#!/bin/bash

# mcgmail :  a script to locally back-up all your IMAP eMails
# 	v 0.6.2 version -- with Thunderbird & Evolution integration
#	January 2014
# 	COPYRIGHT Arrow-SEO.com 2014

# You are not permitted to sell this script, but you can use it, alter it, copy 
# it or distribute it, providing that you do not delete this copyright notice, 
# and you do not remove any reference or links to Arrow-SEO.com

# For support or questions, please visit: http://arrow-seo.com/
# McGmail is offered without any warranty of any kind.  Use at your own discretion.  

# recommended steps for use:
# 1.	Run back-up on your system drive!  
# 2.	File any older eMails from your INBOX to appropriate sub-directories. 
# 3.	Empty your IMAP mail trash -- unless you want to back-up your trash.  ?
# 4.	Run McGmail as follows
#			$ mcgmail email-client emailaddress@yourdomain.com 
#		where email client is the name of the program you use to down-load and
#		view eMails, like Thunderbird or Sylpheed, etc., and emailaddress is 
#		your eMail address that you wish to back-up long-term.    

# McGmail will create individual MH format eMail files and mirror directory 
#	tree under your main IMAP folder.  It will move any eMail older than 
#	120 days into this mirror archive tree to the matching directory, 
#	renaming it in the process by receive date and sender's eMail address.  
#	This MAY take up more space on your main drive, but these days, drive space
# 	is cheap.  

# As many people keep eMails older than 120 days in their main INBOX, 
#   McGmail currently skips any eMail still filed under the INBOX directory.

#	To change the age in the script, search for 'AGE="120"' below and alter the 
#	the number accordingly.  

# a NOTE about mbox format:  McGmail relies on a very well written script
#	called 'mbox2mh' by Valentin Koch <vale@canada.com> to convert mbox format
#	to MH format.  You can get info and d/l it at: 
#	http://vale.thus.ch/software/mbox2mh.html
#	If you are using Thunderbird, etc. 'mbox2mh' must be installed in either
#	/usr/bin or ~/bin to work completely.  

# a NOTE on Outlook *.pst files:  McGmail relies on readpst and libpst packages 
#	to convert MS Outlook *.pst data files to MH format eMails.  These functions
#	have not yet been tested with McGmail.  If you are using Outlook on your PC, 
#	please install these packages in their appropriate Windows folder for 
#	McGmail to work with your eMail client.  (and please let us know your results!)

PROGNAME=$(basename $0)

usage () {
	cat <<- EOF
	Usage 1: $PROGNAME CLIENT ADDRESS

	Where:

	CLIENT is the preferred eMail client used to retrieve IMAP eMails
	ADDRESS is the IMAP eMail address to archive messages from.  

	EOF
	return
}


# test for NO value entered w/ command
if [[ $# == 0 ]]; then
	usage >&2
	exit
fi

# display running script and create logfile
exec > >(tee -a ~/Desktop/"McGmail-log_$(date +%F_%T%Z).log") 2>&1

# locate specified mail client's home folder, eMail address to be archived:  
CLIENT="$1"
	echo "eMail client program is $CLIENT."
	CLIENT=$(echo "$CLIENT" | tr "[:upper:]" "[:lower:]")		
		
ADDRESS="$2"
	echo "eMail address to archive is $ADDRESS."

# to alter the age at which eMails get archived, change the numeric value below:
AGE="120"
#AGE="0"
ARCHDATE=$(date -Rd -"$AGE"days)
	echo "McGmail will now archive eMails older than \"$ARCHDATE.\""

# to alter the minimum threshold count, change the value below:
MIN="25"
	echo "McGmail will look for $MIN eMails or more when filtering potential directories."

# reset IFS to allow for directory/file names w/ spaces, odd characters, etc:
OIFS=$IFS
IFS=$'\n'


# ----------------------------------------
# this section IDs pertinent directories.  

# identify the Current Working Directory before we start messing around:  
CWD=$(pwd)

# cd $CWD   to return to starting directory

# find CLIENT $USER folder:  
CTLIST=$(find ~ -iname "*$CLIENT*" 2>/dev/null)
echo "The CTList contains: $CTLIST"

for i in $CTLIST; do

	if [ $(rgrep "$i" -e "$ADDRESS" | grep -iv archives | wc -l ) -ge "$MIN" ]; then
		CLIENT_FOLDER="$i"

		if [ -d "$CLIENT_FOLDER" ]; then
			echo "The path to the CLIENT DIRECTORY is $CLIENT_FOLDER"
#		elif [ $(echo "$CLIENT") == "outlook" ]; then
#			CLIENT_FOLDER="~/AppData/Local/Microsoft"
#			if [ -d "$CLIENT_FOLDER" ]; then
#				echo "The path to Outlook's user directory is $CLIENT_FOLDER"
		fi

	elif [ $(rgrep "$i" -e "$ADDRESS" | grep -iv archives | wc -l) -lt "$MIN" ]; then
		echo "\"$i\" is not the correct folder."
	
	else
		echo "ERROR:  no qualfying CLIENT folder.  CHECK eMail address or client." >&2
		exit 1
	fi		

done

if [ -z "$CLIENT_FOLDER" ]; then
	echo "ERROR:  CLIENT folder is blank.  Try lowering the \$MIN value in script." >&2
	exit 1
fi
		
# find the main INBOX folder for specified $ADDRESS:
IBLIST=$(find $CLIENT_FOLDER -type d -iname "*inbox*" | grep -iv archives) 2>/dev/null
echo "IBList contains: $IBLIST"

for i in $IBLIST; do
	if [ $(rgrep "$i" -e "$ADDRESS" | grep -iv archives | wc -l) -ge "$MIN" ]; then
		INBX_FOLDER="$i"
		if [ -d "$INBX_FOLDER" ]; then
			echo "The path to the INBOX folder for $ADDRESS is $INBX_FOLDER."		
		else
			echo "$i does not actually exist.  Go figyah!" 
		fi
	
	elif [ $(rgrep "$i" -e "$ADDRESS" | grep -iv archives | wc -l) -lt "$MIN" ]; then
		echo "$i is not the inbox folder you're looking for..."  
		
	else 
		echo "Error:  unable to locate INBOX folder." >&2
		exit 1
	fi
done

if [ -z "$INBX_FOLDER" ]; then
	echo "The INBX Folder value is BLANK.  Please check the name of \$CLIENT." >&2
	exit 1
fi

# identify the main $eMAIL_FOLDER for $ADDRESS
eMAIL_FOLDER=$(cd "$INBX_FOLDER"; cd ..; pwd)
	if [ -d "$eMAIL_FOLDER" ]; then
		echo "The path to the eMail folder is $eMAIL_FOLDER"
	elif [ ! -d "$eMAIL_FOLDER" ]; then
	# could / should this code be simplified if effective?
		if [ $(grep -im 1 -d recurse "$ADDRESS" "$IMAP_FOLDER" | grep -iv "archives" | wc -l) -eq 0 ]; then
			echo "Please check the name of the eMail address." >&2 
			exit 1
#		elif [ $(echo "$CLIENT" == "outlook" ]; then
#			smtp_name=$(grep -r "smtp" "$IMAP_FOLDER" | grep -m 1 -iv archives | sed -e 's_^.*\(y\).__' | sed -e 's_ .*$__')
#			SMTP_FOLDER=$(mkdir "$IMAP_FOLDER"/"$smtp_name")
#			if [ -d "$SMTP_FOLDER" ]; then
#				echo "$CLIENT's initial server folder is at $SMTP_FOLDER."
#			fi
#			echo "Creating temp directory for Outlook mail."
#			mkdir "$IMAP_FOLDER"/"$ADDRESS"
		fi
	fi

# identify the $SMTP_FOLDER for $ADDRESS
SMTP_FOLDER=$(cd "$eMAIL_FOLDER"; cd ..; pwd)
	if [ -d "$SMTP_FOLDER" ]; then
		echo "$CLIENT's SMTP folder is at $SMTP_FOLDER"	
#		elif [ $(echo "$CLIENT" == "outlook" ]; then
#			smtp_name=$(grep -r "smtp" "$IMAP_FOLDER" | grep -m 1 -iv archives | sed -e 's_^.*\(y\).__' | sed -e 's_ .*$__')
#			SMTP_FOLDER=$(mkdir "$IMAP_FOLDER"/"$smtp_name")
#			if [ -d "$SMTP_FOLDER" ]; then
#				echo "$CLIENT's initial server folder is at $SMTP_FOLDER."
#			fi
#			echo "Creating temp directory for Outlook mail."
#			mkdir "$IMAP_FOLDER"/"$ADDRESS"
	else
		echo "Please manually check for $SMTP_FOLDER in the directory tree."  >&2
		exit 1
	fi

# identify the imap folder for $ADDRESS:
IMAP_FOLDER=$(cd "$SMTP_FOLDER"; cd ..; pwd)
	if [ -d "$IMAP_FOLDER" ]; then
		echo "The path to $CLIENT's folder containing the IMAP directories is $IMAP_FOLDER"
#	elif [ $(echo "$CLIENT") == "outlook") ]; then
#		IMAP_FOLDER=$(find "$CLIENT_FOLDER" -type d -iname "outlook*")
#		if [ -d "$IMAP_FOLDER" ]; then
#			echo "The path to Outlook's main directory is $IMAP_FOLDER"
#		fi
	else 
		echo "ERROR:  Please check the path to the imap sub-directory." >&2
		exit 1
	fi

# check for MBOX formatting; extract to MH format
if [ $(find "$IMAP_FOLDER" -type f -iname "*.mbox" -o -iname "*.msf" | wc -l) -gt 0 ]; then
	echo "This is an mbox format eMail client." 
	mkdir "$IMAP_FOLDER"/"$ADDRESS"
	mbox2mh -r "$SMTP_FOLDER" "$IMAP_FOLDER"/"$ADDRESS"
	mv "$IMAP_FOLDER"/"$ADDRESS" "$SMTP_FOLDER"
	echo "Moving temp directory of mail files."

	eMAIL_FOLDER=$(find "$CLIENT_FOLDER" -type d -iname "$ADDRESS" | grep -iv tags | grep -iv archives)

		if [ -d "$eMAIL_FOLDER" ]; then
			echo "The path to the temp eMail folder is $eMAIL_FOLDER."
		fi
else
	echo "$CLIENT is not mbox format.  Moving on..."
fi

# check for Outlook .pst format; extract to MH format
# call readpst
#if [ $(find "$IMAP_FOLDER" -type f -iname "*.pst" | wc -l) -gt 0 ]; then 
#	echo "This is an Outlook eMail client."
#	readpst -M "Outlook.pst" "$IMAP_FOLDER"/"$ADDRESS"
#	mv "$IMAP_FOLDER"/"$ADDRESS" "$SMTP_FOLDER"
#	echo "Moving temp directory of Outlook mail."
#	eMAIL_FOLDER=$(find "$CLIENT_FOLDER" -type d -iname "$ADDRESS" | grep -iv tags | grep -iv archives)
#	if [ -d "$eMAIL_FOLDER" ]; then
#		echo "The path to the temp eMail folder is now $eMAIL_FOLDER."
#	fi

cd $SMTP_FOLDER
echo "Now in the SMTP_FOLDER."

# ----------------------------------------
# this section assigns the array values

# commands to create MIRROR archive directories
# note:  if a directory already exists, it will be skipped:

# create an array with the ORIGINAL directory values:

inbx_folder=$(find "$eMAIL_FOLDER" -type d -iname "*inbox*")
echo "The little inbox folder is at $inbx_folder"	

boxarray=$(ls -R "$eMAIL_FOLDER" | grep "$inbx_folder" | sed -e "s_^\($SMTP_FOLDER\)/__" | tr ':' '/')

# create a MIRROR of each directory in the /archive/ tree:
for i in $boxarray; do 
	mkdir -vp "$IMAP_FOLDER"/archives/"$i"
done

# create the $mailarray of all eMails to be archived:
mailarray=$(find "$eMAIL_FOLDER" -type f)

#echo $mailarray

# steps to refine $mailarray to skip rm & b-up of regular eMail files 
# in the main INBOX directory.
inbarray=$(find "$inbx_folder" -maxdepth 1 -type f)
sntarray=$(find $(find "$inbx_folder" -type d -iname sent) -type f)

#echo "Proceeding to refine '\$mailarray' to exclude eMails in main INBOX and Sent folder."

# to back-up eMails in main INBOX directory, comment out the lines above 
# and the line below.  

mailarray=$(echo "$mailarray" | grep -iEv "$inbarray" | grep -iEv "$sntarray")
# echo $mailarray
echo "We're dealing with " $(echo "$mailarray" | wc -l) "eMails."

echo "'\$mailarray' refined.  Proceeding with eMail archiving."

# pwd

# ---------------------------------------
# this section processes and files the MH format eMails:

for i in $mailarray; do 

	# step 1: assign the date variables for $i from eMail header, 
	idate=$(less -f "$i" | grep -im 1 "Date:\ " | sed -e 's_^.*\(ate: \)__')
	echo "$idate"
	idt=$(date -d "$idate" +%F_%T%Z)
	echo "$idt"

	# step 2A: check ( eMail date ) < ( current date - 120 days )
	if (( $(date -ud "$idate" +%s) < $(date -ud "$ARCHDATE" +%s) )); then 
		echo "$i needs to be archived."

		# step 2B: check naming convention:
		if [[ -f "$i" && -n "$(echo $i | grep -iEv '*[0-9]{4}-[0-9]{2}-[0-9]{2}_[0..23]?[0..59]?[0..59]*')" ]]; then 
			echo "The name of eMail $i needs to be changed"
				
				# step 3: pull the sender from the header, assign it to the variable $idn:
				idn=$(less -f "$i" | grep -iEm 1 "From:\ " | sed -e 's_^.*\(rom\).*<\|^.*\(rom:\).__' | sed -e 's_@.*$__')
				echo "$idn"
				
				# step 4:  construct name from parts
				inewname=$(echo "$idt"@"$idn")
				echo "$inewname"
				
				# step 5:  confirm name matches format + filter out non-eMails (such as hidden .blah-blah files)
				if [ -n "$(echo \"$inewname\" | grep -iE '*[0-9]{4}-[0-9]{2}-[0-9]{2}_[0..23]?[0..59]?[0..59]*')" ]; then

					# step 6:	set the parent directory path for "$i", move & rename "$i" to "$newname"
						pardir=$(echo "$i" | sed -e 's_/[^/]*$__' | sed -e "s_^\($SMTP_FOLDER\)/__")
						echo "The path to pardir is \"$pardir\""
						cp -vn "$i" "$IMAP_FOLDER"/archives/"$pardir"/"$inewname"

				else
					echo "The file is not an eMail. Moving on to next file..." >&2
					exit 1 
				fi	

		else 
			echo "File name $i is correct.  Exit name change" >&2
#			exit 1
		fi 

	else 
		echo "Eegads.  The eMail $i is much too young for the old-eMails' home." >&2
	fi

done

# clean up the temp directory for .mbox format
if [ "$CLIENT" == "thunderbird" -o "$CLIENT" == "outlook" ]; then 
# future fix: auto-detect clients that need these steps
	echo "The client is $CLIENT.  Deleting temp directory at $SMTP_FOLDER/$ADDRESS."
	rm -r "$SMTP_FOLDER"/"$ADDRESS"
else 
	echo "$CLIENT is not Thunderbird or Outlook."
fi


# -------------------------------------------------------------
# clean-up section:

# return Internal Field Seperator to original value: 
IFS=$OIFS

# return to starting directory: 
cd $CWD

# make /archives/log directory & move log file into /archive/log directory
mkdir "$IMAP_FOLDER"/archives/logs/
echo "A log file of this back-up can be found in $IMAP_FOLDER/archives/logs/ for reference."  
mv ~/Desktop/McGmail-log_*.log "$IMAP_FOLDER"/archives/logs/

exit