Shell Programming and Scripting

Don,

I had essentially pointed out the same thing as you, but the data in my example didn't show up properly aligned. After a few attempts, I removed it. The OP probably noticed the post while it was still there for a few minutes.

Code:

$
$
$ cat f79
egy dfdsf ketto sada
alma:a korte*szilva narancs
$
$ sed 's/^\([a-zA-Z]\+\)\([^a-zA-Z]\+\)\([a-zA-Z]\+\)\([^a-zA-Z]\+\)\([a-zA-Z]\+\)\(.*\)/\5\2\3\4\1\6/' f79
ketto dfdsf egy sada
korte:a alma*szilva narancs
$
$

Although, if the separators are known and the actual words will only consist of "a-zA-Z" (no numbers or underscores, for instance), then you could do something like this -

Code:

$
$ sed 's/^\([a-zA-Z]\+\)\([* ,:]\+\)\([a-zA-Z]\+\)\([* ,:]\+\)\([a-zA-Z]\+\)\(.*\)/\5\2\3\4\1\6/' f79
ketto dfdsf egy sada
korte:a alma*szilva narancs
$
$

durden_tyler posted an answer, but tis answer was deleted at the time...or I'm not sure about what happened with that.
And sorry, I'm verry tired I made a terrible mistake...your version is correct.
so finally : alma:a korte*szilva narancs need to be changed to: korte:a alma*szilva narancs

thanks for understanding, and again, apologies for my bad english

---------- Post updated at 08:39 PM ---------- Previous update was at 08:07 PM ----------

Thank you durden_tyler! Your first example works perfectly!

Applying the specification literally, I'd propoese

Code:

$ cat file
korte:a alma*szilva narancs
$ sed -r 's/([[:alpha:]]+)([^[:alpha:]]+)([[:alpha:]]+)([^[:alpha:]]+)([[:alpha:]]+)/\5\2\3\4\1/' file
alma:a korte*szilva narancs

Hoppla - should have read the entire thread - still I think this one will respect locale settings.

T720,
Your specification is not clear as to whether empty strings of alphabetic characters constitute a field. (If the first character on a line is not alphabetic, is the 1st field empty? If there are two non-alphabetic characters together, is there an empty field between them?)

The script durden_tyler provided assumes that there are no empty fields and that all sequences of one or more non-alphabetic characters count as a single field separator. (It also assumes that + is a special character in a sed RE acting to specify one or more of the previous RE. Some versions of sed do that, but it is not portable and does not meet the requirements of the POSIX standards and the Single UNIX Specifications.)

The following sed script will do what you want if empty strings are allowed and each non-alphabetic character is to be treated as a separator:
When given the input:

Code:

egy dfdsf ketto sada
alma:a korte*szilva narancs
*save3rd remainder
first$second|third\fourth
1x23no change

it produces the output:

Code:

ketto dfdsf egy sada
korte:a alma*szilva narancs
rd*save3 remainder
third$second|first\fourth
1x23no change