readvars() -- a multiline "read" implementation for /bin/sh

#!/bin/sh

# readvars [-d] [-f filter] [-F field:filter] [-L limit] [-n "var ..."] \
#          [-s sep [-c]] [-t var] [--] cmd ...
#
# For each line of output generated by ``cmd ...'', set a series of numbered
# variables (default "line1", "line2", etc.). Also sets a variable containing
# the number of lines processed (default "lines"). If ``cmd ...'' is "-", NULL,
# or missing, lines are read instead from stdin.
#
# If `-d' is given, debugging information is provided on stderr.
#
# If ``-f filter'' is given, only lines matching filter regular expression are
# processed.
#
# If ``-F field:filter'' is given, only lines where numbered field matches the
# filter regular expression are processed. If field is zero, acts the same as
# ``-f filter''.
#
# If ``-L limit'' is given, process at-most limit number of lines. When
# combined with ``-f filter'', process at-most limit number of lines matching
# filter regular expression. Default value is `0' and to process all lines.
#
# If ``-n "foo bar baz rest"'' is given, the first word on the first line is
# assigned to variable "foo1", 2nd-word 1st-line to "bar1", 3rd-word 1st-line
# to "baz1", and any words remaining on line-1 are assigned to "rest1". The
# last variable name always contains the remainder of the line. Default value
# if not present is simply "line".
#
# If ``-s sep'' is given, use sep as the separator between fields. If `-c' is
# given, the space between sequential separators are counted as empty words.
#
# If ``-t variable'' is given, the total number of lines processed is assigned
# to "variable" (default value is "lines")..
#
# If `--' is given, any further potential option flags are taken as part of
# command to execute.
#
#
# The return status is that of ``cmd [args ...]''.
#
readvars()
{
    local OPTIND=1 OPTFLAG __readvars_flag
    local __readvars_count_nulls=
    local __readvars_debug=
    local __readvars_filter=
    local __readvars_Ffilter=
    local __readvars_Ffilter_field=
    local __readvars_limit=0
    local __readvars_sep=
    local __readvars_total=
    local __readvars_vars=

    while getopts cdf:F:L:n:s:t: __readvars_flag; do
        case "$__readvars_flag" in
        c) __readvars_count_nulls=1 ;;
        d) __readvars_debug=1 ;;
        f) __readvars_filter="$OPTARG" ;;
        F) __readvars_Ffilter="${OPTARG#*:}"
           __readvars_Ffilter_field="${OPTARG%%:*}" ;;
        L) __readvars_limit="$OPTARG" ;;
        n) __readvars_vars="$OPTARG" ;;
        s) __readvars_sep="$OPTARG" ;;
        t) __readvars_total="$OPTARG" ;;
        esac
    done
    shift $(( $OPTIND - 1 ))

    eval "$( awk \
            -F "${__readvars_sep:-$IFS}" \
            -v IFS="$IFS" \
            -v count_nulls="${__readvars_count_nulls:-0}" \
            -v debug="${__readvars_debug:-0}" \
            -v filter="$__readvars_filter" \
            -v Ffilter="$__readvars_Ffilter" \
            -v Ffilter_field="${__readvars_Ffilter_field:-0}" \
            -v sep="${__readvars_sep:-$IFS}" \
            -v total="${__readvars_total:-lines}" \
            -v vars="${__readvars_vars:-line}" \
        -- '
        ############################################ FUNCTIONS

        function esc(str) { gsub(/'\''/, "&\\\\&&", str); return str }

        function set(k, v)
        {
            parent_eval = sprintf("%s='\''%s'\''", k, esc(v))
            if (debug) print "DEBUG:", parent_eval > "/dev/stderr"
            print parent_eval
        }

        function seti(key, val) { set(key i, val) }

        function extract_value(line, varnum)
        {
            match(line, sprintf(count_nulls ? "^[^%s]*[%s]?" : \
                "^[%s]*[^%s]+", sep, sep))
            if (varnum == nvars) {
                value = substr(line, RSTART)
                sub(sprintf("[%s]*$", sep), "", value)
            } else
                value = substr(line, RSTART, RLENGTH)
            sub(sprintf(count_nulls ? \
                "[%s]?$" : "^[%s]*", sep), "", value)
            seti(varlist[varnum], value)
            return substr(line, RSTART + RLENGTH)
        }

        function setivars(vars, line)
        {
            for (v = 1; v <= nvars; v++)
                line = extract_value(line, v)
        }

        function match_filters(line)
        {
            return (line ~ filter && $Ffilter_field ~ Ffilter)
        }

        ############################################ MAIN

        BEGIN {
            nvars = split(vars, varlist, "[" IFS "]+")

            cmd = "eval"
            if (ARGC <= 1 || (ARGC == 2 && ARGV[1] == "-")) {
                ARGC = 2
                ARGV[1] = "cat"
            }
            for (a = 1; a < ARGC; a++)
                cmd = cmd " '\''" esc(ARGV[a]) "'\''"

            retval = 0
            while (cmd | getline)
            {
                if (match_filters($0)) {
                    if (++i > (limit > 0 ? limit : i)) {
                        retval = close(cmd)
                        exit
                    }
                    setivars(varlist, $0)
                }
            }
            retval = close(cmd)
            exit
        }

        END {
            set(total, i == "" ? 0 : i)
            printf "( exit %u )", int(retval / 256)
        }
    ' "$@" )"
}

### BEGIN UNIT-TEST CODE
#
#readvars "$@"
#echo "?=[$?]"
#
#args=line total=lines
#while getopts cdf:L:n:s:t: flag; do
#   case "$flag" in
#   n) args="$OPTARG" ;;
#   t) total="$OPTARG" ;;
#   esac
#done
#shift $(( $OPTIND - 1 ))
#
#eval count=\"\$$total\"
#echo "$total=[$count]"
#n=1
#while [ $n -le $count ]; do
#   for arg in $args; do
#       eval value=\"\$$arg$n\"
#       echo "$arg$n=[$value]"
#   done
#   n=$(( $n + 1 ))
#done
#
### END UNIT-TEST CODE

#
# Examples:
#

echo ">>> Example 1. List of users in passwd(5)"
readvars -cs: -n"user rest" < /etc/passwd
echo "?=[$?]"
n=1
while [ $n -le ${lines:-0} ]; do
    eval user=\"\$user$n\"
    echo "$user"
    n=$(( $n + 1 ))
done
echo

echo ">>> Example 2. Get homedir of a specific user in passwd(5)"
PASSWD_FIELDS="user hash uid gid gecos home shell"
unset home1
readvars -cs: -n"$PASSWD_FIELDS" -F1:'^root$' -L1 < /etc/passwd
echo "?=[$?]"
echo "root's home directory is: ${home1:-**UNKNOWN**}"
echo

echo ">>> Example 3. List primary group membership names for each user"
PASSWD_FIELDS="user hash uid gid gecos home shell"
GROUP_FIELDS="group chkgrp_hash group_gid members"
readvars -t nusers -cs: -n"$PASSWD_FIELDS" < /etc/passwd
echo "?=[$?]"
readvars -t ngroups -cs: -n"$GROUP_FIELDS" < /etc/group
echo "?=[$?]"
g=1
while [ $g -le ${ngroups:-0} ]; do
    eval group=\"\$group$g\" group_gid=\"\$group_gid$g\"
    eval group_name_$group_gid=\"\$group$g\"
    g=$(( $g + 1 ))
done
u=1
while [ $u -le ${nusers:-0} ]; do
    eval user=\"\$user$u\" user_gid=\"\$user_gid$u\"
    eval user_group=\"\$group_name_$user_gid\"
    echo "user=[$user] user_group=[$user_group]"
    u=$(( $u + 1 ))
done