counting non blank lines per block of a file, having several blocks separated by a blank line

Question

Hi, I'd like to count lines in a file that consists of several blocks, say 3, each with a different number of lines. Each block is separated by a blank line. Is there a one line solution? So far here is what I have:

awk '(NR>4) && NF!=0 {++count} END {print count}' filename > outfile

This obviously counts all non-blank lines (and gets rid of a 4-line header). I now have to include a for loop and after each run it should print the number of lines.

So if I have 100 non-blank lines, and the first block contains 20 lines, the second 50 and the third 30 lines, the ideal output would be 20 50 30

All my effort so far had syntax errors.

Thanks for your help Tom

Answer 1

awk 'NR>4 {if ($0 ~ /./ ) { mylines=mylines+1 } else { printf("%d ",mylines) ; mylines=0 } }
      END { if ($0 ~ /./) { printf("%d ",mylines) } }' <FILENAME>

would do it.

Note: I'm using printf because you had specified the output as "20 50 30" which is on one line.

Edit: just recognized, we must skip the first 4 lines.

Answer 2

if i get you correctly (next time show examples)

$ cat file
#Surface 0 of 1 surfaces

# Contour 0, label:    0.138
 462  370.107  0.137889
 461.82  370  0.137889
skipping lines
 463  370.529  0.137889
 462  370.107  0.137889

 570  448.082  0.137889
 569.772  448  0.137889
skipping lines
 571  448.272  0.137889
 570  448.082  0.137889

 569  465.332  0.137889
 568.299  465  0.137889
skipping lines
 570  465.554  0.137889
 569  465.332  0.137889

$ awk 'NR==3{ RS=""; FS="\n"}NR>3{print NF}' file
5
5
5

So basically, at record 3 just before reaching record 4, set the record separator to blank and field separator to newlines. This is because we don't want to touch the RS and FS variables for the header lines. AFter the 3rd line, we need to change the RS and FS so that we get desired result. ie, a record ends with a blank line and all fields are separated by newlines "\n", essentially, counting NF will get us the total number of lines in one record.

Answer 3

awk 'BEGIN{count=0}\
        { if(NF==0) {if(NR>4)print count;count=0} \
          else count++ ;}' test.txt

Answer 4

Here is a version of my file: It starts with a blank line:

#Surface 0 of 1 surfaces

# Contour 0, label:    0.138
 462  370.107  0.137889 
 461.82  370  0.137889 
skipping lines
 463  370.529  0.137889 
 462  370.107  0.137889 

 570  448.082  0.137889 
 569.772  448  0.137889 
skipping lines
 571  448.272  0.137889 
 570  448.082  0.137889 

 569  465.332  0.137889 
 568.299  465  0.137889 
skipping lines
 570  465.554  0.137889 
 569  465.332  0.137889 

Yes there is a blank line at the end

THX