CODEWITHSUNDEEP
bashawksed
bosona
bosona

Reputation: 43

bash - How to delete block of lines which length is given in another file and doesn't match patterns

Say I have a pattern - file1 of this content

48     0      1000
50     0      3000

and file2 that I want to edit of the following content:

48     0      1000
bla
...
bla
49     0      2000
bla
...
bla
50     0      3000
bla
...
bla

Now, I want to delete lines that

  1. do not match pattern

  2. and all the following block lines which length is given by number in the third column.

So the result is:

48     0      1000
bla
...
bla
50     0      3000
bla
...
bla

and wc -l < result is 4002.

My idea is to take a loop through file2, start in the first line and take a loop through file1. If match, then move line marker over a number that is in column 3 nl=awk 'NR==1{print $3}', if not delete sed -i 'lm,lm+$nl d' file2

Thank you.

Upvotes: 0

Views: 94

Answers (2)

bosona
bosona

Reputation: 43

Here is the code in Fortran:

  nr = 0
  do
   read(1,*,iostat=io) i
   if (io/=0) exit
   nr = nr + 1
  end do
  rewind(1)
  allocate( I1(nr),I2(nr) )
  do j = 1, nr
    read(1,*) I1(j), I2(j)
  end do

   largeN = 10000
   nb = 0
   do
! Nbl...no. in the third column
   read(2,*,iostat=io) int1, int2, Nbl
       match = .false.
       do j = 1, nr
         if ( (int1.eq.I1(j)).and.(int2.eq.I2(j)) ) then
! avoid duplicities, change matched arrays to sth. that is not contained in columns 1 and 2 in file1
           I1(j) = largeN
           I2(j) = largeN
           match = .true.
! output in unit=3
       write(3,*) int1, int2, Nbl
         end if
       end do
       if (io/=0) exit
       nb = nb + 1
       do i = 1, Nbl
         read(2,*) x 
       if(match) write(3,*) x
       end do
     end do
  End

Upvotes: 0

Casimir et Hippolyte
Casimir et Hippolyte

Reputation: 89557

You can use this command:

awk 'NR==FNR{a[$0]=1;next;}/^([[:digit:]]+[[:blank:]]+){2}[[:digit:]]+$/{f=a[$0]}f' file1 file2

details:

NR==FNR {    # when the first file is processed
    a[$0]=1  # store each line in an array
    next     # jump to the next line
}
# file2: if a line has the format
/^([[:digit:]]+[[:blank:]]+){2}[[:digit:]]+$/ {
    f=a[$0]  # then set the flag f to a[$0] (1 if it exists, 0 if not)
}
f # when f is set to 1, print the line.

Upvotes: 1

Related Questions