16 files changed, 322 insertions, 128 deletions

diff --git a/www/harvest/Makefile b/www/harvest/Makefile
index 8861d70d970f..7c4c7e486921 100644
--- a/www/harvest/Makefile
+++ b/www/harvest/Makefile
@@ -6,28 +6,48 @@
 #
 
 PORTNAME=	harvest
-PORTVERSION=	1.5
+PORTVERSION=	1.8.1
 CATEGORIES=	www
-MASTER_SITES=	ftp://ftp.tardis.ed.ac.uk/pub/harvest/ \
-		ftp://nic.nasa.gov/software/harvest-dist/ \
-		ftp://ftp.sunet.se/pub/nir/harvest-uk/ \
-		ftp://gd.tuwien.ac.at/infosys/indexers/uk-harvest/ \
-		http://www3.uniovi.es/~mirror/harvest-net/
-EXTRACT_SUFX=	-src.tar.gz
+MASTER_SITES=	${MASTER_SITE_SOURCEFORGE}
+MASTER_SITE_SUBDIR=	${PORTNAME}
 
 MAINTAINER=	ports@FreeBSD.org
 
-USE_REINPLACE=	yes
+USE_BZIP2=	yes
 USE_PERL5=	yes
+USE_REINPLACE=	yes
+HAS_CONFIGURE=	yes
+CONFIGURE_ARGS=	--build=${MACHINE_ARCH}-portbld-freebsd${OSREL} \
+		--prefix=${PREFIX}/${PORTNAME} \
+		--without-wish 
+
+STRIP=		# none
 
-pre-patch:
-	@${REINPLACE_CMD} -e 's,<malloc.h>,<stdlib.h>,' \
-		${WRKSRC}/src/configure ${WRKSRC}/src/broker/broker.h \
-		${WRKSRC}/src/gatherer/essence/main.c \
-		${WRKSRC}/src/gatherer/include/essence.h \
-		${WRKSRC}/components/gatherer/standard/dvi2tty/dvi2tty.h
+post-patch:
+	@${GREP} -lR "malloc.h" ${WRKSRC} | ${XARGS} ${REINPLACE_CMD} -e \
+		's|malloc\.h|stdlib.h|g'
 
-pre-install:
-	${MKDIR} ${PREFIX}/harvest
+post-install:
+.for file in bin/broker bin/gather lib/broker/BrokerRestart \
+	lib/broker/agrep lib/broker/brkclient lib/broker/dumpregistry \
+	lib/broker/glimpse lib/broker/glimpseindex.bin \
+	lib/broker/glimpseserver lib/broker/swish lib/ftpget \
+	lib/gatherer/HTML-lax.sum lib/gatherer/HTMLurls lib/gatherer/catdoc \
+	lib/gatherer/cksoif lib/gatherer/cleandb lib/gatherer/consoldb \
+	lib/gatherer/dvi2tty lib/gatherer/enum lib/gatherer/essence \
+	lib/gatherer/expiredb lib/gatherer/fileenum lib/gatherer/ftpenum \
+	lib/gatherer/gatherd lib/gatherer/gdbmutil lib/gatherer/gifinfo \
+	lib/gatherer/gopherenum-breadth lib/gatherer/gopherenum-depth \
+	lib/gatherer/hexbin lib/gatherer/httpenum-breadth \
+	lib/gatherer/httpenum-depth lib/gatherer/mergedb \
+	lib/gatherer/mkindex lib/gatherer/newsenum lib/gatherer/prepurls \
+	lib/gatherer/print-attr lib/gatherer/print-c-comments \
+	lib/gatherer/ps2txt-2.1 lib/gatherer/pstext lib/gatherer/rast \
+	lib/gatherer/rmbinary lib/gatherer/rtf2html lib/gatherer/sgmls \
+	lib/gatherer/sgmlsasp lib/gatherer/skim lib/gatherer/staturl \
+	lib/gatherer/template2db lib/gatherer/wp2x lib/gatherer/xls2csv \
+	lib/md5 lib/urlget lib/urlpurge
+	@${STRIP_CMD} ${PREFIX}/${PORTNAME}/${file}
+.endfor
 
 .include <bsd.port.mk>
diff --git a/www/harvest/distinfo b/www/harvest/distinfo
index b8547bf3951d..8881e1c932d5 100644
--- a/www/harvest/distinfo
+++ b/www/harvest/distinfo
@@ -1 +1 @@
-MD5 (harvest-1.5-src.tar.gz) = 708fd9ac95fd7606ee35d569a0c33a4c
+MD5 (harvest-1.8.1.tar.bz2) = 9425a081d215686f55bf13efecd02bd0
diff --git a/www/harvest/files/patch-aa b/www/harvest/files/patch-aa
deleted file mode 100644
index 90e5aae10b26..000000000000
--- a/www/harvest/files/patch-aa
+++ /dev/null
@@ -1,20 +0,0 @@
---- components/gatherer/standard/skim/skim.c.orig	Sat Mar 30 18:54:27 2002
-+++ components/gatherer/standard/skim/skim.c	Sat Mar 30 18:54:45 2002
-@@ -15,7 +15,7 @@
- 
- int Argc;
- char **Argv;
--FILE *InputFile = stdin;
-+FILE *InputFile;
- 
- main(argc, argv)
- int argc;
-@@ -25,6 +25,8 @@
- 	int Skipping = 0, nlines = 0;
- 	char *Ptr;
- 	char *EndLine;
-+
-+	InputFile = stdin;
- 
- 	/* Set up globals */
- 	Argc = argc;
diff --git a/www/harvest/files/patch-ac b/www/harvest/files/patch-ac
deleted file mode 100644
index a525ee60884b..000000000000
--- a/www/harvest/files/patch-ac
+++ /dev/null
@@ -1,12 +0,0 @@
---- ./src/gatherer/essence/file/print.c.orig	Sun Oct 20 17:44:09 2002
-+++ ./src/gatherer/essence/file/print.c	Sun Oct 20 17:44:18 2002
-@@ -77,8 +77,8 @@
- warning(f, a)
-      char *f, *a;
- {
--	extern int errno, sys_nerr;
- #if !defined(__FreeBSD__) && !defined(__NetBSD__)
-+	extern int errno, sys_nerr;
- 	extern char *sys_errlist[];
- #endif
- 	int myerrno;
diff --git a/www/harvest/files/patch-ad b/www/harvest/files/patch-ad
deleted file mode 100644
index c3b35674caee..000000000000
--- a/www/harvest/files/patch-ad
+++ /dev/null
@@ -1,11 +0,0 @@
---- src/broker/Glimpse/index.c.old	Fri Oct 25 19:52:18 2002
-+++ src/broker/Glimpse/index.c	Fri Oct 25 19:52:34 2002
-@@ -192,7 +192,7 @@
- 		GL_GlimpseSrvPort = -1;
- 		return;
- 	}
--	if (GL_GlimpseSrvPort <= IPPORT_USERRESERVED) {
-+	if (GL_GlimpseSrvPort <= 5000) {
- 		/* choose a random port number between 16384-30000 */
- #if defined(HAVE_SRAND48) && defined(HAVE_LRAND48)
- 		srand48(time(NULL));
diff --git a/www/harvest/files/patch-ae b/www/harvest/files/patch-ae
deleted file mode 100644
index 7cdc80ec63b3..000000000000
--- a/www/harvest/files/patch-ae
+++ /dev/null
@@ -1,12 +0,0 @@
---- components/broker/standard/WWW/DisplayObject.c.old	Tue Oct 29 00:51:14 2002
-+++ components/broker/standard/WWW/DisplayObject.c	Tue Oct 29 00:51:30 2002
-@@ -214,8 +214,8 @@
- char *xstrerror(n)
- int n;
- {
--	extern int sys_nerr;
- #if !defined(__FreeBSD__) && !defined(__NetBSD__)
-+	extern int sys_nerr;
- 	extern char *sys_errlist[];
- #endif
- 
diff --git a/www/harvest/files/patch-components::broker::standard::WWW::DisplayObject.c b/www/harvest/files/patch-components::broker::standard::WWW::DisplayObject.c
new file mode 100644
index 000000000000..1622f8058eed
--- /dev/null
+++ b/www/harvest/files/patch-components::broker::standard::WWW::DisplayObject.c
@@ -0,0 +1,12 @@
+--- components/broker/standard/WWW/DisplayObject.c.orig	Mon Feb 11 07:57:16 2002
++++ components/broker/standard/WWW/DisplayObject.c	Wed Nov 13 09:01:03 2002
+@@ -215,8 +215,8 @@
+ char *xstrerror(n)
+ int n;
+ {
+-	extern int sys_nerr;
+ #ifdef NEED_SYS_ERRLIST
++	extern int sys_nerr;
+ 	extern char *sys_errlist[];
+ #endif
+ 
diff --git a/www/harvest/files/patch-components::broker::swish::Makefile.in b/www/harvest/files/patch-components::broker::swish::Makefile.in
new file mode 100644
index 000000000000..c63d91aba762
--- /dev/null
+++ b/www/harvest/files/patch-components::broker::swish::Makefile.in
@@ -0,0 +1,11 @@
+--- components/broker/swish/Makefile.in.orig	Wed Nov  5 03:14:43 1997
++++ components/broker/swish/Makefile.in	Wed Nov 13 10:11:37 2002
+@@ -15,7 +15,7 @@
+ prefix = @prefix@
+ bindir = $(prefix)/lib/broker
+ 
+-INSTALL_PROG = @INSTALL@
++INSTALL_PROG = @INSTALL_PROGRAM@
+ 
+ TEMPLATEDIR = ../../../src/common
+ TEMPLATESRC = $(srcdir)/$(TEMPLATEDIR)
diff --git a/www/harvest/files/patch-components::gatherer::catdoc::charsets::Makefile.in b/www/harvest/files/patch-components::gatherer::catdoc::charsets::Makefile.in
new file mode 100644
index 000000000000..c3c855fb0f37
--- /dev/null
+++ b/www/harvest/files/patch-components::gatherer::catdoc::charsets::Makefile.in
@@ -0,0 +1,29 @@
+--- components/gatherer/catdoc/charsets/Makefile.in.orig	Sat Jun  8 17:12:36 2002
++++ components/gatherer/catdoc/charsets/Makefile.in	Wed Nov 13 10:03:59 2002
+@@ -1,5 +1,7 @@
+ SHELL = /bin/sh
+ INSTALL = @INSTALL@
++INSTALL_BIN = @INSTALL_PROGRAM@
++INSTALL_LIB = @INSTALL_DATA@
+ 
+ srcdir = @srcdir@
+ top_srcdir = @top_srcdir@
+@@ -29,13 +31,13 @@
+ clean:
+ install: install-dirs install-spc install-charsets
+ install-spc: tex.spc ascii.spc tex.rpl ascii.rpl
+-	    $(INSTALL) -m 644 $(srcdir)/ascii.spc $(installroot)$(LIB_DIR)/ascii$(SPEC_EXT)
+-	    $(INSTALL) -m 644 $(srcdir)/tex.spc $(installroot)$(LIB_DIR)/tex$(SPEC_EXT)
+-	    $(INSTALL) -m 644 $(srcdir)/ascii.rpl $(installroot)$(LIB_DIR)/ascii$(REPL_EXT)
+-	    $(INSTALL) -m 644 $(srcdir)/tex.rpl $(installroot)$(LIB_DIR)/tex$(REPL_EXT)
++	    $(INSTALL_LIB) $(srcdir)/ascii.spc $(installroot)$(LIB_DIR)/ascii$(SPEC_EXT)
++	    $(INSTALL_LIB) $(srcdir)/tex.spc $(installroot)$(LIB_DIR)/tex$(SPEC_EXT)
++	    $(INSTALL_LIB) $(srcdir)/ascii.rpl $(installroot)$(LIB_DIR)/ascii$(REPL_EXT)
++	    $(INSTALL_LIB) $(srcdir)/tex.rpl $(installroot)$(LIB_DIR)/tex$(REPL_EXT)
+ install-charsets:
+ 	for i in $(srcdir)/*.txt; do\
+-	   $(INSTALL) -m 0644 $$i $(installroot)$(LIB_DIR);\
++	   $(INSTALL_LIB) -m 0644 $$i $(installroot)$(LIB_DIR);\
+ 	done
+ install-dirs:
+ 	$(top_srcdir)/mkinstalldirs $(installroot)$(LIB_DIR)
diff --git a/www/harvest/files/patch-components::gatherer::catdoc::doc::Makefile.in b/www/harvest/files/patch-components::gatherer::catdoc::doc::Makefile.in
new file mode 100644
index 000000000000..086e2e9bed99
--- /dev/null
+++ b/www/harvest/files/patch-components::gatherer::catdoc::doc::Makefile.in
@@ -0,0 +1,22 @@
+--- components/gatherer/catdoc/doc/Makefile.in.orig	Thu Jan 31 05:00:58 2002
++++ components/gatherer/catdoc/doc/Makefile.in	Wed Nov 13 10:00:45 2002
+@@ -5,6 +5,8 @@
+ srcdir = @srcdir@
+ top_srcdir = @top_srcdir@
+ INSTALL = @INSTALL@
++INSTALL_BIN = @INSTALL_PROGRAM@
++INSTALL_LIB = @INSTALL_DATA@
+ VPATH = @srcdir@
+ prefix = @prefix@
+ exec_prefix = @exec_prefix@
+@@ -20,8 +22,8 @@
+ clean:
+ install: catdoc.1 xls2csv.1
+ 	../mkinstalldirs $(installroot)$(mandir)
+-	$(INSTALL) -m 644 catdoc.1 $(installroot)$(mandir)/catdoc$(manext)
+-	$(INSTALL) -m 644 xls2csv.1 $(installroot)$(mandir)/xls2csv$(manext)
++	$(INSTALL_LIB) catdoc.1 $(installroot)$(mandir)/catdoc$(manext)
++	$(INSTALL_LIB) xls2csv.1 $(installroot)$(mandir)/xls2csv$(manext)
+ # Following rules are primarily for developers use. I doubt that you
+ # need to rebuild these versions of documentation
+ distclean: catdoc.1 catdoc.txt catdoc.ps xls2csv.1 xls2csv.txt xls2csv.ps
diff --git a/www/harvest/files/patch-components::gatherer::catdoc::src::Makefile.in b/www/harvest/files/patch-components::gatherer::catdoc::src::Makefile.in
new file mode 100644
index 000000000000..09a7d7e8c3d5
--- /dev/null
+++ b/www/harvest/files/patch-components::gatherer::catdoc::src::Makefile.in
@@ -0,0 +1,36 @@
+--- components/gatherer/catdoc/src/Makefile.in.orig	Sat Jun  8 17:13:17 2002
++++ components/gatherer/catdoc/src/Makefile.in	Wed Nov 13 10:02:19 2002
+@@ -12,13 +12,15 @@
+ libdir = @prefix@/lib/gatherer/catdoc-lib
+ mandir = @mandir@
+ INSTALL = @INSTALL@
++INSTALL_BIN = @INSTALL_PROGRAM@
++INSTALL_LIB = @INSTALL_DATA@
+ WISH = @WISH@
+ CC = @CC@
+ build_targets = @buildtargets@
+ install_targets = @installtargets@
+ # Flags to pass to your C compilier
+ # remove -O2 on HP/UX 9. It is known to cause problems
+-FLAGS=-Wall -O2 -g
++CFLAGS = @CFLAGS@
+ 
+ #Compile-time configuration of catdoc itself. Edit according to taste
+ 
+@@ -77,12 +79,12 @@
+ install: @installtargets@
+ install-catdoc:catdoc xls2csv
+ 	$(top_srcdir)/mkinstalldirs $(installroot)$(bindir)
+-	$(INSTALL) -m 755 catdoc $(installroot)$(bindir)/catdoc
+-	$(INSTALL) -m 755 xls2csv $(installroot)$(bindir)/xls2csv
+-#	$(INSTALL) -m 755 msxlsview.sh $(installroot)$(bindir)/xlsview
++	$(INSTALL_BIN) catdoc $(installroot)$(bindir)/catdoc
++	$(INSTALL_BIN) xls2csv $(installroot)$(bindir)/xls2csv
++#	$(INSTALL_BIN) msxlsview.sh $(installroot)$(bindir)/xlsview
+ install-wordview: wordview
+ 	$(top_srcdir)/mkinstalldirs $(installroot)$(bindir)
+-	$(INSTALL) -m 755 wordview $(installroot)$(bindir)/wordview
++	$(INSTALL_BIN) wordview $(installroot)$(bindir)/wordview
+ wordview: wordview.tcl
+ 	echo "#! $(WISH)" >wordview
+ 	echo set charset_lib "\"$(LIB_DIR)\"">>wordview
diff --git a/www/harvest/files/patch-ab b/www/harvest/files/patch-src::common::util::strerror.c
index 1e39ca0f6c6b..947fc4433193 100644
--- a/www/harvest/files/patch-ab
+++ b/www/harvest/files/patch-src::common::util::strerror.c
@@ -1,5 +1,5 @@
---- ./src/common/util/strerror.c.orig	Sun Oct 20 17:43:50 2002
-+++ ./src/common/util/strerror.c	Sun Oct 20 17:43:59 2002
+--- src/common/util/strerror.c.orig	Mon Mar 17 07:16:32 1997
++++ src/common/util/strerror.c	Wed Nov 13 08:58:11 2002
 @@ -89,8 +89,10 @@
  char *strerror(n)
       int n;
diff --git a/www/harvest/files/patch-src::gatherer::essence::file::print.c b/www/harvest/files/patch-src::gatherer::essence::file::print.c
new file mode 100644
index 000000000000..bf01e67fa518
--- /dev/null
+++ b/www/harvest/files/patch-src::gatherer::essence::file::print.c
@@ -0,0 +1,12 @@
+--- src/gatherer/essence/file/print.c.orig	Sat Dec 13 01:02:15 1997
++++ src/gatherer/essence/file/print.c	Wed Nov 13 08:56:34 2002
+@@ -78,8 +78,8 @@
+ warning(f, a)
+      char *f, *a;
+ {
+-	extern int errno, sys_nerr;
+ #ifdef NEED_SYS_ERRLIST
++	extern int errno, sys_nerr;
+ 	extern char *sys_errlist[];
+ #endif
+ 	int myerrno;
diff --git a/www/harvest/files/patch-src::gatherer::essence::pp_rules.y b/www/harvest/files/patch-src::gatherer::essence::pp_rules.y
new file mode 100644
index 000000000000..9bef0cf1cd4b
--- /dev/null
+++ b/www/harvest/files/patch-src::gatherer::essence::pp_rules.y
@@ -0,0 +1,10 @@
+--- src/gatherer/essence/pp_rules.y.orig	Mon Mar 17 07:18:10 1997
++++ src/gatherer/essence/pp_rules.y	Wed Nov 13 09:06:33 2002
+@@ -65,6 +65,7 @@
+ 		;
+ 
+ condition	: condattr condop condstr	{ $$ = mk_cond ($1, $2, $3); }
++		;
+ 
+ condattr	: word				{ $$ = $1; }
+ 		;
diff --git a/www/harvest/pkg-descr b/www/harvest/pkg-descr
index 27e9159c01a2..3f36c43b944a 100644
--- a/www/harvest/pkg-descr
+++ b/www/harvest/pkg-descr
@@ -1,19 +1,8 @@
-This is Harvest Indexer version 1.5, an updated version of the Indexing Code
-originally developed by the Harvest project (http://harvest.transarc.com/).
-
-As they are no longer developing the code this version contains many 
-improvements and bug fixes over 1.4pl2, the last official release.
-
-If you are interested in contributing patches or ideas, please mail
-harvest@tardis.ed.ac.uk.
-
-Information on new releases will be sent to the harvest-announce mailing
-list. To subscribe send mail to harvest-announce-request@tardis.ed.ac.uk
-containing the message "subscribe"
-
-Bug reports can be mailed to harvest-bugs@tardis.ed.ac.uk
-
-Documentation: Information on this version of the Harvest software
-is available from http://www.tardis.ed.ac.uk/harvest/
-
-WWW: http://www.tardis.ed.ac.uk/harvest/
+Harvest is a system to collect information and make it searchable
+using a Web interface. It can collect information using HTTP, FTP,
+NNTP, and local files. Supported formats include HTML, DVI, PS,
+fulltext, mail, man pages, news, troff, WordPerfect, C sources, and
+many more. Adding support for new formats is easy due to Harvest's
+modular design.
+
+WWW: http://harvest.sourceforge.net/
diff --git a/www/harvest/pkg-plist b/www/harvest/pkg-plist
index 0006cbc740d4..c207dae8be21 100644
--- a/www/harvest/pkg-plist
+++ b/www/harvest/pkg-plist
@@ -4,31 +4,45 @@ harvest/bin/Gatherer
 harvest/bin/Harvest
 harvest/bin/broker
 harvest/bin/gather
-harvest/bin/gzip
-harvest/bin/hostname
-harvest/bin/perl
-harvest/bin/whoami
 harvest/brokers/Attributes.html
 harvest/brokers/adminhelp.html
+harvest/brokers/images/harvest-ic1.gif
+harvest/brokers/images/harvest-ic1.png
+harvest/brokers/images/harvest-ic2.gif
+harvest/brokers/images/harvest-ic2.png
+harvest/brokers/images/harvest-lg.gif
+harvest/brokers/images/harvest-lg.png
+harvest/brokers/images/harvest-spike.gif
+harvest/brokers/images/harvest-spike.png
+harvest/brokers/images/harvest-weight.gif
+harvest/brokers/images/harvest-weight.png
+harvest/brokers/images/index.html
 harvest/brokers/queryhelp.html
 harvest/brokers/skeleton/Makefile.in
 harvest/brokers/skeleton/Query.cgi.in
 harvest/brokers/skeleton/RunBroker.in
-harvest/brokers/skeleton/RunUpdate.in
 harvest/brokers/skeleton/admin/Collection.conf.in
 harvest/brokers/skeleton/admin/admin.html.in
 harvest/brokers/skeleton/admin/broker.conf.in
+harvest/brokers/skeleton/query-glimpse-modern.html.de.in
+harvest/brokers/skeleton/query-glimpse-modern.html.es.in
+harvest/brokers/skeleton/query-glimpse-modern.html.in
 harvest/brokers/skeleton/query-glimpse.html.in
+harvest/brokers/skeleton/query-swish-modern.html.in
+harvest/brokers/skeleton/query-swish.html.in
+harvest/brokers/skeleton/query-wais-modern.html.in
 harvest/brokers/skeleton/query-wais.html.in
 harvest/brokers/soifhelp.html
-harvest/cgi-bin/BrokerAdmin.cgi
-harvest/cgi-bin/BrokerQuery.pl.cgi
-harvest/cgi-bin/DisplayObject.cgi
-harvest/cgi-bin/HarvestGather.cgi
-harvest/cgi-bin/cgiparse.pl
-harvest/cgi-bin/lib/BrokerQuery.cf
+harvest/cgi-bin/BrokerAdmin.pl.cgi
+harvest/cgi-bin/displaySOIF.cgi
 harvest/cgi-bin/lib/Sample.cf
+harvest/cgi-bin/lib/classic.cf
+harvest/cgi-bin/lib/german.cf
+harvest/cgi-bin/lib/modern.cf
+harvest/cgi-bin/lib/search.cf
+harvest/cgi-bin/lib/spanish.cf
 harvest/cgi-bin/nph-search.cgi
+harvest/cgi-bin/search.cgi
 harvest/gatherers/example-1/RunGatherer
 harvest/gatherers/example-1/example-1.cf
 harvest/gatherers/example-2/RunGatherer
@@ -61,42 +75,48 @@ harvest/gatherers/example-5/example-5.cf
 harvest/gatherers/example-5/mydomain-host-filter
 harvest/gatherers/example-5/mydomain-url-filter
 harvest/lib/broker/BrokerRestart
-harvest/lib/broker/RunBrokers.sh
 harvest/lib/broker/agrep
 harvest/lib/broker/brkclient
 harvest/lib/broker/dumpregistry
 harvest/lib/broker/glimpse
 harvest/lib/broker/glimpseindex
+harvest/lib/broker/glimpseindex.bin
 harvest/lib/broker/glimpseserver
 harvest/lib/broker/info-to-html.pl
 harvest/lib/broker/mkbrokerstats.pl
-harvest/lib/ctime.pl
+harvest/lib/broker/swish
+harvest/lib/chat2.pl
 harvest/lib/dateconv.pl
+harvest/lib/ftp.pl
 harvest/lib/ftpget
 harvest/lib/gatherer/Archive.sum
 harvest/lib/gatherer/Binary-raw.sum
 harvest/lib/gatherer/Binary.sum
-harvest/lib/gatherer/C-raw.sum
 harvest/lib/gatherer/C.sum
 harvest/lib/gatherer/CHeader.sum
 harvest/lib/gatherer/Configuration.sum
 harvest/lib/gatherer/Dvi.sum
 harvest/lib/gatherer/FAQ.sum
 harvest/lib/gatherer/FullText.sum
+harvest/lib/gatherer/GifImage.sum
 harvest/lib/gatherer/HTML-lax.sum
+harvest/lib/gatherer/HTML-sum.pl
 harvest/lib/gatherer/HTML.sum
 harvest/lib/gatherer/HTMLurls
+harvest/lib/gatherer/HTTP-Query.sum
 harvest/lib/gatherer/HomeHTML.sum
 harvest/lib/gatherer/IMS-DOCUMENT.sum
+harvest/lib/gatherer/MSWord.sum
 harvest/lib/gatherer/Mail.sum
 harvest/lib/gatherer/ManPage.sum
 harvest/lib/gatherer/NewsArticle.sum
 harvest/lib/gatherer/NewsGroup.sum
 harvest/lib/gatherer/Object.sum
+harvest/lib/gatherer/Pdf.sum
 harvest/lib/gatherer/PostScript.sum
 harvest/lib/gatherer/RCS.sum
 harvest/lib/gatherer/RFC.sum
-harvest/lib/gatherer/RunGatherers.sh
+harvest/lib/gatherer/RTF.sum
 harvest/lib/gatherer/SCCS.sum
 harvest/lib/gatherer/SGML.sum
 harvest/lib/gatherer/SourceDistribution-raw.sum
@@ -110,25 +130,64 @@ harvest/lib/gatherer/Troff-me.sum
 harvest/lib/gatherer/Troff-ms.sum
 harvest/lib/gatherer/Troff.sum
 harvest/lib/gatherer/URL-filter-default
+harvest/lib/gatherer/WordPerfect.sum
 harvest/lib/gatherer/bycontent.cf
 harvest/lib/gatherer/byname.cf
 harvest/lib/gatherer/byurl.cf
+harvest/lib/gatherer/catdoc
+harvest/lib/gatherer/catdoc-lib/8859-1.txt
+harvest/lib/gatherer/catdoc-lib/8859-2.txt
+harvest/lib/gatherer/catdoc-lib/8859-3.txt
+harvest/lib/gatherer/catdoc-lib/8859-4.txt
+harvest/lib/gatherer/catdoc-lib/8859-5.txt
+harvest/lib/gatherer/catdoc-lib/8859-6.txt
+harvest/lib/gatherer/catdoc-lib/8859-7.txt
+harvest/lib/gatherer/catdoc-lib/8859-8.txt
+harvest/lib/gatherer/catdoc-lib/8859-9.txt
+harvest/lib/gatherer/catdoc-lib/ascii.replchars
+harvest/lib/gatherer/catdoc-lib/ascii.specchars
+harvest/lib/gatherer/catdoc-lib/cp2250.txt
+harvest/lib/gatherer/catdoc-lib/cp2251.txt
+harvest/lib/gatherer/catdoc-lib/cp2252.txt
+harvest/lib/gatherer/catdoc-lib/cp2253.txt
+harvest/lib/gatherer/catdoc-lib/cp2254.txt
+harvest/lib/gatherer/catdoc-lib/cp2255.txt
+harvest/lib/gatherer/catdoc-lib/cp2256.txt
+harvest/lib/gatherer/catdoc-lib/cp2257.txt
+harvest/lib/gatherer/catdoc-lib/cp2258.txt
+harvest/lib/gatherer/catdoc-lib/cp437.txt
+harvest/lib/gatherer/catdoc-lib/cp850.txt
+harvest/lib/gatherer/catdoc-lib/cp852.txt
+harvest/lib/gatherer/catdoc-lib/cp855.txt
+harvest/lib/gatherer/catdoc-lib/cp857.txt
+harvest/lib/gatherer/catdoc-lib/cp860.txt
+harvest/lib/gatherer/catdoc-lib/cp861.txt
+harvest/lib/gatherer/catdoc-lib/cp862.txt
+harvest/lib/gatherer/catdoc-lib/cp863.txt
+harvest/lib/gatherer/catdoc-lib/cp864.txt
+harvest/lib/gatherer/catdoc-lib/cp865.txt
+harvest/lib/gatherer/catdoc-lib/cp866.txt
+harvest/lib/gatherer/catdoc-lib/cp869.txt
+harvest/lib/gatherer/catdoc-lib/cp874.txt
+harvest/lib/gatherer/catdoc-lib/koi8-r.txt
+harvest/lib/gatherer/catdoc-lib/koi8-u.txt
+harvest/lib/gatherer/catdoc-lib/tex.replchars
+harvest/lib/gatherer/catdoc-lib/tex.specchars
+harvest/lib/gatherer/catdoc-lib/us-ascii.txt
+harvest/lib/gatherer/catdoc-lib/x-mac-cyrillic.txt
 harvest/lib/gatherer/cksoif
 harvest/lib/gatherer/cleandb
 harvest/lib/gatherer/consoldb
-harvest/lib/gatherer/deroff
 harvest/lib/gatherer/dvi2tty
 harvest/lib/gatherer/enum
 harvest/lib/gatherer/essence
 harvest/lib/gatherer/expiredb
-harvest/lib/gatherer/extract-perl-procs
 harvest/lib/gatherer/fileenum
 harvest/lib/gatherer/folddb
 harvest/lib/gatherer/ftpenum
 harvest/lib/gatherer/ftpenum.pl
 harvest/lib/gatherer/gatherd
 harvest/lib/gatherer/gdbmutil
-harvest/lib/gatherer/get-include-files
 harvest/lib/gatherer/gifinfo
 harvest/lib/gatherer/gopherenum-breadth
 harvest/lib/gatherer/gopherenum-depth
@@ -137,10 +196,9 @@ harvest/lib/gatherer/httpenum-breadth
 harvest/lib/gatherer/httpenum-depth
 harvest/lib/gatherer/in.gatherd
 harvest/lib/gatherer/kill-gatherd
-harvest/lib/gatherer/l2h.pl
+harvest/lib/gatherer/l2h
 harvest/lib/gatherer/magic
 harvest/lib/gatherer/mergedb
-harvest/lib/gatherer/mkcompressed
 harvest/lib/gatherer/mkgathererstats.pl
 harvest/lib/gatherer/mkindex
 harvest/lib/gatherer/mktemplate
@@ -155,6 +213,7 @@ harvest/lib/gatherer/quick-sum
 harvest/lib/gatherer/quick-sum.cf
 harvest/lib/gatherer/rast
 harvest/lib/gatherer/rmbinary
+harvest/lib/gatherer/rtf2html
 harvest/lib/gatherer/sgmls
 harvest/lib/gatherer/sgmls-lib/HTML/HTML.decl
 harvest/lib/gatherer/sgmls-lib/HTML/HTML.sum.LaTeX.tbl
@@ -168,58 +227,107 @@ harvest/lib/gatherer/sgmls-lib/HTML/html-1.dtd
 harvest/lib/gatherer/sgmls-lib/HTML/html-1s.dtd
 harvest/lib/gatherer/sgmls-lib/HTML/html-3.dtd
 harvest/lib/gatherer/sgmls-lib/HTML/html-32.dtd
+harvest/lib/gatherer/sgmls-lib/HTML/html-4fs.dtd
+harvest/lib/gatherer/sgmls-lib/HTML/html-4strict.dtd
+harvest/lib/gatherer/sgmls-lib/HTML/html-4trans.dtd
 harvest/lib/gatherer/sgmls-lib/HTML/html-hj.dtd
 harvest/lib/gatherer/sgmls-lib/HTML/html-mcom.dtd
 harvest/lib/gatherer/sgmls-lib/HTML/html-s.dtd
 harvest/lib/gatherer/sgmls-lib/HTML/html.dtd
-harvest/lib/gatherer/sgmls-lib/HTMLPLUS/HTMLplus.dtd
 harvest/lib/gatherer/sgmls-lib/IMS/IMS.decl
 harvest/lib/gatherer/sgmls-lib/IMS/IMS.dtd
 harvest/lib/gatherer/sgmls-lib/IMS/IMS.sum.tbl
 harvest/lib/gatherer/sgmls-lib/ISOlat1-no-convert.sgml
 harvest/lib/gatherer/sgmls-lib/ISOlat1.sgml
+harvest/lib/gatherer/sgmls-lib/README
 harvest/lib/gatherer/sgmls-lib/catalog
+harvest/lib/gatherer/sgmls-lib/html-3.2/HTML32.cat
+harvest/lib/gatherer/sgmls-lib/html-3.2/HTML32.dtd
+harvest/lib/gatherer/sgmls-lib/html-3.2/ISOlat1.ent
+harvest/lib/gatherer/sgmls-lib/html-4.0/HTML4.cat
+harvest/lib/gatherer/sgmls-lib/html-4.0/HTML4.decl
+harvest/lib/gatherer/sgmls-lib/html-4.0/HTMLlat1.ent
+harvest/lib/gatherer/sgmls-lib/html-4.0/HTMLspecial.ent
+harvest/lib/gatherer/sgmls-lib/html-4.0/HTMLsymbol.ent
+harvest/lib/gatherer/sgmls-lib/html-4.0/frameset.dtd
+harvest/lib/gatherer/sgmls-lib/html-4.0/loose.dtd
+harvest/lib/gatherer/sgmls-lib/html-4.0/strict.dtd
+harvest/lib/gatherer/sgmls-lib/html-4.01/HTML4.cat
+harvest/lib/gatherer/sgmls-lib/html-4.01/HTML4.decl
+harvest/lib/gatherer/sgmls-lib/html-4.01/HTMLlat1.ent
+harvest/lib/gatherer/sgmls-lib/html-4.01/HTMLspecial.ent
+harvest/lib/gatherer/sgmls-lib/html-4.01/HTMLsymbol.ent
+harvest/lib/gatherer/sgmls-lib/html-4.01/frameset.dtd
+harvest/lib/gatherer/sgmls-lib/html-4.01/loose.dtd
+harvest/lib/gatherer/sgmls-lib/html-4.01/strict.dtd
+harvest/lib/gatherer/sgmls-lib/html/ISOlat1.sgml
+harvest/lib/gatherer/sgmls-lib/html/catalog
+harvest/lib/gatherer/sgmls-lib/html/html-0.dtd
+harvest/lib/gatherer/sgmls-lib/html/html-1.dtd
+harvest/lib/gatherer/sgmls-lib/html/html-1s.dtd
+harvest/lib/gatherer/sgmls-lib/html/html-s.dtd
+harvest/lib/gatherer/sgmls-lib/html/html.decl
+harvest/lib/gatherer/sgmls-lib/html/html.dtd
+harvest/lib/gatherer/sgmls-lib/htmlplus/HTMLplus.dtd
 harvest/lib/gatherer/sgmlsasp
 harvest/lib/gatherer/skim
 harvest/lib/gatherer/staturl
 harvest/lib/gatherer/stoplist.cf
 harvest/lib/gatherer/template2db
 harvest/lib/gatherer/unshar
+harvest/lib/gatherer/uudecode
+harvest/lib/gatherer/wp2x
+harvest/lib/gatherer/wp2x-lib/accent.cfg
+harvest/lib/gatherer/wp2x-lib/ascii.cfg
+harvest/lib/gatherer/wp2x-lib/gml.cfg
+harvest/lib/gatherer/wp2x-lib/html.cfg
+harvest/lib/gatherer/wp2x-lib/latex.cfg
+harvest/lib/gatherer/wp2x-lib/sample.cfg
+harvest/lib/gatherer/wp2x-lib/script.cfg
+harvest/lib/gatherer/wp2x-lib/tex.cfg
+harvest/lib/gatherer/wp2x-lib/troff.cfg
+harvest/lib/gatherer/wp2x.sh
 harvest/lib/gatherer/wrapit
+harvest/lib/gatherer/xls2csv
 harvest/lib/gopherget.pl
 harvest/lib/harvest-check.pl
 harvest/lib/lsparse.pl
+harvest/lib/md5
 harvest/lib/newsget.pl
 harvest/lib/socket.ph
 harvest/lib/soif-mem-efficient.pl
 harvest/lib/soif.pl
-harvest/lib/test-socket-ph.pl
 harvest/lib/timelocal.pl
 harvest/lib/urlget
 harvest/lib/urlpurge
-@dirrm harvest/bin
+@dirrm harvest/lib/gatherer/wp2x-lib
+@dirrm harvest/lib/gatherer/sgmls-lib/htmlplus
+@dirrm harvest/lib/gatherer/sgmls-lib/html-4.01
+@dirrm harvest/lib/gatherer/sgmls-lib/html-4.0
+@dirrm harvest/lib/gatherer/sgmls-lib/html-3.2
+@dirrm harvest/lib/gatherer/sgmls-lib/html
 @dirrm harvest/lib/gatherer/sgmls-lib/IMS
 @dirrm harvest/lib/gatherer/sgmls-lib/HTML
-@dirrm harvest/lib/gatherer/sgmls-lib/HTMLPLUS
 @dirrm harvest/lib/gatherer/sgmls-lib
+@dirrm harvest/lib/gatherer/catdoc-lib
 @dirrm harvest/lib/gatherer
 @dirrm harvest/lib/broker
 @dirrm harvest/lib
-@dirrm harvest/cgi-bin/lib
-@dirrm harvest/cgi-bin
-@dirrm harvest/man/man1
-@dirrm harvest/man
-@dirrm harvest/gatherers/example-1
-@dirrm harvest/gatherers/example-2/lib
-@dirrm harvest/gatherers/example-2
-@dirrm harvest/gatherers/example-3/lib
-@dirrm harvest/gatherers/example-3
-@dirrm harvest/gatherers/example-4/bin
+@dirrm harvest/gatherers/example-5
 @dirrm harvest/gatherers/example-4/lib
+@dirrm harvest/gatherers/example-4/bin
 @dirrm harvest/gatherers/example-4
-@dirrm harvest/gatherers/example-5
+@dirrm harvest/gatherers/example-3/lib
+@dirrm harvest/gatherers/example-3
+@dirrm harvest/gatherers/example-2/lib
+@dirrm harvest/gatherers/example-2
+@dirrm harvest/gatherers/example-1
 @dirrm harvest/gatherers
+@dirrm harvest/cgi-bin/lib
+@dirrm harvest/cgi-bin
 @dirrm harvest/brokers/skeleton/admin
 @dirrm harvest/brokers/skeleton
+@dirrm harvest/brokers/images
 @dirrm harvest/brokers
+@dirrm harvest/bin
 @dirrm harvest