#!/bin/sh
# shar:	Shell Archiver  (v1.22)
#
#	Run the following text with /bin/sh to create:
#	  Readme
#	  mklist
#	  mylist.cover
#	  mylist.response
#	  srch.cgi
#
sed 's/^X//' << 'SHAR_EOF' > Readme &&
XTitleSrch -  a CGI script to search for HTML titles or do grep searches
X=======================================================================
X
XOverview:
X
XThis package consists of two perl scripts and two customizable HTML
Xdocs (a coversheet and a response).  The first script, mklist,
Xproduces a file consisting of title/URL pairs (or for grep searches
Xtitle/URL/filepath).  This file can be edited or entirely built by
Xhand if you desire.  The CGI script, srch.cgi, searches the title
Xfield of all lines in this file and returns a response HTML document
Xin which it inserts an unordered list of all URLs matching the query
Xwhen it encounters a line containing only the special token $MATCHES.
XAlso the token $QUERY in the HTML response is replaced by the search
Xterm.  If the list file has three fields per line, the last being the
Xpath to the file, then a grep search of the file is done instead of
Xlooking for matches only in the title.
X
X
XHere are the steps to get srch.cgi running.
X
X1. Edit the mklist script entering your hostname and data root
Xdirectory.  Then produce a list file with this script. This is a
Xfile each line of which looks like
X
XTitle<tab>URL\n        (for title searches)
X
Xor
X
XTitle<tab>URL<tab>path2file\n  (for grep searches)
X
XBy default mklist produces (on standard output) a list file for title
Xsearches in the directory (relative to the data root) which is its
Xfirst argument.  If it is given the -r option it will recursively do
Xall subdirectories of that directory.  If a directory is not given
Xthe root is used.  To find the title it checks the first 5 lines of 
Xevery text file looking for <title> or <TITLE>.  The beginning and
Xending title tags should be on the same line.
X
XExamples:  "mklist > mylist" -> list for all files in root dir
X	   "mklist -r > mylist -> list for all files in root and its subdirs
X	   "mklist -r /dir/dir2 > mylist" -> list for all files in 
X					root/dir/dir2 and its subdirectories
X
X
XIf you use the "-g" option it produces the three field file which indicates
Xto srch.cgi to do a complete grep search of the files rather than just
Xsearch the titles.
X
X2.  Install the "mylist" file you made in step 1 in some directory on
Xyour server along with the mylist.cover and mylist.response files.
XYou will probably want to edit these two files to customize your
Xinformation.  There is nothing magic about the name "mylist", but if
Xthe listfile is named "foo" the cover and response files must be named
X"foo.cover" and "foo.response" and must be in the same directory.
X
X3.  Install srch.cgi in your cgi-bin. The URL pointing to this script
Xshould then be
X
Xhttp://host/cgi-bin/srch.cgi/path2/mylist
X
Xwhere "/path2/mylist" is whatever your server will translate to the
Xcorrect path to the mylist file and place in the CGI environment 
Xvariable PATH_TRANSLATED.
X
XThat's it.
X
X
XJohn Franks 	Dept of Math. Northwestern University
X		john@math.nwu.edu
X
X
X
SHAR_EOF
chmod 0644 Readme || echo "restore of Readme fails"
sed 's/^X//' << 'SHAR_EOF' > mklist &&
X#!/usr/net/bin/perl
X# Mklist version 0.1
X# Usage: mklist [-rtbg] [-R rootdir] [-h host] dir
X# -r means recursively do subdirectories
X# -t include all text files
X# -b include binary and text files
X# -g full text grep search of files rather than title only
X# dir is starting directory relative to root (if empty use root)
X
X# This program produces a file consisting of lines of the form
X# title<tab>url\n   or   title<tab>url<tab>path2file\n 
X# if the -g option is used.  It assumes that the only address translation
X# from a  URL to the file path is prepending the root directory.
X
X# This program has been placed in the public domain by its author
X# John Franks
X
X# You must edit the following fields filling in you root directory and host.
X
X$host="host.univ.edu";
X$rootdir="/data/root";
X
X##########################################################################
Xrequire "getopts.pl";
X&Getopts( "rtbgR:h:");
X$opt_t |= $opt_b;  #Option -b implies option -t
X
X$host= $opt_h ? $opt_h : $host;
X$rootdir= $opt_R ? $opt_R : $rootdir;
X$startdir=$ARGV[0];
X
X&dodir($startdir);
X
Xsub dodir {
X	local($dir) = @_;
X	local($path) = $rootdir.$dir;
X	local($file);
X	opendir( DIR, $path) || die "Can't open $path";
X	local( @filenames) = readdir(DIR);
X	closedir(DIR);
X
X	foreach $file (@filenames) {
X		next if ($file =~ /^\./);
X		if ( $file =~ /\.html$/) {
X			&do_file( $dir, $file, "h");
X		} elsif ( $opt_t && (-T "$path/$file")) {
X			&do_file( $dir, $file, "t");
X		} elsif ( $opt_b && (-B "$path/$file")
X				&& !(-d "$path/$file")) {
X			&do_file( $dir, $file, "b");
X		}
X		&dodir( "$dir/$file") if (( -d "$path/$file") && $opt_r );
X		
X	}
X}
X
Xsub do_file {
X	local( $dir, $file, $type) = @_;
X	local( $filepath) = $rootdir.$dir."/".$file;
X
X	if ( $type eq "b" ) {
X		$title = "Binary file: $file";
X	} elsif ( $type =~ /[ht]/ ) {
X		open( TEXT_FILE, $filepath)
X				|| die "Can't open $filepath\n";
X		local( $line) = "";
X		local( $linenum) = 1;
X		$title = "Text file: $file";
X		$foundtitle = 0;
X		while ( $line = <TEXT_FILE>) {
X			last if ($linenum > 5);
X			$linenum++;
X			if ($line =~ s/^.*<title>//i ) {
X				chop( $line);
X				$line =~ s!</title>.*$!!i;
X				$title = $line;
X				$title =~ s/^\s*//;
X				$title =~ s/\s*$//;
X				$foundtitle = 1;
X				last;
X			}
X		}
X		close( TEXT_FILE);
X	}
X	if ( ($type =~ /[bt]/) || $foundtitle) {
X		if ( $opt_g ) {
X			printf( "%s\thttp://%s%s/%s\t%s\n",
X				$title, $host, $dir, $file, $filepath);
X		} else {
X			printf( "%s\thttp://%s%s/%s\n",
X				$title, $host, $dir, $file);
X		}
X	}
X}
SHAR_EOF
chmod 0755 mklist || echo "restore of mklist fails"
sed 's/^X//' << 'SHAR_EOF' > mylist.cover &&
X<HTML>
X<HEAD>
X<ISINDEX>
X<TITLE>Search Titles</TITLE>
X</HEAD>
X<BODY>
X<H2>Search Titles on this server</H2>
X
XThe search is case insenstive.  Any Perl regular expression is an
Xallowable search term. 
X
X</BODY>
X</HTML>
SHAR_EOF
chmod 0644 mylist.cover || echo "restore of mylist.cover fails"
sed 's/^X//' << 'SHAR_EOF' > mylist.response &&
X<HTML>
X<HEAD>
X<ISINDEX>
X<TITLE>Document Search</TITLE>
X</HEAD>
X<BODY>
X<H2>Results of Title Search</H2>
X
XHere are the matches for the search term <b>'$QUERY':</b>
X<P>
X
X$MATCHES
X
XYou may repeat your search with a new search term.  The search is case
Xinsenstive.  Any Perl regular expression is an allowable search term.
X<P>
X</BODY>
X</HTML>
SHAR_EOF
chmod 0644 mylist.response || echo "restore of mylist.response fails"
sed 's/^X//' << 'SHAR_EOF' > srch.cgi &&
X#!/usr/net/bin/perl
X# Srch.cgi version 0.1
X# This is a CGI script which receives a query and pathinfo specifying
X# a file consisting of lines of the form Title<tab>URL. It prints a
X# response file to stdout replacing the token $QUERY with the query passed
X# to it an the line containing $MATCHES with an unorderd list of URLs
X# for matching documents.  If the query is empty it prints a coversheet.
X# If the list file contains entries of the form Title<tab>URL<tab>filepath,
X# then the file referenced by filepath is grepped for the query string instead
X# of seeking a match in the title field.
X
X# This program has been placed in the public domain by its author
X# John Franks
X
X
X$pathinfo = $ENV{PATH_INFO};  # Get the argument from the URL
X$listfile = $ENV{PATH_TRANSLATED};  # Get the argument from the URL
X$coverfile = $listfile.".cover";
X$responsefile = $listfile.".response";
X$query = $ENV{QUERY_STRING};
X$query =~ s/%([\da-f]{1,2})/pack(C,hex($1))/eig;
X
Xif ( $query eq "") {
X	open (COVER, $coverfile) 
X		|| &punt( "Couldn't open cover file $coverfile.\n");
X	print "Content-type: text/html\n\n";
X	print "Got here\n";
X	while ( $line = <COVER> ) {
X		print $line;
X	}
X	exit( 0);
X}
X
Xopen (LISTFILE, $listfile) || &punt( "Couldn't open list file: $listfile.\n");
Xopen (RESPONSE, $responsefile) || &punt( "Couldn't open response file.\n");
X
Xprint "Content-type: text/html\n\n";
X
Xwhile ( $line = <RESPONSE> ) {
X	$line =~ s/\$QUERY/$query/;
X	if ( $line =~ /^\$MATCHES/) {
X		&matches;
X	}
X	else {
X		print $line;
X	}
X}
X
Xsub matches 
X{
X	local( $foundmatch);
X	$foundmatch = 0;
X	LINE:
X	while ( $listline = <LISTFILE> ) {
X		chop $listline;
X		$listline =~ s/\s*#.*$//;
X		next LINE if ( $listline eq "");
X
X		( $title, $url, $filepath ) = split( /\t/, $listline, 3);
X		if ( $filepath ) {
X			open( GREPFILE, $filepath)
X				|| &punt( "Couldn't open file to grep.\n");
X			while ( <GREPFILE> ) {
X				if ($_ =~ /$query/i) {
X					if ( !$foundmatch) {
X						print "<UL>\n";
X						$foundmatch = 1;
X					}
X					printf( "<LI><A HREF=\"%s\"> %s</A>\n",
X						$url, $title);
X					next LINE;
X				}
X			}
X		}
X		else {
X			if ( $title =~ /$query/i) {
X				if ( !$foundmatch) {
X					print "<UL>\n";
X					$foundmatch = 1;
X				}
X				printf( "<LI><A HREF=\"%s\"> %s</A>\n",
X					$url, $title);
X			}
X		}
X	}
X	if ( $foundmatch) {
X		print "</UL>\n<P>\n";
X	} else {
X		print "<B>Sorry no matches were found.</B> <P>\n";
X	}
X}
X
Xsub punt
X{
X	print "Content-type: text/html\n\n";
X	printf( "<HEAD><TITLE>500 %s</TITLE></HEAD>\n", @_);
X	printf( "<BODY><H2>Error code 500</H2>\n%s\n</BODY>\n", @_);
X	exit( 1);
X}
SHAR_EOF
chmod 0755 srch.cgi || echo "restore of srch.cgi fails"
exit 0