[SOLVED] Removing specific lines from a file (multiple conditions)

Sushukka · 01-22-2015, 07:10 AM

Here is the problem. I have automatically generated file with looks something like this:
------------------------------------------------------
// Comment
// more comments
// blabla
javascript code
.
.
.
"number|Dedicated|SnameXX|number|number|number"
"number|Dedicated|SnameXX|number|number|number"
"number|Dedicated|SnameNN|number|number|number"
"number|Dedicated|SnameYY|number|number|number"
"number|Dedicated|SnameZZ|number|number|number"
...the list continues...
...more code...
.
.
.
);
------------------------------------------------------

What I need to do is to remove all other Sname lines than SnameXX. So in above example I need to spare all the other lines in the file but SnameNN, SnameYY and SnameZZ need to be deleted. All these Sname lines have also "Dedicated" written in them (other fields cannot be used for grouping). With grep I can isolate the "bad" lines with: grep -i -v SnameXX file.js | grep Dedicated. From here I'm stucked. What I have understood, having "inverse" selection in sed requires some strange tricks. In any normal programming language, you would need one loop running through the file with two if statements. However, I would like to believe that there are some neat solution to solve this with only command line tools. I'm playing with RHEL, but this level of fiddling probably doesn't have any os dependencies.

veerain · 01-22-2015, 07:42 AM

Well your script successfully selects required lines, then what with sed?

J Martin Rushton · 01-22-2015, 08:05 AM

AWK is standard on (AFAIK) all *nixes, so consider:

Code:

awk '
  /|Dedicated|SnameXX|/ { print ; next }
  /|Dedicated|Sname/ { next }
  { print }
'

The first test looks for Dedicated|SnameXX lines which you want and therefore they are printed to standard output and processing starts with the next line.

The next line looks for a SnameYY or SnameZZ line and if found just goes on to the next line.

All other lines are output unchanged.

Miati · 01-22-2015, 11:05 AM

Quote:

Originally Posted by Sushukka

SnameNN, SnameYY and SnameZZ need to be deleted .. Sname lines have also "Dedicated" written in them

use egrep?

Code:

egrep -v "SnameNN|SnameYY|SnameZZ" file | grep "Dedicated"

output

Code:

"number|Dedicated|SnameXX|number|number|number"
"number|Dedicated|SnameXX|number|number|number"

Slax-Dude · 01-22-2015, 11:59 AM

This should work:

Code:

cat yourfilename | sed '/Sname[NN|YY|ZZ]/d'

If you are happy with the output, then this will mod the file directly:

Code:

sed -i '/Sname[NN|YY|ZZ]/d' yourfilename

Sushukka · 01-22-2015, 01:36 PM

Thanks for the quick answers! However, the tricky part (which I didn't manage to mention well enough) is that there is no pattern with SnameYY, SnameZZ or SnameNN. Only pattern is found in SnameXX which need to be saved. The list could look like this:

Code:

"number|Dedicated|prodName1|number|number|number"  << save
"number|Dedicated|testName3|number|number|number"  << save
"number|Dedicated|devName45|number|number|number"  << save
"number|Dedicated|someblatext|number|number|number" 
"number|Dedicated|ipsumlorem|number|number|number"   
"number|Dedicated|x1vr23d|number|number|number"

So basically in pseudo code:

Code:

do until end of the file
  if substr(line) = "Dedicated" then
     if substr(line) != "name" then
        do nothing
  else
     line > output
end

J Martin Rushton · 01-23-2015, 04:17 AM

Your pseudo code is missing an else. If the line does contain "Dedicated" and does contain "name", then what should be done? You appear to be dealing with 3 forms of line:

*Dedicated*name* print
*Dedicated* ignore
* print

If this is correct, then a simple case statement in BASH is probably the easiest solution. Most importantly though, a clear and precise statement of the problem would help.

I used the following data file:

Code:

qwertDedicatedtyuinameghjk
qwertDedicatedtyui
sdfghj

and put it through the following script:

Code:

while read line;
do
    case $line in
        *Dedicated*name*)       echo $line ;;
        *Dedicated*)            ;;
        *)                      echo $line ;;
    esac
done

The first and third lines were printed as expected.

Sushukka · 01-23-2015, 04:27 AM

Yep, you're right. I just forgot the else statement from the pseudo code. Late evening here, long day, blabla, my mistake.

But again, that's the solution. Many thanks! Wasn't sure about the syntax (hence here in general noob area). Better to have a simple couple of lines of code than cryptic parameter jungle with one or many ready made utilities.

Slax-Dude · 01-23-2015, 08:29 AM

This should delete all lines that contain "Dedicated" except those that also contain "name":

Code:

cat yourfilename | sed -e '/Dedicated/{/name/!d;}'