Home / ubuntu / Questions / 1284034
Accepted
Lorenz Keel
Asked: 2020-10-16 04:23:45 +0800 CST

Using awk for a bash script that defines a "minimum packages set"

  • 772

I would like to write a script with the following requirements:

  • in input, get a list of packages available via apt. In there you may have some packages that are automatically installed by other packages in list.
  • in output, provide the same list without the packages that depends from other packages in list.

In other terms, I want to do in bash what the user Francois G did in this answer

Maybe something like this already exists, but sometimes I like to write scripts to improve my bash-scripting and also for fun.

In my mind, I already designed the script, but I have a technical issue. Let's assume that I have the dependency list in this format (it's the way apt-rdepends put it):

useless-line-1
useless-line-2
useless-line-3
item-1
  fixed-string substring-1-1
  fixed-string substring-1-2
  fixed-string substring-1-3
item-2
  fixed-string substring-2-1
  fixed-string substring-2-2
item-3
item-4
  fixed-string substring-4-1
  fixed-string substring-4-2
  fixed-string substring-4-3
  fixed-string substring-4-4

I want to extract the paragraph related to item-1 i.e.:

  fixed-string substring-1-1
  fixed-string substring-1-2
  fixed-string substring-1-3

I'm not an awk expert, but I think that it can suit my purposes. I'm not able to "build" the correct command. Due to the fact that item-2 may be not known, I tried:

# extract text between item-1 and the next line that starts without blank
$ awk '/item-1/,/^[A-Za-z0-9]/' deplist
item-1

but item-1 already fits the condition ^[A-Za-z0-9], so it's not good. Moreover, I want to exclude item-1 and item-2 from the output.

Which is the best way to extract that portion of data?

scripts command-line bash awk

1 Answers

  1. Best Answer

    You could do something "stateful" ex.

    $ awk -v item='item-1' '$0 !~ /^[[:blank:]]/{p=0} $0 ~ "^" item {p=1} p' deplist
item-1
  fixed-string substring-1-1
  fixed-string substring-1-2
  fixed-string substring-1-3

    How this works:

    • set p=0 whenever we match a line that stars with anything other than horizontal whitespace (you could use your original slightly more specific ^[A-Za-z0-9] here)

    • set p=1 if we match the desired ^item

    • print whenever p==1

    Essentially "turn printing on when we match the desired item and turn it off when we match any other item".

    You'll need a bit of extra logic to skip the matched line:

    $ awk -v item='item-1' '
      $0 !~ /^[[:blank:]]/{p=0} {m = $0 ~ "^" item ? 1 : 0} m {p=1} p && !m
  ' deplist
  fixed-string substring-1-1
  fixed-string substring-1-2
  fixed-string substring-1-3

    Here, we perform the same match but save the result in variable m; then set p=1 when m is true (this part is identical to what we had before); we then only print when both p==1 and m==0 i.e. skip the line where the actual match occurs.

    • 1