# Blosxom Plugin: Find
# Author: Fletcher T. Penney
#		advanced search concept and keywords code by Eric Sherman
#		Recent Searches feature based on code by Marc Nozell
# Version: 0.9
package find;

# --- Configurable variables -----
# None yet - may add ability to change search behaviors, such as
# always starting at the root level of your datadir while staying in a subdirectory

$keywords_tag = "meta-keywords:" unless defined $keywords_tag;

$do_local_search = 0;	# Perform search relative to the current page, 
						# not the whole site.  If set to 1, this will
						# override the advanced search option

$show_advanced = 0;		# Set to 1 to always show the advanced form

$show_debug = 0;		# display more info about search terms for debugging

$default_to_and = 0;		# Set to 1 to always do "and" searches by default

$match_whole_words = 0;		# Set to 1 to only match whole words by default

$log_searches = 1;		# Log search queries to a file?
$log_filename = "$blosxom::plugin_state_dir/queries";	# Where should I log?


$max_previous_searches = 10;	# Maximum old queries to display

$search_writebacks = 1;		# Should I also search writebacks?
$writeback_dir = "$blosxom::plugin_state_dir/writeback";
$writeback_ext = "wb";


$search_filenames = 1;		# Should I also search filenames?

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


$results = "";
$recentsearches = "";

use CGI qw/:standard/;


sub start {


	# Figure out the current path and flavour for the form
	$path_withflavour = $blosxom::path_info;
	if ($path_withflavour !~ s/\.[^\.]*$//) {
		$path_withflavour =~ s/\/$//;
		$path_withflavour .= "\/index";
		$path_withflavour =~ s/^([^\/])/$1/;
	}
	$path_withflavour =~ s/^\/*//;
	$path_withflavour.="\.$blosxom::flavour";

	# Insert this html code only if advanced form is indicated
	$advancedform = qq!<br />Search:<br />
<input checked type="radio" name="domain" value="all" />Entire Site
<input type="radio" name="domain" value="topic" />This Topic Only
<br />Match:<br />
<input checked="checked" type="radio" name="type" value="any" />Any
<input type="radio" name="type" value="all" />All
<br />
<input checked="checked" type="radio" name="match" value="any" />Partial
<input type="radio" name="match" value="whole" />Whole Words only
! 
		if ((param('advanced_search')) || $show_advanced);

	# This is the basic form

$searchform = qq!<form method="get" action="$blosxom::url/index.$blosxom::flavour">
	<div>
		<input type="text" name="find" size="15" value=""/>
		<input type="submit" value="Search" />
		<input type="hidden" name="plugin" value="find"/>
		<input type="hidden" name="path" value="$blosxom::path_info"/>
		<br/>
		<a href="$blosxom::url/$path_withflavour?advanced_search=1">Advanced Search</a>
$advancedform
</div></form>!;

	1;
}

sub filter {
	# Check that writebacks are working
	#$search_writebacks = 0 if ( $writeback::writeback_dir eq "");

	my ($pkg, $files_ref) = @_;
	my @files_list = keys %$files_ref;
	if (param('find')) {
		my $terms = param('find');
		$searchpath = "$blosxom::datadir/" . param('path');
		$do_local_search = 1 if (param('domain') eq "topic");

		$match_whole_words =1 if (param('match') eq 'whole');
		$match_whole_words =0 if (param('match') eq 'any');

		my $searchtype = param('type');
		$default_to_and = 0 if ($searchtype eq 'any');

		my @requiredterms, @forbiddenterms;
		

		$results = "These pages matched: $terms";

		if ($log_searches eq 1) {
			if ( !-e $log_filename ) {
				open (LOG, ">>$log_filename");
				chmod (0666, "$log_filename");
			} else {
				open (LOG, ">>$log_filename") or warn "Error in find logging file."
			}
			print LOG "$terms\n";
			close (LOG);
		}

		$terms = " " . $terms;	# Add a space for pattern matching reasons


		# Handle double quotations (exact phrases)
		$terms =~ s/\"([^\"]+)\"/\[\{$1\}\]/g;
		while ($terms =~ s/\[\{([^\}]*)\s+([^\}]*)/\[\{$1\\s\+$2/g) {
		}
		$terms =~ s/\[\{/(/g;
		$terms =~ s/\}\]/)/g;
		# Any left over quotes were "odd-numbered"
		$terms =~ s/\"//g;
		
		# Handle parentheses
		while ($terms =~ s/\(([^\)]*)\s+([^\)]*)\)/\($1\|$2\)/g) {
		}

		# Strip trailing spaces to prevent empty terms
		# Don't strip leading spaces yet!
		$terms =~ s/\s+$//;
		
		# Convert English to symbols
		# The "OR"'s will wait til the end
		# Handle "NOT"'s
		$terms =~ s/\s+not\s+/ \-/ig;
		# Handle "AND"'s and convert to "+", unless preceded by "-"
		$terms =~ s/\s+(\([^\)]+\))\s+and\s+/ \+$1 \+/ig;
		$terms =~ s/\-(\([^\)]+\))\s+and\s+/\-$1 \+/ig;
		$terms =~ s/\s+([^\)]+)\s+and\s+/ \+$1 \+/ig;
		$terms =~ s/\-([^\)]+)\s+and\s+/\-$1 \+/ig;
		$terms =~ s/\+\-/\-/g;	# Fix if the second term already had "-"
		
		$results = "These pages matched: $terms" if ($show_debug eq 1);	
	
		# If doing "all" search, then every term is required
		# Will not override terms already set to "NOT"
		$terms =~ s/\s+\+?([\(\)\|\w]+)/ \+$1/g if (($searchtype eq "all") || ($default_to_and eq 1));

		# Extract all required terms ("AND"  terms)
		while ($terms =~ s/\s+\+([\(\)\|\\\+\w]+)//){
			$theterm = $1;
			$theterm = "\\b$theterm\\b" if ($match_whole_words eq 1);
			push(@requiredterms,$theterm);
			$results.="<br>Required Term: $theterm" if ($show_debug eq 1);
		}

		# Extract all "forbidden" terms ("NOT" terms)
		while ($terms =~ s/\s+\-([\(\)\|\\\+\w]+)//){
			$theterm = $1;
			$theterm = "\\b$theterm\\b" if ($match_whole_words eq 1);
			push(@forbiddenterms,$theterm);
			$results.="<br>Forbidden Term: $theterm" if ($show_debug eq 1);
		}

		# Strip "OR"'s with only one term
		while ($terms =~ s/^\s*or\s+//i) {}
		while ($terms =~ s/\s+or\s*$//i) {}
		

		# Now cleanup for regexp's
		$terms =~ s/^\s+//;	#Strip leading and trailing spaces
		$terms =~ s/\s+$//;
		# Finally, convert all the "OR" terms to a single regexp
		$terms =~ s/\s+(or\s+)?/\|/ig;
		$terms =~ s/(\s)\+/$1/g;	# Loose '+' will crash regexp

		# Handle whole word matching on remainder
		$terms = "\\b$terms\\b" if ($match_whole_words eq 1);

		# Debugging Aids
		$results.="<br>Remainder regexp: $terms<br>" if ($show_debug eq 1);
		$results.="Search path: $searchpath <br>" if ($show_debug eq 1);

		# Quit now if nothing to search for
		if (($terms eq "") & (scalar(@requiredterms) eq 0) & (scalar(@forbiddenterms) eq 0)) {
			$results = "";
			return 0;
		}

		foreach $file (@files_list) {
		#	next;		# Enable this line to debug terms parsing only
			if ($do_local_search eq 1) {
				# Limit search to the current path only
				if ($file !~ /^$searchpath/) {
					delete $files_ref->{$file};
					next;
				}
			}
			my $keep = 0;
			my $delete = 0;
			open (FILE, "<$file") or next;
			my $contents = "";
			my $pastHeader = 0;
			while ($line = <FILE>) {
				if (!$pastHeader) {
					# include keywords
					if ($line =~ /^$keywords_tag/i) {
						$line =~ s/^$keywords_tag(.*)$/\1/;
					}
					# don't read other meta- tags
					elsif ($line =~ /^meta-/i) {
						next;
					}
					# if reached the header, say so
					elsif ($line =~ /^\s.*$/) {
						$pastHeader = 1;
					}
				}
				$contents .= $line;
			}

			close (FILE);
			
			# Now scan writebacks for this story
			if ( $search_writebacks == 1) {
				my $writeback_file = $file;
				$writeback_file =~ s/$blosxom::datadir/$writeback_dir/;
				$writeback_file =~ s/$blosxom::file_extension$/$writeback_ext/;

				if (open (FILE, "<$writeback_file")) {
					while ($line = <FILE>) {
						# We'll just appened writebacks to the story
						$contents .= $line;
					}	
					close (FILE);		
				}
			}
			
			# If searching filenames, append that to the story for 
			# searching as well
			
			if ($search_filenames == 1) {
				$contents.=$file;
			}
			
			# If we match any "OR" terms flag file for keeping
			$keep = 1 if ($contents =~ /$terms/si);
						
			# If we match required terms, keep, else delete for sure
			foreach (@requiredterms) {
				if ($contents =~ /$_/si) {
					$keep =1;
				} else {
					$delete = 1;
				}
			}

			# If we match forbidden terms, then delete
			foreach (@forbiddenterms) {
				if ($contents =~ /$_/si) {
					$delete =1;
				} 
			}

			# Remove file if marked for delete or not marked to keep
			delete $files_ref->{$file} if (($delete eq 1) or ($keep eq 0));
		}
	}

	1;
}

sub getrecentsearches {
	if ( open(LOG, "< $log_filename")) { 
		my @searches = <LOG>; 
		close(LOG); 
		@searches = reverse(@searches); 
		$recentsearches = "<ul>";
		for ($count = 0; $count < $max_previous_searches; $count++) { 
			$recentsearches .= '<li>' . $searches[$count] . '</li>';
		}
		$recentsearches .= "</ul>";
	} else {
		warn "Couldn't open $log_filename: $!\n" if ($log_searches == 1);
	}
	1;
}

sub head {
	getrecentsearches();
	1;
}


1;

__END__

=head1 NAME

Blosxom Plug-in: find

=head1 DESCRIPTION

Find searches through the available articles and filters out those that do not match the submitted search terms.  To use it, simply place $find::searchform in your template, and it will create a search box that automatically calls the search routine.  It performs a boolean "OR" search by default, or you can use regular expressions for more complicated search terms.

This plugin is capable of handling the following search terms

term1 term2; term1 or term2
	These match any page with term1 OR term2
	
term1 and term2; +term1 +term2
	These match any page with both term1 AND term2
	
term1 not term2; term1 -term2
	This matches pages with term1 that DO NOT contain term2
	
term1 not (term2 term3)
	This matches pages with term1 that DO NOT contain term2 OR term3

"term1 term2 term3"
	This matches the exact phrase, term1 term2 term3

" pen "
	This will match the word "pen", but not the word "pencil".
	
You can also use regular expressions within your search terms to further refine your searches, creating a very powerful search engine.

Additionally, you can include the most recent search requests in your blog.  Add $find::recentsearches in your template.  By default, the last 10 searches will be shown in an unordered list.  You can change $max_previous_searches to alter the number displayed.

=head1 AUTHORS

Fletcher T. Penney - http://fletcher.freeshell.org

Eric Sherman		<enkidu@enkidu.bloggedup.com>
Marc Nozell		<marc@nozell.com> http://www.nozell.com/blog

=head1 LICENSE

This source is submitted to the public domain.  Feel free to use and modify it.  If you like, a comment in your modified source attributing credit for my original work would be appreciated.

THIS SOFTWARE IS PROVIDED AS IS AND WITHOUT ANY WARRANTY OF ANY KIND.  USE AT YOUR OWN RISK!