Sed replace the second last line of the file

Question

I want to replace the second last line of the file, I know what to use $ last for the last line, but I don’t know how to say the second line from the end.

parallel ( { ignore(FAILURE) { build( "Build2Test", BUILDFILE: "", WARFILE: "http://maven.example.com/130602.0.war", STUDY: "UK", BUG: "33323" ) }}, ) 

I want to replace }}, with }} short one, I want to remove the comma , but there are many other codes in this file, so I can not use pattern matching. I need to use the second line from the end of the file.

Answer 1

The following should work (note that on some systems you may need to delete all comments):

 sed '1 { # if this is the first line h # copy to hold space d # delete pattern space and return to start } /^}},$/ { # if this line matches regex /^}},$/ x # exchange pattern and hold space b # print pattern space and return to start } H # append line to hold space $ { # if this is the last line x # exchange pattern and hold space s/^}},/}}/ # replace "}}," at start of pattern space with "}}" b # print pattern space and return to start } d # delete pattern space and return to start' 

Or compact version:

 sed '1{h;d};/^}},$/{x;b};H;${x;s/^}},/}}/;b};d' 

Example:

 $ echo 'parallel ( { ignore(FAILURE) { build( "Build2Test", BUILDFILE: "", WARFILE: "http://maven.example.com/130602.0.war", STUDY: "UK", BUG: "33323" ) }}, )' | sed '1{h;d};/^}},$/{x;b};H;${x;s/^}},/}}/;b};d' parallel ( { ignore(FAILURE) { build( "Build2Test", BUILDFILE: "", WARFILE: "http://maven.example.com/130602.0.war", STUDY: "UK", BUG: "33323" ) }} ) 

Answer 2

If you know how to change the Nth line, just format the file first, for example. This is not as many professionals as other sed solutions, but it works ... :)

 tail -r <file | sed '2s/}},/}}/' | tail -r >newfile 

eg. from next input

 }}, }}, }}, }}, }}, 

the above does

 }}, }}, }}, }} }}, 

tail -r are the BSD equivalent of the Linux tac command. On Linux, use tac for OS X or Freebsd using tail -r . The bot does the same: it prints the file in the revered line order (the last line prints as the first).

Answer 3

cancel the file, work with the second line, then flip the file:

 tac file | sed '2 s/,$//' | tac 

to save the result back to the "file", add it to the command

  > file.new && mv file file.bak && mv file.new file 

Or use ed script

 ed file <<END $-1 s/,$// w q END 

Answer 4

This may work for you (GNU sed):

 sed '$!N;$s/}},/}}/;P;D' file 

Save two lines in the template space and at the end of the file, replace the required template.