CODEWITHSUNDEEP
shellawksedcut
tubby
tubby

Reputation: 2144

split line using string as delimiter from shell

I have a file which looks like

60 -> 36
48 -> 11
60 -> 59
35 -> 29
27 -> 76

I would like to split this file into two separate files called source and destination such that the source file only contains elements before ' -> ' and the destination file, the part after.

I tried using cut as follows

cut -d' -> ' -f1 input > source
cut -d' -> ' -f2 input > destination

But cut gives me this error

cut: the delimiter must be a single character

Upvotes: 0

Views: 3597

Answers (6)

NeronLeVelu
NeronLeVelu

Reputation: 10039

sed '1!H;1h;$!d
g;s/ ->[ 0-9]*//g;w Source
g;s/[ 0-9]* ->//g;w Destination
' YourFile

Write source by removing all end of line, than destination removing head of line after loading the file in memory and used it for each substition

Upvotes: 0

Cyrus
Cyrus

Reputation: 88553

Try to use a whitespace as delimiter: " ". And in second case use -f3.

Or with GNU sed:

sed -ne 'h;s/ .*//w source' -e 'g;s/.* //w destination' input

Upvotes: 1

Ed Morton
Ed Morton

Reputation: 203149

awk '{print $1 > "source"; print $3 > "destination"}' input

Upvotes: 3

dawg
dawg

Reputation: 103714

awk is your best bet for a regex type split.

Given:

$ echo "$tgt"  
60 -> 36
48 -> 11
60 -> 59
35 -> 29
27 -> 76

You can split the input with awk on a regex:

$ echo "$tgt" | awk -F " -> " '{print $1}'
60
48
60
35
27
$ echo "$tgt" | awk -F " -> " '{print $2}'
36
11
59
29
76

And redirect into two files as desired.

Upvotes: 1

Siguza
Siguza

Reputation: 23830

As far as I understand it, cut is incapable of using more than a single character as delimiter.

Using sed, it should work like this:

sed -E 's/^(.*) -> .*$/\1/g' input > source
sed -E 's/^.* -> (.*)$/\1/g' input > destination

Upvotes: 0

josifoski
josifoski

Reputation: 1726

Something like this might be (gnu sed) 

sed 's/^\([^ ]*\)/\1/' <file >source  
sed 's/.* \(.*\)/\1/' <file >destination

Upvotes: 0

Related Questions