#!/bin/sh
#----------------------------------------------------------------------------------------
# /usr/bin/archmaildir - a simple mailbox archiver for maildir mailboxes
#
# Copyright (c) 2020-2025  The Eisfair Team, team(at)eisfair(dot)org
#
# Creation:     2020-06-14  jed
# Last Update:  $Id$
#
# Usage:        archmdir [-h|--version]
#               archmdir MODE [OPTIONS] -d <date> <mailbox> [<mailbox>, ...]
#               archmdir MODE [OPTIONS] -o <days> <mailbox> [<mailbox>, ...]
#
#               The available modes and options are described below
#               in the 'show_help' function.
#
# Example:      archmaildir --copy --date-header --date 2017-12-31 --extension 2017 \
#                 --archive-path /home/ralf/.imapmail/.Archiv/.Ablage --format maildir
#                 "/home/ralf/.imapmail/.Ablage.Allgemein"
#
#   Target dir: /home/ralf/.imapmail/.Archiv/.Ablage/.Allgemein\2e2017
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#----------------------------------------------------------------------------------------

. /var/install/include/eislib
. /var/install/include/jedlib

if [ "$1" != '--version' ]
then
    # skip loading to prevent collision
    # with jedlib internal version function
    . /var/install/include/jedlib
fi

# debug mode: true/false
#debug=true
if ${debug:-false}
then
    exec 2>/tmp/$(basename ${0})-trace$$.log
    set -x
    ask_debug=true
    export ask_debug
fi

#----------------------------------------------------------------------------------------
# show command help
#----------------------------------------------------------------------------------------
show_help ()
{
    echo 'Archmdir: a simple mailbox archiver for maildir mailboxes.'
    echo 'Archive all selected messages from the specified mailbox(es).'
    echo
    echo 'usage:'
    echo '  archmdir [-h|--version]'
    echo '  archmdir MODE [OPTIONS] -d <date> <mailbox> [<mailbox>, ...]'
    echo '  archmdir MODE [OPTIONS] -o <days> <mailbox> [<mailbox>, ...]'
    echo
    echo 'MODES'
    echo '  -a, --archive                 archive mode: archive selected messages'
    echo '  -k, --kill                    kill mode: delete selected messages'
    echo '  -l, --list                    list mode: list what messages will be'
    echo '                                selected'
    echo '  -y, --copy                    copy mode: copy selected messages'
    echo
    echo 'OPTIONS'
    echo '  -d, --date <date>             treshold date for messages'
    echo '  -D, --date-header             use 'Date:' header to age the messages'
    echo '  -e, --extension <extension>   suffix for the archive mailbox, "none" means no'
    echo '                                extension (default: archived)'
    echo '      --format <format>         specify mailboxes format (default: maildir)'
    echo '                                !! This option only exists for archmbox command'
    echo '                                !! line compatibility and is without any function.'
    echo '  -h, --help                    prints help'
    echo '      --keep-flagged            keep flagged messages'
    echo '      --keep-unread             keep unread messages'
    echo '  -o, --offset <days>           offset (in days) from today for treshold date'
    echo '                                -1 means "all messages"'
    echo '  -p, --archive-path <directory>'
    echo '                                where to store archive mailbox (full path)'
    echo '  -r, --reverse                 offset means messages newer rather than older'
    echo '      --simulate                simulate the choosen action only'
    echo '  -t, --tmpdir <directory>      directory to store temporary files (full path)'
    echo '                                (default: "/tmp")'
    echo '      --time <time>             treshold time to refine treshold date'
    echo '  -v, --verbose                 enable verbose mode'
    echo '      --version                 prints version number'
    echo
    echo '<date> must be supplied using the following format: yyyy-mm-dd'
    echo '<time> must be supplied using the following format: hh:mm:ss (24h)'
    echo '<mailbox> must be specified with full path.'
}

#========================================================================================
# main
#========================================================================================

pgmname=`basename "$0"`

action=''
compare_operator='lt'
date_header=0
date_str=''
date_secs=''
date_offset=''
dest_path=''
file_extension='archived'
keep_flagged=0
keep_unread=0
simulate_mode=0
src_files=''
time_str=''
tmp_path='/tmp'
verbose_mode=1

CMD_LINE="$0 $*"

while [ $# -gt 0 ]
do
    case $1 in
        # MODES
        -a|--archive )
            action='archive'
            shift
            ;;
        -y|--copy )
            action='copy'
            shift
            ;;
        -k|--kill )
            action='kill'
            shift
            ;;
        -l|--list )
            action='list'
            shift
            ;;
        # OPTIONS
        -d|--date )
            date_str="$2"
            shift; shift
            ;;
        -D|--date-header )
            date_header=1
            shift
            ;;
        -e|--extension )
            file_extension="$2"
            shift; shift
            ;;
        --format )
            # This option only exists for archmbox command
            # line compatibility and is without any function.
            shift; shift
            ;;
        -h|--help )
            show_help
            exit 0
            ;;
        --keep-flagged )
            keep_flagged=1
            shift
            ;;
        --keep-unread )
            keep_unread=1
            shift
            ;;
        -o|--offset )
            date_offset="$2"
            shift; shift
            ;;
        -p|--archive-path )
            dest_path="$2"
            shift; shift
            ;;
        -r|--reverse )
            compare_operator='gt'
            shift
            ;;
        --simulate )
            simulate_mode=1
            shift
            ;;
        --time )
            time_str="$2"
            shift; shift
            ;;
        -t|--tmpdir )
            tmp_path="$2"
            shift; shift
            ;;
        -v|--verbose )
            if [ "$2" = '2' ]
            then
                verbose_mode=2
            fi
            shift; shift
            ;;
        --version )
            echo "Archmdir v`grep '^# Last Update: *$Id:' ${pgmname} | sed 's/^.*\$Id: \([^ ]*\) \([0-9]*\) .*$/\2/'`"
            exit 0
            ;;
        * )
            if [ -z "$1" ]
            then
                echo "ERROR: At lease one mailbox must be specified."
                exit 1
            else
                # mailbox name given, check if it
                # starts with an absolut path (/)
                echo "$1" | grep -q '^ */'

                if [ $? -ne 0 ]
                then
                    echo "ERROR: Mailbox '$1' must be specified with an absolute path."
                    exit 1
                else
                    # absolute path given
                    # 1. escape all kind of special characters, e.g. '\2e' by '\\2e' etc.
                    # 2. unescape curly brackets '{}'
                    # 3. unescape curly brackets '()'
                    # 4. unescape whitespaces ' '
                    # 5. unescape ampersands '&'
                    # 6. unescape commas ','
                    if [ -z "${src_files}" ]
                    then
                        src_files="`printf '%q\n' "$1" | sed -e 's|\\\{|{|g' -e 's|\\\}|}|g' -e 's|\\\(|(|g' -e 's|\\\)|)|g' -e 's|\\\ | |g' -e 's|\\\&|\&|g' -e 's|\\\,|,|g'`\n"
                    else
                        src_files="${src_files}\n`printf '%q\n' "$1" | sed -e 's|\\\{|{|g' -e 's|\\\}|}|g' -e 's|\\\(|(|g' -e 's|\\\)|)|g'  -e 's|\\\ | |g' -e 's|\\\&|\&|g' -e 's|\\\,|,|g'`\n"
                    fi
                fi
            fi
            shift
            ;;
    esac
done

# action   : archive
# date     : 2017-09-05
# src_file : /home/ralf/.imapmail/.Ablage.Allgemein
# dest_path: /home/ralf/.imapmail/.Archiv.Ablage

# check date/offset validity
if [ -z "${date_str}" -a -z "${date_offset}" ]
then
    echo "ERROR: You must specify a date (-d) OR an offset (-o)."
    exit 1
elif [ -n "${date_str}" -a -n "${date_offset}" ]
then
    echo "ERROR: You cannot use a date (-d) AND an offset (-o) at the same time."

    exit 1
fi

if [ -n "${date_str}" ]
then
    if ! is_date "${date_str}"
    then
        echo "ERROR: Invalid date given, must be in the format YYYY-MM-DD."
        exit 1
    else
        if [ -n "${time_str}" ]
        then
            if ! is_time "${time_str}"
            then
                echo "ERROR: Invalid time given, must be in the format HH:MM:SS."
                exit 1
            else
                date_secs=`date +'%s' -d "${date_str} ${time_str}"`
            fi
        else
            date_secs=`date +'%s' -d "${date_str}"`
        fi
    fi
else
    if [ -n "${time_str}" ]
    then
        echo "ERROR: A time (--time) can only be used if a date (-d) has been specified."
        exit 1
    fi
fi

if [ -n "${date_offset}" ]
then
    if [ ${date_offset} -lt -1 ]
    then
        echo "ERROR: Offset must be a positive integer (or -1 for all messages)."
        exit 1
    elif [ "${date_offset}" = '-1' ]
    then
        # show all
        date_secs=0
    else
        date_secs=`date +'%s' -d "${date_offset} days ago"`
    fi
fi

# check if an archive path has been given
if [ \( "${action}" = 'archive' -o "${action}" = 'copy' \) -a -z "${dest_path}" ]
then
    echo "ERROR: An archive path (-p) is required to archive or copy nessages."
    exit 1
fi

if [ -n "${src_files}" ]
then
    # source file(s) given
  # echo "SRCFILES: ${src_files}"

    tmp_file=${tmp_path}/.${pgmname}-tmp.$$

    rm -f ${tmp_file}

    # read path list separated by '\n' (newline) to handle spaces in names correctly
    printf "${src_files}" | while IFS=$'\n' read -r SRCFILE
    do
        echo "${SRCFILE}" | grep -q '\.INBOX$'

        if [ $? -eq 0 ]
        then
            # remove /.INBOX from mailbox name
            search_path=`echo "${SRCFILE}" | sed 's|/\.INBOX$||'`
        else
            search_path="${SRCFILE}"
        fi

        # make sure that the directory existance check works as expected
        # 1. replace '\2e' by '&&'
        # 2. remove all occurencies if '\\'
        # 3. replace '&&' by '\2e'
        search_path="`echo "${search_path}" | sed -e 's#\\\2e#\&\&#g' -e 's/\\\//g' -e 's#\&\&#\\\2e#g'`"

        if [ -d "${search_path}" ]
        then
            # source file/directory exists
            # read file list separated by '\n' (newline) to handle spaces in names correctly
            find "${search_path}/cur" "${search_path}/new" "${search_path}/tmp" -name "[0-9]*" -printf "%p\n" 2>/dev/null | \
                sort | while IFS=$'\n' read -r SFILE
            do
                msg_path=`dirname "${SFILE}"`
                msg_file=`basename "${SFILE}"`

                if [ ${date_header} -eq 1 ]
                then
                    # read date from msg file header
                    msg_date_secs=`grep "^Date: " "${msg_path}/${msg_file}" | head -n1 | sed -e 's/^ *Date: *//' -e "s/\"/'/g" | date +'%s' -f -`
                else
                    # take date from msg file name
                    msg_date_secs=`echo "${msg_file}" | cut -d. -f1`
                fi

                # compare_operator: lt - messages older than the specified date
                #                   gt - messages newer than the specified date
                if [ "${date_offset}" = '-1' -o "${msg_date_secs}" -${compare_operator} ${date_secs} ]
                then
                    msg_date_str=`echo "@${msg_date_secs}" | date +'%Y-%m-%d' -f -`

                    # msg_file: 1504128331.00000012.farragut:2,SR
                    #
                    # message flag information starts with "2,". Each
                    # character after the comma is an independent flag.
                    #
                    # Flags:
                    # - 'D' (draft)  : the user considers this message a draft; toggled
                    #                  at user discretion.
                    # - 'F' (flagged): user-defined flag; toggled at user discretion.
                    # - 'P' (passed) : the user has resent/forwarded/bounced this
                    #                  message to someone else.
                    # - 'R' (replied): the user has replied to this message.
                    # - 'S' (seen)   : the user has viewed this message, though perhaps
                    #                  he didn't read all the way through it.
                    # - 'T' (trashed): the user has moved this message to the trash;
                    #                  the trash will be emptied by a later user action.

                    process_flag=1
                    if [ ${keep_flagged} -eq 1 ]
                    then
                        # check if 'F' flag (flagged) has been set
                        echo "${msg_file}" | grep -q -E ',[^,]*F[^,]*$'

                        if [ $? -eq 0 ]
                        then
                            # 'F' flag set, skip file
                            process_flag=0
                        fi
                    fi

                    if [ ${keep_unread} -eq 1 ]
                    then
                        # check if 'S' flag (seen) has been set
                        echo "${msg_file}" | grep -q -E ',[^,]*S[^,]*$'

                        if [ $? -ne 0 ]
                        then
                            # 'S' flag not set, skip file
                            process_flag=0
                        fi
                    fi

                    if [ ${process_flag} -eq 1 ]
                    then
                        # process message
                        # msg_date_str  = 2017-08-30
                        # msg_date_secs = 1504128331
                        #
                        # msg_path      = /home/ralf/.imapmail/.Ablage.Allgemein/[cur|new|tmp]
                        #               = /home/ralf/.imapmail/[cur|new|tmp]
                        #
                        # msg_file      = 1504128331.00000012.farragut:2,SR
                        #
                        # SRCFILE       = /home/ralf/.imapmail/.Ablage.Allgemein
                        #               = /home/ralf/.imapmail/.INBOX
                        echo "${msg_date_str}|${msg_date_secs}|${msg_path}|${msg_file}|${SRCFILE}/`basename "${msg_path}"`" >> ${tmp_file}
                    fi
                fi
            done
        fi

      # echo "ACTION: ${action}"

        case ${action} in
            archive|copy|kill )
                # archive, copy or delete a message
                if [ -s ${tmp_file} ]
                then
                    # mesaage list to process exists
                    idx=1
                    while read -r LINE
                    do
                        _old_ifs=${IFS}
                        IFS='|'
                        set ${LINE}

                        msg_date_str="$1"      # 2017-07-11
                        msg_date_secs="$2"     # 1499771722
                        msg_path="$3"          # /home/ralf/.imapmail/.Ablage.Allgemein/cur
                        msg_file="$4"          # 1499771723.00000001.server:2,S
                        msg_org_path="$5"      # /home/ralf/.imapmail/.Ablage.Allgemein/cur

                        IFS=${_old_ifs}

                        case ${action} in
                            archive|copy )
                                # file_extension='archived'
                                # dest_path     = /home/ralf/.imapmail/.Archiv/.Ablage
                                # msg_org_path  = /home/ralf/.imapmail/.Ablage(.Allgemein)(/cur)
                                # msg_path      = /home/ralf/.imapmail/.Ablage.Allgemein/cur
                                # dest_folder   = .Allgemein\2earchived/cur
                                # dest_path_with_folder= /home/ralf/.imapmail/.Archiv.Ablage.Allgemein\2earchived/cur
                                #
                                #  or
                                #
                                # file_extension= 'archived'
                                # dest_path     = /home/ralf/.imapmail/.Archiv
                                # msg_org_path  = /home/ralf/.imapmail/(.INBOX)(/cur)
                                # msg_path      = /home/ralf/.imapmail/cur
                                # dest_folder   = .INBOX\2earchived/cur
                                # dest_path_with_folder= /home/ralf/.imapmail/.Archiv.INBOX\2earchived/cur
                                #
                                # 1.1 strip of sub directory: '/home/ralf/.imapmail/.Ablage.Allgemein/cur' -> '/home/ralf/.imapmail/.Ablage.Allgemein'
                                # 1.2 remove directory path : '/home/ralf/.imapmail/.Ablage.Allgemein'     -> '.Allgemein' 
                                # 2.  grep sub directory    : '/home/ralf/.imapmail/.Allgemein/cur' -> '/cur'       
                                # 3.  build path + folder   : '/home/ralf/.imapmail/.Archiv/.Ablage' + '.Allgemein' + '\2e' + 'archived'' + '/cur'

                                dest_folder_part1=`echo "${msg_org_path}" | sed -e 's|^\(.*\)\/[^\/]*$|\1|' -e 's|^.*\.|.|'`
                                dest_folder_part2=`echo "${msg_org_path}" | sed -e 's|^.*\(\/[^\/]*\)$|\1|'`
                                dest_path_with_folder="${dest_path}${dest_folder_part1}\2e${file_extension}${dest_folder_part2}"
                                dest_path_without_folder="${dest_path}${dest_folder_part1}\2e${file_extension}"

                                if [ ${simulate_mode} -eq 1 ]
                                then
                                    # only simulate action
                                    if [ "${action}" = 'archive' ]
                                    then
                                        echo "${idx} - MOVE: ${msg_path}/${msg_file}"
                                        echo "           -> ${dest_path_with_folder}/"
                                    else
                                        echo "${idx} - COPY: ${msg_path}/${msg_file}"
                                        echo "           -> ${dest_path_with_folder}/"
                                    fi
                                else
                                    # make sure the default folder structure exists
                                    for DNAME in cur new tmp
                                    do
                                        if [ ! -d "${dest_path_without_folder}/${DNAME}" ]
                                        then
                                            # create folder
                                            mkdir -p "${dest_path_without_folder}/${DNAME}"
                                        fi
                                    done

                                    if [ ! -d "${dest_path_with_folder}" ]
                                    then
                                        # create destination folder
                                        mkdir -p "${dest_path_with_folder}"
                                    fi

                                    # run command
                                    if [ "${action}" = 'archive' ]
                                    then
                                        # move files to archive folder
                                        mv "${msg_path}/${msg_file}" "${dest_path_with_folder}/"
                                    else
                                        # copy files to archive folder
                                        cp "${msg_path}/${msg_file}" "${dest_path_with_folder}/"
                                    fi
                                fi
                                ;;
                            kill )
                                if [ ${simulate_mode} -eq 1 ]
                                then
                                    # only simulate action
                                    echo "${idx} - DELETE: ${msg_path}/${msg_file}"
                                else
                                    # delete files from source folder
                                    rm -f "${msg_path}/${msg_file}"
                                fi
                                ;;
                        esac

                        idx=`expr ${idx} + 1`
                    done < ${tmp_file}

                    # output summary
                    echo
                    echo "Mailbox '`echo "${msg_path}" | sed -e 's|/cur||' -e 's|/new||' -e 's|/tmp||'`': `expr ${idx} - 1` messages"
                    echo

                    if [ "${action}" != 'kill' ]
                    then
                        echo "Archive mailbox: `echo "${dest_path_with_folder}" | sed -e 's|/cur||' -e 's|/new||' -e 's|/tmp||'`'"
                    fi
                fi
                ;;

            list )
                # list mailbox content
                echo

                # print header
                if [ ${verbose_mode} -eq 2 ]
                then
                    echo '       DATE'
                fi

                echo 'ID     FROM                       SUBJECT'
                echo '--------------------------------------------------------------------------------'

                techo --begin 5 2 26 1 43

                idx=1
                while read -r LINE
                do
                    _old_ifs=${IFS}
                    IFS='|'
                    set ${LINE}

                    msg_date_str="$1"
                    msg_date_secs="$2"
                    msg_path="$3"
                    msg_file="$4"

                    IFS=${_old_ifs}

                    msg_date=`grep "^Date: "    "${msg_path}/${msg_file}" | sed -e 's/^ *Date: *//'    -e "s/\"/'/g"` # -e 's/^\(.\{25\}\).*/\1/'`
                    msg_from=`grep "^From: "    "${msg_path}/${msg_file}" | sed -e 's/^ *From: *//'    -e "s/\"/'/g"` # -e 's/^\(.\{25\}\).*/\1/'`
                    msg_subj=`grep "^Subject: " "${msg_path}/${msg_file}" | sed -e 's/^ *Subject: *//' -e "s/\"/'/g"` #  -e 's/^\(.\{42\}\).*/\1/'`

                    # print message information
                    if [ ${verbose_mode} -eq 2 ]
                    then
                        techo --row "${idx}" "" "${msg_date}"
                        techo --row ""       "" "${msg_from}" "" "${msg_subj}"
                    else
                        techo --row "${idx}" "" "${msg_from}" "" "${msg_subj}"
                    fi

                    idx=`expr ${idx} + 1`
                done < ${tmp_file}

                techo --end

                # output summary
                echo
                echo "Mailbox '`echo "${msg_path}" | sed -e 's|/cur||' -e 's|/new||' -e 's|/tmp||'`': `expr ${idx} - 1` messages"
                echo
                ;;

            * )
                # no action specified
                show_help
                ;;
        esac

        rm -f ${tmp_file}
    done
fi

#========================================================================================
# end
#========================================================================================
exit 0