#!/bin/bash

#Get the folder with the pdf documents in it.
while :
do 
    echo -e "What is the directory with the pdfs?"
    read PDFDIR
    #PDFDIR=/home/btmccrea/Downloads/pdfs/
    if [ -d $PDFDIR ]; then
        break
        else 
        echo -e "Directory not found."
    fi
done
#Get the folder for the output file. 
while :
do 
    echo -e "What is the output directory for the pdf's with missing addresses?"
    read OUTPUTFOLDER
    #OUTPUTFOLDER=/home/btmccrea/Documents/output/
    if [ -d $PDFDIR ]; then
        break
        else 
        echo -e "Directory not found."
    fi
done

#Create the csv and headers for the Addresses.  It is tab separated.
touch "$OUTPUTFOLDER"Addresses.csv
echo -e "File\tUser\tAddress" > "$OUTPUTFOLDER"Addresses.csv
#FILES=$(ls $PDFDIR)
cd $PDFDIR
#echo $FILES
#*.pdf represensts All files in the current directory with a pdf extention.
for f in *.pdf ;
    do 
        #pdftotext converst the file to text and places it in the file name temp
        pdftotext "$f" temp
        #assigns the lines with relevant values to their variable
        USER=$(sed -n '13p' temp)
        ADDRESS1=$(sed -n '14p' temp)
        ADDRESS2=$(sed -n '15p' temp)
        #Adds the row of variables to the csv.  >> appends to the file. \t represents tabs
        echo -e "$f\t$USER\t$ADDRESS1 $ADDRESS2" >> "$OUTPUTFOLDER"Addresses.csv
    done
exit