namelist_check

#!/bin/bash
#################################################################
# Check an input file against namelists in a given source code
# 
# namelist_check -s SOURCE DIR -i INPUTFILE
#
# 	SOURCE DIR - absolute path to source directory
#	INPUTFILE  - absolute path to input file
#
# General plan of action:
# 	make two sets of arrays: namelists from code, and parameters in input file
#	compare the two sets of arrays
#	if some parameters don't appear in the namelists, say what they are
#
# v2 - Commandline options for sourcedir and inputfile
#    - Can probably handle namelists defined in several files (not checked)
#    - Rudimentary error handling
#
# v1 - Currently no form of error handling
#    - Input file hard coded
#    - Relies on namelists all being in one file in code
#
# Peter Hill 2011
# Feel free to distribute, but please keep this attribution
#################################################################

source_loc=
input_loc=

usage=" namelist_check -s SOURCE DIR -i INPUTFILE

 	SOURCE DIR - absolute path to source directory
	INPUTFILE  - absolute path to input file
"

while getopts :s:i: OPT; do
    case $OPT in
	s|+s) # Source location
	    source_loc="$OPTARG"
	    ;;
	i|+i) # Input file location
	    input_loc="$OPTARG"
	    ;;
	*)
	    echo $usage
	    exit 2
    esac
done
shift $(( OPTIND - 1 ))
OPTIND=1

if [[ -z $source_loc ]]
then
    echo "No source given"
    echo "$usage"
    exit
fi
if [[ -z $input_loc ]]
then
    echo "No input file given"
    echo "$usage"
    exit
fi

#################################################################
# Make a file with namelists on individual lines

# Go to source dir
cd $source_loc

# Temp files
nml_working_tot=`mktemp`

# Find where the namelists are
nml_file=`grep -i -l -E "namelist /[a-zA-Z_]*/" *`

# Check if there are any results!
if [[ -z  $nml_file ]]
then
    echo "No namelists found! Check source"
    echo "$source_loc"
fi

# Loop over results of grep
i=0
for i_file in $nml_file
do
# Temp files
    nml_working[$i]=`mktemp`
    nml_working2[$i]=`mktemp`
# Make a new file so we don't screw up original if it all goes terribly wrong
# Probably relies on namelists only being in one file
    cp $i_file ${nml_working[$i]}

# Put multi-line FORTRAN statements onto one line
    while grep -q -E "&$" ${nml_working[$i]}
    do
	sed -i '
    # look for FORTRAN line continuation characters (plus trailing whitespace)
    /&[ \t]*$/ {
    # read in next line
	N
    # delete new line and FORTRAN &
    	s/&[ \t]*\n//g
    }' ${nml_working[$i]}
    done

# Get just the namelists out
    grep -i -E "namelist /[a-zA-Z_]*/" ${nml_working[$i]} > ${nml_working2[$i]}

# Clean up file
# Remove forward slashes and commas from namelist names and put everything in lower case
    cat ${nml_working2[$i]} | tr '!,/' ' ' | tr '[:upper:]' '[:lower:]' > ${nml_working[$i]}
# Remove leading word "NAMELIST"
    sed -i "s/namelist//" ${nml_working[$i]}
# Get rid of multiple spaces
    sed -i "s/  */ /g" ${nml_working[$i]}

# Put all namelists into one file
    cat ${nml_working[$i]} >> $nml_working_tot
    let i+=1
done
i=0

# List of namelists and their parameters in code
for index in `cat $nml_working_tot | awk '{print NR'}`
do
    code_nml[$index]=`cat $nml_working_tot | awk 'NR=='$index' {print $1}'`
    code_para[$index]=`cat $nml_working_tot | awk 'NR=='$index' {print $0}'`
done
code_nml_arr=${code_nml[@]}
#echo $code_nml_arr	# uncomment for sanity check

#################################################################
# Make a similar file for the input file

# Temp file
inputbak=`mktemp`
inputbak2=`mktemp` 

# Prep input file
cp $input_loc $inputbak
cat $inputbak | tr '=' ' ' > $inputbak2


# Remove preamble
if [[ $(head -1 $inputbak2 | grep -o -E "^/") != '/' ]]
then
    sed -i '1,/\// d' $inputbak2
fi

# Get namelist names and input variables from input file
cat $inputbak2 | awk '{print $1'} > $inputbak

# Put namelists onto one line
cat $inputbak | tr '\n\r' ' ' > $inputbak2

# Make sure each namelist is on its own line
cat $inputbak2 | tr '&' '\n' > $inputbak

# Clean up file: get rid of stray punctuation, switch to lower case
cat $inputbak | tr -d '!/,' > $inputbak2
cat $inputbak2 | tr '[:upper:]' '[:lower:]' > $inputbak

# List of namelists and their parameters in input file
for index in `cat $inputbak | awk '{print NR'}`
do
    input_nml[$index]=`cat $inputbak | awk 'NR=='$index' {print $1}'`
    input_para[$index]=`cat $inputbak | awk 'NR=='$index' {print $0}'`
done
input_nml_arr=${input_nml[@]}
#echo $input_nml_arr	# uncomment for sanity check
#echo ${input_para[@]}
#################################################################
# Snippet to compare two arrays
# Stolen (30 Aug 2011) from http://stackoverflow.com/questions/2312762/compare-difference-of-two-arrays-in-bash/2315541#2315541
# Credit to ghostdog74
arr_comp(){
    a1="$1"
    a2="$2"
    awk -va1="$a1" -va2="$a2" '
     BEGIN{
       m= split(a1, A1," ")
       n= split(a2, t," ")
       for(i=1;i<=n;i++) { A2[t[i]] }
       for (i=1;i<=m;i++){
            if( ! (A1[i] in A2)  ){
                printf A1[i]" "
            }
        }
    }'
}
#################################################################
# Compare the two arrays

# Compare namelists
nml_comp=( $(arr_comp "$input_nml_arr" "$code_nml_arr") )
if [[ -z ${nml_comp[@]} ]]
then
    echo "Namelists all correct!"
else
    echo "Following namelists not in code: ${nml_comp[@]}"
fi

# Loop over input namelists
for index in `seq ${#input_nml[@]}`
do
    # Clean comparision variable
    para_comp=
    # Find right code namelist
    for index2 in `seq ${#code_nml[@]}`
    do
	if [[ "${input_nml[$index]}" = "${code_nml[$index2]}" ]]
	then
	    # Compare input parameters to those in code namelist
	    input_para_temp=${input_para[$index]}
	    code_para_temp=${code_para[$index2]}
	    para_comp=( $(arr_comp "$input_para_temp" "$code_para_temp" ) )
	    if [[ -z ${para_comp[@]} ]]
	    then
		echo "Namelist ${input_nml[$index]} is correct"
	    else
		echo "Following input parameters in namelist ${input_nml[$index]} are not in code: ${para_comp[@]}"
	    fi
	fi
    done
done

# Clean up tempory files
rm ${nml_working[@]}  ${nml_working2[@]} $nml_working_tot $inputbak $inputbak2

# Last line