#!/bin/bash

MYVERSION=0.20080606

# If you improve this script, please read the comments at the
# end of this file.  Thanks.

# usage/help message
usage_message() {
cat >&2 <<EOT

Usage:  ${0##*/} [options] [file...]

Input files should be text files containing Message-ID headers.
E.g., they may be single messages or entire mailbox files.
If no files are given, stdin is read.
Output will be web archive URLs, one per line.

Options:

  -a google|freenix            specify an archive;  default is google
  -b firefox|konqueror|opera   send the output URLs to browser
  -c                           append to output file;  default is to overwrite
  -d                           avoid duplicate URLs;  can be cpu intensive
  -f                           force URLs to browser even if
                                 there are lots of them
  -h                           print this help text and exit
  -o filename                  send output URLs to named file
  -q                           suppress output of URLs to stdout

Options may be used in any combination, but -c does nothing without -o
and -f does nothing without -b.

EOT
}

# initialize a few variables
BROWSERCOMMAND=""  # Set default to "firefox", "konqueror", or
                   # "opera -newpage" if you want.
BROWSERLIMIT=20    # This is the max number of URLs which will
                   # be sent to the browser by default.
                   # Override with -f or see FORCE, below.
FORCE=no           # Set to yes if you want to send to browser
                   # always, no matter how many URLs there are.
                   # *** recommend leave as no! ***
OUTPUTFILE=""      # Change to e.g. "./urllist.txt" to set a default.
APPEND=no          # Change to yes if you always want to append
                   # URLs to the output file rather than overwriting.
ARCHIVE=google     # Set default archive to google or freenix.
DUPEKILL=no        # Change to yes if you never want dupe URLs.
QUIET=no           # Can be changed to yes, but there's no -v
                   # switch to turn output back on.

# formats and exit codes for errors
ERRFMT=${0##*/}:\ error:\ %b\\\n    # printf format for errors
WRNFMT=${0##*/}:\ warning:\ %b\\\n  # printf format for warnings
E_ARGS=65                           # something wrong with the arguments
E_INPUTFILE=3                       # input file not usable
E_OUTPUTFILE=4                      # output file not usable
E_EMPTYOUT=5                        # no MIDs found, so no output
E_OVERLIMIT=6                       # browser called, but too many URLs
                                    #   to send to browser
E_SED=7                             # sed exited with an error
                                    #  this should never happen
                                    #  if it does, this script needs debugging

# parse the arguments
while getopts hqa:b:o:fdcV ARGS 2>/dev/null; do
	case $ARGS in
		\?) printf "$ERRFMT" "I cannot make sense of those parameters" >&2
		   usage_message
		   exit $E_ARGS
		   ;;
		h) usage_message; exit 10; ;;
		b) if [[ $OPTARG == firefox ]]; then BROWSERCOMMAND=firefox
		   elif [[ $OPTARG == konqueror ]]; then BROWSERCOMMAND=konqueror
		   elif [[ $OPTARG == opera ]]; then BROWSERCOMMAND=opera\ -newpage
		   else
				printf "$ERRFMT" "$OPTARG is not a supported browser" >&2
				usage_message
				exit $E_ARGS
		   fi
		   ;;
		a) # if a bad arg was passed with -a, there's no error
  		   # the default browser specified above is used
		   [[ $OPTARG == google ]] && ARCHIVE=google
		   [[ $OPTARG == freenix ]] && ARCHIVE=freenix
		   ;;
		o) OUTPUTFILE=$OPTARG; ;;
		c) APPEND=yes; ;;
		d) DUPEKILL=yes; ;;
		q) QUIET=yes; ;;
		f) FORCE=yes; ;;
		V) echo -e "\e[1;33m${0##*/}: version $MYVERSION\e[0m"; exit 10; ;;
	esac
done

# having caught the options, move to other parameters
shift $((OPTIND - 1))

# some error checking on input files
# not necessary, but prevents some errors sed would throw
for INPUT in "$@"; do
	[[ ! -f $INPUT ]] && [[ ! -c $INPUT ]] && {
		printf "$ERRFMT" "$INPUT is not a file I can use as input" >&2
		usage_message
		exit $E_INPUTFILE
		}
	[[ -r $INPUT ]] || {
		printf "$ERRFMT" "I have no read access to $INPUT" >&2
		usage_message
		exit $E_INPUTFILE
		}
done

# some error checking on output file, if -o given
# not necessary, but prevents errors later
[[ -e $OUTPUTFILE ]] && {
	[[ ! -f $OUTPUTFILE ]] && [[ ! -c $OUTPUTFILE ]] && {
		printf "$ERRFMT" "$OUTPUTFILE is not a file I can use for output" >&2
		usage_message
		exit $E_OUTPUTFILE
		}
	[[ -w $OUTPUTFILE ]] || {
		printf "$ERRFMT" "I have no write access to $OUTPUTFILE" >&2
		usage_message
		exit $E_OUTPUTFILE
		}
	}

# sed commands used to URL-encode the MID fields
#   I think these are all the characters which can be
#   in MIDs but must be encoded in URLs.
#   See RFC 2822, section 3.4.2.
#   Note that the < > and @ characters are encoded by
#   the last three commands.
URLENC_SED='s~%~%25~g; s~!~%21~g; s~#~%23~g; s~\$~%24~g; s~'\''~%27~g; s~&~%26~g; s~\*~%2a~g; s~\+~%2b~g; s~-~%2d~g; s~/~%2f~g; s~\?~%3f~g; s~\^~%5e~g; s~_~%5f~g; s~`~%60~g; s~{~%7b~g; s~|~%7c~g; s~}~%7d~g; s/~/%7e/g; s~<~%3c~g; s~>~%3e~g; s~@~%40~g'

# make the list of URLS
[[ $ARCHIVE == google ]] && {
	URLLIST=$(sed -n "\~^Message-ID: <[^<>@]\+@[^<>@]\+>[[:space:]]\?\$~I{s~^Message-ID:~~I; s~[[:space:]]~~g; $URLENC_SED; s~^%3c\(.*\)%3e$~http://groups.google.com/groups?as_umsgid=\1~; p}" "$@" ) || {
		printf "$ERRFMT" "*** exiting due to a sed error ***" \
		                 "***  please, please debug me   ***" >&2
		exit "$E_SED"
		}
	}


[[ $ARCHIVE == freenix ]] && {
	URLLIST=$(sed -n "\~^Message-ID: <[^<>@]\+@[^<>@]\+>[[:space:]]\?\$~I{s~^Message-ID:~~I; s~[[:space:]]~~g; $URLENC_SED; s~.*~http://howardk.freenix.org/msgid.cgi\?STYPE=msgid\&MSGI=\0~; p}" "$@" ) || {
		printf "$ERRFMT" "*** exiting due to a sed error ***" \
		                 "***  please, please debug me   ***" >&2
		exit "$E_SED"
		}
	}

# quit now if nothing found
[[ -z $URLLIST ]] && {
	printf "$WRNFMT" "no Message-ID headers were found" >&2
	exit $E_EMPTYOUT
	}

# destroy dupes if -d was given
#   sed commands modified from example at Eric Pement's archive
#   <http://www.student.northpark.edu/pemente/sed/sed1line.txt>.
#   note:  this scales poorly.  for me, it's fine with a few
#   hundred MIDs, but takes few minutes of heavy cpu usage
#   with several thousand MIDs.
[[ $DUPEKILL == yes ]] && { 
	URLLIST=$(echo -e "$URLLIST" | sed -n 'G; s~\n~&&~; \~^\(.*\n\).*\n\1~d; s~\n~~; h; P' )
	(( $? )) && {
		printf "$ERRFMT" "*** exiting due to a sed error ***" \
		                 "***  please, please debug me   ***" >&2
		exit "$E_SED"
		}
	}

# send the URLs to stdout, file, and/or browser
[[ $QUIET == no ]] && echo -e "$URLLIST"

[[ -n $OUTPUTFILE ]] && {
	[[ $APPEND == yes ]] &&
		echo -e "$URLLIST" >> "$OUTPUTFILE" ||
		echo -e "$URLLIST" > "$OUTPUTFILE"
	}
[[ -n $BROWSERCOMMAND ]] && {
	URLCOUNT=$(echo "$URLLIST" | wc -l)
	[[ $URLCOUNT -gt $BROWSERLIMIT ]] && [[ $FORCE == no ]] && {
		printf "$WRNFMT" \
			"there are $URLCOUNT URLs, too many to send to the browser" \
			"running me with -f would send them anyway" >&2
		exit $E_OVERLIMIT
		}
	echo -e "$URLLIST" | xargs $BROWSERCOMMAND &>/dev/null &
	}

exit 0

##################################################################
##                                                              ##
##   This script is provided as-is, with no warranty.  Use it   ##
##   however you want.  No rights are reserved.                 ##
##                                                              ##
##   If you improve the script, please mail your changes to     ##
##   boxcars at gmx dot net.                                    ##
##                                                              ##
##################################################################