sed/awk help in removing the section of file if it contains regex

simplybharath · 04-22-2016, 02:02 AM

Hi i need to remove these contents from a file only if it contains line related_pin : "sbss_pclk_div4_5_SUBRATE_20checkpin1"

timing () {
related_pin : "sbss_pclk_div4_5_SUBRATE_20checkpin1" ;
timing_type : min_pulse_width ;
mode(etm_modes, "FUNC_SUB_20");
/* comment : from another design; */
fall_constraint( scalar ){
values ( "0.024838");
}

} /* end of arc sbss_pclk_div4_5_SUBRATE_20checkpin1

how to delete this section only if body of the section contains such line ?

grail · 04-22-2016, 03:00 AM

Please show what you have tried?

simplybharath · 04-22-2016, 03:27 AM

Hi Grail

i was doing it in vim , but there are lot of files and lot of regex expressions which was time consuming , there are multiple timing () lines in my file so i tried using tac filename | sed /} \/\* end of arc sbss_pclk_div4_5_SUBRATE_20checkpin1/timing \(\) {/d' | tac , which seems to be non-working

HMW · 04-22-2016, 05:30 AM

Ok. Consider this as infile:

Code:

timing () {
related_pin : sbss_pclk_div4_1_SUBRATE_20checkpin1 ;
timing_type : min_pulse_width ;
mode(etm_modes, FUNC_SUB_20);
/* comment : from another design; */
fall_constraint( scalar ){
values ( 0.024838);
}

} /* end of arc sbss_pclk_div4_1_SUBRATE_20checkpin1

timing () {
related_pin : sbss_pclk_div4_5_SUBRATE_20checkpin1 ;
timing_type : min_pulse_width ;
mode(etm_modes, FUNC_SUB_20);
/* comment : from another design; */
fall_constraint( scalar ){
values ( 0.024838);
}

} /* end of arc sbss_pclk_div4_5_SUBRATE_20checkpin1



timing () {
related_pin : sbss_pclk_div4_5_SUBRATE_10checkpin1 ;
timing_type : min_pulse_width ;
mode(etm_modes, FUNC_SUB_20);
/* comment : from another design; */
fall_constraint( scalar ){
values ( 0.024838);
}

} /* end of arc sbss_pclk_div4_5_SUBRATE_10checkpin1

What we want to remove is in bold.

After i run this little script (simplyb.sh, 5 lines) I get this result:

Code:

./simplyb.sh 
timing () {
related_pin : sbss_pclk_div4_1_SUBRATE_20checkpin1 ;
timing_type : min_pulse_width ;
mode(etm_modes, FUNC_SUB_20);
/* comment : from another design; */
fall_constraint( scalar ){
values ( 0.024838);
}

} /* end of arc sbss_pclk_div4_1_SUBRATE_20checkpin1




timing () {
related_pin : sbss_pclk_div4_5_SUBRATE_10checkpin1 ;
timing_type : min_pulse_width ;
mode(etm_modes, FUNC_SUB_20);
/* comment : from another design; */
fall_constraint( scalar ){
values ( 0.024838);
}

} /* end of arc sbss_pclk_div4_5_SUBRATE_10checkpin1

The block with "sbss_pclk_div4_5_SUBRATE_20checkpin1" is gone.
But... it would be helpful to see more of the actual file.

I used grep and sed mainly.

Best regards,
HMW

simplybharath · 04-22-2016, 05:54 AM

Hi HMW

your assumptions are right , my basic problem is how to remove the section of text from a file delineated by some start and end regex only if a line inside that section of text matches another regex , so i want to delete below section only if this line matches related_pin : sbss_pclk_div4_1_SUBRATE_20checkpin1 ; , i was trying to find a generic solution for such thing

timing () {
related_pin : sbss_pclk_div4_1_SUBRATE_20checkpin1 ;
timing_type : min_pulse_width ;
mode(etm_modes, FUNC_SUB_20);
/* comment : from another design; */
fall_constraint( scalar ){
values ( 0.024838);
}

} /* end of arc sbss_pclk_div4_1_SUBRATE_20checkpin1

HMW · 04-22-2016, 06:12 AM

Quote:

Originally Posted by simplybharath

Hi HMW

your assumptions are right , my basic problem is how to remove the section of text from a file delineated by some start and end regex only if a line inside that section of text matches another regex , so i want to delete below section only if this line matches related_pin : sbss_pclk_div4_1_SUBRATE_20checkpin1 ; , i was trying to find a generic solution for such thing

timing () {
related_pin : sbss_pclk_div4_1_SUBRATE_20checkpin1 ;
timing_type : min_pulse_width ;
mode(etm_modes, FUNC_SUB_20);
/* comment : from another design; */
fall_constraint( scalar ){
values ( 0.024838);
}

} /* end of arc sbss_pclk_div4_1_SUBRATE_20checkpin1

Ok. If the text looks EXACTLY like above, then my script obviously does the job.

I used grep to get the starting line number (hint: ((startline=startline-1))), and then sed to remove lines beginning from there until the end of the block.

Best regards,
HMW

simplybharath · 04-22-2016, 08:18 AM

Hi HMW

can you let me know the contents of the script ?

Regards
Bharath

grail · 04-22-2016, 09:17 AM

You have been given suggestions, so at least make an attempt and I am sure HMW will gladly show you how he would have done it.

simplybharath · 04-22-2016, 09:52 AM

i tried getting the line no using grep -n , but can sed delete can be used like this sed /<line no>/,/end line regex/ ? if not how to get the end line no assuming there are multiple matches for that particular regex as such line no is relative to start line ?

Regards
Bharath

pan64 · 04-22-2016, 10:13 AM

the usual way (at least for me) to set the delimiter as record separator and you only need to keep / drop the specified records. This can be easily implemented in awk:

Code:

awk -v RS=timing '/pattern/' inputfile

you only need to adjust it a bit....

HMW · 04-22-2016, 11:52 AM

Quote:

Originally Posted by simplybharath

i tried getting the line no using grep -n , but can sed delete can be used like this sed /<line no>/,/end line regex/ ? if not how to get the end line no assuming there are multiple matches for that particular regex as such line no is relative to start line ?

Regards
Bharath

I'm sorry, but I simply don't understand what you mean. It is probably the language barrier, but I cannot make head or tails out of this.

I'll try to break down my method for you anyway:

Find the line with the string you search for
Based on that line, set a starting point
From the starting point, remove the lines you want to remove

That's how I did it, and it works with your example. You still, however, have not showed the content of the actual file - only what you want to remove from said file - so my method may or may not work properly in real life.

Best regards,
HMW