#!/bin/bash

# DoNG - Do News Groups
#
# A script to par2 recover and auto-extract newsgroup downloads.
#
# usage: dong [-n] [stubs]
# -n 		don't extract
# -l num        Don't process stubs with more than num files (default 100)
# -c num        Don't process more than num stubs (default unlimited)
# -e glob       Pass files that match glob to par2 to see if they have more data
# stubs		enough of a filename to uniquely identify an archive set
#		if not given, a set is generated from par2 files in the current directory


DONEDIR="finished"
TRASHDIR="trash"
LOGFILE="$(basename $0).log"
TMP="/tmp/$(basename $0).$$"

# Generate list of stubs that have associated par2 files in the current directory.
# eg:  foo.par2,  foo.vol01+02.PAR2, etc  -->  foo

stubs () {
 ls *[Pp][Aa][Rr]2						|
  sed -r 's/(\.[Vv][Oo][Ll][0-9]+\+[0-9]+)?(_copy_[0-9]+)?\.[Pp][Aa][Rr]2$//'	|
  uniq |
   while read stub; do
     files=("$stub"*)
     if (( ${#files[@]} > "$fileslimit" )) ;then
       echo "Stub \"$stub\" matches too many files -- try -l ${#files[@]} if it really is just one file set." >&2
     else
       echo "$stub"
     fi
   done |
   if [[ -z "$stubslimit" ]];then
     cat
   else
     head -n "$stubslimit"
   fi
}


# Generate a single par2 file from a stub

stub2parfile () {
  ls -- "$1"*[Pp][Aa][Rr]2 | head -n1
}


# Use par2 to check a stub
# Repair files if possible
# Note: Must return par2-success-status as exit code

parcheck () {
 stub="$1"
 echo -n "$stub  -  "
 par2 r "$(stub2parfile "$stub")" "$stub"*  $extra > "$TMP"
 ret=$?
 sed -r 's/.*\r//' "$TMP" >> "$LOGFILE"
 awk '
  BEGIN	{ RS="[\n\r]" }
  /Loaded [0-9]+ new packets including [0-9]+ recovery blocks/	{ recovery += $6}
  /There are a total of [0-9]+ data blocks/			{ total = $6	}
  /You have [0-9]+ out of [0-9]+ data blocks available/ 	{ have = $3	}
  /All files are correct, repair is not required/		{ have = total	}
  END	{ print ( 100 * ( have + recovery )) / total  "% complete"; }' "$TMP"
 return $ret
}


# Extract a rar file $1 to directory $2

handlerar () {
 origdir="`pwd`"
 cd -- "$2"
 unrar x -- "$origdir/$1" >> "$origdir/$LOGFILE"
 ret=$?
 cd -- "$origdir"
 return $ret
}


# Extract a zip file $1 to directory $2

handlezip () {
 unzip -d "$2" "$1" >> "$LOGFILE"
}


# The file was split directly, and par2 already reassembled it.

handleparsplit () {
 mv "$1" "$2"
}


# Nothing is known about the packing method.  Just move the files 
# described in par2's output.

handlefiles () {
 < "$1" filesinpar | tr \\n \\0 | xargs -0 bash -c '
   dest="$1"
   shift
   mv "$@" "$dest"' -s "$2"
}


# Handle a completed download

handlefinisheddownload () {
 stub="$1"

 if $extract; then
  destdir="$DONEDIR/$stub"
  mkdir -p -- "$destdir"

  echo -n "Extracting files...  "

  # Test for rar file
  files=("$stub"*.[Rr][Aa][Rr])
  if [[ -e "${files[0]}" ]];then 
   handler=handlerar
  else

   # Test for zip fie
   files=("$stub"*.[Zz][Ii][Pp])
    if [[ -e "${files[0]}" ]];then 
     handler=handlezip
   else

    # Assume the stub is the file
    if [[ -e "$stub" ]];then
     files=("$stub")
     handler=handleparsplit
    else

     # Add more filetype checks here...

     # Unknown archive type.  Just move the par-protected files.
     files=("$TMP")
     handler=handlefiles
    fi
   fi
  fi

  $handler "${files[0]}" "$destdir"
  ret=$?

  if [[ $ret == 0 ]];then
   echo "done!"
   mkdir -p -- "$TRASHDIR"
   mv -- "$stub"* "$TRASHDIR"
  else
   echo "extract failed!"
   fi
  return $ret

 else
  # Don't extract

  # Don't do anything?
  :
 fi
}


usage () {
 echo "$(basename "$0") [-n] [-c stubcount] [-e extrafiles] [-l limitperstub] [stub [stub ...]]" >&2
 exit 1
}


# MAIN

extra=''
fileslimit=100
stubslimit=''
extract=true
while getopts c:e:l:nh opt;do
 case $opt in
  c)	stubslimit="$OPTARG";;
  e)	extra="$OPTARG";;
  l)	fileslimit="$OPTARG";;
  n)	extract=false;;
  h)	usage;;
 esac
done
shift `expr $OPTIND - 1`

{ echo -n "Starting $(basename "$0") run at "; date; } >> "$LOGFILE"

{
 if [[ $# -eq 0 ]];then
  stubs
 else
  while [[ $# -gt 0 ]];do
   echo "$1"
   shift
  done
 fi
} | {
 while read stub;do
  parcheck "$stub"
  if [[ $? -eq 0 ]];then
   handlefinisheddownload "$stub"
  fi
 done
}

{ echo -n "Finished $(basename "$0") run at "; date; } >> "$LOGFILE"

rm "$TMP"