing:
% ./s2
Environment: LC_ALL = C, LANG = C
(Versions displayed with local utility "version")
OS, ker|rel, machine: Linux, 2.6.26-2-amd64, x86_64
Distribution : Debian GNU/Linux 5.0.8 (lenny)
bash GNU bash 3.2.39
msort 8.44
perl 5.10.0
-----
Input data file data1, and expected output:
==> data1 <==
[1] C request
[1:1]
[1:4]
[1:2]
[1:3]
[1] C response
[2:3]
[2] C request
[2:1]
[2:2]
[2] C response
==> expected-output.txt <==
[1] C request
[1:1]
[1:2]
[1:3]
[1:4]
[1] C response
[2] C request
[2:1]
[2:2]
[2:3]
[2] C response
-----
Helper perl scripts:
==> p1 <==
#!/usr/bin/env perl
# @(#) p1 Augment header and trailer lines with integers.
while ( <> ) {
s/\[(\d+)\]/[\1:0]/ if /request/;
s/\[(\d+)\]/[\1:32000]/ if /response/;
print;
}
==> p2 <==
#!/usr/bin/env perl
# @(#) p2 Remove integers from head and trailer lines.
while ( <> ) {
s/:\d+// if /request|response/;
print;
}
-----
Results:
[1] C request
[1:1]
[1:2]
[1:3]
[1:4]
[1] C response
[2] C request
[2:1]
[2:2]
[2:3]
[2] C response
Files f1 and f2 show the intermediate form of the data and can be discarded for production work.
See the link in the script for msort if it is not in an accessible repository. Fedora, Debian, FreeBSD have it, for example.
Best wishes ... cheers, drl
With the sample data, that sort command is not performing a numerical sort. Since the first character of the line is not the beginning of a numeric string, all lines compare equal, as if they all had a leading zero, and then they are compared lexicographically to break the tie. [10:1] will incorrectly precede [2:1] .
The T test will always succeed -- since the n command will always reset the tested condition -- so it could be replaced with d (since -n is in effect), in which case the sed script would be portable (instead of GNU specific).
The following is a reimplementation of drl's suggestion. Instead of using perl and msort to decorate-sort-dedecorate, it uses sed and sort:
sed '/req/s/]/:0]/; /res/s/]/:9999]/; s/[][]//g' infile |
sort -nt: -k1,1 -k2,2 |
sed '/req/s/:0//; /res/s/:9999//; s/[:[:digit:]]*/[&]/'
Regards,
1 Like
sed '/req/s/]/:0]/; /res/s/]/:9999]/; s/[][]//g' infile |
sort -nt: -k1,1 -k2,2 |
sed '/req/s/:0//; /res/s/:9999//; s/[:[:digit:]]*/[&]/'
Hi Alister,
If you don't mind can you please explain your sed command and also guruprasadpr's example
sed -n '/request/{p;s/.*//;x;p;d;}; /response/{p;n;h;T}; /request/!{1h;1!H;}'
In this way i'll better understand/ learn each sed option.
Thanks a lot!
Guys,
am i asking too much for a bit of explanation of the options below?
I'd very much appreciate if anyone is willing to explain the options so i can better understand the whole line/ learn it as well.
Best rgds
I gave up waiting for someone to help me out understand the above sed logic/ options.
Thx anyway.