43

if i have a .gz file on unix which has certain number of lines. How could i count the lines on unix without uncompressing it.

61

You can obviously not count newlines if the file is still compressed.

But you can decompress to a stream, and count the newlines in that stream, without ever writing the (decompressed) file to disk. That would go something like so:

zcat file.gz | wc -l

zcat for decompress & cat, wc for wordcount. See man pages for both if you want to know more.

EDIT

If you do not have zcat, zcat is just another name for gunzip -c.

  • 7
    On Unices where gzip is distinct from compress, you want gzcat. – coneslayer Apr 27 '10 at 21:56
7

This also seems to work - grep for the number of line-endings in the file

zgrep -Ec "$" file.gz
  • This gives a different (much higher) answer for me than piping to wc -l – Stop Harming Monica Mar 9 '18 at 17:03
5

If you want to do it quickly, I recommend using 'pigz' (which IIRC stands for "Parallel Implementation of GZip"). I just had a similar situation where I wanted to count the number of lines in a bunch of gzip'ed files and here was my solution:

for x in *.gz; do unpigz -p 8 -c $x | wc -l && echo $x; done

Which gave me the number of lines and the file it counted from on alternating lines, using 8 processors. It ran quickly!

  • 1
    Or if unpigz is not available, simply with for x in *.fastq.gz; do zcat "$x" | wc -l && echo $x; done – Calimo Nov 20 '15 at 22:34
2

Use this command:

gzgrep -c $ filename.gz

The command gzgrep behaves the same as grep but on gzip compressed files. It decompress the file on the fly for the regex matching.

In this case -c instruct the command to output number of matched lines and the regex $ matches end of line so it matches every line or the file.

The final result is identical to gzip -dc filename.gz | grep -c $.

  • Is gzgrep available on other systems than Solaris? – pabouk Nov 21 '14 at 9:43
  • 1
    No. On other systems, command would be zgrep -c $ filename.gz – Ravi K M May 11 '16 at 8:08
  • 1
    Although one might intuitively think this is better than zcat+wc, when I time them, they take the same amount of time. – ngọcminh.oss May 10 '18 at 12:05
1

If you're okay with a rough estimate rather than an exact count, and actually extracting the whole file or zgrepping it for line endings would both take much too long (which was my situation just now), you can:

zcat "$file" | head -1000 > 1000-line-sample.txt
ls -ls 1000-line-sample.txt "$file"

then the approximate line count is 1000 * (size of $file) / (size of 1000-line-sample), as long as your data is fairly homogeneous per line.

0

gzip -cd <file.gz> | wc -l

This worked for me.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.