Append lines which do not contain "|" to the previous line

One way is to use Awk to implement the following algorithm:

• Track the previous line in prev
• If the line contains |, and it's not the first line, then print prev. After that, store the current line in prev
• If the line doesn't contain |, then append it to prev
• At the end of the script, print prev

For example like this:

awk '/\|/ { if (NR > 1) print prev; prev=$0 } !/\|/ { prev = prev $0 } END { print prev }' input 

Use | as the field separator: if the line contains a | then the NF variable will be greater than one.

awk -F'|' 'NR > 1 && NF > 1 {print ""} {printf "%s", $0} END {print ""}' file 

awk '/\|/ { if (printed==1) print ""; else printed=1; printf "%s",$0; next; }; { printf " %s",$0 }; END { print ""; }' inputfile 

Or, if you don't care about the leading newline, shorter:

awk '/\|/ { printf "\n%s",$0; next; }; { printf " %s",$0 }; END { print ""; }' inputfile 

Awk is my go-to utility for text manipulation:

awk '/\|/ && NR>1 { printf "\n" } !/\|/ && NR>1 { printf " " } { printf $0 } END { printf "\n" }' file 

1. Print a newline character if the line contains | (but not for the first line):
   /\|/ && NR>1 { printf "\n" }
2. Otherwise, print a space (but not for the first line):
   !/\|/ && NR>1 { printf " " }
3. Print the entire line, without a trailing newline character:
   { printf $0 }
4. Print a newline at the end of the file:
   END { printf "\n" }

One more awk:

awk -F'|' 'NR>1{printf prev (NF>1?"\n":" ")}{prev=$0}END{print prev}' file 

Testing

$ cat file1 1|1|this is test file line1 2|2|this is test file line2 3|3|this is test file line3 4|4|this is test file line4 5|5|this is test file line5 6|6|this is test file line6 $ awk -F'|' 'NR>1{printf prev (NF>1?"\n":" ")}{prev=$0}END{print prev}' file1 1|1|this is test file line1 2|2|this is test file line2 3|3|this is test file line3 4|4|this is test file line4 5|5|this is test file line5 6|6|this is test file line6 

I would do this using Perl:

perl -e '$f=1;while(<>){chomp;print(($f?"":/\|/?"\n":" ").$_);$f=0}' 

Alternatively, you can use this sed script:

sed -E ':s;N;s/\n([^\n|]*)$/ \1/;bs' 

It is a loop that

• N reads a line and appends it to the pattern space
• s/\n([^\n|]*)$/ \1/ if the line in the pattern space does not contain |, replace the newline with a space
• bs jump back to the :s label