| |
| |
| |
|
Comments:
<0> hi all. can someone help me, please? i need create script in sed or awk which find some config line and replace his. for example config file contains CONF1="192.168.0.1,root,,database" and i need replace this text. i am beginner. please help
<0> i tried out=`cat config.cfg|grep CONF1` && sed -e 's/$out/new_configuration/g' > config.cfg but it's not works
<1> if you have gnu sed, the -i option may help
<1> sed -i 's/CONF1=.*/CONF1="some new config"/' config.cfg
<2> but remeber to use backups
<0> BearPerson, Thanks a lot
<1> np
<3> how do I delete all string that have more than 3 characters?
<4> /^....*$/d
<3> doesn't work
<3> no
<3> working =)
<4> :)
<5> llm0: sed 's/[^ ]\{4,\}\s//g' or similar (not optimal, but i'm no expert either)
<5> does not delete work at the end of a line. hm
<5> sed 's/\<[^ ]\{4,\}\> *//g' file maybe somebit better
<6> hi everyone
<4> hi
<7> is something like: "sed -i.BAK 's|foo|bar|g' *" valid for all files in a directory?
<8> depends on your shell's interpretion of "*", but usually yes ;)
<7> and to do two find/replaces at once, can I do this: "sed -i 's|foo|bar|g' -i 's|foo2|bar2|g' *" ?
<8> -i 1. works only on GNU sed, as sed is and always was a STREAM editor, and 2. i doubt you can use it per-expression
<7> ah.. nevermind. it's "sed -i -e 's|foo|bar|g' -e 's|foo2|bar2|g' files"
<8> if you want to edit text files, why don't you use a file editor, rathen than a stream editor?
<7> because i want to make two changes to 100 files at once. :)
<8> -e 's|foo|bar|g;s|foo2|bar2|g' ?
<8> with "use a file editor" i didn't mean "use a file editor you can't script"
<8> i only said "use a file editor"
<8> ed(1)
<8> ex(1)
<8> even vim(1) is scriptable
<8> sed on multiple files raises problems, sometimes. if it works for you, okay.
<9> hi
<9> I'm attempting to parse a large data file (in this format http://paste.lisp.org/display/25822). Any ideas how I could transform this into CSV format? Any help would be appreciated.
<10> rellis: freshmeat.net search: convert cvs ?
<9> That's way too obvious, I don't like it =p
<10> rellis: hehe, ok, lets obfuscate it then ;)
<9> hehe
<9> Hmmm, I ran that search a couple different ways but I don't seem to see any projects involving reformatting data like that.
<10> rellis: ok, i dont know what cvs format looks like, html?
<9> oh, csv is comma seperated variables
<9> like... company name, address, phone number, work number, more info
<9> instead of how it is in the link i posted
<10> rellis: ok, if you post a simple example line of text for before and after sed edit, we might try it
<9> alright, let me conjur one up real quick
<10> ok
<9> gnubien: http://paste.lisp.org/display/25823
<10> ok
<9> Did that better explain what I'm trying to do?
<9> Trying to get it in some format in which I can import into a DB, csv just seems the simplest.
<10> rellis: why the double ,, after one phone number?
<9> because the field phone_main2 is blank
<10> ok, i see
<9> on the second line i accidently put ,, before unknown
<9> that should be 1 comma
<10> ok
<10> rellis: easiest first step is to ask in #bash to see if anyone has done it before which i have not; looks like its do-able but complex
<9> Alright, thanks.
<9> I'll hit them up.
<10> rellis: ask: anyone converted a text file to cvs format? or the like someone may have a faq on it already
<9> Okay, cool. Ya, it seemed like something someone may have already needed to do to me as well...
<10> rellis: yea, be courteous and you will get some info hopefully
<10> rellis: wait patiently for 5 minutes or so to see if anyone has a brain-storm ;)
<10> rellis: if your doing this for homework i want beer and pizza emailed to me first ;)
<10> rellis: delete all newlines: sed -e :1 -e '/"$/!{N;s/\n/ /;b1' -e '}' filename this will make the file all one line
<10> rellis: the insert /n at the "Company Name" to break it into lines with every company name's info all on one line
<10> rellis: then sed the lines to the format you need, tada ;)
<9> I'm being very courteous, it seems like everyone in that channel is twelve years old =/
<9> heh
<10> rellis: thats if you need to use sed exclusively
<10> rellis: yea, i threw you to the lions ;) thats why i said be courteous
<9> hehe
<9> always...
<10> rellis: they are offering some good ideas and a few abuses to spice up the mix ;)
<9> gnubien: Hehe, ya.
<9> gnubien: It's IRC, I think a thick skin is a requirement.
<10> rellis: yes it is
<9> gnubien: Oh, and this is definitely not homework.
<9> heh
<10> oops, no beer and pizza for me then
<9> hehe
<10> rellis: sed -e :1 -e '/"$/!{N;s/\n/ /;b1' -e '}' filename |sed 's/COMPANY NAME/\nCOMPANY NAME/g' #all company name data should then be on one line
<10> rellis: sed -e :1 -e '/"$/!{N;s/\n/ /;b1' -e '}' cvs.conv |sed 's/COMPANY NAME/\nCOMPANY NAME/g' |sed 's/COMPANY NAME\|Main:\|Fax:\|Member Count://' #removes the line titles
<10> cvs.conv is my test file
<10> rellis: you a total newbie at sed?
<10> rellis http://www.student.northpark.edu/pemente/sed/sed1line.txt AND sed.sf.net
<11> how do i match cpu and not cpus? using something like s/^cpu$/name/g doesn't seem to work
<10> hell``: echo "match cpu and not cpus" |sed '/cpu[^s]/s/cpu/CPU/' #match CPU and not cpus
<9> gnubien: Sorry, I had a quick staff meeting =/
<11> gnubien: is the first '/cpu[^s]/' there telling it what not to match?
<9> gnubien: I'm going to play around with a bit more and get the addresses to always be on two lines, then work with it from there.
<9> gnubien: Having the data formatted as it is makes it much more difficult than it should be, I think.
<10> hell``: yes, anything with the [^something] means not this
<10> hell``: the caret in the box means: not this
<11> gnubien: i understand that but i mean't how you have it setup at the beginning of the sed line
<10> rellis: if you can use awk you can selective print the words you want and use the , seprartor too
<11> gnubien instead of something like s/blah/this/g you have /cpus[^s]/s/this/blah/g
<10> hell``: post an example line of text before and after the sed edit
<9> gnubien: Thanks for the help :)
<10> rellis: sure, have fun ;)
<11> gnubien: what you have is correct and working for me, i'm just trying to understand how you have it written, im used to seeing something like 'sed s/this/now/g' and you have this other part '/cpus[^s]/s/this/now' at the beginning
<11> so im guessing cpus[^s] at the beginning is telling it what not to match
<10> hell``: ok, sed '/cpus[^s]/s/this/now' means: search for regex cpus but not "cpuss" and if you find it substitute the word "this" for "that" on the line that you found "cpus"
<10> hell``: so you probably want sed '/cpu[^s]/s/this/now' which searches for "cpu" but not "cpus"
<11> ok i understand now
<11> thank you
<10> hell``: sure, email me some beer and pizza if thats homework ;)
<11> lol
<11> no i completed my homework a couple years ago :)
<10> hell``: http://www.student.northpark.edu/pemente/sed/sed1line.txt AND sed.sf.net
<11> gnubien: so you could do: s/cpu[^s]/this/g and it should do the same thing?
<10> hell``: s/cpu[^s]/this/that/g works, you post creates: echo "match cpu and not cpus" |sed 's/cpu[^s]/this/g' #match thisand not cpus
<11> ah
<10> hell``: http://www.grymoire.com/Unix/Sed.html
<9> Is there a way using sed to match a pattern and if the line matches that pattern merge it with the line directly above it?
<12> rellis: there is
<12> using the hold space for the previous line or something
<12> afaik, it's in eric pement's one liners
<12> or, grymoire's url
<12> http://www.grymoire.com/Unix/Sed.html#uh-0
<12> and google eric pement sed , look for his one-liners page
<13> how can i append text to the end of each line ? a\ doesn't work because it creates a new line .
Return to
#sed
or
Go to some related
logs:
#perl
install vmware debian etch
#web
fc5 configure lirc
#php
#linux
#ldap
#css
find symbolic links linux
xorg.conf tridentfb
|
|