#!/bin/bash

version=0.28
program=${0##*/}
progdir=${0%/*}
if [ "$progdir" = "$program" ]; then progdir="."; fi
hostname=$(hostname)

ext=${progdir##*/}

# ----------------------------------------------------------------------
function usage() {
    cat <<EOF
NAME
	$program - http 404 handler for a directory of ${ext}ized documents.

SYNOPSIS
	$program [OPTIONS] ARGS

DESCRIPTION
	This is an HTTP server error handler which handles missing ${ext}ized
        versions of IETF documents in the /$ext/ directory.  If a document is
        requested which is not present in $ext format in the directory, but
        the document is available in text version in the document repository,
        a $ext version is generated using the rfcmarkup tool.

        This program is designed to be run by the http server as a cgi program
        when a missing document exception is raised.  It is not designed to be
        run standalone from the command line; it takes a very limited number
        of switches only to inform the user of its version and this usage
        information:

EOF
    echo -e "OPTIONS"
    if   [ "$(uname)" = "Linux" ]; then
        egrep "^[	]+[-][A-Za-z| -]+\*?\)[	]+[A-Za-z].+#" $0 | tr -s "\t|" "\t," | sed -r -e 's/\)[ \t]+([A-Z]+)=\$2[^#]*#/=\1\t/' -e 's/\)[^#]*#/\t/'
    else
        egrep "^[	]+[-][A-Za-z| -]+\*?\)[	]+[A-Za-z].+#" $0 | sed 's/\|.*\$2[^#]*#/	/'| sed -E 's/\|.*\)[^#]*#/	/'
    fi
    cat <<EOF

FILES

AUTHOR
	Written by Henrik Levkowetz, <henrik@levkowetz.com>

COPYRIGHT
	Copyright 2006 Henrik Levkowetz.

	This program is free software; you can redistribute it and/or modify
	it under the terms of the GNU General Public License as published by
	the Free Software Foundation; either version 2 of the License, or (at
	your option) any later version. There is NO WARRANTY; not even the
	implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
	PURPOSE. See the GNU General Public License for more details.
	
EOF

}

# ----------------------------------------------------------------------
function die() {
    echo -e "\n$program: error: $*" > /dev/stderr
    exit 1
}

function note() { 
	if [ -n "$VERBOSE" ]; then echo -e "$*"; fi
}

# ----------------------------------------------------------------------
function version() {
	echo -e "$program $version"
}

# ----------------------------------------------------------------------
trap 'echo "$program($LINENO): Command failed with error code $? ([$$] $0 $*)"; exit 1' ERR


# ----------------------------------------------------------------------
# Option parsing

# Options
shortopts=hV
longopts=help,version

# Default values

if   [ "$(uname)" = "Linux" ]; then
    args=$(getopt -o "$shortopts" --long "$longopts" -n '$program' -- $SV "$@")
    if [ $? != 0 ] ; then die "Terminating..." >&2 ; exit 1 ; fi
    eval set -- "$args"
    sed="sed -r"
else
    # Darwin, BSDs
    args=$(getopt -o$shortopts $SV $*)
    if [ $? != 0 ] ; then die "Terminating..." >&2 ; exit 1 ; fi
    set -- $args
    sed="sed -E"
fi

while true ; do
    case "$1" in
	-h| --help)	usage; exit;;		# Show this help, then exit
#	-v| --verbose)	VERBOSE=1;;		# Be more talkative
	-V| --version)	version; exit;;		# Show program version, then exit
	--)		shift; break;;
	*) die "Internal error, inconsistent option specification: '$1'";;
    esac
    shift
done

# ----------------------------------------------------------------------
# Re-define die() and trap with cgi versions

function die() {
    cat <<-EOF
	Status: $1 $2
	Content-Type: text/html

	<!--#set var="TITLE" value="$2" -->
	<!--#include virtual="/error/include/top.html" -->

	    The requested document ($base) was not found on this server.

	  <!--#if expr="$HTTP_REFERER" -->

	    The link on the
	    <a href="<!--#echo encoding="url" var="HTTP_REFERER" -->">referring
	    page</a> seems to be wrong or outdated. Please inform the author of
	    <a href="<!--#echo encoding="url" var="HTTP_REFERER" -->">that page</a>
	    about the error.

	  <!--#else -->

	    If you entered the URL manually please check your
	    spelling and try again.

	  <!--#endif -->

	  <!--#if expr="$extra" -->
	    <p>
	    Additional information: "$extra"
	    </p>
	  <!--#endif -->

	<!--#include virtual="/error/include/bottom.html" -->
	EOF
    exit 0
}

# ----------------------------------------------------------------------
# usage()
function usage() {
    cat <<-EOF
	Status: 302 Usage
	Location: https://$SERVER_NAME/$ext/

	EOF
    exit 0
}


# ----------------------------------------------------------------------
# trap
trap 'die 404 "Content generation failed (on script line $LINENO) with error code $?"' ERR

# ----------------------------------------------------------------------
# Check preconditions
[ "$1" ] && export PATH_INFO="${1#/$ext}"
[ "$REDIRECT_URL" ] && export PATH_INFO="${REDIRECT_URL#/$ext}"
[ "$PATH_INFO" ] || usage
[ "$SERVER_NAME" ] || export SERVER_NAME="$(hostname)"

# ----------------------------------------------------------------------
# The program itself


# Get the base name
base="${PATH_INFO#/}"
base=${base//[^A-Za-z0-9_.+ -]/_}

# fix RFC case
base=${base/RFC/rfc}

# Document names don't contain spaces -- collapse them...
base=${base/rfc /rfc}
base=${base/draft /draft-}

orig=$base
base=${base%%/*}
base=${base%%[^A-Za-z0-9-]*}
if [ "$orig" != "$base" ]; then
    cat <<-EOF
	Status: 302 See Alternative
	Location: https://$SERVER_NAME/$ext/$base
	
	EOF
    exit
fi

# Check if it's a plain number
if [ ${base} -gt 0 ]; then base=rfc${base}; fi 2>/dev/null

# Determin path from beginning of string
if [ ${base:0:3} = "rfc" ]; then path=/home/ietf/rfc; fi
if [ ${base:0:3} = "std" ]; then path=/home/ietf/rfc/std; fi
if [ ${base:0:3} = "fyi" ]; then path=/home/ietf/rfc/fyi; fi
if [ ${base:0:3} = "bcp" ]; then path=/home/ietf/rfc/bcp; fi
if [ ${base:0:7} = "charter" ]; then path=/www/tools.ietf.org/charter; fi
if [ ${base:0:5} = "draft" ]; then path=/home/ietf/id; fi
if [ ${base:0:3} = "ion" ]; then path=/home/ietf/ion/approved; fi

if [ ${base:0:5} = "draft" -o ${base:0:3} = "ion" -o ${base:0:7} = "charter" ]; then
    file="$base.txt"
    if [ ! -f "$path/$file" ]; then
	declare -a names=($(wget -q -O - https://datatracker.ietf.org/api/v1/doc/document?name__startswith=$base | jq '.objects[]|.name+"-"+.rev' | tr -d '"'))
	if   [ ${#names[@]} = 0 ]; then
		cat <<-EOF
		Status: 404 Not Found

		EOF
		exit
	elif [ ${#names[@]} = 1 ]; then
		base=${names[0]//\"/}
		file="$base.txt"
	elif [ "${names[0]%-[0-9][0-9]}" = "$base" ]; then
		base=${names[0]}
		file="$base.txt"
	else
		cat <<-EOF
		Status: 302 See This
		Location: https://$SERVER_NAME/id/$base

		EOF
		exit
	fi
    fi
fi

#if [ -d "/www/tools.ietf.org/wg/$base" ]; then
#    cat <<-EOF
#	Status: 302 Found
#	Location: https://$SERVER_NAME/wg/$base
#
#	EOF
#    exit
#fi
#[ "$path" ] || die 404 "Not found."
if [ ! "$path" ]; then
    sleep 1
    path=/home/ietf/id;

    file=$(wget --prefer-family=IPv4 -S -O /dev/null https://tools.ietf.org/id/$base 2>&1 | grep "^Location:" | awk '{ sub(".*/", ""); print $1; }' )
    if [ "$file" ]; then
	base="${file%.*}"
    else
	cat <<-EOF
		Status: 302 See Other
		Location: https://$SERVER_NAME/id/$base

		EOF
	exit
    fi
fi

if [ "$path" = "/home/ietf/rfc" -o \
     "$path" = "/home/ietf/rfc/bcp" -o \
     "$path" = "/home/ietf/rfc/std" -o \
     "$path" = "/home/ietf/rfc/fyi" ]; then
    tail=${base#???}
    tail=${tail#0}
    tail=${tail#0}
    tail=${tail#0}
    base="${base:0:3}$tail"
    file="$base.txt"
fi

[ -f "$path/$file" ] || die 404 "Not found: ($file)"

urld="${REQUEST_URI%/*}"
urld="${urld#/}"
urlb="${REQUEST_URI##*/}"
[ "$urld" ] || urld=$ext

#if [ -f "$path/$base.$ext" ]; then
#    cp $path/$base.$ext .
if [ ! -f "$base.$ext" ]; then
    sleep 5

    unset QUERY_STRING
    SCRIPT_NAME="/$urld"
    #/www/tools.ietf.org/tools/rfcmarkup/rfcmarkup "static=true&doc=$base" > $base.$ext 2>&1
    if   [ "$ext" = "html" ]; then
	/www/tools.ietf.org/tools/rfcmarkup/rfcmarkup "robots=index,follow&staticpath=true&topmenu=true&doc=$base.txt" > $base.$ext 2>&1
	content_type="text/html"
    elif [ "$ext" = "abnf" ]; then
	/www/tools.ietf.org/tools/aex/aex $path/$base.txt > $base.$ext 2>&1	
	content_type="text/plain"
	if [ $(stat -c %s $base.$ext) = "0" ]; then echo -e ";\n; No ABNF found in $base.txt\n;" > $base.$ext; fi
    fi
    if head $base.$ext | grep -qs "^Location: "; then
	echo -e "Status: 302 Available\n"
	head $base.$ext
	mv $base.$ext $base.$ext.location
	exit
    fi
    if tail $base.$ext | grep -qs "<h3> &nbsp; &nbsp; Sorry, couldn't find"; then
	rm $base.$ext
    fi
fi

if [ -f "$base.$ext" ]; then
  if [ "$urlb" = "$base" ]; then
    cat <<-EOF
	Status: 200 OK
	Content-Type: $content_type; charset=UTF-8
	
	EOF
    cat	$base.$ext
  else
    cat <<-EOF
	Status: 302 Found
	Location: https://$SERVER_NAME/$urld/$base
	Content-Type: text/html; charset=UTF-8
	
	<html>	
	<body>
	   <p>
	      The document is available here: <a href="http://$SERVER_NAME/$urld/$base">http://$SERVER_NAME/$urld/$base</a>.
	   </p>
	</body>
	</html>
	EOF
  fi
else
    die 404 "Not Found [$base.$ext] "
fi