regex - sed -e to display only a block of text - can you explain the command? -

regex - sed -e to display only a block of text - can you explain the command? -

i've managed build sed command accomplish need help of colleagues nobody can explain why works!

can advise please?

my text file is:

this test configuratoin first test dn: cn=configuration line1 can line2 can linex can end of story no end of story

my sed command is:

sed -e '/./{h;$!d;}' -e 'x;/dn: cn=configuration/!d'

the output is:

dn: cn=configuration line1 can line2 can linex can

i understand h = hold (append newline, append contents of pattern space, hold space) , {} allows multiple commands.

any guidance, much appreciated.

first analysis

first, expressions given sed through -e alternative can grouped single script*, so

sed -e '/./{h;$!d;}' -e 'x;/dn: cn=configuration/!d' my_file

is equivalent to

sed -e '/./{h;$!d;};x;/dn: cn=configuration/!d' my_file

now let's split see how works:

sed -e '/./{ # non-empty lines (containing @ to the lowest degree 1 char): { h; # happend line hold space, $!d; # , if not lastly line of file, delete it. # go next line of input. }; # } x; # reach point empty lines , lastly line of file # “x” command swaps hold , pattern spaces /dn: cn=configuration/!d # if pattern space not contain # “dn: cn=configuration”, delete it. ' my_file

! means: “perform command on lines not matching preceding pattern”.

*at to the lowest degree gnu sed, don't know other versions.

more details

/./{h;$!d;} non-empty lines, save them in sed hold space, , delete pattern space (except lastly line). go next line of input (the d command ends treatment of current line).

x when meet empty line, lastly line of file (empty or not), swap hold space (which contains saved non-empty lines) , current pattern space. after swapping, saved lines in pattern space, , hold space empty.

/dn: cn=configuration/!d @ pattern space (i.e. saved lines): contain, on line, string dn: cn=configuration? if no, delete pattern space (and both hold , pattern spaces blank). if yes, nothing, print pattern space (i.e. saved lines). loop line of input.

putting of together, have: each stanza containing no empty lines, save hold space; on empty line (or @ end of file), content hold space, , search dn: cn=configuration string inside. if stanza not contain string, delete it; else, print (default behaviour of sed).

sum , conclude

so script prints stanzas containing string dn: cn=configuration. remove blank line @ top of output , forcefulness string on first line of stanza with:

sed -e '/./{h;$!d;};x;s/^\n//;/^dn: cn=configuration/!d' my_file # changes: ^^^^^^^ ^

(s/^\n// remove first empty line in pattern space, ^ anchor string @ origin of stanza).

don't hesitate inquire more details in comments if remains unclear.

regex bash sed edit ksh