I appreciate your help with this problem. I like to eliminate everything that is not a specific pattern from a string.
For example, below I like to eliminate everything that is not "5TTGTC".
But as seen here ^5TTGTC is not right. I used different combinations of ^(), ^{}, ^[], but none gave me what I am looking for. Appreciate your feedback!
echo ".,..,...+5TTGTC...+5TTGCC.+5TTGTC,,.,.,,.,+5ttgtc,.,,.,.+5TTGTC.+5TTGTC,..+5TTGTC" | sed 's/^5TTGTC//g'
Thanks in advance
You may use the following command if you want case sensitivity:
echo ".,..,...+5TTGTC...+5TTGCC.+5TTGTC,,.,.,,.,+5ttgtc,.,,.,.+5TTGTC.+5TTGTC,..+5TTGTC" | sed -r 's/(5TTGTC)|[,.A-Za-z+0-9]/\1/g'
The code above prints:
5TTGTC5TTGTC5TTGTC5TTGTC5TTGTC
The regular expression used above uses alternation to capture what you are interested in.
We match and capture what we are interested in (5TTGCC) and we match everything that is not the substring, in this case characters ,.A-Za-z+0-9.
You can check the behaviour of the regex here.
As pointed out by #EdMorton, the command can be simplified to:
echo ".,..,...+5TTGTC...+5TTGCC.+5TTGTC,,.,.,,.,+5ttgtc,.,,.,.+5TTGTC.+5TTGTC,..+5TTGTC" | sed -r 's/(5TTGTC)|./\1/g'
You can try this here.
For compatibility across sed versions the -r flag can be replaced by the -E flag.
You don't make it very clear what you are trying to achieve.
One way to get where you are trying to go could be the -o option in grep.
echo ".,..,...+5TTGTC...+5TTGCC.+5TTGTC,,.,.,,.,+5ttgtc,.,,.,.+5TTGTC.+5TTGTC,..+5TTGTC" | grep -o '5TTGTC'
Output:
5TTGTC
5TTGTC
5TTGTC
5TTGTC
5TTGTC
You can then change 5TTGTC into a pattern, e.g. grep -o '[0-9]TT[AG]GTC'
With any sed:
$ echo ".,..,...+5TTGTC...+5TTGCC.+5TTGTC,,.,.,,.,+5ttgtc,.,,.,.+5TTGTC.+5TTGTC,..+5TTGTC" |
sed 's/#//g; s/5TTGTC/#/g; s/[^#]//g; s/#/5TTGTC/g'
5TTGTC5TTGTC5TTGTC5TTGTC5TTGTC
With any awk:
$ echo ".,..,...+5TTGTC...+5TTGCC.+5TTGTC,,.,.,,.,+5ttgtc,.,,.,.+5TTGTC.+5TTGTC,..+5TTGTC" |
awk -v str='5TTGTC' '{gsub(str,"\n"); gsub(/[^\n]/,""); gsub(/\n/,str)}1'
5TTGTC5TTGTC5TTGTC5TTGTC5TTGTC
I am trying to filter a long html page, for leaving only fingerprints which have a consistent structure. for example:
DCD0 5B71 EAB9 4199 527F 44AC DB6B 8C1F 96D8 BF60
i know how to do it by using standrd command line commands as grep, cut and head/tail, but is there more elegant way to do it with sed? the shell comman i use is long and not looking so nice.
thank you
grep is the right tool for extracting strings from a file based on regular expression matching:
grep -Eo '([A-F0-9]{4}[[:space:]]){9}[A-F0-9]{4}' file.html
Here is a sed command tested with GNU sed 4.2.2:
sed -nr '/(([[:xdigit:]]){4} ?){10}/p' file
It matches and prints
10 groups that are made of
4 hexdigits
followed by an optional space
With GNU sed:
sed -E 's/.*(([A-F0-9]{4}[[:space:]]){9}[A-F0-9]{4}).*/\1/' file
I would like to use (GNU) sed to do a simple search and replace. The issue is that I'm searching for a special character and it might be the reason it failed for me.
The input is:
^A9=139^A35=V^A34=9^A49=xxxx^A52=20140527-06:18:43.759^A5
and I want to replace the ^A with ;. I used:
sed -i '/s/^A/;/g' file.log
but I didn't get anything.
Your command should be,
sed -i 's/\^A/;/g' file
Command you tried,
sed -i '/s/^A/;/g' file.log
| |
| |______________You have to escape this special character. Because in general(regex) it means the starting point.
[No need to use `/` before s]
Example:
$ sed 's/\^A/;/g' file
;9=139;35=V;34=9;49=xxxx;52=20140527-06:18:43.759;5
^ has a special meaning with regular expressions. Use \^ (or potentially \\^, depending on how bash escapes things, I never quite remember it).
Does anybody know which version of sed is required to get option -i to work? I am on CentOS5 and I am getting this error.
If you're going to be using -i with sed you're doing it wrong. sed is a stream editor and it should be used to edit streams, not files, as -i wants to do.
If you want to edit a file, you should be using ed. ed is a line editor and it should be used to edit files. IMO, that's the tool you want to be using.
btw, -i is a GNUism. from the wikipedia:
GNU sed added several new features. The best-known is in-place editing of files (i.e., replace the original file with the result of applying the sed program), which was later included in BSD sed too. This feature is nowadays often used instead of ed scripts: for example,
I don't think you can get -i to work then.
I think this other SO question may help you out:
sed -i + what the same option in SOLARIS
Perhaps the solution isn't as nice as sed -i, however.
im trying the following sed command, but i have no luck with special chars:
echo "x#asdf" | sed "s/\([^-]\)#/\1\n/g"
x
asdf
but if i use some special char in test.txt
echo "ä#asdf" | sed "s/\([^-]\)#/\1\n/g"
ä#asdf
why ?
this works:
echo "ü#asdf" | sed "s/ü/-/g"
-#asdf
but this doesnt:
echo "ü#asdf" | sed "s/[ü]/-/g"
ü#asdf
I'm not sure about this, because your sed commands work ok for me (gnu sed 4.1.5), but try invoking sed this way:
$ LANG=de_DE.UTF-8 sed ...
See this post for more information: Why does sed fail with International characters and how to fix?.
If this doesn't work, it may help to upgrade to gnu sed 4.2, if you can. The NEWS file says "multibyte processing fixed" for 4.2 but does not go into further detail.