#!/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/Downloads/output"
    if [ -d $OUTPUTFOLDER ]
    then
        break
        else 
        echo -e "Directory not found."
    fi
done

#Create the csv and headers for the Addresses.  It is tab separated.
mkdir $OUTPUTFOLDER/Non_Blank
touch $OUTPUTFOLDER/Non_Blank/Addresses.csv
echo -e "File,User,Street,City,State,Zip" > $OUTPUTFOLDER/Non_Blank/Addresses.csv

#Create csv with all the files with missing addresses
mkdir $OUTPUTFOLDER/Missing_Addresses
touch $OUTPUTFOLDER/Missing_Addresses/Missing_Addresses.csv

echo -e "File\tUser" > $OUTPUTFOLDER/Missing_Addresses/Missing_Addresses.csv

cd $PDFDIR
#FOLDER=$(ls | grep *.pdf)
#echo $FOLDER
#*.pdf represensts All files in the current directory with a pdf extention.
for f in $PDFDIR/*
do 
    echo "Processing file $f."
    #pdftotext converst the file to text and places it in the file name temp
    pdftotext "$f" - > TEMP
    #assigns the lines with relevant values to their variable
    USER=$(sed -n '12p' TEMP)
    ADDRESS1=$(sed -n '13p' TEMP)
    ADDRESS2=$(sed -n '14p' TEMP)
    echo -e "User is $USER"
    echo -e "Address is $ADDRESS1 $ADDRESS2"
    if [ -z "$ADDRESS1" ]
    then 
        #Move the file with the Missing addresses to the Missing Address folder and make an note in its csv.
        echo -e "Copying file to Missing_Addresses"
        cp "$f" "$OUTPUTFOLDER"/Missing_Addresses/
        echo -e "$f\t$USER" >> "$OUTPUTFOLDER"/Missing_Addresses/Missing_Addresses.csv
        #Adds the row of variables to the csv.  >> appends to the file. \t represents tabs
    else
    # Split Addresses 
        CITY=$(echo $ADDRESS2 | cut -d, -f1)
        ADDRESS2=$(echo $ADDRESS2 | sed -e "s/$CITY, //")

        STATE=$(echo $ADDRESS2 | cut -d" " -f1)
        ZIP=$(echo $ADDRESS2 | cut -d" " -f2)
        
        echo -e "$f,$USER,$ADDRESS1,$CITY,$STATE,$ZIP" >> $OUTPUTFOLDER/Non_Blank/Addresses.csv
        cp "$f" $OUTPUTFOLDER/Non_Blank/
        echo -e "Copying file to Non_Blank"
    fi
done
exit