Writing a bash script without using awk?

Question

We are asked to write a shell script which takes in input of player records as standard inputs.

Sample input:

id|name|time
23|Jordan|45:17
14|Jason|4:50
12|Bryan|
24|Cody|00:12
33|kobe|41
55|rocky|0

And we need to read each record ( skip header ) in our script and then output each corresponding record by converting time in seconds and changing delimiter from '|' to ' ' ( space ).

As we can see in above sample testcase , some records have empty time field ( 3rd record ), for those record time in seconds will be considered 0 .

Sample Case Output:

23 Jordan 2717
14 Jason 290
12 Bryan 0
24 Cody 12
33 kobe 2460
55 rocky 0

my_solution_script.sh

#!/bin/bash


read -r header
while IFS="|" read -r pid pname time || [[ -n $pid ]]
do  
    min=$(cut -d ':' -f 1 <<< "$time")
    sec=$(cut -d ':' -f 2 <<< "$time")
    ((min*=60))
    ((min+=sec))
    
    echo "$pid $pname $min"
done

Wrong Output:

23 Jordan 2717
14 Jason 290
12 Bryan 0
24 Cody 12
33 kobe 2501
55 rocky 0

As we can see above script is giving wrong output for 5th record .

How can I fix the above script , to get the correct output in every case?

I think there might be a simpler solution possible using awk , but I don't have much idea about 'awk scripting' right now, so I am looking for a way to solve this question using basic shell commands , nevertheless awk command solutions are also welcome.

Thank You.

Answer 1

The problem is that cut -d: -f2 <<< "$time" returns all of $time when it doesn't contain a : delimiter. So for kobe you're calculating 41*60+41 instead of just 41*60 .

So you need to check whether $time contains a : before trying to get the seconds.

read -r header
while IFS="|" read -r pid pname time || [[ -n $pid ]]
do  
    min=$(cut -d ':' -f 1 <<< "$time")
    if [[ $time =~ : ]]
        sec=$(cut -d ':' -f 2 <<< "$time")
    else
        sec=0
    fi
    ((min*=60))
    ((min+=sec))
    
    echo "$pid $pname $min"
done

Answer 2

With GNU awk:

awk 'NR>1{$3=$3*60+$4; NF=3; print}' FS='[|:]' file

Output:

23 Jordan 2717
14 Jason 290
12 Bryan 0
24 Cody 12
33 kobe 2460
55 rocky 0

NF=3 limits GNU awk's print to three columns.

---

See: 8 Powerful Awk Built-in Variables – FS , OFS, RS, ORS, NR , NF , FILENAME, FNR

Answer 3

Could you please try following. Written and tested in https://ideone.com/9RkGvJ

awk '
BEGIN{
  FS="|"
}
FNR==1{  next  }
{
  split($3,arr,":")
  $3=(arr[1]*60)+arr[2]
}
1;
' Input_file

Explanation: Setting field separator as | for all lines. Then checking FNR==1 in which putting next will skip that line. Then on each line splitting 3rd column with : separator and re-creating 3rd field which has 1st element of array multiply with 60 and adding it's 2nd element to it get seconds value in 3rd column. Then mentioning 1 will print lines.

Answer 4

An immediate fix is to set minutes to zero if there is no colon in the value. You can avoid the ugly and moderately expensive external processes entirely.

sec=${time#*}
min=${time%:"$sec"}
min=${min:-0}

This uses the shell's built-in parameter expansion facility to pick apart the value. Briefly, ${time#pattern} returns the value of $time without any prefix which matches pattern ; the % operator does the same for suffixes.

Using an Awk script is almost certainly a better idea; you should be able to learn the basics in less than an hour, perhaps already enough to solve this problem yourself. Here's a quick and dirty untested attempt.

awk -F '|' 'NR>1 && ($3 ~ /:/) {m = s = $3;
    sub(/:.*/, "", m); sub(/.*:/, "", s);
    $3 = m*60+s } 1'

Answer 5

bash:

{
  read header
  while IFS='|' read -r id name time; do
    IFS=':' read -r mins secs <<<"$time"
    echo "$id $name $((60 * 10#$mins + 10#$secs))"
  done
} < file

We're using the pattern IFS=delim read -r field1 field2... twice here to do the parsing.

The 10# in the arithmetic expression is to force the values to be interpreted as base-10 numbers. Otherwise, 08 and 09 will be interpreted as invalid octal numbers due to the leading zero.

Answer 6

Assuming the other fields won't have an embedded colon, you can parse min & secs on the read, then use parameter parsing default zeroes for empty min or secs. You can also do all the math in one pass, inside the echo .

read -r header
while IFS="|:" read -r pid pname min secs || [[ -n $pid ]]
do echo "$pid $pname $(( 10#${secs:-0} + 10#${min:-0}*60 ))"
done

If names can have colons, this doesn't work.

As pointed out, leading zeroes would also cause issues, so I added a base-selection indicator ( 10# ) to assure base-10 math. c.f. https://mywiki.wooledge.org/ArithmeticExpression