#! /bin/sh
##########################################################################
# Title      :	urlgetopt - decode "urlencoded" forms data (RFC 1866)
# Author     :	Heiner Steven <heiner.steven@odn.de>
# Date       :	1999-06-27
# Category   :  CGI, File Conversion
# Requires   :	-
# SCCS-Id.   :	@(#) urlgetopt	1.6 05/04/20
##########################################################################
# usage
#    urlgetopt [-l] [-p prefix] [urlencoded-data]
#        -l: print fields on a separate line
#
# Description
#    Data entered within a HTML-FORM is transfered to the HTTP-Server
#    using the "application/x-www-form-urlencoded" encoding type
#    This script decodes "form-urlencoded" data in a form
#    suitable to be used within shell scripts
#
#    Example:
#     The form-urlencoded string
#	  name=Heiner+Steven&org=Heiner%27s+SHELLdorado
#
#     will be decoded to
#	  name='Heiner Steven'
#	  org='Heiner'"'"'s SHELLdorado'
#
#    usage example (i.e. from within a CGI script):
#
#	eval `urlgetopt -p FORM_ -l "$QUERY_STRING"`
#
#    Without "eval":
#
#	OIFS=$IFS
#	IFS="\n"
#	set -- `urlgetopt -l "$QUERY_STRING"`
#	IFS=$OIFS
#	echo $# arguments:
#	for arg
#	do
#	    echo "<$arg>"
#	done
#
# Notes
#    o	Assumes form-data in the format "name=value", but should
#	handle other values as well
#    o	An apostrophe ' within apostrophes cannot be written as '\'' (in sh,
#	bash, ksh88), so we use the work around ' "'" '. Using ksh93,
#	we could just use $'\''.
#    o	If the last character of a value in an assignment is an
#	apostrophe ('), the printed assignment will contain two
#	superfluous apostrophes ('').
#
# Portability
#	Solaris 9 "nawk"
#	Linux 2.4 "awk" (= GNU awk) 3.1.0
#
# Changes:
# 2005-04-20 claudio	allow values with leading whitespace (thanks to
#			Claudio Jolowicz <claudio@jolowicz.com>) (1.6)
# 2002-03-26 heiner	Handle a "-n" argument correctly (1.4)
#			Thanks to Brian Hiles <bsh@rawbw.com> for
#			pointing this out.
##########################################################################

PN=`basename "$0"`			# Program name
VER="1.6"

# Set the following variable to shorten startup time. Otherwise the
# directories from the PATH variable will be searched for a suitable AWK
# implementation
#: ${NAWK=nawk}			# Set this variable to speed up startup

usage () {
    echo >&2 "$PN - decode \"urlencoded\" CGI form data, $VER
usage: $PN [-lF] [-p prefix] [urlencoded-string]
    -l:	print fields on a separate line
    -p: prefix for generated environment variable names
    -F: force output of invalid assignments

Example:
    eval \`urlgetopt -l \"\$QUERY_STRING\"\`"
    exit 1
}

# We use "getopts" instead of "getopt" to preserve whitespace within
# arguments.

VarPrefix=
LongFormat=no
EvalCheck=true
while getopts :hFlp: opt
do
    case "$opt" in
    	F)	EvalCheck=false;;
	h)	usage;;
	l)	LongFormat=yes;;
	p)	VarPrefix=$OPTARG;;
	?)	usage;;
    esac
done
shift `expr $OPTIND - 1`

# We need a new AWK program supporting the "gsub()" function.
# Most AWKs (i.e. "gawk") do support it, but some older NAWKs do not.
# Search for a "gsub" capable NAWK using the current PATH.
if [ X"$NAWK" = X ]
then
    for path in `echo "$PATH" | sed 's|^:|./ |;s|:$| ./|;s|:| |g'`
    do
	for awk in $path/*awk
	do
	    [ -f "$awk" -a -x "$awk" ] || continue
	    case "$awk" in
		*.*) continue;		# ignore "script.awk"
	    esac
	    # Now check for the "gsub()" function
	    result=`echo "UU" | $awk '{ gsub (/U/, "X"); print }' 2>/dev/null`
	    [ X"$result" = X"XX" ] || continue
	    NAWK=$awk; break 2		# Found!
	done
    done
    : ${NAWK:=awk}
fi

if [ $# -gt 0 ]
then			# Process arguments, if specified...
    # Special handling: BSD derived "echo" implementations
    # will not echo "-n", but omit the trailing newline. "echo --"
    # doesn't work as expected, too, so we add a non-dash character
    # as the first character to the "echo", and remove it later on.

    echo X"$@" |
	sed "s/^X//"
else			# ...otherwise read standard input
    cat
fi |
    $NAWK -F'[&]' '
	BEGIN {
	    FieldSep = ("'"$LongFormat"'" == "yes") ? "\n" : " "
	    VarPrefix = "'"$VarPrefix"'"
	    evalcheck = ("'"$EvalCheck"'" == "true")

	    Hex ["0"] =  0; Hex ["1"] =  1; Hex ["2"] =  2; Hex ["3"] =  3;
	    Hex ["4"] =  4; Hex ["5"] =  5; Hex ["6"] =  6; Hex ["7"] =  7;
	    Hex ["8"] =  8; Hex ["9"] =  9; Hex ["A"] = 10; Hex ["B"] = 11;
	    Hex ["C"] = 12; Hex ["D"] = 13; Hex ["E"] = 14; Hex ["F"] = 15;
	    squote = sprintf ("%c", 39)
	    exitcode = 0
	}
	{
	    gsub (/\+/, " ");

	    # Parse assignment separated by "&", e.g. "a=b&c=d&..."

	    for ( field=1; field<=NF; ++field ) {
		# Substitute %HH with the "real" character
		if ( $field ~ /%[0-9A-F][0-9A-F]/ ) {
		    newfield = ""
		    for ( i=1; i<=length ($field); i++ ) {
			if ( substr ($field, i, 1) == "%" ) {
			    dec = Hex [substr ($field, i+1, 1)] * 16 + \
				    Hex [substr ($field, i+2, 1)]
			    newfield = sprintf ("%s%c", newfield, dec)
			    i += 2;
			} else {
			    newfield = newfield substr ($field, i, 1);
			}
		    }
		    $field = newfield
		}
		#print "after % processing " $0

		# Decode "var=value" pairs.
		#
		# If "evalcheck" is true, we ensure that the result is a
		# valid variable assignment:
		#   1.	it has the form varname=value
		#   2.	the variable name must be a valid shell
		#   	identifier: start with letter or underscore, followed
		#	only be letters, underscores or digits

		if (evalcheck && !match ($field, /^[a-zA-Z_][a-zA-Z_0-9]*=/)) {
		    print "invalid assignment: " $field | "cat >&2"
		    exit (exitcode=1);
		}

		# Now parse assignments of the form "a=b".

		if ( $field ~ /\=/ ) {
		    newfield  = ""
		    equalseen = 0
		    fieldlength = length ($field)
		    for ( i=1; i<=fieldlength; i++ ) {
			s = substr ($field, i, 1)
			if ( s == "=" ) {
			    if ( !equalseen ) s = s squote
			    equalseen = 1
			} else if ( equalseen ) {	# value
			    if ( s == squote ) {
				# Special handling: since an apostrophe
				# (in this example represented by a dot
				# .) cannot be quoted using a backslash,
				# we use the following work around: we
				# replace it with "."

				if ( i<fieldlength ) {
				    s = squote "\"" squote "\"" squote
				} else {
				    # Special handling: if the last character
				    # is a single quote, we do not want to add
				    # "squote" to the end, because this would
				    # result in two trailing superfluous
				    # apostrophes.
				}
			    }
			}
			newfield = newfield s
		    }
		    if ( s != squote ) {
			$field = newfield squote
		    } else {
			# Special handling: last character was an apostrophe:
			$field = newfield "\"" squote "\""
		    }
		} else if ( evalcheck ) {
		    # We cannot get here...
		    print "invalid assignment: " $field | "cat >&2"
		    exit (exitcode=1)
		}
		#print "after = processing " $0
	    }

	    for ( i=1; i<=NF; ++i ) {
		printf ("%s%s", VarPrefix, $i)
		if ( i<NF ) printf (FieldSep); else printf ("\n");
	    }
	}
	END {
	    exit (exitcode)
	}
    '