#!/usr/bin/perl

#########################################################################
#site Search
#Version 2.00b
#copyright  1997-1998 Krishnan Jayakrishnan.
#All Rights Reserved
#------------------------------------------------------------------------
#copyright Notice:-
#site Search is a shareware program and may be used , modified or
#distributed  by anyone free of charge. site Search is provided on an
#"as-is" basis and Krishnan Jayakrishnan makes no warranties either
#expressed or implied with respect to the software's merchantablity or
#fitness for any particular purpose. By using site Search, you agree to
#indemnify Krishnan Jayakrishnan from any and all incidental or
#consequential damages that may arise from its use. Please obtain a
#written permission before selling this program.In all cases this
#copyright Notice  must remain intact.
#------------------------------------------------------------------------
#Contact information
#krishnan@bayou.uh.edu
#
#Krishnan . J
#Allied Geophysical Labs
#University of Houston
#Houston TX77204-4231
#------------------------------------------------------------------------
#Version History
#Intial Release on September 17th 1997                  Version 1.00
#2nd Release on September 23rd 1997                     Version 1.01
#3rd Release on September 25th 1997                     Version 1.02
#4th Release on January 21nd 1998                     	Version 2.00b
#
#The latest version is available at http://www.agl.uh.edu/~saljxk/site_Search
#------------------------------------------------------------------------
#Thanks To,
#
#B. G. Mahesh (www.mahesh.com),
#Darron Hutchison (tm)(www.webmakers.com.au),
#Cinnamon Chu, Tina Ma, 
#Bob Lenaerts, Patrice Serrano, 
#V.A.
#------------------------------------------------------------------------


open(READ,"site_Search.conf ") or &Error_Module("1");
require("./site_Search.conf");
require ("cgi-lib.pl") or &Error_Module("2");

$version = "2.00b";
@months = (January,February,March,April,May,June,July,August,September,October,November,December);

&Check_Configurations	if($check_script eq "1");
&Clean_Scratch          if($clean_scratch eq "1");
&Get_Search_Terms;
&Get_Files_to_Search	if($skip_search ne "1");
&Search_Files;
&Display_Results1 if ((scalar(@file_names_to_display) == 0) ||
                (scalar(@file_names_to_display) <= int($get_hits)) ||
                ($get_hits =~ /all/i) ||
                ($get_hits eq "") ||
		($multi_display eq "0"));
		
&Display_Results2;

#		-:PRIMARY MODULES:-			#
#########################################################
#    Module Checks for errors in configuration file	#
#########################################################
sub Check_Configurations{
print &Header;

&Error_Module("3")	if(!(-r "site_Search.desc") && ($desc1 eq "1"));
&Error_Module("4")	if(!(-d "$base_path") || !(-r "$base_path"));
&Error_Module("5")	if(scalar(@filetypes) == 0);   

if ($multi_display eq "1"){
&Error_Module("6")	
	if(($scratch eq "") || !(-d "$base_path/$scratch") || !(-w "$base_path/$scratch") || !(-r "$base_path/$scratch"));
}

if ($record_usage eq "1"){
&Error_Module("7")	if ((-e "site_Search.usage") && !(-w "site_Search.usage"));
}

&Error_Module("8")	if ($path_to_searchform eq "");
&Error_Module("9")	if ((($output_form eq "1") && ($form_to_use !~ /\binternal\b/)) && !(-r "$form_to_use"));

($name_of_URI = "http:\/\/" . $ENV{'HTTP_HOST'}) if (length($name_of_URI) == 0);

print

"<P>No errors were Found. However, the script only runs a simple check through the configuration file.
If you notice the script not working as intended please go through the files,
<BR><BR><STRONG><LI>readme.html<BR><LI>FAQ.html<BR><BR></STRONG> before 
sending me a mail (krishnan\@bayou.uh.edu).</P>
<P>To run the script set the variable \"\$check_script\" to \"0\" in the configuration file.</P>"
if ($errors ne "1");
print &Bottom;
die;
}

#################################################
#       Module Cleans the Scratch directory     #
#################################################
sub Clean_Scratch{
my @list;
@list = <$base_path/$scratch/[1-9]*.html>;
$^T = time;
 
foreach (@list){
        unlink($_) if (int(100*($age = -M)) > int(100*($empty_scratch)));
}
}

#################################################
#         Module Parses the Form Output         #
#################################################
sub Get_Search_Terms{
        
&ReadParse(*input);
$search_terms = ($input{'Searchterms'});
@get_search_terms = split(/ /,$search_terms);
$get_case = ($input{'Case'});
$get_construct = ($input{'Construct'});

if (defined $input{'OUTPUT_FORM'}){
	$skip_search = $skip_hits = "1";
	$get_hits = "ALL";
	open(READ,$input{'List_File'});
	@files_to_be_searched = <READ>;
}


$get_hits = ($input{'Hits'}) if ($skip_hits ne "1");
&no_search_terms if (scalar(@get_search_terms) == 0);
&record_usage if($record_usage eq "1");
&get_desc if ($desc1 eq "1");
}

#################################################
#    Module Gathers the Files to be Searched    #
#################################################
sub Get_Files_to_Search{
my $file;

@files_from_scandir = &scandir($base_path);


foreach $file (@files_from_scandir){
                push(@files_from_scandir,&scandir($file)) if((-d $file) && (&dirs_avoid($file) ne "0"));
}

foreach $file (@files_from_scandir){
        push(@files_to_be_searched,$file) if((&scantype($file) == 1) && (&files_avoid($file) ne "0"));
}

foreach $file (@files_to_include){
        push(@files_to_be_searched,$file);
}
}

#################################################
#    Module Searches Files for Search Term(s)   #
#################################################
sub Search_Files{
my $count;

foreach $file_name (@files_to_be_searched){
	my $file_match_y = 0;
	my $str_length = 0;
	$mod_time = (stat($file_name))[9];
        $mod_time{$file_name} = &Get_Date;
        open(READ,"$file_name");
        my @file_chars = <READ>;
        $file_to_search = join("",@file_chars);
        $title_array{$file_name} = "$1"
        if ($file_to_search =~ /<TITLE>([^>]+)<\/TITLE>/i);
        $desc_array{$file_name} = "$1"
        if (($file_to_search =~ /<[^>]*META[^>]+NAME\s*=[ "]*description[ "]+CONTENT\s*=\s*"(([^>"])*)"[^>]*>/i) && ($desc2 == "1"));
	$file_to_search =~ s/<([^>])*>//gs;
	@list_of_lines = split(/[.!?]/g,$file_to_search) if(($output_line eq "1") && ($smart eq "0"));
        @list_of_lines = split(/[a-z]['")]*[.!?]+['")]*\s/g,$file_to_search) if(($output_line eq "1") && ($smart eq "1"));
 	$_ = $file_to_search;

	if ($get_construct eq "As a phrase"){
        	$get_search_terms = join(" ",@get_search_terms);
		push(@file_names_to_display,$file_name) if (($count = /\b$get_search_terms\b/o)  && ($get_case eq "Sensitive"));
		push(@file_names_to_display,$file_name) if (($count = /\b$get_search_terms\b/io)  && ($get_case eq "Insensitive"));
        	$line_array{$file_name} = &get_line(@list_of_lines) if  (($output_line eq "1") && ($count > 0));
		$hits_array{$file_name} = &get_hits() if (($output_hits eq "1") && ($count > 0));
	}
	
	if ($get_construct eq "Any search term"){
		 foreach $get_search_term (@get_search_terms)
                        {
		push(@file_names_to_display,$file_name) if (($count = /\b$get_search_term/) && ($get_case eq "Sensitive"));
		push(@file_names_to_display,$file_name) if (($count = /\b$get_search_term/i) && ($get_case eq "Insensitive"));
		last if($count > 0);	
                        }
                $line_array{$file_name} = &get_line(@list_of_lines) if(($output_line eq "1") && ($count > 0));
                $hits_array{$file_name} = &get_hits() if (($output_hits eq "1") && ($count > 0));
		}

	if ($get_construct eq "All search terms"){		
                $str_length = @get_search_terms;
                foreach $get_search_term (@get_search_terms){
		$file_match_y++ if (($count = /\b$get_search_term/) && ($get_case eq "Sensitive"));
		$file_match_y++ if (($count = /\b$get_search_term/i) && ($get_case eq "Insensitive"));
		}
    		push(@file_names_to_display,$file_name) if ($file_match_y == $str_length);
		$line_array{$file_name} = &get_line(@list_of_lines) if (($output_line eq "1") && ($count > 0));
		$hits_array{$file_name} = &get_hits()  if (($output_hits eq "1") && ($count > 0));
                }
} 
}

#################################################
#       Modules Display the Search Results      #
#################################################
sub Display_Results1{
my $file_name;

        &No_Matches if (scalar(@file_names_to_display) == 0);
        print &Header;
        print &Top($title_for_search_page);
        print "<CENTER>Search Conditions<HR>";
        print "Case:$get_case | Construct:$get_construct | Hits Per Page:$get_hits | Terms:$search_terms<BR><BR>";
        print "<B>".scalar(@files_to_be_searched)." File(s) Searched ".scalar(@file_names_to_display)."  Match(es)
                Found</B><BR><HR>";
        print "</CENTER>";
                        
        foreach $file_name (@file_names_to_display)
        {
                my $file1 = substr($file_name,rindex($file_name,"/")+1);
                my $file2 = substr($file_name,length($base_path));
                my $mod_time = $mod_time{$file_name};

                print "<BR><TR><TD><B><LI>File Name:</B><A HREF=\"$name_of_URI$file2\">$file1</A><BR>";
                print "<I><B>Modified:</B>$mod_time</I><BR>";
                print "<B>Title: </B>$title_array{\"$file_name\"}<BR>";
                print "<B>Description:</B> $desc_array{\"$file_name\"}<BR>" if (($desc1 eq "1") || ($desc2 eq "1"));
                print "$hits_array{\"$file_name\"}<BR>" if ($output_hits eq "1");
                print "$line_array{\"$file_name\"}" if ($output_line eq "1");
                print "</TD></TR>";
        }
        print "</TABLE><BR>";
        print "<CENTER><A HREF=\"$path_to_searchform\">[Search Again]</A></CENTER>";

	print &output_form if ($output_form eq "1");
	
	print "<HR><A HREF=\"http://www.agl.uh.edu/~saljxk/site_Search\">
	<FONT SIZE=-1>site Search Version $version copyright\&copy\; 1997, Krishnan Jayakrishnan</FONT></A>";
	print &Bottom;
	die;
}

### ... Module 2
sub Display_Results2{
my $counter = 0;
my $counter2;
my $files_displayed = 0;
my $files_to_display = int($get_hits);
my $output_fileno = 1;
my $hits = int($get_hits);

	while($counter < scalar(@file_names_to_display)){
		if($files_displayed == 0){
			print &Header;
			$output = "STDOUT";
			$next_file = $$ . "_$output_fileno" . "_of_" . "$hits" . ".html";
			}else{
			$output = "WRITE";
			open(WRITE,">$base_path/$scratch/$next_file");
			$output_fileno++;
			$next_file = $$ . "_$output_fileno" . "_of_" . "$hits" . ".html";
			}
print $output &Top($title_for_search_page);
print $output "<CENTER>Search Conditions<HR>";
print $output "Case:$get_case | Construct:$get_construct | Hits Per Page:$get_hits | Terms:$search_terms<BR><BR>";
print $output "<B>".scalar(@files_to_be_searched)." File(s) Searched ".scalar(@file_names_to_display)."  Match(es)
                Found</B><BR><HR>";
print $output "<P>Match(es) " . ($files_displayed+1) . " to " . ($files_displayed+$hits) . " </P>"
		if (($counter + $hits) < scalar(@file_names_to_display));
print $output "<P>Match(es) " . ($files_displayed+1) . " to " . scalar(@file_names_to_display) . " </P>"
		if (($counter + $hits) > scalar(@file_names_to_display));
print $output "</CENTER>";
print $output "<FORM METHOD=\"GET\" ACTION=\"$name_of_URI/$scratch/$next_file\">" 
		if (($counter + $hits) < scalar(@file_names_to_display));

 		for($counter2 = $files_displayed;$counter2 < $files_to_display;$counter2++)
                {
                        last if(length($file_names_to_display[$counter2]) == 0);
                        $file1 = substr($file_names_to_display[$counter2],rindex($file_names_to_display[$counter2],"/")+1);
                        $file2 = substr($file_names_to_display[$counter2],length($base_path));
                        $mod_time = $mod_time{$file_names_to_display[$counter2]};
			print $output "<BR><TR><TD><B><LI>File Name:</B><A HREF=\"$name_of_URI$file2\">$file1</A><BR>";
                        print $output "<I><B>Modified:</B>$mod_time</I><BR>";
			print $output "<B>Title:</B>$title_array{\"$file_names_to_display[$counter2]\"}<BR>";
			print $output "<B>Description:</B>$desc_array{\"$file_names_to_display[$counter2]\"}<BR>" 
					if (($desc1 eq "1") || ($desc2 eq "1"));
			print $output "$hits_array{\"$file_names_to_display[$counter2]\"}<BR>" if ($output_hits eq "1");
                	print $output "$line_array{\"$file_names_to_display[$counter2]\"}" if ($output_line eq "1");
        		print $output "</TD></TR>";
        
                }

		print $output "<P><CENTER><INPUT TYPE=\"SUBMIT\" VALUE=\"GET NEXT $hits HITS\"></CENTER></P>"
		if (($counter + $hits) < scalar(@file_names_to_display));

		print $output "<BR><CENTER><A HREF=\"$path_to_searchform\">[Search Again]</A></CENTER>";
		print $output "</FORM>";

		print $output &output_form if ($output_form eq "1");

		print $output "<HR><A HREF=\"http://www.agl.uh.edu/~saljxk/site_Search\">
		<FONT SIZE=-1>site Search Version $version copyright\&copy\; 1997, Krishnan Jayakrishnan</FONT></A>";
		print $output &Bottom;

	$files_displayed += $hits;
	$counter += $hits;
	$files_to_display += $hits;
}
}

#########################################################
#		-:SECONDARY MODULES:-			#
#########################################################
#___Module Prints Error Messages to Help Rectify___#
sub Error_Module
{
        $errors = "1";
        print &Header if ($_[0] eq "1");
        print "<FONT SIZE=\"3\" FACE=\"HELVETICA\"><BR>";
        print "<P> A Configuration file <STRONG> site_Search.conf </STRONG> could not be located in the current<BR>
                directory, either the file is not present or does not have read permissions. To get help rectifying<BR>
                the error, read the \"Errors\" section, (Error-1) of the \"readme.html\" file.<BR>" if ($_[0] eq "1");
        print "<P> Could not load the cgi-lib.pl library. To get help rectifying the error, read<BR>
                read the \"Errors\" section, (Error-2) of the \"readme.html\" file.<BR>" 	if ($_[0] eq "2");
        print "<P> The description file <STRONG> site_Search.desc </STRONG> could not be opened for reading. There<BR>
                could be a permission problem. To get help rectifying the error, read the \"Errors\" section, (Error-3) of<BR>
                of the \"readme.html\" file.<BR>" 						if ($_[0] eq "3");
        print "<P> There is a problem with the base-path specified to start the search. To get help rectifying the error,<BR>
                read the \"Errors\" section, (Error-4) of the \"readme.html\" file.<BR>" 	if ($_[0] eq "4");
        print "<P> No filetypes have been specified to search. To get help rectifying the error, read the \"Errors\" section,<BR>
                (Error-5) of the \"readme.html\" file.<BR>" 					if ($_[0] eq "5");
        print "<P> There is a problem with the scratch directory specified. To get help rectifying the error, read the <BR>
                \"Errors\" section, (Error-6) of the \"readme.html\" file.<BR>" 		if ($_[0] eq "6");
        print "<P> There is a problem recording the script usage. To get help rectifying the error, read the \"Errors\" <BR>
                section, (Error-7) of the \"readme.html\" file.<BR>" 				if ($_[0] eq "7");
        print "<P> A HTTP Path to the search form was not specified. To get help rectifying the error, read the \"Errors\" <BR>
                section, (Error-8) of the \"readme.html\" file.<BR>" 				if ($_[0] eq "8");
        print "<P> No Valid form was found at $form_to_use . To get help rectifying the error, read the \"Errors\" <BR>
                section, (Error-9) of the \"readme.html\" file.<BR>" 				if ($_[0] eq "9");
        print "</FONT>";
        print &Bottom;
 }

#_____Module Returns the Output Document Type______#
sub Header
{
        return "Content-type: text/html\n\n";
}   

#_____Module Returns the Output Document Head_______#
sub Top
{
        my ($title) = @_;
        return <<"ENDPRINT";
        <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
        <HTML>
        <HEAD>
        <TITLE>$title</TITLE>
        <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
        <META NAME="GENERATOR" CONTENT="site_Search Version $version">
        <META NAME="Author" CONTENT="Krishnan J. (Sys adm, Webmaster) AGL">
        <META NAME="Description" CONTENT="site Search Version -- $version results ">
	<META NAME="Keywords" CONTENT="site Search, web site, perl, cgi, shareware,freeware, search, site, web, search engine">
        </HEAD>
        <BODY TEXT="#000000" BGCOLOR="#FFFFFF">
        <H2><CENTER><FONT SIZE="3" FACE="HELVETICA"><U>$title</U></FONT></CENTER></H2>
ENDPRINT
}

#_____Module Returns the Output Document Bottom______#
sub Bottom
{
        return "</BODY>\n</HTML>\n";
}

#____Module Scans Directories for all Files_____#
sub scandir   
{
        $directory_to_scan = ($_[0]);
        return(<$directory_to_scan/*>);
}

#__Module Returns Only Files of Specified Type__#
sub scantype
{       
        foreach $type (@filetypes){
                return 1 if((rindex $_[0], $type) >= 0);
        }
}

#____Module Checks for Validity of Directory____#
sub dirs_avoid
{
my $check_dir = ($_[0]);
my $dir; 
 
        foreach $dir (@directories_to_avoid){
                return 0 if($dir eq $check_dir);
        }
}       

#______Module Checks for Validity of File______#
sub files_avoid
{
my $check_file = ($_[0]);
my $file;

        foreach $file (@files_to_avoid){
                return 0 if($file  eq $check_file);
        }
}

#_Module Prints Message if no Search Terms are Entered_#
sub no_search_terms{
print &Header;
print &Top("No Search Terms Entered");

print<<"ENDPRINT";
        <CENTER><BR><P>No search terms were entered for search.
	Please use your browser's <B>back</B> button to go back to the form and search again or do a
        <A HREF="$path_to_searchform">New Search.</A></P></CENTER><HR>
        <FONT SIZE=-1>
	<A HREF=\"http://www.agl.uh.edu/~saljxk/site_Search\"><FONT SIZE=-1>
        site Search Version $version copyright\&copy\; 1997, Krishnan Jayakrishnan<FONT></A>
ENDPRINT
print &Bottom;
die;
}

#_________Module Records Script Usage___________#
sub record_usage{
        open(WRITE,">>site_Search.usage");
        $date = scalar localtime;
        chomp($date);
        print WRITE $date ."|".$ENV{'REMOTE_ADDR'}."|".$ENV{'REMOTE_HOST'}."|"."$search_terms\n";
        close(WRITE);
}

#_________Module Gathers Descriptions___________#
sub get_desc{
my $file_name;
my $description;

open(READ,"site_Search.desc");
while(<READ>){
	($file_name,$description) = split(/[sep]/,$_);
	$desc_array{"$file_name"} = "$description";
}
}

#_______Module Returns the File Mod_Date________#
sub Get_Date{
        $mod_time = time unless ($mod_time);
        ($day,$month,$year) = (localtime($mod_time))[3,4,5];
        $date = "$day $months[$month] 19$year";
        return $date;
}
        
#_Module Returns the Line with the Search Term(s)_#
sub get_line{
my @array = @_;
my $line;
my $count = 0;
my $sub_count = 0;
my $search_term;

if ($get_construct eq "As a phrase"){
        foreach $line (@array){
                $_ = $line;
             	$count++;
	   	$sub_count = s/\b$get_search_terms\b/\<B\>\<FONT COLOR=\"\#DC143C\"\>$get_search_terms\<\/FONT\>\<\/B\>/ig;
                return  "<FONT SIZE=\"2\" FACE=\"HELVETICA\">$array[$count-2].$_.$array[$count]</FONT><BR>" if ($sub_count > 0 );
                }
}
foreach $line (@array){
	foreach $search_term (@get_search_terms){
                $_ = $line;
                $count++;
                $sub_count = s/\b$search_term\b/\<B\>\<FONT COLOR=\"\#DC143C\"\>$search_term\<\/FONT\>\<\/B\>/ig;
                return  "<FONT SIZE=\"2\" FACE=\"HELVETICA\">$array[$count-2].$_.$array[$count]</FONT><BR>" if ($sub_count > 0 );
                }
}
}

#_________Module Returns the no of Hits__________#
sub get_hits{
$_ = $file_to_search;
my @hits;
my $hits;
my $search_term;
my $line;
my @lines;

if ($get_construct eq "As a phrase"){
	((@hits) = /\b$get_search_terms\b/og)  if ($get_case eq "Sensitive");
        ((@hits) = /\b$get_search_terms\b/iog) if ($get_case eq "Insensitive");
	$hits = int(scalar(@hits));
	
	return "<FONT SIZE=\"2\" FACE=\"HELVETICA\"><LI>$hits Occurence(s) of the search term <B>$get_search_terms</B></FONT><BR>";
}
foreach $search_term (@get_search_terms){
($hits = (@hits = /\b$search_term/g)) if ($get_case eq "Sensitive");
($hits = (@hits = /\b$search_term/ig)) if ($get_case eq "Insensitive");
push(@lines,"<FONT SIZE=\"2\" FACE=\"HELVETICA\"><LI>$hits  occurence(s) 
of the search term <B>$search_term</B> </FONT>"); 
}

$line = join("",@lines);
return $line."<BR>";
}

#__Module Prints a Message if no Matches are Found___#
sub No_Matches
{
print &Header;
print &Top("No Matches Found");
                
print "<CENTER>Case:$get_case | Construct:$get_construct | Hits Per Page:$get_hits | Terms:$search_terms</CENTER><BR><BR>";
        print<<"ENDPRINT";
        <CENTER><BR><P>No successful matches were found after the search.
	Please use your browser's <B>back</B> button to go back to the form and search again or do a
        <A HREF="$path_to_searchform">New Search.</A></P></CENTER><HR>
        <A HREF=\"http://www.agl.uh.edu/~saljxk/site_Search\"><FONT SIZE=-1>
        site Search Version $version copyright\&copy\; 1997, Krishnan Jayakrishnan<FONT></A>
ENDPRINT
print &Bottom;
die;
}

#_Module Outputs a Form to Perform Additional Searches_#
sub output_form{
my $files = scalar(@file_names_to_display);
my $list_of_files = $$ . "_list_" . ".html";

open(LIST,">$base_path/$scratch/$list_of_files");

foreach $file (@file_names_to_display){
chomp($file);
print LIST "$file\n";
}	
if ($form_to_use !~ /\binternal\b/){
open(READFORM,"$form_to_use");
@lines = <READFORM>;
return @lines;
}

return <<"ENDPRINT";
<HR>
<FORM METHOD="GET" ACTION="$program_to_use">
<TR>
<TD>
<LEFT><B>$files</B> file(s) matching your search criteria found.<BR>
Narrow the search by searching <B>$files File(s)?</B></LEFT> </TD>
</TR>

<TR>
<TD>
<DIV ALIGN=LEFT><INPUT TYPE="TEXT" SIZE=20 MAXLENGTH=100 NAME="Searchterms"></DIV> </TD> </TR>
<TR>
<TD>Case:<SELECT NAME="Case">
<OPTION VALUE="Insensitive">Insensitive 
<OPTION VALUE="Sensitive">Sensitive
</SELECT>
Construct:<SELECT NAME="Construct">
<OPTION VALUE="As a phrase">As a Phrase
<OPTION VALUE="Any search term">Any Search Term
<OPTION VALUE="All search terms">All Search Terms
</SELECT>
</TD> </TR>

<TR>
<TD>
<DIV ALIGN=LEFT><INPUT TYPE="SUBMIT" NAME="SUBMIT" VALUE="GO Search!!"><INPUT TYPE="RESET" VALUE="Reset"> </DIV></TD>
<INPUT TYPE="HIDDEN" NAME="List_File" VALUE="$base_path/$scratch/$list_of_files">
<INPUT TYPE="HIDDEN" NAME="OUTPUT_FORM" VALUE="OK">
</TR>
</TABLE>
</FORM>
ENDPRINT
}