parse_args.awk

#!/usr/bin/env awk -f

# Parse option args with an OPTSTRING
#
# Author: Allex Wang ([email protected])
# Last Modified: Tue Aug 24, 2021 20:08
#
# Usage: echo "<OPTS>|<ARGS>" | parse_args
# eg> echo "-c 10 --welcome 'hi allex' -f |-c:,-v,--welcome:" | awk -f parse_args.awk
#
#   while read -r -d $'\0' k v; do
#     case "$k" in
#       grep) grep="$v" ;;
#       __ARGV_REMAINS__) eval set -- "$v" ;;
#     esac
#   done < <(echo "$* |--grep:" | awk -f parse_args.awk)
#
# GistID: 69e0673a54dd1c3fb786d272f719ad89
# GistURL: https://gist.github.com/allex/69e0673a54dd1c3fb786d272f719ad89

function tokenize(s, start_idx, tokens) {
  len = length(s);
  i = 0;
  sb = "";
  c = "";
  open = "";
  n = start_idx - 1;

  if (len < start_idx) {
    return 0
  }

  while (n <= len) {
    is_last = n == len
    c = s[n++];

    if (c == "'" || c == "\"") {
      level++;
      if (open == "") {
        open = c;
      } else if (open == c) {
        open = "";
      }
    }

    if (c == " " && open == "") {
      tokens[++i] = sb;
      sb = "";
      continue;
    }

    sb = sprintf("%s%s", sb, c);

    if (is_last) {
      tokens[++i] = sb;
    }
  }

  return i
}

function split_opts (s, tokens) {
  split(s, arr, "");
  return tokenize(arr, 1, tokens);
}

function error (msg) {
  print msg > "/dev/stderr"
}

function debug (msg) {
  print msg > "/dev/stderr"
}

BEGIN { FS="|"; EX_ERR=0; }
{
  if (NF != 2) {
    error("Usage: echo \"<OPTS>|<ARGS>\" | parse_args \neg> echo \"-c 10 --welcome 'hi allex' -f |-c:,-v,--welcome:\" | parse_args");
    EX_ERR = 254;
    exit;
  }

  gsub(/^[ \t]+|[ \t]+$/, "", $1);
  gsub(/^[ \t]+|[ \t]+$/, "", $2);

  opts_str = $2;
  args_str = $1;

  n = split(opts_str, a, /,/);
  for (i = 1; i <= n; i++) {
    if (match(a[i], /:$/) > 0) opt_tps[substr(a[i], 0, RSTART - 1)] = 1;
    else opt_tps[a[i]] = 0;
  }

  n = split_opts(args_str, a);
  j = 0;
  for (i = 1; i <= n; i++) {
    opt = a[i];
    if (opt !~ /-+.*/ || opt_tps[opt] == "") {
      remains[j++] = opt;
      continue;
    }
    opt_t = opt_tps[opt];
    if (opt_t == 0 && i < n && a[i + 1] !~ /-+.*/) {
      error("error: invalid val for opt: " opt)
      EX_ERR = 253;
      exit 1;
    }
    sub(/^-+/, "", opt);
    if (opt_t == 0) {
      args[opt] = "TRUE";
    } else {
      args[opt] = a[++i];
    }
  }

  for (k in args) printf "%s\n%s\0", k, args[k]

  # append remains args
  sb = ""
  for (k in remains) {
    printf k
    sb = sprintf("%s %s", sb, remains[k]);
  }
  printf "%s\n%s\0", "__ARGV_REMAINS__", sb
}
END { exit EX_ERR }