How to find a specific string sequence in files

puntino · 11-12-2012, 11:47 AM

Hello,
I'd like to know how to find a specific string sequence in files.
For instance, suppose that the file1 holds a sequence like the following
...
ext3_write_super
ext3_ordered_writepage
journal_blocks_per_page
journal_start
kmem_cache_alloc
block_write_full_page
generic_make_request
journal_dirty_data
journal_stop
ext3_write_inode
ext3_bmap
generic_block_bmap
ext3_write_super
ext3_ordered_writepage
journal_blocks_per_page
kmem_cache_alloc
...

I would like to know the line number and count all the sequence like:
ext3_ordered_writepage
journal_blocks_per_page
journal_start
kmem_cache_alloc

the sequence must be matched exactly in file1. Wrt to file 1, the sequence
occurs twice.

I've tried with

Code:

grep -nFf sequence_to_find file1

but it's not what I'm looking for.
in "sequence_to_find" there are the former sequences.
I'd like to have an output like:
count 2
2 ext3_ordered_writepage
3 journal_blocks_per_page
4 journal_start
5 kmem_cache_alloc
13 ext3_ordered_writepage
14 journal_blocks_per_page
15 journal_start
16 kmem_cache_alloc

Is there any way to do that?
Thank you in advance

ntubski · 11-12-2012, 01:51 PM

Quote:

Originally Posted by puntino

I've tried with

Code:

grep -nFf sequence_to_find file1

but it's not what I'm looking for.

What is wrong with that output? Is it just the colons you'd like removed?

Code:

grep -nFf sequence_to_find file1 | tr : ' '

colucix · 11-12-2012, 03:16 PM

Hi ntubski! I think the problem is that the -f option causes grep to match every single line in the pattern file, not the exact sequence as a whole.

@puntino: here is a tricky solution in awk:

Code:

BEGIN {

  RS = "ext3_ordered_writepage\njournal_blocks_per_page\njournal_start\nkmem_cache_alloc"
  
  split(RS,pattern,"\n")
  
}

{

  _[++c] = n += gsub(/\n/,"")

}

END {

  print "count", NR - 1
  
  for ( i = 1; i < c; i++ ) {
    start = _[i] +1 -(i-1) +4*(i-1)
    print start, pattern[1]
    print start+1, pattern[2]
    print start+2, pattern[3]
    print start+3, pattern[4]
  }
  
}

I think there are more elegant solutions using perl, python or ruby, but I prefer to leave them to the experts of these languages!