使用SED在搜索条件后替换变化的字符串

Question

Inside a shell script amidst error checking loops, I have a command that contains a variable that must be updated once an hour. The command as it exists currently is (the line begins with a \\t):

        /home/user/therm/therm -d /dev/ttyUSB0 -t host -O 44.0

In this case, I'm trying to replace the number that's after "-O". This is a temperature in Fahrenheit, so it has the possibility of being a negative number or a positive number, and one, two, or three digits followed by a single decimal place. So I have to be able to replace a variable that could be "-15.0", "7.2", "104.5", or any other combination - with a new variable that could also be any of the above.

I can force the "-O" setting to be the last thing on the line, which helps some. But I can't count on anything else on that line except the file name, the "-d /dev/ttyUSB0" and the "-O" always being there. In this particular shell script there are no other instances of "-O", but I might want to verify that the line has either the file name or the "-d /dev/ttyUSB0" on it to avoid replacing a flag on some other line in the script.

The new variable is parsed (via xmllint and sed) from an XML file grabbed via wget once an hour. This shell script I need to edit is ran by cron every 5 minutes, only receiving the new string once an hour. The new string can be passed to sed via a variable, or as a file with only the number and a \\n in it.

Any thoughts or brilliant ideas would be appreciated. Thanks.

Answer 1

You can use this sed command:

# string to be replaced
r=123.4
sed "s/\(-O *\)[^[:blank:]]*/\1$r/" file
/home/user/therm/therm -d /dev/ttyUSB0 -t host -O 123.4

Pattern -O *[^[:blank:]]* will find literal -O followed by 0 or more spaces and then the non-space string.

Answer 2

这似乎有效

awk '{$NF=($NF-32)/1.8;print}'