The sed command

adelinewss · Jun 2, 2000

Hi, I need to delete the content of files which start with /* and end with */.  Like example:- /* Today is Friday Now is 3p.m. */ And also I need to delete the line from the start point with "//".  Like example:- Today is Friday.  //Now is 3p.m. I want to delete //Now is 3p.m. onwards.   May I know how to do this with sed or other UNIX command? Thks, Adeline

AndyBo · Jun 2, 2000

You could do something for this in perl.  (I used to do something similar to strip customer dependent source code from files, so this should work.) #!/usr/bin/perl $Input = "/path/to/input/file"; $Output = "/path/to/new/output/file"; open(INPUT, "<$Input&quot

; open(OUTPUT, ">$Output&quot

; # Loop thru the input file assigning each line to # $Current_Line to avoid messing around with $_. while ($Current_Line = <INPUT&gt

{     # If the current line starts with a "comment start"     # indicator, set the "in a comment" flag.     if ( $Current_Line =~ /\/\*/ ) {         $In_A_Comment = 1;     }     # Are we in a comment section?     if ( $In_A_Comment == 1 ) {         # See if we are at the end of the comment section.         if ( $Current_Line =~ /\*\// ) {             # Reset "in a comment" indicator.             $In_A_Comment = 0;             # Skip this line.             next;         } else {             # In a comment, so skip current line             next;         }     }     # For all lines, strip data following (and including)     # "//" from input lines.     $Current_Line =~ s/\/\/.*//; } # End "while" Hope this helps.

MikeLacey · Jun 3, 2000

like the perl script Andy - you should put that as a FAQ in Perl (mind you - <grin> - you don't have a 'print OUTPUT "$Current_Line\n";' line - or am I missing something?) If you don't have perl - think you can use the C ProProcessor (cpp) and that will strip C style comments (not C++ // comments though). Mike <a href=mailto:michael.j.lacey@ntlworld.com>michael.j.lacey@ntlworld.com</a> <a href=

http://www.cargill.com/>

Cargill's Corporate Web Site</a> Please -- Don't send me email questions without posting them in Tek-Tips as well. Better yet -- Post the question in Tek-Tips and send me a note saying "Have a look at so-and-so in the thingy forum would you?"

AndyBo · Jun 5, 2000

Doh!  You're right, Mike, I missed out a crucial little line about actually sending the stuff we do want to keep to the new file! After the line:     $Current_Line =~ s/\/\/.*//; you want to add:     print OUTPUT $Current_Line; Also, if you want to be tidy you could add close(INPUT); and close(OUTPUT); at the very end of the script. Thanks for the heads up Mike - I guess that will teach me to post replies late on a Friday afternoon

By the way, I'm still looking at sed/awk solutions to this, and if I get one I'll post it here.

MikeLacey · Jun 6, 2000

You're right about the close statements. I used to not bother - but got bitten with a perl script called once a second by cron that opened multiple files. Worked fine - for months. Upgraded to HPUX 10. (something - what I was on before 10.20) and *BANG* as the whole system ran out of file handles. Whoops..... So - now I tend to close files when I'm done with them Mike <a href=mailto:michael.j.lacey@ntlworld.com>michael.j.lacey@ntlworld.com</a> <a href=

http://www.cargill.com/>

Cargill's Corporate Web Site</a> Please -- Don't send me email questions without posting them in Tek-Tips as well. Better yet -- Post the question in Tek-Tips and send me a note saying "Have a look at so-and-so in the thingy forum would you?"

bigscouse · Jun 7, 2000

Getting rid of the // is easy using sed: - sed '/\/\//d' <oldfile>newfile If you want to delete the total content of the file then this will work in korn shell: - head -1 oldfile ¦ grep "^\/\*" > testfile if [ -s testfile ] then tail -1 oldfile ¦ grep "\*\/$" > testfile if [ -s testfile ] then > oldfile echo "file initialised" else echo "file end ok" fi else echo "file start ok" fi Note; in grep and sed the $ is the metacharacter for end of line and the ^ is for start of line, the \ negates the metacharacter so that it is read as it is. Ged Jones <a href=mailto:gedejones@hotmail.com>gedejones@hotmail.com</a> <a href= > </a>

Caliban · Jun 7, 2000

It is not pretty, but: sed -e '/^\/\*/d' -e '/\*\/$/d' -e 's/\/\/[   -z]*//' -e '/^$/d' $INFILE > $OUTFILE should work.  The character in the [   -z] test is a tab.

Caliban · Jun 7, 2000

That is -- if I understand what you want.  My quick test produced: # cat t /* comment 1 comment 2 */ text text 2 // comment3 // comment4 text 3 text 4 //       comment4 # sed -e '/^\/\*/d' -e '/\*\/$/d' -e 's/\/\/[   -z]*//' -e '/^$/d' t > tt # cat tt text text 2 text 3 text 4

MikeLacey · Jun 8, 2000

Wow! I *like* that Caliban... Obfuscation rules, but it's *very* cool Mike <a href=mailto:michael.j.lacey@ntlworld.com>michael.j.lacey@ntlworld.com</a> <a href=

http://www.cargill.com/>

Cargill's Corporate Web Site</a> Please -- Don't send me email questions without posting them in Tek-Tips as well. Better yet -- Post the question in Tek-Tips and send me a note saying "Have a look at so-and-so in the thingy forum would you?"

adelinewss · Jun 10, 2000

Thanks a lot for your help.  It's greatly appreciated.

bigscouse · Jun 13, 2000

Sorry, I made a slight omission, to delete all text lines from // onward use: - sed '/\/\/.*/d' <oldfile>newfile I MISSED THE .* Ged Jones <a href=mailto:gedejones@hotmail.com>gedejones@hotmail.com</a> <a href= > </a>

bigscouse · Jun 13, 2000

I realy like the one line asnwer, think I will borrow the idea, however I believe the problem is to get rid of ALL text in files starting with ^/* AND ending with */& not just any line starting /* and any ending */. Ged Jones <a href=mailto:gedejones@hotmail.com>gedejones@hotmail.com</a> <a href= > </a>

bigscouse · Jun 13, 2000

Silly me again: - sed 's/\/\/.*//' <oldfile>newfile It's hard to think when collegues keep pestering you isn't it. Ged Jones <a href=mailto:gedejones@hotmail.com>gedejones@hotmail.com</a> <a href= > </a>

adelinewss · Jul 3, 2000

This is not applicable if the content of file is as followed:- /* this is a test delete this row */ or /******************************** this is a test delete this row *********************************/ It will give result as followed:- this is a test delete this row Any idea?

bigscouse · Jul 10, 2000

The following checks if thefirst line starts /*, if so and it ends */ then it is emptied but you could replace > with rm. head -1 oldfile ¦ grep "^\/\*" > testfile if [ -s testfile ] then tail -1 oldfile ¦ grep "\*\/$" > testfile if [ -s testfile ] then > oldfile echo "file initialised" else echo "file end ok" fi else echo "file start ok" fi Note: About the sed option, loos like you missed a \/   s/    : - s substitutes one string with another, \/\/  : - \ tells the shell not to interpret what follows           so in this instance it looks for // .*    : - wild card, means all subsequent charecters              will be used. //    :- This sets the replacement string, as none was          entered then the 1st string is deleted. I think this need to be incorporated in the script like this: - head -1 oldfile ¦ grep "^\/\*" > testfile if [ -s testfile ] then tail -1 oldfile ¦ grep "\*\/$" > testfile if [ -s testfile ] then > oldfile   # or touch oldfile or rm oldfile etc.. echo "file contents deleted" else echo "file end ok" sed '/\/\/.*/d' <oldfile>newfile fi else echo "file start ok" sed '/\/\/.*/d' <oldfile>newfile fi Ged Jones <a href=mailto:gedejones@hotmail.com>gedejones@hotmail.com</a> <a href= > </a>

Guest_imported · Jul 17, 2000

Ok, You guys seem to like this sorta thing, so I'll give you one you can answer for me. History: I am trying to write a script for some AIX 4.3.2 OS RS/6000's.  I am using the generic "machine_info" script to gather the info.  It basically does lsvg, lslv, lspv, lslpp etc.. and writes it to a file you specify.  What I'm trying to do is break down each level of the script into different files for each (i.e., software inventory = lslpp -l > software.xls)  I put them in *.xls files and delimit them with semi-colons (

then I can put them in a microsludge Excel spreadsheet or an Access Db table for my DR inventory.  The problem I'm having is there is a section where I do "lspv -l hdisk#" to list physical volumes by disk.  The output rendered looks like: hdisk14: LV NAME  LPs PPs  DISTRIBUTION      Mount point enbkup01  542 542   109..108..108..108..109/edbackup where the "109/edbackup" run together, I'm trying to delimit with a (

, but the number is not always "9" and the letter isn't always "e".  How can I put a delimiter in there using a generic integer followed by a slash followed by a generic letter?  I've done this with a sed statement but I have to change the statement for each different iteration of the same problem.  I don't know perl well enough to figure this out, although I've learned quite a bit trying to get a solution.  Any help would be greatly appreciated.                Jon Jon Zimmer <a href=mailto:b0rg@pcgeek.net>b0rg@pcgeek.net</a> <a href= > </a> You WILL be assimilated, or we can write a script to parse you out.

ElgisRamon · Jul 17, 2000

Try a head, a tail command and 2 (two) cut commands: AllOutput=$(lspv -l hdisk #¦ head -1) Header=$(echo $AllOutput¦head -1) Body=$(echo $AllOutput¦tail +2) FirstPartOfBody=$(echo $Body¦cut -f1 -d/) SecondPartOfBody=$(echo $Body¦cut -f2- -d/) Now you have All the data before the / in FirstPartOfBody and all the data after the / in SecondPartOfBody... I hope it works...

Caliban · Jul 24, 2000

Since the first slash of the mountpoint is the first slash of the output line, a simple: sed 's/\/;\//' should work.

unixxxx · Oct 20, 2000

I have question about sed. If I have a file and I want to delete all lines in the file except the lines that start with the word "HELLO". How should I make a script using sed command? [sig][/sig]

vti · Feb 26, 2001

hi
I have a big problem,maybe it's really easy for you .I have a data like :

37
94
176
209
262
320
357
375
421

there is two blanks before each line .I want to put
kill -9 before each line but i don't know how to do?
Could you please help me?

Regards

Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

The sed command

Programmer

MIS

MIS

MIS

MIS

MIS

MIS

MIS

MIS

Programmer

MIS

MIS

MIS

Programmer

MIS

New member

MIS

MIS

MIS

Technical User

Similar threads

Log in

Part and Inventory Search

Sponsor