#!/bin/sh

# Downloads a gopher file or directory to the current directory
# Only uses port 70 for now, as I've never seen a gopher server
# use anything else.
#
# If called on a directory, it will download all of the urls that
# link to files within the directory or at deeper levels.
#
# So for example if a directory has a gophermap that links to stuff
# on other servers or outside the directory, they will be ignored.
#
# Warning: this overwrites files without any prompt

me=`basename "$0"`
if [ "$#" -ne 1 ]; then
    echo "usage: $me gopher://domain.abc/path/to/file"
    exit
fi

pushd () {
    command pushd "$@" > /dev/null
}

popd () {
    command popd "$@" > /dev/null
}

selfpath="$0"
url="$1"

url="${url#*//}" # strip protocol

domain=$(domain "gopher://$url")
path="/${url#*/}" # strip domain

echo $path

map="$(mktemp)"
echo "$path" | nc $domain 70 > "$map"

cat "$map" |
{
# we need these brackets do to everything in the subshell and
# retain the value of $quit

quit=0

while read line
do
    # this is not a gophermap, therefore use the output as a
    # single file download
    if ! [[ $line =~ ^[0135789ghiIds\;cM.].*$ ]]
    then
        filename="$(basename "$path")"
        cp "$map" "$filename"
        quit=1
        break
    fi
done

if [[ $quit -eq 1 ]]; then
    exit
fi

cat "$map" | while read line
do
    # these are all the prefixes for filetypes and directories
    if [[ $line =~ ^[0159ghIds\;cM].*$ ]]
    then
        split=()
        IFS=$'\t' read -r -a split <<< "$line"

        # the link name becomes the selector if the selector is
        # omitted
        selector="${split[0]}"

        if [[ "${split[1]}" ]]; then
            selector="${split[1]}"
        fi

        # no domain means current domain
        # ignore urls outside current domain
        if ! [[ "${split[2]}" ]] || \
             [[ "${split[2]}" = "$domain" ]]
        then
            isdir=0
            if [[ $line =~ ^[1].*$ ]]
            then
                isdir=1
            fi

            # extract relative path by removing $path from the
            # selector
            relative="$(echo $selector | sed -e "s|$path||g")"

            # Recurse directory or download file if the file is
            # at a deeper level than the current directory
            case $selector/ in
              $path*)
                  [[ $isdir -eq 1 ]] && \
                      mkdir -p "./$relative" && \
                      pushd "./$relative"

                  # !!!! recursion !!!!
                  $selfpath "$(echo "$url/$relative" | \
                      sed -e 's|//|/|g')"

                  [[ $isdir -eq 1 ]] && popd
                  ;;
            esac
        fi
        # end domain check
    fi
    # end type check
done
}
# end cat "$map"