使用sed从流中分离模式

Question

I have a file that has log entries on each line like:

request country_US url=http://example.com/us/id=6546456 {response_time 1000 msec} 
request country_UK url=http://example.com/uk/id=1242423 {response_time 60 msec} 

Right now I'm using sed to process this file and separate the response time like so:

sed -e 's/.*\(response_time \S\+\).*/\1/p' -e 's/'

That transforms each line to

response_time 1000
response_time 60

Right now I would like to separate also the country prefix and get processed lines like so:

US 1000
UK 60

How should i modify my sed command? Thanks!

Answer 1

尝试这个：

sed -e 's/^request country_\([^ ]*\).*response_time \([0-9]*\) msec.*$/\1 \2/g'

Answer 2

You could modify your sed expression a bit to get the desired result. Saying:

sed -r 's/.*country_(\S+).*response_time ([0-9]+).*/\1 \2/' filename

would return the following for your sample input:

US 1000
UK 60

Answer 3

Using awk

awk '/response_time/ {split($2,a,"_");print a[2],$(NF-1)}' file
US 1000
UK 60

---

If all line does have response_time you can just do:

awk '{split($2,a,"_");print a[2],$(NF-1)}' file

---

Or just:

awk -F" |_"  '{print $3,$7}' file
US 1000
UK 60

Answer 4

Another possibility which is pretty much in the sence of your original version is:

sed -e 's/.*country_\(\S\+\).*response_time\( \S\+\).*/\1\2/' file
US 1000
UK 60

For the future I can only recommend http://vimregex.com/ . I always found it being extremely useful.