Lines of code

Recently, tonyg asked on comp.lang.ada for a code counter. There were a couple of recommendations, including gnat metric; however, my favourite is this shell script:

#!/bin/bash -f
t=0
for f in $*; do
  n=`sed -e "s/\".*\"//" -e "s/--.*//" <$f | tr -Cd \; | wc -c`
  t=`expr $t + $n`
  echo "$f$n"
done
echo "total loc $t"

Quite close to line noise sense 3.

gnat metric does a whole lot more, which isn't surprising; on my current project,

  $ gnat metric -P ../tools/tools.gpr ../tools/normalize_xmi*.ad?

results in

  Line metrics summed over 37 units
    all lines            : 4210
    code lines           : 2866
    comment lines        : 822
    end-of-line comments : 1
    comment percentage   : 22.31
    blank lines          : 522

  Average lines in body: 23.73

  Element metrics summed over 37 units
    all statements      : 830
    all declarations    : 811
    logical SLOC        : 1641

   23 type declarations in 17 units

   2 public subprograms in 2 units

   101 subprogram bodies in 18 units

  Average cyclomatic complexity: 2.61

For comparison, the shell script above says the total [source] lines of code is 1587; close enough to gnat metric's "logical SLOC" for government work, I think.

I should add, I've no particular interest in measuring lines of code! who's counting, after all!