Counting chars from right - shell script - shell

Question

I have a temperature sensor connected to a *nix system and the typical output is something like:
pi#raspberrypi $ cat /sys/bus/w1/devices/28-00000202070c/w1_slave
c3 00 4b 46 7f ff 0d 10 12 : crc=12 YES
c3 00 4b 46 7f ff 0d 10 12 t=12187
The result comes without any comma, but is assumed that is always coming with 3 digits after the comma, so in this example would be 12.187º.
I have implemented a filter that places a comma after the second char, and it works most of the time:
grep t= | awk '{print $10;}'| sed 's/t\=//g' | sed 's/\([0-9][0-9]\)\([0-9]\)/\1,\2/g'
However, during winter, temperature drops below 10º and my filter returns values like 95,32º (when it should be 9,532º).
Is there any way of counting characters from the right, so I could always count with the 3 digital characters (and avoiding this problem in temperatures below 10º)?
Thanks,
Joaoabs

Answer 1

awk can handle floating point operations:
awk '/t=/ { sub(/t=/,"",$NF); print $NF/1000}' /sys/bus/w1/devices/28-00000202070c/w1_slave

Answer 2

If I understand you correctly, then what you want to do is :
sed 's/\([0-9][0-9][0-9]\)$/,\1/g'
$ in a regex means 'the end' so this searches for 3 digits right at the end of a string and replaces them with comma+the found digits.
(Note: This should be the last part of your pipe, with the beginning unchanged.)

Answer 3

Using awk
awk -F= '/t=/ {print $NF/1000}' /sys/bus/w1/devices/28-00000202070c/w1_slave
12.187
Or store it to a variable:
temp=$(awk -F= '/t=/ {print $NF/1000}' /sys/bus/w1/devices/28-00000202070c/w1_slave)
echo "$temp"
12.187

Answer 4

don't do grep|awk|sed|sed..., try this:
...|awk -F't=' -v OFS='t=' 'NF>1{$2=sprintf("%'\''d",$2)}7'
test with your data:
kent$ echo "c3 00 4b 46 7f ff 0d 10 12 : crc=12 YES
c3 00 4b 46 7f ff 0d 10 12 t=12187"|awk -F't=' -v OFS='t=' 'NF>1{$2=sprintf("%'\''d",$2)}7'
c3 00 4b 46 7f ff 0d 10 12 : crc=12 YES
c3 00 4b 46 7f ff 0d 10 12 t=12,187
awk has the format pattern in 'printf' for your requirement. just use it.
if you just want to have the t= value:
awk -F't=' -v OFS='t=' 'NF>1{printf "%'\''d\n",$2}'
with same input:
kent$ echo "c3 00 4b 46 7f ff 0d 10 12 : crc=12 YES
c3 00 4b 46 7f ff 0d 10 12 t=12187"|awk -F't=' -v OFS='t=' 'NF>1{printf "%'\''d\n",$2}'
12,187