#! /usr/bin/perl 

# This script makes the index listing used sort describe entries by
# category.  It gets most of its information from where utilities
# reside in the GNO ftp hierarchy.
#
# Devin Reade, January 1998
#
# $Id: mkindex,v 1.4 1999/07/03 16:55:42 gdr-ftp Exp $
#

$dirroot = "/ftp/pub/apple2/gs.specific";
$dirlist = "gno orca";
$describe = '/usr/local/bin/describe';

# This is a list of describe entries which we want to force into a
# given category.  This should only be used when a utility doesn't have
# an entry of its own in the .index.src files as a result of being 
# archived with another utility.  For example, we fudge "descc" and
# "descu" because they are archived in with "describe".

$fudgelist{"cclean"}		= "gno/programming";	# with occ
$fudgelist{"ccprep"}		= "gno/programming";	# with occ
$fudgelist{"cjpeg"}		= "gno/graphics";	# part of jpeg tools
$fudgelist{"djpeg"}		= "gno/graphics";	# part of jpeg tools
$fudgelist{"jpegtran"}		= "gno/graphics";	# part of jpeg tools
$fudgelist{"rdjpgcom"}		= "gno/graphics";	# part of jpeg tools
$fudgelist{"wrjpgcom"}		= "gno/graphics";	# part of jpeg tools
$fudgelist{"descc"}		= "gno/doc.utils";	# with describe
$fudgelist{"descu"}		= "gno/doc.utils";	# with describe
$fudgelist{"fortune"}		= "gno/games";		# resolve fortuna-[abc]
$fudgelist{"newuserv"}		= "gno/sysadmin";	# with newuser
# $fudgelist{""}		= "";

# Actually, another possibility is that the file is not up for ftp in the
# pre-packaged directories, but is rather part of the GNO base distribution.
$fudgelist{"aroff"}		= "gno/doc.utils";
$fudgelist{"asml"}		= "gno/programming";
$fudgelist{"assemble"}		= "gno/programming";
$fudgelist{"aw30"}		= "gno/doc.utils";
$fudgelist{"binprint"}		= "gno/editors";
$fudgelist{"blist"}		= "gno/editors";
$fudgelist{"cal"}		= "gno/games";
$fudgelist{"cat"}		= "gno/editors";
$fudgelist{"center"}		= "gno/text.utils";
$fudgelist{"chtyp"}		= "gno/file.manip";
$fudgelist{"cksum"}		= "gno/archive";
$fudgelist{"cmp"}		= "gno/text.utils";
$fudgelist{"cmpl"}		= "gno/programming";
$fudgelist{"colcrt"}		= "gno/doc.utils";
$fudgelist{"compile"}		= "gno/programming";
$fudgelist{"ctags"}		= "gno/programming";
$fudgelist{"cu"}		= "gno/network";
$fudgelist{"df"}		= "gno/sys";
$fudgelist{"dial"}		= "gno/network";
$fudgelist{"duplex"}		= "gno/network";
$fudgelist{"eps"}		= "gno/sys";
$fudgelist{"expr"}		= "gno/scripting";
$fudgelist{"false"}		= "gno/scripting";
$fudgelist{"fmt"}		= "gno/editors";
$fudgelist{"ftp"}		= "gno/network";
$fudgelist{"gsh"}		= "gno/shells";
$fudgelist{"head"}		= "gno/editors";
$fudgelist{"help"}		= "gno/doc.utils";
$fudgelist{"inetd"}		= "gno/network";
$fudgelist{"init"}		= "gno/daemons";
$fudgelist{"kill"}		= "gno/sys";
$fudgelist{"link"}		= "gno/programming";
$fudgelist{"list"}		= "gno/doc.utils";
$fudgelist{"logger"}		= "gno/scripting";
$fudgelist{"login"}		= "gno/daemons";
$fudgelist{"lseg"}		= "gno/programming";
$fudgelist{"mkfs"}		= "gno/sysadmin";
$fudgelist{"mkso"}		= "gno/sysadmin";
$fudgelist{"phone"}		= "gno/mail";
$fudgelist{"pwd"}		= "gno/sys";
$fudgelist{"rcp"}		= "gno/network";
$fudgelist{"remote"}		= "gno/network";
$fudgelist{"removerez"}		= "gno/file.convert";
$fudgelist{"rlogin"}		= "gno/network";
$fudgelist{"rsh"}		= "gno/network";
$fudgelist{"rtf2text"}		= "gno/file.convert";
$fudgelist{"sed"}		= "gno/editors";
$fudgelist{"sendmail"}		= "gno/mail";
$fudgelist{"sleep"}		= "gno/scripting";
$fudgelist{"su"}		= "gno/sysadmin";
$fudgelist{"syslogd"}		= "gno/daemons";
$fudgelist{"sz"}		= "gno/comm";
# $fudgelist{"tr"}		= "gno/text.utils";
# $fudgelist{"trek"}		= "gno/games";
$fudgelist{"true"}		= "gno/scripting";
$fudgelist{"uniq"}		= "gno/text.utils";
$fudgelist{"uptimed"}		= "gno/daemons";
$fudgelist{"wc"}		= "gno/text.utils";
$fudgelist{"who"}		= "gno/sys";
$fudgelist{"whois"}		= "gno/sys";
$fudgelist{"yes"}		= "gno/scripting";

#
# Get the list of .index.src files.  We will look in here for references
# to the describe database.
#
open(fp, "(cd $dirroot; find $dirlist -name .index.src -print)|") ||
    die("couldn't get file list");
@indexfiles = <fp>;
close(fp);

# extract all the program names from the db source file.
while(<>) {
    chop;
    if (/^Name:\s+(.*)/) {
	$name = $1;
	$name =~ s/\s+$//;
	if ($name =~ /\s+/) {
	    printf(stderr "%s:%d: Multi-word program name \"%s\".  Skipped.\n",
		   $ARGV, $., $name);
	} else {
	    $name =~ tr/A-Z/a-z/;
	    $namelist{$name} = ':';
	}
    }
}


foreach $f (@indexfiles) {
    chop($f);
    if (open(fp, "$dirroot/$f")) {
	$f2 = $f;
	$f2 =~ s,/.index.src$,,;
	while ($_ = <fp>) {
	    if (/%%describe%%([^%]+)%%/) {
		$ref = $1;
		$ref =~ tr/A-Z/a-z/;
		if (defined($namelist{$ref})) {
		    ($namelist{$ref} .= "$f2:") 
			unless ($namelist{$ref} =~ /:$f2:/);
		} else {
		    printf(stderr
			   "%s/%s:%d: Warning: Reference to program (\"%s\") ".
			   "not in database.  Entry skipped.\n",
			   $dirroot, $f, $., $ref);
		}
	    }
	}
    } else {
	printf("couldn't open %s/%s: file skipped", $dirroot, $f);
    }
}

# print out the results.
@keylist = keys(%namelist);
foreach $k (@keylist) {
    $s = $namelist{$k};
    if ($s eq ':') {
	if (defined($fudgelist{$k})) {
	    $s = ':' . $fudgelist{$k} . ':';
	} else {
	    $s = ':unsorted:';
	    $result = system("$describe $k > /dev/null") / 256;
	    if ($result == 0) {
		printf(stderr
		       "WARNING: '$k' is showing up as 'unsorted', but has ".
		       "a describe(1) entry.\n\tThis probably means that the ".
		       ".index.src file is not using an\n\texisting".
		       "describe entry.\n", $k);
	    }
	}
    } elsif (defined($fudgelist{$k})) {
	printf(stderr 
	       "WARNING: %s should no longer be in the fudge list\n", $k);
    }
    printf("%s\t%s\n", $k, $s);
}