#!/usr/bin/sh
###################################################################################
# Script name: pf
#
# Show SAS proc freq output
#
# example: 
# > pf adsl "armcd*sex / list missing"
###################################################################################

# Base name of this script
typeset basename=${0##*/}

# Lower case left-justified dataset name
typeset -L dataset

# Check if there are too many parameters
if [[ $# -gt 4 ]];then
print "Too many paramters, please refer to usage by typing pf at the prompt..." 
exit 5
fi

# Check if the first argument is linesize, or that the first and second argument is '-a'
if [[ $1 = "-a" ]]; then
print "Please use the '-a' option only after you have specified the dataset and the numeric cut-off..."
exit 2
elif [[ $2 = "-a" ]]; then
typeset all="yes"
set $1 "" $3 $4
elif [[ $3 = "-a" ]] || [[ $4 = "-a" ]]; then
typeset all="yes"
else 
typeset all="no"
fi

# Defaults the linesize of the SAS output to 80, or set to some other value if there are user inputs
if  echo "$1" | grep -q -E '^-[0-9]+$'; then
print "Please use the linesize option only after you have specified at least one other parameter..." 
exit 3
elif echo "$2" | grep -q -E '^-[0-9]+$'; then
typeset linesize=`echo $2 | sed "s/-//g"`
set $1 "" $3 $4
elif echo "$3" | grep -q -E '^-[0-9]+$'; then
typeset linesize=`echo $3 | sed "s/-//g"`
elif echo "$4" | grep -q -E '^-[0-9]+$'; then
typeset linesize=`echo $4 | sed "s/-//g"`
else
typeset linesize=80
fi

# Create unique and temporary log and sas program names
log=$(mktemp -d ~/)$basename.log
sasfile=$(mktemp -d ~/).sas


# Usage description 
function usage {
cat << END_OF_HELP
SYNTAX:  1. pf dataset "tables" [-linesize]  OR  2. pf dataset [number] [-a] [-linesize]  

DESCRIPTION:  1. Produce one-way to n-way frequency and contingency tables of the
		 specified SAS dataset (like SAS "proc freq") if the "tables" statement
		 is specified as a command line argument, OR
              2. Display pertinent variable attributes (e.g. the type, the number of 
	         level, the number of missing levels) if the "tables" argument is 
		 missing
		 
OPTIONS:  number - Displays only variables with number of levels <= number
	      -a - Produce one-way frequency tables for all variables selected. 	    
       -linesize - Allows you to specify the linesize of the SAS output. Default
                   is 80. 

USAGE:  This tool will search for the first occurence of the specified SAS dataset 
	in the following order:
		1. current directory         	
	You can use this tool on only one SAS dataset at a time.

EXAMPLE: #1: pf dm
	 #2: pf dm 10 
	 #3: pf dm 10 -a 
	     OR
         #4: pf dm "race*sex/list missing" 
	 #5: pf dm "race*sex/list missing" edit
exiting...
END_OF_HELP
}

# If no dataset is specified, show usage
if [[ "$1" = "" ]]; then
	usage
	exit 4
fi


# The first parameter is the file path
typeset filepath=`echo $1 | sed "s/.ssd01//;s/.sas7bdat//;s/.*\.//"`
typeset library=$(dirname $filepath)
typeset dataset=$(basename $filepath)


# Save the present working directory
typeset pwd=`pwd`

# Search for the SAS dataset, first in $pwd, then in $pwd/libraries, and finally in sdtmroot
if [ -f $filepath.ssd01 ] || [ -f $filepath.sas7bdat ]; then 
   if [[ $library = "." ]]; then 
   	typeset sasdir="$pwd"
   else 
   	typeset sasdir="$library"
   fi
elif [ -f libraries/$dataset.ssd01 ] || [ -f libraries/$dataset.sas7bdat ]; then 
   typeset sasdir="$pwd/libraries"
else
   sdtmroot
   find_file=$(find `pwd` \( -name $dataset.ssd01 -o -name $dataset.sas7bdat \) -print -quit)
   if [[ $find_file != "" ]] && [[ `pwd` != $pwd ]]; then
   sasdir="${find_file%/*}"
   else
   echo "Dataset: $dataset.ssd01 or $dataset.sas7bdat not found.  Exiting . . . "
   exit 1
   fi
fi


# The second parameter refers to the table statement
typeset tables="$2"

# Begin constructing the sas program
echo "libname LIBRARY '$sasdir' access=readonly;"                                 >> $sasfile
echo "options nocenter nofmterr pagesize=max linesize=$linesize nonumber nodate;" >> $sasfile
echo 'title; footnote;'                                                           >> $sasfile
echo "proc freq data=LIBRARY.$dataset;"                                           >> $sasfile
echo "tables $tables ; run;"                                                       >> $sasfile


echo "Running proc freq on dataset: $dataset ($sasdir)"
sas9 -nodms -noautoexec -stdio $sasfile 2>$log
grep -F -e 'ERROR' -e 'WARNING' <$log
rm -f $sasfile $log