#!/usr/local/bin/perl
#
#	TxtIndex  - Create look-up index file for /src/txt directory.
#	V1.1 pl 00 -- 25 Nov 1997
#
#   Written by
#	Jouko Valta   (jopi@x2ftp.oulu.fi)
#
#   Locate files on FTP and text directories and find matching pairs.
#
#
#   Revision History
#	V1.0 pl 00 --  8 Apr 1997
#	 Initial revision.
#
#	V1.1 pl 00 -- 25 Nov 1997
#	 Misc options, comments, etc.
#	 Added LongDescription Module Files.
#	 Sort the Keywords file.
#
#
 
$Usage = "\
    Options:\
	 -h	Show help page and exit\
	 -k	Sort Keywords file as well\
	 -nk	Suppress sorting the Keywords file\
	-c,-nw	Don't update index file\
	 -w	Write updated index file\n\n";


# -----------------------------------------------------------------------------

## Default Filenames

$LOGHOME	= "/pc/log";
$TEXTDIR	= "/src/txt";
$FTPDIR		= "/ftp-service/ftp/pub/msdos";
$INDEXDIR	= $LOGHOME . "/programming";	# Re-formatted TXT files.

$TARGET		= $TEXTDIR . "/00index.txt";

$FIND_PROGS	= "/bin/find $FTPDIR/programming -type f -print | /bin/cut -c40-";
$FIND_MSDOS	= "/bin/find $FTPDIR/arcers $FTPDIR/virus -type f -print | /bin/cut -c28-";
$SORT		= "/bin/sort";


## Defaults

$CntOnly	= 0;
$SortKeys	= 0;
$DoLDesc	= 1;


## Command line options
while ($_ = $ARGV[0], /^-/) {
	shift;
	last if /^--$/;

	if (/^-h/)	{ print $Usage; exit 2; }

	if (/^-k/)	{ $SortKeys = 1; next; }	# Update keywords.
	if (/^-nk/)	{ $SortKeys = 0; next; }
	if (/^-c|^-nw/)	{ $CntOnly = 1; next; }		# No index update.
	if (/^-w/)	{ $CntOnly = 0; next; }
	if (/^-l/)	{ $DoLDesc = 1; next; }		# Check LDM Files.
	if (/^-nl/)	{ $DoLDesc = 0; next; }
	if (/^-/)	{ die "Unsupported argument near $_\n Stopped"; }
}


# -----------------------------------------------------------------------------
#
# Initialize Counters
#

$FtpFiles	= 0;
$TxtFiles	= 0;			# Txt Files
$MatchLess	= 0;			# No mainfile
$MatchOne	= 0;
$MatchMany	= 0;			# Multiple mainfiles

$LDescFiles	= 0;			# Long Description Module Files
$LDescOrphan	= 0;
$LDMatchOne	= 0;
$LDMatchMany	= 0;			# Multiple mainfiles


# -----------------------------------------------------------------------------
# Check Long Description Module File matches
#
#  *.txt  *.lsm
# others => error
#


if ($DoLDesc) {
    open (T , "/bin/find $INDEXDIR -type f -print |") ||
	die "Can't read $INDEXDIR";
    print "\nChecking files on $INDEXDIR.\n";

    while (<T>) {
	chop;
#	print "$_\n";

	if (/^\s*$/) { next line; }			# Empty line
	if (/^00|\/00|00\_*index|index.html/) { next line; } # Index file
	if (/rejected\.\d+/) { next line; }		# Info file

	# Takes the last slash by default.
	if (/\S*\/(\S+\/)(\S+)$/) {			# File in LDM directory
	    $dir  = $1;
	    $file = $2;

	    $txt = get_txt_name ($file);		# '-' for no txt

#    print "$dir\t$file\t$txt\n";

	    if ($dir ne "programming/") {
		$IndexCategory{$txt} = $dir;		# '-' for no txt
	    }
	    $IndexName{$txt}     = $file;		# '-' for no txt

	    # Update Count

	    ++$IndexMatch{$txt};			# '-' for no txt
	    ++$LDescFiles;
	}


#	if (!$IndexMatch{$_}) {				# '-' for no txt
#	    printf "No mainfile for  $_.\n";
##	    printf OF (" -\t\t\t-\t\t%s\n", $_);
#	    ++$LDescOrphan;
#	}
#	elsif ($IndexMatch{$_} > 1) {			# '-' for no txt
#	    printf "Multiple matches for  $_.\n";	# Multiple mainfiles
##	    ++$LDMatchMany;
#	}
#	else {
#	    ++$LDMatchOne;
#	}

    }  # while

    close T;
    print "\n";

}  # Do LDesc


$IndexCategory{"-"} = "";				# '-' for no txt
$IndexName{"-"}     = "-";

$date = `/bin/date`;


# -----------------------------------------------------------------------------
# Get the FTP directory listing ...
#

open (foo, "$FIND_PROGS |$SORT |") || die "Can't excute '$FIND_PROGS | $SORT'";
if (! $CntOnly) {
    open (OF, ">$TARGET");
    print "\nCreating $TARGET\n";
    print OF "% IndexFile Format v1.0  -- File created $date\n";    # Double LF
}
print "\nChecking files on $FTPDIR.\n";
process_input();

open (foo, "$FIND_MSDOS |$SORT |") || die "Can't excute '$FIND_MSDOS | $SORT'";
process_input();


# -----------------------------------------------------------------------------
#
# Check TXT File matches
# Count matching files for each textfile found.
#

open (T , "/bin/ls -1 $TEXTDIR |") || die "Can't read $TEXTDIR";
print "Checking for matches on $TEXTDIR.\n";

while (<T>) {
    chop;
    ++$TxtFiles;

    if (!$TxtMatch{$_}) {				# '-' for no txt
	printf "No mainfile for  $_.\n";

	printf OF (" -\t\t\t-\t\t%-8s", $_);
	if ($IndexName{$_}) {
	    printf OF ("\t%s%s\n",  $_, $IndexCategory{$_}, $IndexName{$_});
	} else {
	    printf OF ("\t-\n");
	}

	++$MatchLess;
    }
    elsif ($TxtMatch{$_} > 1) {				# '-' for no txt
	printf "Multiple matches for  $_.\n";		# Multiple mainfiles
	++$MatchMany;
    }
    else {
	++$MatchOne;
    }
}  # while
close T;


# -----------------------------------------------------------------------------
#
# Print Summary
#

printf "\n\n";
printf "\t%4d  files on FTP archive. (Index files not included.)\n", $FtpFiles;
printf "\t%4d  txtfiles.\n", $TxtFiles;
($DoLDesc) &&
    printf "\t%4d  long description text files.\n", $LDescFiles;
printf "\n";

printf "\t%4d  textfiles don't have a main file.\n",   $MatchLess;
printf "\t%4d  textfiles shared by multiple files.\n", $MatchMany;
printf "\t%4d  files uniquely have a txt file.\n", $MatchOne;
printf "\t%4d  files don't have a txt file.\n",    $TxtMatch{"-"};
printf "\n";

if ($DoLDesc) {
    printf "\t%4d  long text files don't have a main file.\n",   $LDescOrphan;
    printf "\t%4d  long text files shared by multiple files.\n", $LDMatchMany;
    printf "\t%4d  files uniquely have a long txt file.\n", $LDMatchOne;
    printf "\t%4d  (txt) files don't have a long txt file.\n",    $IndexMatch{"-"};
    printf "\n";
}

printf OF "\n";
close OF;
chmod 0664, $TARGET;

print "Done.\n";


# -----------------------------------------------------------------------------
#
# For each mainfile, define the corresponding textfile name and count
# matches.
#

sub process_input {
    line: while (<foo>) {

#	print $_;

	if (/^\s*$/) { next line; }			# Empty line
	if (/^00|\/00|00\_*index|index.html/) { next line; } # Index file
	if (/rejected\.\d+/) { next line; }		# Info file

	# Takes the last slash by default.
	if (/(\S+)\/(\S+)/) {			# File in any FTP directory
	    $dir  = $1;
	    $file = $2;

	    $txt = get_txt_name ($file);		# '-' for no txt

	    printf OF (" %-16s\t%-8s\t%-8s\t%s%s\n",
		       $dir, $file, $txt, $IndexCategory{$txt}, $IndexName{$txt});

	    # Update Count

	    ++$TxtMatch{$txt};				# '-' for no txt
	    ++$FtpFiles;
	}
    }  # while
}  # process_input


# -----------------------------------------------------------------------------

sub get_txt_name {
    local ($mainfile, $foo) = @_;
    local ($txtfile, $ext) = split (/\./, $mainfile);

#   $descfile = $txtfile . ".txt";
    $txtfile = $txtfile . ".txt";
#   print "$mainfile $txtfile\n";

    if (-f "$TEXTDIR/$txtfile") {
	return $txtfile;
    }

    return "-";
}


# -----------------------------------------------------------------------------
#
# NOTE: 'textindex' and 'longindex.pl' expect .txt ot .lsm extensions only.
# (Counting on 'dosadm')
#
# This is not very reliable method -- should use FIND to work out subdirs etc.
#

sub get_desc_name {
    local ($mainfile, $foo) = @_;
    local ($descfile, $ext) = split (/\./, $mainfile);

#   $descfile = $descfile . ".txt";
#   print "$mainfile $descfile\n";

    if (-f "$INDEXDIR/$descfile.txt") {
	return "$descfile.txt";
    }
    if (-f "$INDEXDIR/$descfile.lsm") {
	return "$descfile.lsm";
    }

    return "-";
}