Sum values in a file with a for-loop

Question

I have a file which contains values in KB and MB. How can I, in a for-loop calculate the sum of the values in MB for both MB and KB ?

For MB, This gives me the values and then I need to sum those:

cat usage_1.12.2015_31.12.2015.xls.csv | grep MB | awk -F '"' '{print $4}'  | awk '{print $1}'

Output:

KB

  student@student-vm:~/Downloads$ cat usage1.12.2015_31.12.2015.xls.csv | grep KB
    "2015/12/28 14:15:46","109 KB",,0
    "2015/12/28 14:13:47","331 KB",,0
    "2015/12/21 11:01:34","762 KB",,0
    "2015/12/20 19:51:40","253 KB",,0
    "2015/12/07 07:18:50","81 KB",,0
    "2015/12/03 14:56:06","407 KB",,0

MB

student@student-vm:~/Downloads$ cat usage1.12.2015_31.12.2015.xls.csv | grep MB
"2015/12/06 15:54:47","10 MB",,0
"2015/12/06 09:43:02","4 MB",,0
"2015/12/05 21:39:04","21 MB",,0
"2015/12/05 09:17:11","18 MB",,0
"2015/12/04 21:57:49","70 MB",,0
"2015/12/04 17:09:45","1 MB",,0
"2015/12/04 16:47:05","17 MB",,0

replace KB by * 1, MB by * 1024, etc. and `eval` the results — amdixon, Jan 01 '16 at 13:41

Tom Fenech · Accepted Answer · 2016-01-01T15:12:06.200

1

I would recommend using awk here:

awk -F'"?,"?' '{ split($2, a, / /); if (a[2] == "KB") a[1] /= 1000; sum += a[1] } 
END { print sum }' file

The -F switch is used to split the input on commas, optionally surrounded by quotes. Split the second field $2 on spaces / / into the array a. Divide the first part a[1] (the number) by 1000 if the second part a[2] is KB. Accumulate a sum of the values and once the file has been fully processed, print the value in the END block.

edited Jan 01 '16 at 15:12

answered Jan 01 '16 at 14:06

Tom Fenech

72,334
12
107
141

The code worked, but I have no idea how it works. :) – Jan 01 '16 at 14:07
1

Would you like me to explain some part of it more clearly? – Tom Fenech Jan 01 '16 at 14:26
That would be helpfull for everyone I guess. – Jan 01 '16 at 14:28

score 0 · Answer 2 · answered Jan 01 '16 at 13:49

First get the column values so that you have just lines like 1 MB, 50 KB etc., then you can do this kind of thing to get the number of bytes:

Replace spaces with empty strings (using sed notation):
```
s/ //
```
Convert megabytes to kilobytes (add three zeros if using ISO units, *1024KB and send the result of step 3 through bc if it's kibibytes; you can also do this for GB, TB etc.):
```
s/MB/000KB/
```
Convert kilobytes to bytes:
```
s/KB/000/
```
Sum the numbers
Convert bytes to megabytes:
```
bc <<< "${sum} / 1000000"
```

Sum values in a file with a for-loop

2 Answers2