UNIX for Dummies Questions & Answers

What's wrong with wc anyway?

I suppose you could also use:-

Code:

grep -c "" file

...amongst other things, but wc is written for the purpose.



Robin

Perhaps the key word here is in the opening post (#1): "exercise". I doubt that there is any value in training anything if you let others do the exercising - you won't grow muscles from watching others lifting weights, so to say.

Normally i would at best ignore such a thread (or close it, if i think it is not only exercise but homework), but one thing I think i can mention without jeopardizing the intention of the exercise at all:

wc is not only counting lines (wc -l) but also words (wc) or characters (wc -c). A script intending to replace wc would have to implement all these functions, IMHO. No?

I hope this helps.

bakunin

Minor correction: -c counts bytes. Characters are counted with -m.

Regards,
Alister

Certain grep versions give RE error.

Code:

grep -c "^"

is a complete RE

---------- Post updated at 12:17 PM ---------- Previous update was at 12:08 PM ----------

The following emulates a wc without arguments:

Code:

awk '{c+=(length+1); w+=NF} END {printf " %-6s %-6s %s%s\n",NR,w+0,c+0,(FILENAME=="-")?"":" "FILENAME}'

The emulation is imperfect in at least two respects.

length counts characters while wc counts bytes.

wc implementations usually delimit words with the space ctype, which is a superset of the default blank class which awk uses to split records. This means that your awk emulation may underestimate the word count if the FS isn't set appropriately.

Code:

$ printf 'one\vtwo\vthree\n'
one
   two
      three

$ printf 'one\vtwo\vthree\n' | wc -w
3

$ printf 'one\vtwo\vthree\n' | awk '{print NF}'
1

$ printf 'one\vtwo\vthree\n' | awk '{print NF}' FS='[[:space:]]+'
3

Regards,
Alister