| |
| |
| |
|
Page: 1 2
Comments:
<0> where can i find the exact syntax for what chars are escaped vs non-escaped in sed regex?
<1> it's plain BRE syntax. you can find it everywhere
<1> don't mix it up with ERE, it's different
<2> PCRE !
<1> and don't get confused by GNU sed interpreting \+ in BRE - that's GNU, not BRE ;)
<3> strange thing, theres a file where i did s/aa$/laalaaa/ and the file actually shrinked in file size..
<0> kaneuneu, so how do i figure out what regex my sed supports?
<0> hahah, talkin to myself :)
<3> :)
<4> kaneuneu: sed --version
<2> it support BRE by default
<0> goldfish, yes i know that
<0> goldfish, i read the faq
<0> and i read the link to the BRE documentation including the ERE documentation
<0> GNU sed version 4.1.5
<2> ok.
<5> kaneuneu: Are you escaping the contents of a shell variable for use in a pattern?
<0> not that i know of why
<0> i'm doing search/replace in multiple files with bash/find/sed on the command line
<0> i just wanted to know the exact syntax of regex's in sed
<0> or at least in my version
<5> man sed I suppose. :)
<5> or info sed or http://www.opengroup.org/onlinepubs/009695399/utilities/sed.html
<0> um, man sed not so much
<2> kaneuneu: http://wooledge.org/mywiki/BashFaq#faq21 -- How can I replace a string with another string in all files?
<2> might also help you.
<5> BRE is simple. ^ is special at the beginning of a pattern, $ is special at the end. dot . star * and bracket [ are special, and backslash is special.
<5> That's it
<5> Actually, * is not special if it is the first character of a (sub) expression.
<5> And the rules inside [] are different.
<6> grep -Rh blah . | sed -e 's/(.*)/\1/g'
<6> sed: -e expression #1, char 11: invalid reference \1 on `s' command's RHS
<6> this is just a simple case..does someone see what's wrong w/ the syntax?
<7> aib: are you sure the grep part outputs anything?
<7> usually what sed says there means that it didnt match anything, it doesnt have a \1 becuase it didnt match anything.
<6> grep won't always output anything. ok
<2> aib: sed -e 's/\(.*\)/\1/g'
<2> although
<2> that's a pretty useless command.
<6> yeah it's not what i'm doing, same idea tho =)
<7> i thought -e in this example was "use extended regex" :)
<2> nope.
<2> -e script, --expression=script
<2> add the script to the commands to be executed
<2> -r, --regexp-extended
<2> use extended regular expressions in the script.
<7> in freebsd -E is extended re :)
<2> you could also use, s/.*/&/g
<7> aib: did it work out though?
<6> not yet
<2> what's the problem?
<7> i think his command line or grep comman disnt performing. he will have to spank it really hard to get some out of it >_<
<6> echo blah | sed 's/()/\1/' reproduces it...wonder why this is happening
<8> aib: like goldfish said you have to use \( and \) if you want to back reference things
<8> (if you use the 'defualt' RE)
<6> ahh
<7> thats true
<6> sorry i missed that suggestion
<6> works=)
<7> heh
<7> or use -r for extended regex
<7> or is that not preferred by the sed community? =)
<7> just asking ^^
<2> Probably not.
<2> Who cares :)
<7> hehe
<6> it works but it's pretty quirky. in the actual RE it won't allow me to do it w/ -r, have to escape
<7> hm
<7> echo blah | sed -E 's/()/\1/' => blah
<7> (-E equal to -r)
<1> rawtaz_: the keyword is not "community", it's "standard software" ;)
<7> =)
<9> does command D act like a loop in sed?
<9> in certain cases
<9> what is the meaning of empty regular expression?
<9> //
<9> match previous regex?
<9> or match previous matched regex
<7> no
<7> matches nothing
<9> then why does sed give this error:
<9> sed: -e expression #1, char 0: No previous regular expression
<9> on an expression sed -n '// p'
<9> i am trying to understand what it does
<2> // is a line address
<9> and which address does it match?
<2> it doesn't match any, afaik
<9> I am trying to demistify this sed expression: sed '/\n/!G;s/\(.\)\(.*\n\)/&\2\1/;//D;s/.//'
<9> which reverses the lines
<9> acts like `rev'
<2> read http://www.grymoire.com/Unix/Sed.html#uh-0
<9> I know all the sed commands etc.
<2> hm, ok.
<9> i know sed very good
<9> but it is hard to understand this one
<9> because of the //D command
<2> where did you get teh command from?
<9> i am going through all the sed one liners
<9> http://sed.sourceforge.net/sed1line.txt
<9> or http://www.pement.org/sed/sed1line.txt
<9> to boost my skills of sed a little more
<9> and came accross that one which is difficult to grasp
<9> can anyone explain it to me?
<2> I bet prec can.
<9> prec, you there?
<2> I haven't played with G or D, i'm afraid.
<9> oh, they are very easy, read http://www.grymoire.com/Unix/Sed.html#uh-0
<9> :)
<2> :)
<9> G appends whatever is in hold space to pattern space
<9> in this case it is always new line
<9> and D deletes everything from pattern space up until \n
<9> i remember there were examples of commands
<9> and maybe there was this one
<9> any sed guru in here?
<7> pkrumins: how will you know unless you ask them a question which wil reveal their true skillz? ;)
<7> (hint: im not the one, though)
<9> I am trying to demistify this sed expression: sed '/\n/!G;s/\(.\)\(.*\n\)/&\2\1/;//D;s/.//'
<9> what it does it acts as `rev'
<7> ooh that one.
<9> yes, the same one
<7> i have no idea really, but if i were to give it a shot, id start with "for each line" (since it matches first on \n). then something.. :)
<9> misteriously it doesnt match first on \n
<9> echo "test" | sed -n '/\n/ p'
<9> does not print anything
<10> I believe it will always execute G?
<7> its an interesting line
<10> hmm... although
<9> iSteve, I am not sure
<10> pkrumins: G will append
<9> i know all the commands, iSteve
<10> pkrumins: so if you get two lines, you get a metch
<10> match
<10> yea yea you said that
<9> tell me about //D
<10> and do you hear me arguing?
<10> //D keeps me confused
<10> it shouldn't work:)
<9> now i hear you arguing
<9> iSteve, what about echo "test" | sed -n '/\n/ p'
<9> why doesnt it print
<10> you heard me arguing? now you hear me saying 'good luck with your problem'
<9> okay, thanks
<7> sed is really interesting
<7> so is awk
<9> if /\n/ matched then the test example should have printed
<11> iSteve: '//D' is supposed to take the last matched pattern ('//') and delete thing to a first new line starting a new cycle with the rest of pattern space ('D')
Return to
#sed
or
Go to some related
logs:
amount of cafeine in tea
+preseed +partman +primary partitions
#linux
ktorrent 2gb
1684370019
emarge command not found
isgenerator type python
ubuntu iproute2 ifconfig deprecated
Invalid object definition type 'hostextinfo'
#suse
|
|