Search between two patterns and print the data from first pattern to second pattern multiple times, excluding non-matching blocks

Question

I have to search for a specific string in the reverse order and print it.

In the below example I want to search from the pattern not in order until the pattern number of.

Example: input file contains:

number of characters a[1] 
reg1 
reg2 
reg3 
info a[1] is not in order

number of characters a[3] 
reg1 
reg2 
reg3 
info a[3] is in order


number of characters a[2] 
reg1 
reg2 
reg3 
info a[2] is not in order

output should be:

number of characters a[1]
reg1 
reg2 
reg3 
info a[1] is not in order

number of characters a[2] 
reg1 
reg2 
reg3 
info a[2] is not in order

Answer 1

Using in order with a newline as the paragraph delimiter, we can do:

awk -v RS='in order\n' '/not/{print $0 "in order"}'

awk treats text separated by the pattern in RS (record separator) as records, and each operation is done on a record. So /not/ tests if the record matches not, and then we print the record ($0) along with the separator text, which was removed by awk.

So:

$ mawk -v RS='in order\n' '/not/{print $0 "in order"}' foo
number of characters a[1]
reg1
reg2
reg3
info a[1] is not in order


number of characters a[2]
reg1
reg2
reg3
info a[2] is not in order

Answer 2

If the blocks have to end with a line that contains is in order or is not in order, we can delete things between them...

$ sed '/is not in order/,/is in order/ {/is not in order/n;d}' file
number of characters a[1] 
reg1 
reg2 
reg3 
info a[1] is not in order


number of characters a[2] 
reg1 
reg2 
reg3 
info a[2] is not in order

Notes:

• /is not in order/,/is in order/ find the lines between is not in order and is in order, inclusive
• {some commands} group these commands
• /is not in order/n skip the line with this pattern from the next command
• d delete the specified lines

It might seem convoluted to find the line and then find it again to skip it, but we can't use an empty line instead, or sed will turn on its operate here flag at the next empty line and keep going until the next is in order, which will delete too much.

Answer 3

another approach:

tac file | awk ' BEGIN {weprint=0 ; rem="not necessary, but for clarity"}
  /is not in order$/ { weprint=1 ;}
  ( weprint == 1)    { print $0  ;   rem="same remark here..."; }
  /^number of/       { weprint=0 ;}
' | tac

which could be shortened if needed...

If you want the separator line: change the last line as

"/^number of/" { print ; weprint=0;}

Answer 4

perl -00 -ne 'print if /not in order/' file

The -00 option reads the file by paragraphs.
The -n adds an implicit loop over all the paragraphs in the file.
Then, print the paragraph if it contains the desired text "not in order".