shellunixawksed

Remove multiple lines from a given match until an empty line

On a unix system I have a file and I need to remove all lines that would match Package: <PKG_NAME> until an empty line is found. Here is an example, I would need to remove terminfo information:

...
Package: kmod-usb-storage
Version: 5.4.218-1
Depends: kernel (= 5.4.218-1-0c02597a113d34441a9bfe9294e3fb84), kmod-scsi-core, kmod-usb-core
Status: install user installed
Architecture: mips_24kc
Installed-Time: 1667822688
Auto-Installed: yes

Package: terminfo
Version: 6.2-1
Depends: libc
Status: install user installed
Architecture: mips_24kc
Installed-Time: 1667816896

Package: libuci-lua
Version: 2020-10-06-52bbc99f-5
Depends: libc, libuci20130104, liblua5.1.5
Status: install user installed
Architecture: mips_24kc
Installed-Time: 1667816896
Auto-Installed: yes
...

Afterwards I need to have no information about terminfo:

...
Package: kmod-usb-storage
Version: 5.4.218-1
Depends: kernel (= 5.4.218-1-0c02597a113d34441a9bfe9294e3fb84), kmod-scsi-core, kmod-usb-core
Status: install user installed
Architecture: mips_24kc
Installed-Time: 1667822688
Auto-Installed: yes

Package: libuci-lua
Version: 2020-10-06-52bbc99f-5
Depends: libc, libuci20130104, liblua5.1.5
Status: install user installed
Architecture: mips_24kc
Installed-Time: 1667816896
Auto-Installed: yes
...

I would be using this functionality inside a shell script, I have tried using sed and awk with no useful results, I couldn't figure out this problem.

Solution

  • You can probably get away with:

     awk '!/terminfo/' RS= ORS='\n\n'

    Setting RS to the empty string causes awk to treat each section as a record ("section" meaning chunk of text delimited by an empty line, often called a "paragraph"), so you just print the ones that don't match the string terminfo. To be more precise, you might want to use:

    awk '!/^Package: terminfo/' RS= ORS='\n\n'

    but that doesn't seem necessary.