CODEWITHSUNDEEP
regexlinuxsed
NarūnasK
NarūnasK

Reputation: 4950

Indent line ranges with sed?

I'm trying here to convert old fashion phpBB code blocks to MARKDOWN using sed.

Please consider following data sample:

cat sed.txt
[code]xxxx-YYY-xxxx[/code]
Some text
[code]yyyy-ZZZ-yyyy[/code]
More text
Bogus code block[/code]
[code]zzzz-XXX-zzzz[/code]

After long trial and error I've ended up with the following strategy:

sed -ne '
  /\[code\].*\[\/code\]/ {
  s#\[/*code\]##g
  s#^#\n\n    #
  s#$#\n\n#p
}' sed.txt | cat -Av
$
$
    xxxx-YYY-xxxx$
$
$
$
$
    yyyy-ZZZ-yyyy$
$
$
$
$
    zzzz-XXX-zzzz$
$
$

This works great, however I find it would be easier and seem more natural to do it this way:

sed -ne '
/\[code\].*\[\/code\]/ {
  s#\[/*code\]#\n\n#g
  s#^#    #p
}' sed.txt | cat -Av
    $
$
xxxx-YYY-xxxx$
$
$
    $
$
yyyy-ZZZ-yyyy$
$
$
    $
$
zzzz-XXX-zzzz$
$
$

But that does not work as expected. Any suggestions why, how to get around this?

Thank you

Upvotes: 0

Views: 445

Answers (2)

potong
potong

Reputation: 58381

This might work for you (GNU sed):

sed -nr 's/^\[(code\])(.*)\[\/\1$/\n\n    \2\n\n/p' file | sed -n l

N.B. In your script you prepend 2 newlines to the beginning of the pattern space and then prepend 4 spaces again, thus the indentation is added infront of the first of the newlines not infront of the text.

Upvotes: 0

NeronLeVelu
NeronLeVelu

Reputation: 10039

sed '/\[code\].*\[\/code\]/ {
   s#\[code]#&    #g
   s#\[/*code\]#\
\
#g
   }' sed.txt

order of substitution is important and changed between your two sample

I also change a bit the behavior, the -n and p are not needed in this text sample (but maybe if coming from a biggest structure) (test on my aix so posix version)

Upvotes: 2

Related Questions