WWW interface to Glimpse search engine

10 files changed, 424 insertions, 0 deletions

diff --git a/www/webglimpse/Makefile b/www/webglimpse/Makefile
new file mode 100644
index 0000000..4cd1f71
--- /dev/null
+++ b/www/webglimpse/Makefile
@@ -0,0 +1,36 @@
+# New ports collection makefile for:    WebGlimpse
+# http://glimpse.cs.arizona.edu/webglimpse/
+# Version required:                     1.*
+# Date created:                         2 Nov 1998
+# Whom:                                 ache
+#
+# $Id: Makefile,v 1.9 1998/10/07 10:15:19 ache Exp $
+#
+
+DISTNAME=       webglimpse-1.6.src
+PKGNAME=        webglimpse-1.6
+CATEGORIES=     www textproc
+MASTER_SITES=   ftp://ftp.cs.arizona.edu/glimpse/
+
+RUN_DEPENDS+=   /usr/local/etc/apache/httpd.conf:${PORTSDIR}/www/apache13
+RUN_DEPENDS+=   glimpse:${PORTSDIR}/textproc/glimpse
+WRKSRC=         ${WRKDIR}/webglimpse-1.6.edu
+
+MAINTAINER=     ache@freebsd.org
+
+USE_PERL5=      YES
+NO_BUILD=       YES
+
+pre-configure:
+.for i in wginstall wginstall.pl
+	cd ${WRKSRC}; cp $i $i.tmp; \
+		sed s=%PERL%=${PERL}= < $i.tmp > $i
+.endfor
+
+do-install:
+	cd ${WRKSRC}; ./wginstall
+	${INSTALL} -d -o ${SHAREOWN} -g ${SHAREGRP} -m 755 \
+		${PREFIX}/share/doc/webglimpse
+	cd ${WRKSRC}/docs && ${INSTALL_DATA} * ${PREFIX}/share/doc/webglimpse
+
+.include <bsd.port.mk>
diff --git a/www/webglimpse/distinfo b/www/webglimpse/distinfo
new file mode 100644
index 0000000..bd817d7
--- /dev/null
+++ b/www/webglimpse/distinfo
@@ -0,0 +1 @@
+MD5 (webglimpse-1.6.src.tar.gz) = 686b35432cac4d447601fa19d9b476b2
diff --git a/www/webglimpse/files/patch-aa b/www/webglimpse/files/patch-aa
new file mode 100644
index 0000000..d36c3d1
--- /dev/null
+++ b/www/webglimpse/files/patch-aa
@@ -0,0 +1,11 @@
+--- wginstall.bak	Fri Apr 24 03:41:34 1998
++++ wginstall	Mon Nov  2 07:18:17 1998
+@@ -3,7 +3,7 @@
+ # Ground-zero install script for webglimpse
+ # Tries to find location of perl and execute wginstall
+ 
+-perl_loc=`which perl`
++perl_loc=%PERL%
+ 
+ if [ -z $perl_loc ]
+ then
diff --git a/www/webglimpse/files/patch-ab b/www/webglimpse/files/patch-ab
new file mode 100644
index 0000000..88a5850
--- /dev/null
+++ b/www/webglimpse/files/patch-ab
@@ -0,0 +1,257 @@
+--- wginstall.pl.orig	Fri Sep 25 19:34:42 1998
++++ wginstall.pl	Mon Nov  2 09:11:33 1998
+@@ -31,7 +31,7 @@
+ 
+ 
+ # Get operating system
+-$LEGALOS = 'linux osf sunos solaris';
++#$LEGALOS = 'linux osf sunos solaris freebsd';
+ $DEFAULTOS = 'linux';
+ $PLATFORM = $ARGV[1] || $^O || '';
+ 
+@@ -46,16 +46,16 @@
+ }
+ 
+ 
+-do {
+-    $PLATFORM = prompt("What OS are you running (must be one of $LEGALOS)?", $PLATFORM);
+-} while ($LEGALOS !~ /$PLATFORM/);
++#do {
++#    $PLATFORM = prompt("What OS are you running (must be one of $LEGALOS)?", $PLATFORM);
++#} while ($LEGALOS !~ /$PLATFORM/);
+ 
+ ###########################################################
+ # Make sure we can find needed system binaries
+ # Since not all systems have which, we make a guess if which fails
+ #
+ # get the paths for perl and glimpse
+-$PERL = $ARGV[0] || `which perl` || '/usr/local/bin/perl';
++$PERL = '%PERL%';
+ $CONVERT_LOC = `which wgconvert` || '/usr/local/bin/wgconvert';
+ $GLIMPSE_LOC = `which glimpse` || '/usr/local/bin/glimpse';
+ $GLIMPSEIDX_LOC = `which glimpseindex` || '/usr/local/bin/glimpseindex';
+@@ -168,9 +168,10 @@
+ }
+ chomp $guess;
+ if (($guess eq '')||($guess eq '.')) {
+-	$guess = '/usr/local/webglimpse';
++	$guess = '/usr/local/www/webglimpse';
+ }
+-$WEBGLIMPSE_HOME = prompt('Webglimpse Home Directory: ', $guess);
++#$WEBGLIMPSE_HOME = prompt('Webglimpse Home Directory: ', $guess);
++$WEBGLIMPSE_HOME = $guess;
+ 
+ # Remove any trailing '/'
+ if ($WEBGLIMPSE_HOME =~ /\/$/) { 
+@@ -211,12 +212,12 @@
+ # Check if .wgsiteconf already exists
+ $HAVECONFIG = 'N';
+ if ( -e "$WEBGLIMPSE_HOME/.wgsiteconf" ) {
+-	$HAVECONFIG = prompt('Existing installation found.  Keep configuration info? (Y/n)', 'Y');
+-	if ($HAVECONFIG =~ /^[yY]/) {
++#       $HAVECONFIG = prompt('Existing installation found.  Keep configuration info? (Y/n)', 'Y');
++#       if ($HAVECONFIG =~ /^[yY]/) {
+ 		$HAVECONFIG = 'Y';
+-	} else {
+-		$HAVECONFIG = 'N';
+-	}
++#       } else {
++#               $HAVECONFIG = 'N';
++#       }
+ }
+ 
+ #######################################################################
+@@ -225,53 +226,53 @@
+ #
+ if ($HAVECONFIG eq 'N') {
+ 
+-	print "\nI will now ask you some questions about your HTTP server.\n";
+-	print "Automatic configuration is supported for Apache servers v1.2 and above\n";
+-	print "If you do not have an Apache server, you may need to edit the file\n";
+-	print "$WEBGLIMPSE_HOME/.wgsiteconf by hand after the installation is completed.\n\n";
+-
++#        print "\nI will now ask you some questions about your HTTP server.\n";
++#        print "Automatic configuration is supported for Apache servers v1.2 and above\n";
++#        print "If you do not have an Apache server, you may need to edit the file\n";
++#        print "$WEBGLIMPSE_HOME/.wgsiteconf by hand after the installation is completed.\n\n";
++#
+ 	#		
+ 	# Try to guess the configuration directory/file from the process list
+ 	#
+-	$allprocs = `ps -a` || '';
+-	$confdir = '';
+-	$conffile = '';
+-	if ($allprocs ne '') {
+-		@allprocs = split(/\n/,$allprocs);
+-		foreach $proc ( @allprocs ) {
+-			if ($proc =~ /httpd/) {
+-				if ($proc =~ / -f (\S+)/) {
+-					$conffile = $1;
+-				}
+-				if ($proc =~ / -d (\S+)/) {
+-					$confdir = $1;
+-				}
+-				last;
+-			}
+-		}
+-	}
+-	$HTTPD_CONF = '';
+-	if ($conffile ne '') {
+-		# Usually the daemon is run as httpd -f /path/httpd.conf
+-		$HTTPD_CONF = $conffile;
+-	}
+-
+- 	if ($confdir ne '') {
+-		if ($confdir =~ /\/$/) {
+-			chop $confdir;
+-		}
+-		if ($HTTPD_CONF eq '') {
+-			$HTTPD_CONF = 'httpd.conf';
+-		}
+-		$HTTPD_CONF = $confdir.'/'.$HTTPD_CONF;
+-	} 
+-
+-	if ($HTTPD_CONF eq '') {
+-		$HTTPD_CONF = '/usr/local/etc/httpd/conf/httpd.conf';
+-	}
++#        $allprocs = `ps -a` || '';
++#        $confdir = '';
++#        $conffile = '';
++#        if ($allprocs ne '') {
++#                @allprocs = split(/\n/,$allprocs);
++#                foreach $proc ( @allprocs ) {
++#                        if ($proc =~ /httpd/) {
++#                                if ($proc =~ / -f (\S+)/) {
++#                                        $conffile = $1;
++#                                }
++#                                if ($proc =~ / -d (\S+)/) {
++#                                        $confdir = $1;
++#                                }
++#                                last;
++#                        }
++#                }
++#        }
++#        $HTTPD_CONF = '';
++#        if ($conffile ne '') {
++#                # Usually the daemon is run as httpd -f /path/httpd.conf
++#                $HTTPD_CONF = $conffile;
++#        }
++#
++#        if ($confdir ne '') {
++#                if ($confdir =~ /\/$/) {
++#                        chop $confdir;
++#                }
++#                if ($HTTPD_CONF eq '') {
++#                        $HTTPD_CONF = 'httpd.conf';
++#                }
++#                $HTTPD_CONF = $confdir.'/'.$HTTPD_CONF;
++#        }
++#
++#        if ($HTTPD_CONF eq '') {
++		$HTTPD_CONF = '/usr/local/etc/apache/httpd.conf';
++#        }
+ 
+ 	# Check with the user if our guess is right.
+-	$HTTPD_CONF = &prompt('Full path to HTTP server config file: ',$HTTPD_CONF);
++#        $HTTPD_CONF = &prompt('Full path to HTTP server config file: ',$HTTPD_CONF);
+ 
+ 	########################################################################
+ 	# Call wgSiteConfig from wgserver.pl to parse server config file
+@@ -351,13 +352,13 @@
+ 
+ 
+ # Prompt user no matter what; we may be overwriting a file here.
+-$CGIBIN_PWD = 
+-	&prompt("Which directory should I copy the WebGlimpse cgi-bin scripts to?\nIf you choose a different\ndirectory than the default, make sure it exists.\n",
+-	"$CGIBIN_PWD");
+-
+-$CGIBIN = 
+-	&prompt("What is the script alias for $CGIBIN_PWD?",
+-		$CGIBIN);
++#$CGIBIN_PWD =
++#        &prompt("Which directory should I copy the WebGlimpse cgi-bin scripts to?\nIf you choose a different\ndirectory than the default, make sure it exists.\n",
++#        "$CGIBIN_PWD");
++
++#$CGIBIN =
++#        &prompt("What is the script alias for $CGIBIN_PWD?",
++#                $CGIBIN);
+ 
+ # If this is news to us, add/change the WGSCRIPTALIAS line to the site config file
+ $wg_added = 0;
+@@ -435,8 +436,8 @@
+ print "cgi-bin *relative* URL: $CGIBIN\n";
+ print "-----------------------------------------------------\n";
+ 
+-print "To continue with these settings, press RETURN.  Otherwise, press ^C to abort.\n";
+-$_=<STDIN>;
++#print "To continue with these settings, press RETURN.  Otherwise, press ^C to abort.\n";
++#$_=<STDIN>;
+ 
+ 
+ # build local copy of httpget and html2txt
+@@ -606,11 +607,11 @@
+ 
+ print "Webglimpse installation complete.\n\n";
+ 
+-print "**NOTE: You may need to edit your web server configuration file \n";
+-print "and specify $CGIBIN_PWD as a cgi-bin directory.\n";
+-print "For Apache, use the ScriptAlias directive.\n\n";
+-print("Press RETURN to continue\n");
+-<STDIN>;
++#print "**NOTE: You may need to edit your web server configuration file \n";
++#print "and specify $CGIBIN_PWD as a cgi-bin directory.\n";
++#print "For Apache, use the ScriptAlias directive.\n\n";
++#print("Press RETURN to continue\n");
++#<STDIN>;
+ 
+ 
+ print "\nSee http://glimpse.cs.arizona.edu/webglimpse for more\n";
+@@ -619,25 +620,25 @@
+ print "You may configure an archive at any time with the command \n";
+ print "$WEBGLIMPSE_HOME/confarc\n\n";
+ 
+-print "To configure one or more archives now, just enter the directory to index or URL to traverse.\n";
+-print "A directory should be a full path starting with   /  \n";
+-print "A URL should be absolute starting with    http://  \n\n";
++#print "To configure one or more archives now, just enter the directory to index or URL to traverse.\n";
++#print "A directory should be a full path starting with   /  \n";
++#print "A URL should be absolute starting with    http://  \n\n";
+ 
+-$archivepath = &prompt("Enter directory or URL to index, return to quit:", '');
+-while($archivepath ne '') {
++#$archivepath = &prompt("Enter directory or URL to index, return to quit:", '');
++#while($archivepath ne '') {
+ 
+ 	# We want to make certain simplified assumptions here.
+ 	# User can get full flexibility by running confarc from
+ 	# command line later.
+-	$type = 'D';
+-	if ($archivepath =~ /^http/) {
+-		$type = 'T';
+-	}
++#        $type = 'D';
++#        if ($archivepath =~ /^http/) {
++#                $type = 'T';
++#        }
+ 
+-	system("$WEBGLIMPSE_HOME/confarc -i $archivepath -t $type -l $archivepath");
++#        system("$WEBGLIMPSE_HOME/confarc -i $archivepath -t $type -l $archivepath");
+ 
+-	$archivepath = &prompt("Enter directory or URL to index, return to quit:", '');
+-}
++#        $archivepath = &prompt("Enter directory or URL to index, return to quit:", '');
++#}
+ 
+ 1;
+ 
+@@ -703,8 +704,8 @@
+ 			print TEMP "\$CGIBIN = \"$CGIBIN\";\n";
+ 		}elsif(/^\$gunzip *=/){
+ 			print TEMP "\$gunzip = \"$gunzip\";\n";
+-		}elsif(/^#!\/usr\/local\/bin\/perl$/){
+-			print TEMP "#!$PERL\n";
++		}elsif(/^#!\/usr\/local\/bin[^ ]*\/perl(.*)/){
++			print TEMP "#!$PERL$1\n";
+ 		}else{
+ 			print TEMP $_;
+ 		}
diff --git a/www/webglimpse/files/patch-ac b/www/webglimpse/files/patch-ac
new file mode 100644
index 0000000..f57516e
--- /dev/null
+++ b/www/webglimpse/files/patch-ac
@@ -0,0 +1,17 @@
+--- Makefile.orig	Wed Oct 22 09:49:26 1997
++++ Makefile	Mon Nov  2 09:21:49 1998
+@@ -19,9 +19,13 @@
+ 
+ 
+ 
+-CC = gcc
++#CC = gcc
+ 
+ OSFCC = cc
++
++freebsd:
++	$(CC) ${CFLAGS} -s -o lib/httpget lib/httpget.c
++	$(CC) ${CFLAGS} -s -o lib/html2txt lib/html2txt.c
+ 
+ linux:
+ 	$(CC) -O -o lib/httpget lib/httpget.c
diff --git a/www/webglimpse/files/patch-ad b/www/webglimpse/files/patch-ad
new file mode 100644
index 0000000..a00e3ab
--- /dev/null
+++ b/www/webglimpse/files/patch-ad
@@ -0,0 +1,21 @@
+--- lib/wgserver.pl.orig	Mon Jun 29 20:45:06 1998
++++ lib/wgserver.pl	Mon Nov  2 08:26:21 1998
+@@ -114,6 +114,7 @@
+ 
+ 	$override = 0;
+ 	$ServerRoot = '';
++	$output = '';
+ 	
+ 	while (<CONF>)	{
+ 		# Trim leading white space 
+@@ -163,8 +164,8 @@
+ 
+ 	close(CONF);
+ 
+-	($ServerRoot ne '') && ($ResourceConfig = $ServerRoot.'/'.$ResourceConfig);	
+-	($ServerRoot ne '') && ($AccessConfig = $ServerRoot.'/'.$AccessConfig);	
++	($ServerRoot ne '') && ($ResourceConfig ne '') && ($ResourceConfig !~ '^/') && ($ResourceConfig = $ServerRoot.'/'.$ResourceConfig);
++	($ServerRoot ne '') && ($AccessConfig ne '') && ($AccessConfig !~ '^/') && ($AccessConfig = $ServerRoot.'/'.$AccessConfig);
+ 
+ 	return $output;
+ }
diff --git a/www/webglimpse/files/patch-ae b/www/webglimpse/files/patch-ae
new file mode 100644
index 0000000..dfc8893
--- /dev/null
+++ b/www/webglimpse/files/patch-ae
@@ -0,0 +1,30 @@
+--- cgi-bin/webglimpse-fullsearch.orig	Mon Nov  2 08:52:04 1998
++++ cgi-bin/webglimpse-fullsearch	Mon Nov  2 08:53:25 1998
+@@ -159,19 +159,19 @@
+ print "<INPUT TYPE=submit VALUE=Submit>\n";
+ print "<br>\n";
+ print "<center>\n";
+-print "<INPUT NAME=case TYPE=checkbox>Case&#160;sensitive\n";
+-print "<!-- SPACES -->&#160;&#160;&#160;\n";
+-# print "<!-- SPACES -->&#160;&#160;&#160;&#160;&#160;&#160;&#160;\n";
+-print "<INPUT NAME=whole TYPE=checkbox>Partial&#160;match\n";
+-print "<!-- SPACES -->&#160;&#160;&#160;\n";
+-print "<INPUT NAME=lines TYPE=checkbox>Jump&#160;to&#160;line\n";
+-print "<!-- SPACES -->&#160;&#160;&#160;\n";
++print "<INPUT NAME=case TYPE=checkbox>Case&nbsp;sensitive\n";
++print "<!-- SPACES -->&nbsp;&nbsp;&nbsp;\n";
++# print "<!-- SPACES -->&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;\n";
++print "<INPUT NAME=whole TYPE=checkbox>Partial&nbsp;match\n";
++print "<!-- SPACES -->&nbsp;&nbsp;&nbsp;\n";
++print "<INPUT NAME=lines TYPE=checkbox>Jump&nbsp;to&nbsp;line\n";
++print "<!-- SPACES -->&nbsp;&nbsp;&nbsp;\n";
+ print "<SELECT NAME=errors align=right>\n";
+ print "<OPTION>0\n";
+ print "<OPTION>1\n";
+ print "<OPTION>2\n";
+ print "</SELECT>\n";
+-print "misspellings&#160;allowed\n";
++print "misspellings&nbsp;allowed\n";
+ print "<br>\n";
+ print "</center>\n";
+ print "Return only files modified within the last <INPUT NAME=age size=5>\n";
diff --git a/www/webglimpse/pkg-comment b/www/webglimpse/pkg-comment
new file mode 100644
index 0000000..c0347e6
--- /dev/null
+++ b/www/webglimpse/pkg-comment
@@ -0,0 +1 @@
+WWW interface to Glimpse search engine.
diff --git a/www/webglimpse/pkg-descr b/www/webglimpse/pkg-descr
new file mode 100644
index 0000000..6a187c4
--- /dev/null
+++ b/www/webglimpse/pkg-descr
@@ -0,0 +1,10 @@
+WebGlimpse adds search capabilities to your WWW site automatically and
+easily. It attaches a small search box to the bottom of every HTML page,
+and allows the search to cover the neighborhood of that page or the whole
+site. With WebGlimpse there is no need to construct separate search pages,
+and no need to interrupt the users from their browsings. All pages remain
+unchanged except for the extra search capabilities. It is even possible
+for the search to efficiently cover remote pages linked from your pages.
+(WebGlimpse will collect such remote pages to your disk and index them.)
+Installation, customization (e.g., deciding which pages to collect and
+which ones to index), and maintenance are easy.
diff --git a/www/webglimpse/pkg-plist b/www/webglimpse/pkg-plist
new file mode 100644
index 0000000..cd7ab9b
--- /dev/null
+++ b/www/webglimpse/pkg-plist
@@ -0,0 +1,40 @@
+share/doc/webglimpse/confarc.html
+share/doc/webglimpse/install_1.6.html
+share/doc/webglimpse/sample.wgsiteconf
+www/cgi-bin/mfs
+www/cgi-bin/webglimpse
+www/cgi-bin/webglimpse-fullsearch
+www/webglimpse/.archives
+www/webglimpse/.wgsiteconf
+www/webglimpse/Makefile
+www/webglimpse/addsearch
+www/webglimpse/confarc
+www/webglimpse/dist/glimpse-eye.jpg
+www/webglimpse/dist/glimpse_filters
+www/webglimpse/dist/wgall.html
+www/webglimpse/dist/wgbox.html
+www/webglimpse/dist/wgfilter-box
+www/webglimpse/dist/wgfilter-index
+www/webglimpse/dist/wgindex.html
+www/webglimpse/getfile
+www/webglimpse/lib/OutputTool.pm
+www/webglimpse/lib/URL.pl
+www/webglimpse/lib/config.pl
+www/webglimpse/lib/ftplib.pl
+www/webglimpse/lib/html2txt
+www/webglimpse/lib/httpget
+www/webglimpse/lib/httpget.c
+www/webglimpse/lib/httpget.pl
+www/webglimpse/lib/siteconf.pl
+www/webglimpse/lib/url_get
+www/webglimpse/lib/url_get.pl
+www/webglimpse/lib/utils.h
+www/webglimpse/lib/webgutils.pl
+www/webglimpse/lib/wgserver.pl
+www/webglimpse/makecron
+www/webglimpse/makenh
+www/webglimpse/rmarc
+@dirrm www/webglimpse/dist
+@dirrm www/webglimpse/lib
+@dirrm www/webglimpse
+@dirrm share/doc/webglimpse