What shell script should I use in Linux to replace a group with n lines of text with a single line?
I have a file like :
a
b
c
*
d
e
f
*
g
h
i
*
and I want to get a file as:
abc
def
ghi
-
Something like 'i=0; while read line; do test $((++i))', but I haven't found any solutionfamedoro– famedoro2014-01-10 09:08:56 +00:00Commented Jan 10, 2014 at 9:08
-
I just saw your answer stackoverflow.com/a/21042715/1983854 . Why don't you give proper input? All the approaches were based on your input which is very far from the one you are using in your answer.fedorqui– fedorqui2014-01-10 11:17:55 +00:00Commented Jan 10, 2014 at 11:17
-
1@fedorqui and besides, his/her answer is the longest and least understandable of all :)holgac– holgac2014-01-10 11:44:56 +00:00Commented Jan 10, 2014 at 11:44
Add a comment
|
8 Answers
sed way:
sed ':a;N;$!ba;s/\n//g' < t | sed 's:*:\n:g'
t is the file you want to change.
references: How can I replace a newline (\n) using sed? Why does sed not replace all occurrences?
The first command replaces \n with nothing. the second replaces * with \n.
sed is a very powerful stream editor tool by the way, knowing it can help you in more ways than you can imagine.
Comments
can use awk
awk '{if ($0=="*"){print s;s=""}else{s=s$0}}' file
an bash way to this is
while read x
do
[ "$x" == "*" ] && echo || echo -n $x
done < file
Comments
Other than awk solution :
tr '\n' ' ' < Input.txt |sed 's/ //g' | tr '*' '\n'
Comments
One awk way:
$ awk '/\*/ {printf "\n"; next} {printf $1}' file
abc
def
ghi
- In case of
*, print a new line. Otherwise, print the character. Usingprintf, so that new line is not added unless specified.
With xargs:
$ xargs -n4 < file | cut -d' ' -f1-3
a b c
d e f
g h i
xargs -n4 < fileprints the file in blocks of 4 columns.cut -d' ' -f1-3prints from 1st to 3rd column.
Comments
If you treat * as the line token, try remove '\n' and replace STAR.
cat <youfile> | tr -d '\n' | tr '*' '\n'
1 Comment
ray
instead of using 2 pipes, you could have said
tr -d '\n' < file | tr '*' '\n'You can do it in the shell without awk, or tr, or anything external too:
#!/bin/bash
while read x
do
if [ "$x" == "*" ]; then
echo $line
line=""
else
line="$line$x"
fi
done
Save as "go", then do:
chmod +x go
./go <inputfile
Comments
I believe tr could be considered the correct tool to replace single characters.
tr -d '\n' < file | tr '*' '\n'
Correct in this context would mean most lightweight, least clumsy, least bug-prone, and least cognitively demanding for the purpose. The other solutions work, but tr has the least hidden costs. When you use a lot of small tools as you do in unix, these small advantages really add up over time.
Comments
Thanks to everybody, I've found this solution which works in case the file is like
field1 2008
field2 703
field3 qeew
field4 80
field5 52
field6 749.11
field1 2009
field2 532
field3 wecwer
field4 82
field5 44
field6 649.97
...
#!/bin/bash
if [ "$1" = "" ]
then
echo "missing file name"
exit 1
fi
N=7
i=0
FULL_ROW=""
SAVEIFS=$IFS
if [ -f $1 -a -r $1 ]; then
LAST_DATA_SCAD=""
while IFS=$(echo -en "|\n\b") read -ra ROW_RECORD; do
if [[ "$FULL_ROW" == "" ]]
then
FULL_ROW=${ROW_RECORD:11}
else
FULL_ROW=$FULL_ROW"|"${ROW_RECORD:11}
fi
i=$((i+1))
if [ $((10#$i)) == $N ]
then
echo $FULL_ROW
FULL_ROW=""
i=0
fi
done < "$1"
else
echo "file not found"
fi
IFS=$SAVEIFS
Comments
