#!/usr/local/bin/./perl
##########################################################################
# Created March/April 1996, Michael D. Smith 
# Part of WebGlimpse (GlimpseHTTP) research with Udi Manber
# Glimpse mailing list: glimpse@cs.arizona.edu
# WebGlimpse home page: http://glimpse.cs.arizona.edu/webglimpse
#
# Modified by Dachuan Zhang, May 22, 1996
# Some bugs fixed.
##########################################################################



##########################################################################
## GLOBALS
##########################################################################
$archivedir = shift(@ARGV);	## Cleaner handling of options as per M. Ernst. --GB 9/24/97
@filelist=();
@THEBOX=();
@THEPAGE=();


##########################################################################
## SETTINGS 
##########################################################################
# to be changed
$WEBGLIMPSE_HOME = "/usr/local/webglimpse";
$CGIBIN = "cgi-bin/webglimpse";
$FULLSEARCH="webglimpse-fullsearch";

# static
$HTMLFILE_RE = "((.s?html)|(.sht)|(.htm))\$";
$WEBGLIMPSE_LIB = "$WEBGLIMPSE_HOME/lib";
$BACKUPEXT=".wgbak";
$MADENH = ".wg_madenh";
$SEARCHBOX = ".wgbox.html";
$SEARCHPAGE_TEMPLATE = ".wgindex.html";
$SEARCHPAGE = "wgindex.html";
$nh_pre=".nh.";




##########################################################################
## ENTRY POINT
##########################################################################

#---------------------------------
# make my libraries more important
unshift(@INC, "$WEBGLIMPSE_LIB");
require "config.pl";

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

## Cleaner handling of options as per M. Ernst. --GB 9/24/97
$removing = 0;
$keep_nh = 0;
$quiet = 0;

while (defined($arg = shift(@ARGV))) {
	if ($arg eq "-r") {
		$removing = 1;
	} elsif ($arg eq "-k") {
		$keep_nh = 1;
	} elsif ($arg eq "-q") {
		$quiet = 1;
	} else {
		die "Bad argument $arg (remaining args @ARGV)";
	}
}
## End option handling change --GB 9/24/97 


if($archivedir eq "") {
	$archivedir = ".";  # make it current dir
}

# try to change the directory to indexdir
$startpwd = `pwd`;
$retval = chdir ($archivedir);
if($retval==0){
	print "Cannot change directory to $archivedir.  Quitting.\n";
	exit -3;
}

# get the 'real' path
$archivepwd = `pwd`;
chomp $archivepwd;

# make sure it has a configuration file
if($removing==0 && &TestConfig($archivepwd)==0){
	print "Cannot find configuration file for archive.  Quitting.\n";
	exit -4;
}

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

$MADENH = "$archivepwd/$MADENH";
$SEARCHBOX = "$archivepwd/$SEARCHBOX";

# get the settings from the configuration file
# there should be no problem opening this file -- we know it exists
# read the settings
# ($title, $archiveurl, $traverse_type, $numhops,$nhhops,$addboxes) =
# &ReadConfig($archivepwd);
   ($title, $archiveurl, $traverse_type, $explicit_only, $numhops,
    $nhhops, $local_limit, $remote_limit, $addboxes, $vhost, $usemaxmem, @urllist) = ReadConfig($archivepwd);

### HUH?  What does this code do?  Can we remove it?
###       here just in case.
# if we're told to do nothing, do nothing.
# Removed next 4 lines by bgopal oct/11/96
#if($addboxes==1 && $removing==0){
#   print "Possible problem with addsearch.  Please investigate (debug msg).\n";
#   exit(0);
#}


$archivetitle = $title;

if($removing==0){
   # generate the wgindex.html file from the .wgindex.html
   &read_search_page($SEARCHPAGE_TEMPLATE);
   &make_search_page($SEARCHPAGE);
}

# if we don't have to addboxes, just abort!
if($addboxes==0) {
   print "No search boxes used\n";
   exit(0);
}

# open the .wg_madenh
open(FILELIST, "$MADENH") || die "Cannot open $MADENH for reading.";

# get the filelist from the file
@filelist = <FILELIST>;
close FILELIST;

# read the search box
if($removing==0){
	&read_search_box($SEARCHBOX);
}

# make the additions to the corefiles
&add_search_box(@filelist);

#----------------------
#change the dir back
chdir($startpwd);






##########################################################################
### PROCEDURES
##########################################################################

##########################################################################
sub read_search_page{
   local($file)=@_;
   
   open(FILE, $file) || die "Cannot open $file for reading.\n";
   
   @THEPAGE = <FILE>;
   close(FILE);
}

##########################################################################
sub make_search_page{
   local($file)=@_;
   local($realline, $line, $newfile);

   open(OUTPUT,">$file") || die "Cannot open $file for writing.\n";

   # output the contents of THEPAGE
   foreach $realline(@THEPAGE){
      $line = $realline;
      # do the substitutions and output
      $line =~ s/\$ARCHIVETITLE/$archivetitle/g;
      $line =~ s/\$ARCHIVEURL/$archiveurl/g;
      $line =~ s/\$ARCHIVEPWD/$archivepwd/g;
      $line =~ s/\$CGIBIN/$CGIBIN/g;
      $line =~ s/\$FILE/$file/g;
      print OUTPUT $line;
   }

   close OUTPUT;
   
}

##########################################################################
sub read_search_box{
	local($file)=@_;

	open(FILE, $file) || die "Cannot open $file for reading.\n";

	@THEBOX = <FILE>;
	close(FILE);
}

##########################################################################
sub add_search_box{
	local(@filelist) = @_;
	local($file, $tempfile,$left,$mid,$right,$did_box);

	# for each file
	foreach $file(@filelist){
		chomp($file);  # remove \n if exists

		# I can ONLY modify .html files
		next if ($file !~ /$HTMLFILE_RE/);

		# stat the file
		my(@statinfo) = stat($file);

		# Changed 9/16/97 to keep existing permissions
		# First we make a backup, then write into the real file.

		$bakfile = "$file.bak";
		system("cp $file $bakfile");

		# open the backup file for reading
		eval {
			open(INPUT, $bakfile);
		};
		if ($@){
			warn "Cannot open file $bakfile: $@\n";
			next;
		}

		eval {
			open(OUTPUT, ">$file");
		};
		if ($@) {
			warn "Cannot open file $file for writing: $@\n";
			system("cp $bakfile $file");
			close(INPUT);
			next;
		}

		$did_box=0;
		$rid_box=0;
		while(<INPUT>){
			# copy until we see either </body>, </html>, or <!--GH_SEARCH-->
			if($did_box==0 && /(.*)(<\/body>|<\/html>|<\!--GH_SEARCH-->)(.*)/i){

				$left = $1;
				$mid = $2;
				$right = $3;
	
				# if we have <!GH_SEARCH>, skip through <!GH_END>
				if($mid =~ /^<\!--GH_SEARCH-->$/i){
					while(<INPUT>){
						/(.*)(<\!--GH_END-->)(.*)/i || next;
						$right=$3;
						last;
					}
					($left eq "") || print OUTPUT "$left\n";
					if ($removing == 1 && $rid_box == 0)	{
						&do_box($file);
						print OUTPUT "<!--GH_SEARCH-->\n";
						print OUTPUT "<!--GH_END-->\n";
						$rid_box = 1;
					} elsif ($removing == 1)	{
						;
					} else	{
						&do_box($file);
						$did_box = 1;
					}	
					($right eq "")||print OUTPUT "$right\n";
				} else	{
					$right = "$mid$right";
					($left eq "") || print OUTPUT "$left\n";
					if ($removing != 1 && $did_box == 0)	{
						&do_box($file);
						$did_box = 1;
					}
					($right eq "")||print OUTPUT "$right\n";
				}
				# set the box var to true
			}else{
				print OUTPUT $_;
			}
		}

		if($did_box==0 && $rid_box==0){
			&do_box($file);
		}
			
		# close the two files
		close OUTPUT;
		close INPUT;

		### TO DO -- don't do this
		# rename the original into .wgbackup
		# system("mv $file $file$BACKUPEXT");

		# copy the temp file into the original
		#system("mv -f $tempfile $file");
		#chmod (0644, $file);  # DO NOT CHANGE THE MODE!!  IT MIGHT ALREADY HAVE BEEN SET FOR A REASON!! --GB 

		# Instead of the above we now just delete the bakup file. -GB 9/16/97
		system("rm $bakfile");
		
		# modify the access time back to the original
		utime(@statinfo[8], @statinfo[9], $file);

	}
}

##########################################################################
sub do_box{
	local($file)=@_;
	local($realline, $line, $newfile);

	# if we're removing, try to delete the neighborhood, too
	if($removing==1){
		# might as well kill the neighborhood if no one will refer to it!
		# prepend the .nh_
		$newfile = $file;
   		$newfile =~ s/([^\/]+)$/$nh_pre$1/;
		if (!$quiet) { print "removing $newfile\n"; }
		if($keep_nh==0){
		   unlink("$newfile");
		}
		return;
	}

	print OUTPUT "<!--GH_SEARCH-->\n";

	# output the contents of THEBOX
	foreach $realline(@THEBOX){
		$line = $realline;
		# do the substitutions and output
		$line =~ s/\$ARCHIVETITLE/$archivetitle/g;
		$line =~ s/\$ARCHIVEURL/$archiveurl/g;
		$line =~ s/\$ARCHIVEPWD/$archivepwd/g;
		$line =~ s/\$CGIBIN/$CGIBIN/g;
		$line =~ s/\$FILE/$file/g;
		print OUTPUT $line;
	}

	print OUTPUT "<!--GH_END-->\n";
}