fetch-mailman-archives

#!/bin/sh

usage()
{
    cat >&2 <<__EOF__
Usage: $(basename $0) -U list_url -s start_year [ -m start_month ] [ -x extension ]

Options:
  -s start_year   -- Year at which to start downloading ML archives.

  -m start_month  -- Month at which to start downloading ML archives. Must
                     be specified by name, e.g. "November."

  -U list_url     -- The URL of the directory containing the ML archives.

  -x extension    -- An optional extension to add to each archive file,
                     e.g. "gz"

__EOF__
  exit ${1-1}
}

log()
{
    echo "$(basename $0): $1" 1>&2
}

while [ $# -gt 0 ]; do
    case $1 in
    -h|--help)
        usage 0
        ;;
    -m)
        shift; STARTMONTH="$1"
        ;;
    -s)
        shift; STARTYEAR="$1"
        ;;
    -U)
        shift; URL="$1"
        ;;
    -x)
        shift; EXT=".$1"
        ;;
    *)
        usage
        ;;
    esac
    shift
done

if [ -z "$URL" ] || [ -z "$STARTYEAR" ]; then
    usage
elif ! expr "$STARTYEAR" : '[1-9][0-9]*' >/dev/null 2>&1; then
    log "invalid year '$STARTYEAR'"
    usage
fi

MONTHS="January
        February
        March
        April
        May
        June
        July
        August
        September
        October
        November
        December"

YEAR=$STARTYEAR
while [ $YEAR -le $(date '+%Y') ]; do
    for MONTH in $MONTHS; do
        [ $YEAR -eq $STARTYEAR -a -n "$STARTMONTH" \
            -a -n "$(echo $MONTHS | grep ${MONTH}.*${STARTMONTH})" ] && continue
        fetch "${URL}/${YEAR}-${MONTH}.txt${EXT}"
        [ $? -eq 0 ] || continue
        case $EXT in
        .gz)
            gunzip ${YEAR}-${MONTH}.txt.gz
            ;;
        esac
        [ $YEAR -eq $(date '+%Y') -a $MONTH = $(date '+%B') ] && break
    done
    YEAR=$((YEAR + 1))
done