#!/bin/bash

VERSIONINFO="codestat  version 1.4,  by  scott arrington"

# created 13 Nov 98 by scott arrington.
#
# 19 Nov 98
#	the test syntax varies between the cygwin32 environment and
#	my linux system (both with bash as /bin/sh).  i think the 
#	difference is which test is being used (internal vs external).
#	also, under the cygwin stuff, really large greps cause sh to
#	have a "GPF", i suppose because the pipe fills up. i don't
#	know if that's fixable in here.
#	something that might be nice would be to make subroutines out of 
#	the different little functional blocks here. 
#	added some command line args and a little progress display.
#	checks to see if there are any files before trying grep, so that 
#	grep won't freeze waiting for stdin. 
#
# 31 Dec 98
#	changed a tab to a space in the output, so that the columns
#	line up properly.
#
# 16 Feb 99
#	forgot to bump the version number on 1.2... changed this to 1.3.
#	converted some parts to functions, making this a bash script.
#	we can now process more than one directory on the command line,
#	and can also suppress the default recursive file-searching.
#	the "working" dots also disappear if your echo supports the -e
#	option.
#
# 17 Feb 99
#	minor bugfixes like variable initialization.
#	added a long report format (-l on the command line).
#
# 12 Oct 99
#	changed highly inefficient code to use loops and such, so now
#	it's quite easy to add filename extensions.  shorter and
#	cleaner code, as well.  might this benefit from a perl rewrite?
#	bumped version number to 1.4
#


# this is empty by default
DEPTHSTR=
# chooses the report style
CODESTAT=codestat


# calculate all of the code stats for the given directory.  
# the first arg should be the desired directory.
#
function codestat
{
	DIR=$1
	echo "--------------------------------------------------------"
	echo "    directory root: $DIR"
	ncodelines=0
	nlines=0
	nfilest=0
	for i in "*.c" "*.cc" "*.cpp" "*.h"; do
		thesefiles=$(find $DIR -name "$i" -print $DEPTHSTR)
		if ! [ -z "$thesefiles" ]; then
			echo -e "\tnumber of $i files:\t\t$(echo $thesefiles | wc -w)"
			let nfiles=$nfiles+$(echo $thesefiles | wc -w)
			let ncodelines=$ncodelines+$(echo $thesefiles | xargs grep ";" | wc -l);
			let nlines=$nlines+$(cat $thesefiles | wc -l);
		fi
	done
	echo -e "\ttotal number of files:\t\t$nfiles"
	echo -e "\t    total number of lines:\t$nlines"
	echo -e "\t    estimated lines of code:\t$ncodelines"
	echo "--------------------------------------------------------"
}


function codestat_4
{
	# initialize the local variables
	NFILES=0
	NLINES=0
	NCODELINES=0
	DIR=$1;

	echo "-----------------------------------------------------------------------------"
	echo "    directory root: $DIR"
	echo "-----------------------------------------------------------------------------"
	echo "    type	# files		# lines		est # statements"
	echo "-----------------------------------------------------------------------------"
	
	for EXT in "*.c" "*.cc" "*.cpp" "*.h"; do
		FILES=`find $DIR -name "$EXT" -print $DEPTHSTR`
		if ! [ -z "$FILES" -a -z "$CSRCS" -a -z "$HEADERS" ]; then
		
			# count the files
			COUNT=`echo $FILES | wc -w | awk '{print $1}'`
			let NFILES=$NFILES+$COUNT
			
			# compute the line stats
			if [ $COUNT != 0 ]; then
				let CODELINECOUNT=`cat $FILES | grep ";" | wc -l | awk '{print $1}'`
				let LINECOUNT=`cat $FILES | wc -l | awk '{print $1}'`
			fi
	
			echo "    $EXT  	$COUNT		$LINECOUNT		$CODELINECOUNT"
			
			let NLINES=$NLINES+$LINECOUNT
			let NCODELINES=$NCODELINES+$CODELINECOUNT
		fi
	done

	echo "   totals 	$NFILES		$NLINES		$NCODELINES"
	echo "-----------------------------------------------------------------------------"
}





#
# display a helpful usage message for those desiring guidance.
#
function usage
{
cat <<EOF

usage: 
    $1 -h
    $1 --help
        print this message and exit

    $1 -V
    $1 --version
        print version information and exit

    $1 [directory [directory2...]]
        Compute statistics for each of the directories on the command 
        line, or the current directory, if no paths given.  
        Subdirectories are descended by default. The "estimated lines of
        code" counts how many semicolons are present in the code files,  
        as an estimate of the number of C statements.  An interesting
        statistic (that this program does not compute) is the ratio of 
        comments to statements, a gauge of how much documentation (or 
        pontification) there is in your code.

EOF
}

#
# advertise yourself
#
function version
{
cat <<EOF
 
$VERSIONINFO
Calculates very basic C/C++ project statistics.

EOF
}


#
#
# print the version info in a banner
function banner
{
cat <<EOF
--------------------------------------------------------
   $VERSIONINFO
--------------------------------------------------------
EOF
}







########################################################
# the buck starts here:

# first check the args.  note that if we want to support more than just
# one command switch, we'll need to put this case statement in a loop.
case "$1" in
	-h|--help)
		usage $0
                exit
		;;
	-V|--version)
		version
                exit
		;;
	-r|--notrecursive)
		DEPTHSTR="-maxdepth 1"
                shift
		;;
	-l|--longreport)
		#CODESTAT=codestat_2
		CODESTAT=codestat_4
                shift
		;;
esac

# start the output...
#banner

# now do the real work.
if [ -z "$1" ]; then
	$CODESTAT `pwd`
else
	while ! [ -z "$1" ]; do
		if [ -d "$1" ]; then
			$CODESTAT "$1"
		else 
			echo "$1": not a directory
		fi
		shift
	done
fi