#!/bin/bash
#Export_Large_Projects tester
if [ $# -lt 2 ]
  then
    echo "USE $0 CSV_FILE SAS_FILE"
    exit 1
fi

declare -a EQ_to_CSV=("info_informacin_de_c_v_0"  "f23_registro_contact_v_1"  "f18_actualizacion_in_v_2" "f3_ecografia_abdomin_v_3" "f3_ecografia_abdomin_v_4" "f3_ecografia_abdomin_v_5" "f10_encuesta_de_segu_v_6" "f22_encuesta_colecis_v_7" "f5_pareamiento_segun_v_8" "f4_examen_fisico_seg_v_9" "f22_encuesta_colecis_v_10" "f12_recoleccion_de_e_v_11" "f25_sospecha_cncer_c_v_12" "f11a_revisin_de_fich_v_13" "f11a_revision_de_fic_v_14" "f20_formularios_falt_v_15" "f19_salida_del_estud_v_16")
declare -a EQ_to_SAS=("info_informacin_de_contacto_complete" "f23_registro_contacto_seguimiento_complete" "f18_actualizacion_informacion_de_participante_complete" "f3_ecografia_abdominal_complete" "f3_ecografia_abdominal_segunda_complete" "f3_ecografia_abdominal_tercera_complete" "f10_encuesta_de_seguimiento_complete" "f22_encuesta_colecistectomia_complete" "f5_pareamiento_segundo_complete" "f4_examen_fisico_segundo_complete" "f22_encuesta_colecistectomia_segunda_complete" "f12_recoleccion_de_especimenes_quirurgicos_complete" "f25_sospecha_cncer_clnico_complete" "f11a_revisin_de_ficha_clnica_basal_complete" "f11a_revision_de_ficha_clinica_segunda_complete" "f20_formularios_faltantes_complete" "f19_salida_del_estudio_complete")
CSV=$1
SAS=$2
TMP_CSV="csv.txt"
TMP_SAS="sas.txt"
NOT_FOUND_SAS=0
NOT_FOUND_CSV=0
FOUND=0
RED='\033[0;31m'
NC='\033[0m'
YELLOW='\033[0;33m'
head -n 1 $CSV | tr  ',' '\n' > $TMP_CSV

for i in "${EQ_to_CSV[@]}"
do
   echo -n ",$i" >> $TMP_CSV
done


sed 's/\$//g' $SAS |grep -v "='" |grep -v informat |grep -v format |tail -n +5 | head --lines=-22 > $TMP_SAS
for i in "${EQ_to_SAS[@]}"
do
   echo  "                $i" >> $TMP_SAS
done


#read csv
echo -e "${RED} In CSV but not in SAS ${NC}"
while read -r lineCSV;
    do
	if grep -q "$lineCSV" $TMP_SAS; then
  	  #echo "found "
	  FOUND=$((FOUND + 1))

	else
	  echo -e "${YELLOW} $lineCSV ${NC}"
	  NOT_FOUND_CSV=$((NOT_FOUND_CSV + 1))
	fi
    done <  $TMP_CSV

#read sas

echo -e "\n${RED}In SAS but not in CSV${NC}"
FOUND=0
NOT_FOUND=0
while read -r lineSAS;
    do
        if grep -q "$lineSAS" $TMP_CSV; then
          #echo "found "
          FOUND=$((FOUND + 1))

        else
          echo -e "${YELLOW} $lineSAS ${NC}"
          NOT_FOUND_SAS=$((NOT_FOUND_SAS + 1))
        fi
    done < $TMP_SAS

echo -e  "\nTotal Found: $FOUND"
echo "Total Not Found in SAS: $NOT_FOUND_SAS"
echo "Total Not Found in CSV: $NOT_FOUND_CSV"

#cleaning
rm -f $TMP_CSV $TMP_SAS