Using bc to total numbers in a file

[balanga]

Does anyone know if it's possible to use bc() to total the numbers in a file, ie something which just has a number on every line, eg:-

27
53
77
143

 

[VladiBG]

awk '{ sum += $1 } END { print sum }' myfile.txt

awk(1)

Add up first column, print sum and average:

{ s += $1 }
END { print "sum is", s, " average is", s/NR }

awk '{ s += $1 } END { print "sum is", s, " average is", s/NR }' myfile.txt

 

[balanga]

Thanks, not familiar with awk().... Would it be possible add up a mix of MB and KB?

 

[covacat]

awk is the correct solution, but


echo $(cat file.txt)|tr ' ' +|bc

 

[aragats]

Would it be possible add up a mix of MB and KB?

Probably not a perfect solution (I always use sed(), not awk() (-; ):

sed 's/[^0-9]*\([0-9]*\)[kK][bB]*.*/\1\*1024/; s/[^0-9]*\([0-9]*\)[mM][bB]*.*/\1\*1024^2/' my_file | bc | echo 0 `sed 's/^/+/'` | bc

However, to avoid an X-Y problem you should tell us more details on your file. In the OP you mentioned only numbers.

 

[balanga]

I originally thought the denominations (is that the right word) would be the same but they are a mix:-

8KB
795KB
164KB
55KB
55KB
94.50MB

I've since come across numfmt which is in GNU's coreutils, but is missing from FreeBSD's version of this.

 

[balanga]

Probably not a perfect solution (I always use sed(), not awk() (-; ):

sed 's/[^0-9]*\([0-9]*\)[kK][bB]*.*/\1\*1024/; s/[^0-9]*\([0-9]*\)[mM][bB]*.*/\1\*1024^2/' my_file | bc | echo 0 `sed 's/^/+/'` | bc

I like using sed(), although not clever enough for scripts like that...

When I ran this it produced

bc: stdin:1: illegal character: : unexpected
bc: stdin:1: illegal character: : unexpected

I just pasted it straight in from your post.

 

[aragats]

If your lines start with a digit and may contain a decimal point, the script can be improved:

sed 's/^\([0-9\.]*\)[kK][bB]*.*/\1\*1024/; s/^\([0-9\.]*\)[mM][bB]*.*/\1\*1024^2/' num | bc | echo 0 `sed 's/^/+/'` | bc

 

[VladiBG]

awk ' { s=substr($1,length($1)-1); if  (s=="GB") $1=( $1 * 1024 * 1024 ); else if (s=="MB") $1=( $1 * 1024 ); else if (s=="KB") $1= ( $1 * 1); sum += $1 } END { print sum/1024 "MB" }' myfile.txt

 

[covacat]

improvement of aragats' version
sed -e 's/[^0-9]*\([0-9]*\)[kK][bB]*.*/\1\*1024/; s/[^0-9]*\([0-9]*\)[mM][bB]*.*/\1\*1024^2/; s/$/+\\/' -e '$s/+\\$//' file.txt |bc

 

[ralphbsz]

When your input file says "MB", does that mean 1.000.000 bytes or 1.048.576 bytes?

 

[Jose]

If you're parsing the output of du(1) keep in mind that utility will already give you a total with the -c option. Also, you won't have to worry about KB and MB suffixes if you don't use the -h option, and look at the -k option.

Edit: You might also find something like sysutils/jdiskreport useful.

 

[balanga]

When your input file says "MB", does that mean 1.000.000 bytes or 1.048.576 bytes?

It's a usage report from my ISP. My allowance is 10GB, so I would guess it's 1.000.000 bytes.

 

[bobmc]

It's a usage report from my ISP. My allowance is 10GB, so I would guess it's 1.000.000 bytes.

A legacy spinning disk has 512 byte sectors so your allowance is probably an even multiple of that.