问题
I need to move the contents of every second line up to the line above such that line2's data is alongside line1's, either comma or space separated works.
Input:
line1
line2
line3
line4
Output:
line1 line2
line3 line4
I've been doing it in vim with a simple recording but vim seems to crash when I tell it to do it 100 000 times... I'm thinking maybe sed would be a good alternative but not sure how to do what I want or maybe there's a better option?
Each line only contains 1 numerical value, I just have a million lines...
回答1:
If I understand correctly, you have:
line1
line2
line3
line4
...
and you want:
line1<SEP>line2
line3<SEP>line4
then you can do it easily with (g)awk like this:
awk 'NR % 2 == 1 { o=$0 ; next } { print o "<sep>" $0 }' INPUTFILE
See it in action here.
Update: if the number of lines is odd, the above will omit the last line (as Martin Stettner pointed out) so this will not:
awk 'NR % 2 == 1 { o=$0 ; next } { print o "<sep>" $0 } END { if ( NR % 2 == 1 ) { print o } }' INPUTFILE
HTH
回答2:
The paste command can do this. Its "-s" option will join consecutive lines; and the "-d" option specifies a list of characters to use as delimiters, repeating them cyclically. Join first with a space, then with a newline, and repeat:
seq 10 | paste -sd" \n" -
回答3:
try this:
sed -rn 'N;s/\n/ /;p' yourFile
test with seq:
kent$ seq 10
1
2
3
4
5
6
7
8
9
10
kent$ seq 10|sed -rn 'N;s/\n/ /;p'
1 2
3 4
5 6
7 8
9 10
awk works too:
awk 'NR%2{printf $0" ";next;}1' yourFile
test
kent$ seq 10|awk 'NR%2{printf $0" ";next;}1'
1 2
3 4
5 6
7 8
9 10
回答4:
Well your example is this in Vim.
:g/^/+t.|-j
But then what about the last line?
Or did you mean this?
:g/^/j
You might also be interested in this Vim script, which makes dealing with large files easier.
http://www.vim.org/scripts/script.php?script_id=1506
回答5:
This might work for you:
sed 'N;s/\n/ /' file
Or
cat file | paste -d' ' - -
Or another couple of ways for the above:
paste -d\ - - <file
paste -sd' \n' file
回答6:
$ seq 10 | sed '2~2G' | awk -v RS='' '{$1=$1; print}'
1 2
3 4
5 6
7 8
9 10
$ paste -d' ' <(sed -n 'p;n' num.txt) <(sed -n 'n;p' num.txt)
1 2
3 4
5 6
7 8
9 10
$ echo -e 'g/^/,+1j\n%p' | ex num.txt
1 2
3 4
5 6
7 8
9 10
$ seq 10 | awk 'NR%2{printf("%s ", $0); next}1'
1 2
3 4
5 6
7 8
9 10
$ seq 10 | sed 'N;s/\n/ /'
1 2
3 4
5 6
7 8
9 10
note: $ seq 10 >num.txt
回答7:
seq 10 | awk 'ORS=NR%2?FS:RS'
This solution uses "ternary operator" to set ORS
ORS= ....... output register separator (will receive =)
NR%2 ....... test if it has division remainder of Number of Register by 2
?FS:RS ..... FS = "space" RS = "\n" (newline)
回答8:
You may utilize xargs for this.
Usually xargs takes as many input elements as possible and executes some command with the elements read as arguments. E.g.
cat file | xargs
would echo
line1 line2 line3 line4
But you can limit the number of lines read with option -n
cat file | xargs -n 2
will have the desired effect of joining every two lines:
line1 line2
line3 line4
If the lines may contain white space, you must specify the input delimiter (newline) explicitly
cat file | xargs -n 2 -d '\n'
And finally, don't use cat this way, say instead
xargs -n 2 -d '\n' <file
or even
xargs -n 2 -d '\n' -a file
来源：https://stackoverflow.com/questions/8545538/how-do-i-join-pairs-of-consecutive-lines-in-a-large-file-1-million-lines-using