Can sed replace multiple occurrences in a single line?

halfpower · 09-14-2019, 12:21 PM

I have the following string:

Code:

ABCABC

I want to output

Code:

A B C A B C

I've tried doing it this way:

Code:

echo "ABCABC" | sed -E "s:(A)(B)(C)(.*):\1 \2 \3 \4:g"

However, this produces

Code:

A B C ABC

In my problem, the substring ABC can occur N times. Is sed capable of producing the desired output?

Turbocapitalist · 09-14-2019, 12:22 PM

How about this?

Code:

sed -e 's:.:& :g'

Firerat · 09-14-2019, 02:09 PM

another way

Code:

foo="ABCABC"
sed -E "s:([ABC]{1}):\1 :g" <<<$foo
A B C A B C 
sed -E "s:([AB]{1}):\1 :g"  <<<$foo
A B CA B C
sed -E "s:([AC]{1}):\1 :g"  <<<$foo
A BC A BC

danielbmartin · 09-14-2019, 04:42 PM

With this InFile ...

Code:

ABCABC
DEFGHIJK

... this sed ...

Code:

sed 's/./\0 /g' $InFile >$OutFile

... produced this OutFile ...

Code:

A B C A B C 
D E F G H I J K

Daniel B. Martin

.

syg00 · 09-14-2019, 06:05 PM

Adding blank char to the end of line always causes subsequent confusion - especially when using an anchor with regex.

Firerat · 09-14-2019, 07:15 PM

true

Code:

foo="ABCABC"
sed -E "s:([ABC]):\1 :g" <<<$foo | cat -A -
A B C A B C $
sed -E "s:([ABC]):\1 :g;s: $::" <<<$foo | cat -A -
A B C A B C$

I realised I didn't need {1}

MadeInGermany · 09-14-2019, 08:13 PM

As was shown in post#2 you don't need to capture all in a (group) - the sed-special & refers to all that was matched by the search.

Code:

sed 's:.:& :g; s: $::'

\0 might be a GNU-sed specific synonym for the & in a standard-sed.

ondoho · 09-15-2019, 09:06 AM

Quote:

Originally Posted by halfpower

I've tried doing it this way:

Code:

echo "ABCABC" | sed -E "s:(A)(B)(C)(.*):\1 \2 \3 \4:g"

However, this produces

Code:

A B C ABC

Too much matching!

Code:

echo "ABCABC" | sed -E "s:(A)(B)(C):\1 \2 \3 :g"

Better?