Delete text between parentheses, but never past empty line

Question

Consider a text file, with lines of text gathered in many blocks, where each block is separated by at least one empty line. Using a Bash one-liner, how do I delete all text from < to either > or \n\n?

To put it differently: Delete everything between each pair of < and >. If a <has no closing >, delete everything until the end of the block (an empty line), but never, ever delete outside the block!

Conceptually, should I physically separate the blocks into objects in a list before parsing for safety, or is this a straight forward linear text parsing job as long as you know what you are doing?

Example text:

This is the first
block of text.
                             <-- empty line
<delete me>
This is the second block.
<delete
here>
<delete this, but
                             <-- empty line
do not delete this>
<delete this too>
Third block here.

(more blocks)

The result should be:

This is the first
block of text.
                             <-- empty line
This is the second block.
                             <-- empty line
do not delete this>
Third block here.

Answer 1

Try awk's paragraph mode:

$ awk -v RS= -v ORS='\n\n' '{gsub(/<[^>]+>?\n?/, "")}1' ip.txt 
This is the first
block of text.

This is the second block.


do not delete this>
Third block here.

• -v RS= this will cause one or more consecutive empty lines to be used as input record separator
• -v ORS='\n\n' set output record separator as two newline characters
• gsub(/<[^>]+>?\n?/, "") delete < followed by non > characters followed by optional > and newline characters
• 1 idiomatic way to print input record contents

Same thing with perl

perl -00 -lpe 'BEGIN{$\="\n\n"} s/<[^>]+>?\n?//g' ip.txt

Answer 2

GNU Awk solution:

awk -v RS='[<>]' '/\n\n/{ sub(/^[^\n]+\n/, ""); print $0 RT }' file

• RS='[<>]' - treat < and > as record separator
• /\n\n/ - if current record contains 2 linebreaks:
  • sub(/^[^\n]+\n/, "") - remove everything till the 1st newline (inclusive)
  • print $0 RT - print current record followed by RT (i.e. >)
  • RT - the record terminator. Gawk sets RT to the input text that matched the character or regular expression specified by RS.

---

The output:

<empty line>   
don't delete this>