#!/bin/sh

# TODO:
# * Look into ssh-hp :). (ie the slowness from ssh may just be from small
#    buffers).
# * bring back stderr, currently there is no notification of errors, not even
#   ENODIR or device full.
# * fcp src1 src2 src3 ... dst/ doesn't work (but glob does).
#   Need to take the last positional arg as dest (and unset it) then use
#   positional args at sources. Go through each one and make sure they are all
#   local because we don't support mixed remote and local.
# * Doesn't work when the remote end is android. Although if I ssh over and run
#   the fcp_ script manually it does work!?! (Maybe something about android
#   tar, saw something else (what?) making a fifo instead of pipe.)
# * There are a couple of calls to eval in which any redirect characters
#   (angle brackets) could lose our output. They need to be escaped.
# * Not sure about the handling of paths ending in slash or not is proper. 
#   (Eg where we rename the dest file.)
#   I assume if it has no slash it the directory being sent should be renamed
#   to the destination name
#   (unless it is an existing directory on the dest host).
# * If the receiving (connecting) nc is started before the sending (server)
#   one it will wait for one second and try again. They are started in the
#   right order but if there are delays in starting the interpreter or
#   something that sleep may have to be changed to something more robust.
# * If you want something more sophisticated look at http://www.slac.stanford.edu/~abh/bbcp/

portno=4365
comp=""

usage() {
  cat <<EOM
Usage: $(basename $0) [-c] [-p portno] [--] [srchost:]srcpath [[dsthost:]dstpath]

Copy files between hosts on a network. Uses tar and netcat (nc) for
transferring data and ssh for setting up the transport.

\`dstpath\` defaults to the current directory, use "-" to get a tarball on
stdout.

Options:
  -c         Tell tar to use gzip compression.
  -p portno  Port used by netcat. Defaults to $portno.
EOM
}

a1="$1"
# Do a seperated gotopt run so we cat get the return code.
getopt -Q -o hcp: -l help -- "${@}" >/dev/null || {
  ret=$?
  usage
  exit $ret
}
eval "set -- $(getopt -o hcp: -l help -- "${@}")"

# Sometimes set and/or getopt leave a -- there when it shouldn't be
[ "$a1" != "--" ] && [ "$1" = "--" ] && shift

while [ -n "$1" ];do
  case "$1" in
    -h|--help)
      usage
      exit
      ;;
    -c) comp="-z" ;;
    -p)
      portno="$2"
      shift
      ;;
    --)
      shift
      break
      ;;
    -*)
      echo "Unrecognized argument: $1"
      usage
      exit 1
      ;;
    *) break ;;
  esac
  shift
done

while [ -n "$1" ];do
  case "$1" in
    *)
      if [ -z "$srcpath" ];then
        srcpath="$1"
      elif [ -z "$dstpath" ];then
        dstpath="$1"
      else
        echo "Too many positional arguments (expected 2)." >&2
        echo "Try escaping any globs in the destination path." >&2
        usage
        exit 1
      fi
      ;;
  esac
  shift
done

if [ -z "$srcpath" ];then
  usage
  exit 1
fi
[ "$srcpath" = "." ] && srcpath="./"
[ "$srcpath" = ".." ] && srcpath="../"
[ "$dstpath" = "." ] && dstpath="./"
[ "$dstpath" = ".." ] && dstpath="../"
[ -z "$dstpath" ] && dstpath="./"

case "$srcpath" in
  *\\:*) continue ;;
  *:*)
    srchost="${srcpath%%:*}"
    srcpath="${srcpath#*:}"
    ;;
esac
case "$dstpath" in
  *\\:*) continue ;;
  *:*)
    dsthost="${dstpath%%:*}"
    dstpath="${dstpath#*:}"
    ;;
esac

sendcmd() {
  dsthost="${1#*@}"
  srcpath="$2"
  srcdir="$(dirname "$srcpath")"
  srcbase="$(basename "$srcpath")"
  dstpath="$3"
  [ "${dstpath%/}" = "$dstpath" ] && dstbase="$(basename "$dstpath")"
  cat <<EOS
#!/bin/sh
srcdir="\$(eval echo "${srcdir}")"
cd "\$srcdir" || echo \$?
# Use find here to expand any globs. Eval doesn't like filenames with parens
# in them. This can leave us with multiple newline seperated files hence we
# pass them to tar below via stdin.
srcbase="\$(find . -maxdepth 1 -name "${srcbase}" -exec basename {} \\;)"
# Fallback in case the above failed for whatever reason.
[ -z "\$srcbase" ] && srcbase="${srcbase}"
if ! [ -d "\$srcdir/\$srcbase" ] && [ -n "$dstbase" ] && [ "$dstbase" != "-" ] ;then
  trans="--transform=s/\${srcbase}\$/$dstbase/"
else
  # dummy argument so there isn't an empty "" passed to tar
  trans=--check-device
fi
tar --help 2>/dev/null | grep -q check-device || trans=v
pv=pv
type pv >/dev/null 2>&1 || pv=cat
# Try nc twice, with a one second timeout. In case we get called before the
# remote end has started up. May need to come up with something more
# sophisticated in situations where we have to wait longer for the initial
# setup.
echo "\$srcbase" | tar -cT- ${comp} "\$trans" |
        \$pv | ( nc -q 0 -w 3 "${dsthost#*@}" $portno 2>/dev/null ||
            { sleep 1 && nc -q 0 -w 3 "${dsthost#*@}" $portno ; } )
EOS
}

recvcmd() {
  dstpath="$1"
  srcbase="$(basename "$2")"
  cat <<EOS
#!/bin/sh
dstpath="\$(eval echo ${dstpath})"
dstbase="\$(basename "\${dstpath}")"
if [ -d "\$dstpath" ] || [ "\${dstpath%/}" != "\$dstpath" ] ;then
  dstdir="\$dstpath"
  # dummy argument so there isn't an empty "" passed to tar
  trans=--check-device
else
  # Strip trailing component and rename the incoming root object to that.
  # I hope that is what was intended. If not make sure dstpath is
  # terminated with a slash (/).
  dstdir="\$(dirname "\$dstpath")"
  trans="--transform=s/^\([\.\/]*\)${srcbase}/\\1\$dstbase/"
fi
tar --help 2>/dev/null | grep -q check-device || trans=-v
pv=pv
type pv >/dev/null 2>&1 || pv=cat
if [ "\$dstbase" != '-' ] ;then
  mkdir -p "\$dstdir" || exit \$?
  cd "\$dstdir" || exit \$?
fi
nc -lp $portno -w 20 | \$pv | {
    if [ "\$dstbase" != '-' ] ;then
      tar -x ${comp} "\$trans"
    else
      cat
    fi
  }
EOS
}

rpc() {
  ssh "$1" "export ff=/tmp/fcp_$$_\$\$_cmd ; cat > \$ff ; nohup sh -c \"sh \$ff && rm -f \$ff\" </dev/null >/dev/null 2>&1 &"
  [ $? -ne 0 ] && echo "Try escaping any ':' in a path if it is not supposed to be a host seperator." >&2
}

myip() {
  # dig and nslookup and hosts aren't on all machines and don't look in
  # /etc/hosts. Getent isn't everywhere and sometimes returns ipv6 addreses on
  # non-ipv6 enabled hosts. getent ahostsv4 I don't know how prevalent that is.
  tip="$(ping -c1 -W0.1 ${1#*@} 2>&1 | tr -d '():' | awk '/^PING/{print $3}')"
  [ -n "$tip" ] &&
    ip route get "$tip" | sed -n 's/^.*src \([^ ]*\).*$/\1/p'
}

if [ -n "$dsthost" ] && [ -n "$srchost" ];then
  recvcmd "$dstpath" "$srcpath" | rpc "$dsthost"
  sendcmd "$dsthost" "$srcpath" "$dstpath" | rpc "$srchost"
  exit $?
elif [ -n "$dsthost" ];then
  recvcmd "$dstpath" "$srcpath" | rpc "$dsthost"
  sendcmd "$dsthost" "$srcpath" "$dstpath" > /tmp/fcp_$$_cmd
  sh /tmp/fcp_$$_cmd
  rm /tmp/fcp_$$_cmd
  exit $?
elif [ -n "$srchost" ];then
  dsthost="$(myip "$srchost")"
  [ -z "$dsthost" ] && {
    echo "Couldn't find route to $srchost" >&2
    exit 1
  }
  recvcmd "$dstpath" "$srcpath" > /tmp/fcp_$$_cmd
  sh /tmp/fcp_$$_cmd &
  sendcmd "$dsthost" "$srcpath" "$dstpath" | rpc "$srchost"
  wait
  rm /tmp/fcp_$$_cmd
  exit $?
else
  srcdir="$(dirname "$srcpath")"
  srcbase="$(basename "$srcpath")"
  if [ -d "$dstpath" ] || [ "${dstpath%/}" != "$dstpath" ];then
    dstdir="$dstpath"
  else
    dstdir="$(dirname "$dstpath")"
  fi
  # TODO: Handle file renames.
  dstpath="`eval echo ${dstpath}`"
  mkdir -p "$dstpath"
  tar -C "$srcdir" -c "$srcbase" | tar -C "$dstdir" -x
  exit $?
fi

