[SOLVED] How to compare negative and decimal numbers

satishchahar · 09-19-2011, 02:14 AM

a=-3.123
b=4.2

if [ $a > $b ]; then
echo "It's OK"
else
echo "It's not OK"
fi

It is showing me wrong result. Please suggest how can I compare negative and decimal numbers all together.

colucix · 09-19-2011, 03:17 AM

The shell cannot do floating point arithmetic. You have to use an external command, like bc. Example:

Code:

a=-3.123
b=4.2

if [[ $(echo "$a > $b" | bc) -eq 1 ]]
then
  echo "It's OK"
else
  echo "It's not OK"
fi

druuna · 09-19-2011, 03:18 AM

Hi,

bc comes to mind. Here's one way:

Code:

#!/bin/sh

a="-3.123"
b="4.2"

case $(echo "a=$a ;b=$b ;r=-1 ;if(a==b)r=0 ;if(a>b)r=1 ;r" | bc)
in
  0) echo $a and $b are equal ;;
  1) echo $a is bigger then $b ;;
  *) echo $a is less then $b ;;
esac

- Linux / Unix Command: bc
- bc -- The Linux Command-line Calculator

Hope this helps.

David the H. · 09-19-2011, 08:14 AM

Not to mention that you got the shell test wrong anyway. ">" is a string comparison operator. It compares the ascii values of the two sides. You need to use options like "-gt" when comparing integer values.

Code:

if [ "$a" -gt "$b" ]; then

http://www.tldp.org/LDP/abs/html/comparison-ops.html

Don't forget to quote your variables too. Although it isn't a problem with integer inputs like this, it would break if you used a string with a space in it.

However, if you're using bash and portability isn't an issue, it's recommended to use ((..)) instead when working with numbers, and [[..]] when you have strings, filenames, or complex tests.

http://mywiki.wooledge.org/ArithmeticExpression
http://mywiki.wooledge.org/BashFAQ/031

And more on floating point arithmetic here:
http://mywiki.wooledge.org/BashFAQ/022

PS: Please use [code][/code] tags around your code, to preserve formatting and to improve readability.

H_TeXMeX_H · 09-19-2011, 08:24 AM

You can also use awk:

Code:

bash-4.1$ a=-3.123
bash-4.1$ b=4.2
bash-4.1$ echo $a $b | awk '{ if ($1 > $2) print "ok"; else print "not ok"}'
not ok
bash-4.1$ echo $b $a | awk '{ if ($1 > $2) print "ok"; else print "not ok"}'
ok

satishchahar · 09-20-2011, 03:29 AM

A file is having last field of each line as a number (eg. -3.213 or 2.1 or whatever else) and I want to compare this number with 4.2 which I am now able to compare using the above commands.

But Can I delete the line instantly (in place) from this file which is having number less than 4.2. How? Please suggest.

druuna · 09-20-2011, 03:46 AM

Hi,

Quote:

Originally Posted by satishchahar

A file is having last field of each line as a number (eg. -3.213 or 2.1 or whatever else) and I want to compare this number with 4.2 which I am now able to compare using the above commands.

But Can I delete the line instantly (in place) from this file which is having number less than 4.2. How? Please suggest.

Too my knowledge you cannot do this directly (in place), but using a temporary file it will work:

Code:

$ cat infile
just a sample line -5
just a sample line 4.1
Just a sample line 4.2
just a sample line 4.3

$ awk '{ if ($NF < 4.2 ) next ; else print }' infile > tempfile

$ cat tempfile
just a sample line 4.2
just a sample line 4.3

$ mv tempfile infile

Hope this helps.

colucix · 09-20-2011, 03:49 AM

Yes, GNU awk can do that:

Code:

awk '{ if ($NF < 4.2 ) next ; else print > FILENAME }' infile

druuna · 09-20-2011, 04:01 AM

@colucix: I stand corrected (and I learned something!) Thanks.

colucix · 09-20-2011, 04:11 AM

You're welcome! Just a little add-on. I remember an old post by radulov (the greatest awk master here at LQ) suggesting another method that probably works even with non-GNU awk implementation (I cannot test it right now). It was:

Code:

( rm infile && awk '{ if ($NF < 4.2 ) next ; else print }' > infile; ) < infile

this one passes the content of infile to a subshell as standard input, so that the standard output can be safely redirected to the file itself. A bit more tricky, but maybe useful in some cases.

satishchahar · 09-20-2011, 06:01 AM

Last query:

If last field is like below, then how to compare and delete line:

$ cat infile
just a sample line with number=-5
just a sample line with number=4.1
Just a sample line with number=4.2
just a sample line with number=4.3

druuna · 09-20-2011, 06:18 AM

Hi,

You can set a different, or multiple record separators:

awk -F"=| " '{ if ($NF < 4.2 ) next ; else print }' infile

The -F"=| " part sets 2 separators a = and a space.

Code:

$ awk -F"=| " '{ if ($NF < 4.2 ) next ; else print }' infile
Just a sample line number=4.2
just a sample line number=4.3

Hope this helps.

theNbomr · 09-20-2011, 01:51 PM

When posting sample data for testing parsers that may be sensitive to formatting, it is a good idea to post the data in [CODE][/CODE] tags to preserve the formatting.
--- rod.

satishchahar · 09-21-2011, 05:20 AM

Hi druuna

Your Idea worked for me to get the data exactly but when I am sending the output the command to EMAIL, I am getting the output in one line

Quote:

$ awk -F"=| " '{ if ($NF < 4.2 ) next ; else print }' infile | mail -s "postgrep" abc@xyz.local

Below is the output I am getting in my mail client:

Quote:

Just a sample line number=4.2 just a sample line number=4.3

druuna · 09-21-2011, 07:42 AM

Hi,

Is the above exactly how you did this? I'm asking because on my side it works as expected (I get a mail with separated lines). The output separators that awk uses aren't changed and lines should be put one below the other, not one after the other.

What is the output of the following (bold) commands:

Code:

$ file infile

$ awk -F"=| " '{ if ($NF < 4.2 ) next ; else print }' infile > outfile

$ file outfile