Emulate group-by in shell script


 
Thread Tools Search this Thread
Top Forums Shell Programming and Scripting Emulate group-by in shell script
# 8  
Old 03-23-2017
Quote:
Originally Posted by mukulverma2408
Thanks Every-one, that is really helpful Smilie

---------- Post updated at 05:55 PM ---------- Previous update was at 05:27 PM ----------



Hi Scrutinizer, thanks for the solution, the problem is I am not very well versed with awk, would be really helpful if you can explain what's happening here.
Sure:

Code:
awk '                     # There are two associative arrays, which use the first field as the index L(ow) and H(high)
  NR==1 {                  
    next                  # Skip header
  }

  !($1 in L) {            # For every line, if no array element exist for the first field in array L,
    L[$1]=$2              # then create one with the second field as value. This becomes the Lowest number
  }

  $2<L[$1] {              # If the second field is lower than the array element in L for the first field, 
    L[$1]=$2              # then replace it with the second field as the new lowest number
  } 

  $2>=H[$1] {             # If the second field is higher or equal to the array element in H for the first field,
    H[$1]=$2              # Then replace it with the second field as the new highest number
  }

  END {                   # After the input file has been read
    for(i in L)           # For all elements in array L (which are the unique field values)
      print i, L[i], H[i] # Print the element name (the "operation" ) followed by the low and high values
  }
' FS=, OFS=,  file        # Specify a comma as both input and output field separators and specify the file name

Note that the order is not guaranteed, so you might want to pipe the output through sort

Last edited by Scrutinizer; 03-24-2017 at 02:16 AM.. Reason: Added skip header
# 9  
Old 03-23-2017
Code:
$
$ cat data.txt
Opeation,Time-Taken
operation1,83621
operation2,72321
operation3,13288
operation2,12312
operation1,12321
operation2,45455
operation2,42543
operation1,87934
operation4,94865
operation5,27383
operation6,322
operation6,93483
operation7,3223
$
$ cat -n group_data.sh
     1  #!/usr/bin/bash
     2  header="y"
     3  while read line
     4  do
     5      if [ ! -z "$header" ] ; then
     6          header=
     7          continue
     8      fi
     9      opr=${line%%,*}
    10      if [ -z "$prev_opr" ] ; then
    11          # set min, max if first data line
    12          min=${line##*,}
    13          max=$min
    14      elif [ "$opr" != "$prev_opr" ] ; then
    15          # if opr changes, print previous line and set min, max
    16          echo "$prev_opr,$min,$max"
    17          min=${line##*,}
    18          max=$min
    19      elif [ "${line##*,}" -gt "$min" ] ; then
    20          # if opr is same, set max if needed
    21          max=${line##*,}
    22      fi
    23      prev_opr=$opr
    24  done <<< "$(sort -t, -k1,1 -k2,2n data.txt)"
    25  echo "$prev_opr,$min,$max"
    26
$
$ ./group_data.sh
operation1,12321,87934
operation2,12312,72321
operation3,13288,13288
operation4,94865,94865
operation5,27383,27383
operation6,322,93483
operation7,3223,3223
$
$

This User Gave Thanks to durden_tyler For This Post:
# 10  
Old 03-24-2017
--
Quote:
Originally Posted by durden_tyler
Code:
[..]
    24  done <<< "$(sort -t, -k1,1 -k2,2n data.txt)"
[..]

This also sorts the header, putting it on an undefined line number. So the shell logic for skipping the header, which assumes it to be on the first line, will not function as intended (it happens to work with the sample)

An alternative would be to use: tail -n +2 data.txt | sort -k1,1 -k2n

(my approach in post #2 also did not take the header into account, it also happens to work but the order was undefined, so I made a change so that it gets skipped)

Quote:
Code:
[..]
    while read line
[..]
      opr=${line%%,*}
[..]
        min=${line##*,}
[..]

An alternative would be to do the splitting of values while reading the line
Code:
while IFS=, read operation value

then the variable expansions would not need to be used..

Last edited by Scrutinizer; 03-25-2017 at 12:04 AM..
This User Gave Thanks to Scrutinizer For This Post:
# 11  
Old 03-24-2017
HI.
Quote:
Originally Posted by Scrutinizer
This also sorts the header, putting it on an undefined line number. So the shell logic for skipping the header, which assumes it to be on the first line, will not function as intended (it happens to work with the sample)
I thought the header was just some line of text that mukulverma2408 included to describe the fields. Given the misspelling, I deleted it from my first response.

However, if it is intended that such data files have headers, then one can invoke datamash to handle those:
Code:
#!/usr/bin/env bash

# @(#) s2       Demonstrate statistics for grouped data, datamash.

# Utility functions: print-as-echo, print-line-with-visual-space, debug.
# export PATH="/usr/local/bin:/usr/bin:/bin"
LC_ALL=C ; LANG=C ; export LC_ALL LANG
pe() { for _i;do printf "%s" "$_i";done; printf "\n"; }
pl() { pe;pe "-----" ;pe "$*"; }
em() { pe "$*" >&2 ; }
db() { ( printf " db, ";for _i;do printf "%s" "$_i";done;printf "\n" ) >&2 ; }
db() { : ; }
C=$HOME/bin/context && [ -f $C ] && $C datamash dixf

FILE=${1-data2}
E=expected-output.txt

pl " Input data file $FILE:"
cat $FILE

pl " Expected output:"
cat $E

pl " Results:"
datamash -t',' --header-in --sort --group=1 min 2 max 2 < $FILE |
tee f1

pl " Verify results if possible:"
C=$HOME/bin/pass-fail
[ -f $C ] && $C || ( pe; pe " Results cannot be verified." ) >&2

pl " Results for producing header on output:"
# datamash -t',' --header-in --header-out --sort --group=1 min 2 # max 2 < $FILE
datamash -t',' -H --sort --group=1 min 2 max 2 < $FILE

pl " Details for datamash:"
dixf datamash

exit 0

producing:
Code:
$ ./s2

Environment: LC_ALL = C, LANG = C
(Versions displayed with local utility "version")
OS, ker|rel, machine: Linux, 3.16.0-4-amd64, x86_64
Distribution        : Debian 8.7 (jessie) 
bash GNU bash 4.3.30
datamash (GNU datamash) 1.0.6
dixf (local) 1.42

-----
 Input data file data2:
Opeation,Time-Taken
operation1,83621
operation2,72321
operation3,13288
operation2,12312
operation1,12321
operation2,45455
operation2,42543
operation1,87934
operation4,94865
operation5,27383
operation6,322
operation6,93483
operation7,3223

-----
 Expected output:
operation1,12321,87934
operation2,12312,72321
operation3,13288,13288
operation4,94865,94865
operation5,27383,27383
operation6,322,93483
operation7,3223,3223

-----
 Results:
operation1,12321,87934
operation2,12312,72321
operation3,13288,13288
operation4,94865,94865
operation5,27383,27383
operation6,322,93483
operation7,3223,3223

-----
 Verify results if possible:

-----
 Comparison of 7 created lines with 7 lines of desired results:
 Succeeded -- files (computed) f1 and (standard) expected-output.txt have same content.

-----
 Results for producing header on output:
GroupBy(Opeation),min(Time-Taken),max(Time-Taken)
operation1,12321,87934
operation2,12312,72321
operation3,13288,13288
operation4,94865,94865
operation5,27383,27383
operation6,322,93483
operation7,3223,3223

-----
Details for datamash:
Code:
datamash        command-line calculations (man)
Path    : /usr/bin/datamash
Version : 1.0.6
Type    : ELF 64-bit LSB executable, x86-64, version 1 (SYSV ...)
Help    : probably available with -h,--help
Repo    : Debian 8.7 (jessie) 
Home    : http://www.gnu.org/software/datamash

Best wishes ... cheers, drl
# 12  
Old 03-24-2017
Thanks a lot for your comments, Scrutinizer. Those are all very valid points.

While writing the script, the biggest source of my consternation was the repetition of this construct
Code:
${line##*,}

in lines 19 and 21.
I would've loved to do this:
1) Extract the 2nd token from the line and assign it to a temp variable and check if $temp is greater than $min.
2) If comparison test is true, i.e. $temp is greater than $min, then assign the $temp to max.

Some languages like C, Perl and (I think) Python allow this.
Kind of like this idiom in K&R, which accomplishes three things: fetch, assignment and comparison.

Code:
 while ((c = getchar()) != EOF) {

Your suggestion of using IFS along with the read function:

Quote:
Originally Posted by Scrutinizer
...An alternative would be to do the splitting of values while reading the line
Code:
while IFS=, read operation value

then the variable expansions would not need to be used..
is a clean way of avoiding that repetition.
Login or Register to Ask a Question

Previous Thread | Next Thread

10 More Discussions You Might Find Interesting

1. Shell Programming and Scripting

Shell Script to Group by Based on Multiple Fields in a file

Hi, I want to know if there is any simple approach to SUM a field based on group by of different fields for e.g. file1.txt contains below data 20160622|XXX1||50.00||50.00|MONEY|Plan1| 20160622|XXX1||100.00||100.00|MONEY|Plan1| 20160623|XXX1||25.00||25.00|MONEY|Plan1|... (3 Replies)
Discussion started by: cnu_theprince
3 Replies

2. Shell Programming and Scripting

Shell script - group by

Hi, I have text file as shown below. root 25 oracle 25 batch 30 griduser 32 admin 35 root 25 oracle 25 batch 30 griduser 32 oracle 25 batch 30 griduser 32 xuser 45 admin 35 I want to group by based on user name, and the output need to be as below. Not necessary the username to be... (10 Replies)
Discussion started by: baladelaware73
10 Replies

3. Shell Programming and Scripting

Emulate fgrep -f in perl

Is there any equivalent of the below requirement in perl fgrep -f file1 file2 > file3 (2 Replies)
Discussion started by: aravindj80
2 Replies

4. Shell Programming and Scripting

Shell Script to ignore # and take corresponding user and group

Hi, I have a following file: role.IMPACT_USER.user=admin role.IMPACT_USER.user=dd12345 role.IMPACT_USER.user=ss76767 #role.IMPACT_USER.user=root #role.IMPACT_USER.group=System role.IMPACT_USER.group=ImpactUser #Description: Allow users to login in to Impact, start and stop service... (5 Replies)
Discussion started by: dbashyam
5 Replies

5. Shell Programming and Scripting

"group by" using shell script?

not sure if it's called "group by" , but what i'm going to do is like this: i have a file below: 192.168.1.10 192.168.1.10 192.168.1.10 192.168.1.11 192.168.1.15 192.168.1.15 192.168.1.20 192.168.1.22 then i hope to get the result like this: 192.168.1.10 : 3 192.168.1.11 : 1... (6 Replies)
Discussion started by: tiger2000
6 Replies

6. Shell Programming and Scripting

Shell script to rename a group of files

Hello, I am having 1800 files in a directory with a specified format, like amms_850o_prod.000003uNy amms_850o_prod.000003u8x amms_850o_prod.000003taP amms_850o_prod.000003tKy amms_850o_prod.000003si4 amms_850o_prod.000003sTP amms_850o_prod.000003sBg amms_850o_prod.000003rvx... (12 Replies)
Discussion started by: atlantis
12 Replies

7. Shell Programming and Scripting

How to emulate ^S/^Q from a script

Hi, I wrote a little menu script that searches through another script you specify and displays step-names and next to it the text of the step. The scripts are converted JCL from mainframe. It alows you to select steps you want and will then create a new script which includes only the steps you... (5 Replies)
Discussion started by: AliceD
5 Replies

8. Shell Programming and Scripting

HELP! Group by in shell script (awk/sed?)

Hello, Could some expert soul please help me with this? I have following file format - task time abc 5 xyz 4 abc 5 xyz 3 ddd 10 ddd 2 I need to generate output as - task ... (5 Replies)
Discussion started by: sncoupons
5 Replies

9. Shell Programming and Scripting

Shell quiz: emulate an associative array

Most shells flavors do not have associative arrays a.k.a. maps. How would you emulate an associative array? I had this problem once and found a working solution, but I don't want to spoil the game hence I wont tell it. Wonder if anyone comes up with something better. (5 Replies)
Discussion started by: colemar
5 Replies

10. Shell Programming and Scripting

Script to emulate ls -lh?

Does anyone have a script they would like to share that emulates "ls -lh" in ksh on Solaris 8? Yeah, I know. Real men don't need that wimpy "h." Well, I'm a wimp. ;) (0 Replies)
Discussion started by: shew01
0 Replies
Login or Register to Ask a Question