Add site exclusions patches. Some cleanup.

12 files changed, 385 insertions, 27 deletions

diff --git a/news/slurp/Makefile b/news/slurp/Makefile
index a50b074..43ba06b 100644
--- a/news/slurp/Makefile
+++ b/news/slurp/Makefile
@@ -3,7 +3,7 @@
 # Date created:		30 Jun 1996
 # Whom:			jraynard
 #
-# $Id: Makefile,v 1.1.1.1 1996/07/27 21:54:11 asami Exp $
+# $Id: Makefile,v 1.2 1996/11/18 11:38:10 asami Exp $
 #
 
 DISTNAME=	slurp-1.10
@@ -13,21 +13,15 @@ EXTRACT_SUFX=	.tar.Z
 
 MAINTAINER=	jraynard@FreeBSD.ORG
 
-#RUN_DEPENDS=	${PREFIX}/etc/innd:${PORTSDIR}/news/inn
-
 NO_WRKSUBDIR=	yes
 MAN1=		slurp.1
 
-pre-build:
-	@(cd ${WRKSRC}; cp conf.h conf.h.orig; \
-	  sed -e 's!/usr/lib/news!${PREFIX}/news/lib!' -e 's!/var/spool!/var!' conf.h.orig > conf.h; \
-	rm -f syslog.h)
-	@cp ${FILESDIR}/dbz.* ${WRKSRC}
+pre-install:
+	${MKDIR} ${PREFIX}/news/lib ${PREFIX}/news/bin
 
 do-install:
-	${MKDIR} ${PREFIX}/news/bin ${PREFIX}/news/lib
-	install -c -o news -g news -m 0555 -s ${WRKSRC}/slurp ${PREFIX}/news/bin/
+	${INSTALL_PROGRAM} ${WRKSRC}/slurp ${PREFIX}/news/bin
 	${INSTALL_MAN} ${WRKSRC}/slurp.1 ${PREFIX}/man/man1
-	install -c -o news -g news -m 0644 ${WRKSRC}/slurp.sys ${PREFIX}/news/lib/	
+	${INSTALL_DATA} ${WRKSRC}/slurp.sys ${PREFIX}/news/lib
 
 .include <bsd.port.mk>
diff --git a/news/slurp/distinfo b/news/slurp/distinfo
index a483612..85cc27d 100644
--- a/news/slurp/distinfo
+++ b/news/slurp/distinfo
@@ -1 +1,2 @@
 MD5 (slurp-1.10.tar.Z) = 0983ba2ecebf8ee83e36807aabf1a0b2
+MD5 (slurp.siteexclude.patch.gz) = c87c3c351c625374f6d57b3166209bd7
diff --git a/news/slurp/files/patch-aa b/news/slurp/files/patch-aa
index 24d8f76..54089e7 100644
--- a/news/slurp/files/patch-aa
+++ b/news/slurp/files/patch-aa
@@ -1,5 +1,5 @@
 --- Makefile.orig	Tue Feb  7 12:37:45 1995
-+++ Makefile	Sat Jun 29 23:13:53 1996
++++ Makefile	Sun Nov  9 18:14:24 1997
 @@ -15,10 +15,10 @@
  #   appropriate header file and library for the news database format that
  #   you have chosen.
@@ -14,16 +14,20 @@
  #DBLIB = -ldbz
  
  #DBINCLUDE = -I/usr/local/include
-@@ -28,7 +28,7 @@
+@@ -28,7 +28,11 @@
  # C compilation flags
  #   Add here any additional compilation flags you need.
  
 -CFLAGS = -O
-+#CFLAGS = -O
++# To enable the site exclusion features, define SITEEXCLUDING.
++#
++# If KEEPMISSINGARTS is defined, slurp will attempt on the next run
++# to fetch articles which were not available on the current download.
++CFLAGS+=-DSITEEXCLUDING -DKEEPMISSINGARTS
  
  
  # Loader flags
-@@ -41,8 +41,8 @@
+@@ -41,8 +45,8 @@
  #   On some systems (e.g. SVR4) you need to include additional socket
  #   libraries. On others (e.g. Linux, BSD) this is not required.
  
@@ -34,16 +38,16 @@
  
  
  # If you don't have the syslog facility then uncomment these two lines
-@@ -54,15 +54,16 @@
+@@ -54,15 +58,16 @@
  # Everything else probably doesn't need changing
  
  SOURCE = slurp.c newnews.c articles.c history.c time.c sockets.c misc.c \
 -	 cache.c hostfiles.c space.c $(FAKESRC)
-+	 dbz.c cache.c hostfiles.c space.c $(FAKESRC)
++	 dbz.c cache.c hostfiles.c siteexclude.c space.c $(FAKESRC)
  
  OBJECT = slurp.o newnews.o articles.o history.o time.o sockets.o misc.o \
 -	 cache.o hostfiles.o space.o $(FAKEOBJ)
-+	 dbz.o cache.o hostfiles.o space.o $(FAKEOBJ)
++	 dbz.o cache.o hostfiles.o siteexclude.o space.o $(FAKEOBJ)
  
  MANIFEST = README slurp.c newnews.c articles.c history.c time.c sockets.c \
  	   misc.c cache.c hostfiles.c space.c fakesyslog.c conf.h slurp.h \
@@ -53,7 +57,7 @@
  
  slurp: $(OBJECT)
  	$(CC) $(LDFLAGS) $(OBJECT) -o slurp $(DBLIB) $(LIBS)
-@@ -71,17 +72,17 @@
+@@ -71,17 +76,17 @@
  	$(CC) $(CFLAGS) $(DBINCLUDE) -c $<
  
  
diff --git a/news/slurp/files/patch-ab b/news/slurp/files/patch-ab
index 8faeeb2..492c056 100644
--- a/news/slurp/files/patch-ab
+++ b/news/slurp/files/patch-ab
@@ -1,11 +1,24 @@
---- conf.h.orig	Sat Jun 29 23:29:13 1996
-+++ conf.h	Sat Jun 29 23:29:25 1996
-@@ -155,7 +155,7 @@
-    undefined. If the required define is not supplied by the compiler
-    (e.g. SVR4, SVR3) then you will need to include it yourself here. */
+--- articles.c.orig	Sun Nov  9 18:14:49 1997
++++ articles.c	Sun Nov  9 18:16:40 1997
+@@ -348,6 +348,10 @@
+ 		case OK_ARTICLE:
+ 			fetch_article ();
+ 			xfrart++;
++#ifdef KEEPMISSINGARTS
++			/* Note that this article has been retrieved and filed */
++			idnode -> used = TRUE;
++#endif /* KEEPMISSINGARTS */
+ 			break;
  
--#define SVR4
-+/* #define SVR4 */
+ 		/* Otherwise must be a protocol error */
+@@ -358,8 +362,10 @@
+ 			exit (4);
+ 		}
  
++#ifndef KEEPMISSINGARTS
+ 	/* Note that this article has been retrieved and filed */
+ 	idnode -> used = TRUE;
++#endif /* KEEPMISSINGARTS */
  
- /* The facility name which syslog reports errors and stats under. If
+ 	/* Print progress if required */
+ 	if (progress_flag)
diff --git a/news/slurp/files/patch-ac b/news/slurp/files/patch-ac
new file mode 100644
index 0000000..e5d5fbf
--- /dev/null
+++ b/news/slurp/files/patch-ac
@@ -0,0 +1,23 @@
+--- conf.h.orig	Tue Feb  7 14:29:19 1995
++++ conf.h	Sun Nov  9 18:13:56 1997
+@@ -57,6 +57,11 @@
+ #define SYSFILE			"/usr/lib/news/slurp.sys"
+ #define HOSTFILE		"/usr/lib/news/slurp."
+ 
++/* These files will be used by the site exclusion feature.  SITEFILE
++   is a list of sites that you do not wish to receive articles from.
++   The IDs of articles thus rejected will be kept in DROPFILE. */
++#define SITEFILE		"/usr/lib/news/excludesites"
++#define DROPFILE		"/usr/lib/news/exclarticles"
+ 
+ /* Speedup sets the number of extra ARTICLE requests that are stacked
+    up at the NNTP server awaiting transmission - aka pipelining.
+@@ -155,7 +160,7 @@
+    undefined. If the required define is not supplied by the compiler
+    (e.g. SVR4, SVR3) then you will need to include it yourself here. */
+ 
+-#define SVR4
++/* #define SVR4 */
+ 
+ 
+ /* The facility name which syslog reports errors and stats under. If
diff --git a/news/slurp/files/patch-ad b/news/slurp/files/patch-ad
new file mode 100644
index 0000000..ee108d1
--- /dev/null
+++ b/news/slurp/files/patch-ad
@@ -0,0 +1,38 @@
+--- newnews.c.orig	Sun Nov  9 18:17:14 1997
++++ newnews.c	Sun Nov  9 18:19:29 1997
+@@ -158,10 +158,16 @@
+ process_id (char *msgid)
+ 	{
+ 	char *cp;
++#ifdef SITEEXCLUDING
++	char *host;
++#endif /*SITEXCLUDING*/
+ 
+ 	/* Modify the message id appropriate to C-News history files */
+ 	if ((cp = strchr (msgid, '@')) != NULL)
+ 		{
++#ifdef SITEEXCLUDING
++		host = cp;
++#endif /*SITEXCLUDING*/
+ 		for (; *cp != '\0'; cp++)
+ 			if (isupper (*cp))
+ 				*cp = tolower (*cp);
+@@ -169,6 +175,18 @@
+ 
+ 	if (debug_flag)
+ 		(void) fprintf (stderr, "-> %s", msgid);
++
++#ifdef SITEEXCLUDING
++	if((host)&&(site_check(host)))
++	{
++		if (debug_flag)
++			(void) fprintf (stderr, " (Excluded host %s)", host);
++		/* Excluded host */
++		site_droparticle(msgid);
++		exclart++;
++		return;
++	}
++#endif /*SITEXCLUDING*/
+ 
+ 	/* If hit maximum cache size, then store in a temporary file */
+ #ifdef MAXCACHE
diff --git a/news/slurp/files/patch-ae b/news/slurp/files/patch-ae
new file mode 100644
index 0000000..9d24487
--- /dev/null
+++ b/news/slurp/files/patch-ae
@@ -0,0 +1,21 @@
+--- patchlevel.h.orig	Sun Nov  9 18:19:34 1997
++++ patchlevel.h	Sun Nov  9 18:20:34 1997
+@@ -9,6 +9,10 @@
+  * $Id: patchlevel.h,v 1.10 1995/02/07 14:33:25 root Exp root $
+  *
+  * $Log: patchlevel.h,v $
++ *
++ * 1.10.pf.b - Add siteexclusion stuff
++ * 1.10.pf.a - Save missing article ids, to try later
++ *
+  * Revision 1.10  1995/02/07  14:33:25  root
+  * Updated for new version
+  *
+@@ -18,6 +22,6 @@
+  *
+  */
+ 
+-#define PATCHLEVEL "1.10"
++#define PATCHLEVEL "1.10.pf.b"
+ 
+ /* END-OF-FILE */
diff --git a/news/slurp/files/patch-af b/news/slurp/files/patch-af
new file mode 100644
index 0000000..19c220e
--- /dev/null
+++ b/news/slurp/files/patch-af
@@ -0,0 +1,57 @@
+--- slurp.c.orig	Sun Nov  9 18:20:54 1997
++++ slurp.c	Sun Nov  9 18:25:16 1997
+@@ -151,6 +151,9 @@
+ int waiting = 0;
+ int dupart = 0;
+ int misart = 0;
++#ifdef SITEEXCLUDING
++int exclart = 0;
++#endif /* SITEEXCLUDING */
+ int xfrart = 0;
+ long totalsize = 0;
+ 
+@@ -627,6 +630,11 @@
+ 	if (open_history ())
+ 		log_sys ("can't open history file %s", HISTORY_FILE);
+ 
++#ifdef SITEEXCLUDING
++	/* Get site names to exclude */
++	site_init();
++#endif /* SITEEXCLUDING */
++
+ 	/* Load start time or ids from hostfile if appropriate */
+ 	read_hostfile ();
+ 
+@@ -697,6 +705,9 @@
+ 	/* Time to say goodbye */
+ 	close_server ();
+ 	close_history ();
++#ifdef SITEEXCLUDING
++	site_close();
++#endif /* SITEEXCLUDING */
+ 
+ 	/* Submit the remaining batch, if present */
+ 	enqueue_batch ();
+@@ -706,12 +717,22 @@
+ 
+ #ifdef SYSLOG
+ 	if (!debug_flag)
++#ifdef SITEEXCLUDING
++		syslog (LOG_INFO, "Processed %d new, %d duplicate, %d missing, %d excluded articles",
++				xfrart, dupart, misart, exclart);
++#else
+ 		syslog (LOG_INFO, "Processed %d new, %d duplicate, %d missing articles",
+ 				xfrart, dupart, misart);
++#endif /* SITEEXCLUDING */
+ 	else
+ #endif
++#ifdef SITEEXCLUDING
++		(void) fprintf (stderr, "Processed %d new, %d duplicate, %d missing, %d excluded articles\n",
++						xfrart, dupart, misart, exclart);
++#else
+ 		(void) fprintf (stderr, "Processed %d new, %d duplicate, %d missing articles\n",
+ 						xfrart, dupart, misart);
++#endif /* SITEEXCLUDING */
+ 
+ #ifdef SPEEDSTATS
+   #ifdef SYSLOG
diff --git a/news/slurp/files/patch-ag b/news/slurp/files/patch-ag
new file mode 100644
index 0000000..864ed21
--- /dev/null
+++ b/news/slurp/files/patch-ag
@@ -0,0 +1,12 @@
+--- slurp.h.orig	Sun Nov  9 18:25:22 1997
++++ slurp.h	Sun Nov  9 18:26:04 1997
+@@ -97,6 +97,9 @@
+ extern int dupart;			/* Number of duplicate articles */
+ extern int misart;			/* Number of missing articles */
+ extern int xfrart;			/* Number of transferred articles */
++#ifdef SITEEXCLUDING
++extern int exclart;			/* Number of excluded articles */
++#endif /* SITEEXCLUDING */
+ extern long totalsize;		/* Total size of articles tranferred */
+ 
+ /* Misc variables */
diff --git a/news/slurp/files/siteexclude.c b/news/slurp/files/siteexclude.c
new file mode 100644
index 0000000..67e4aaa
--- /dev/null
+++ b/news/slurp/files/siteexclude.c
@@ -0,0 +1,168 @@
+/* siteexclude.c - check article id for excluded sites
+ *
+ * Copyright (C) 1995 Peter Fox <fox@roestock.demon.co.uk>
+ *
+ * This code may be freely copied in it's entirety, but the
+ * above notice may not be removed or altered without
+ * reference to the author.
+ * Individuals may make changes for their own use, however
+ * any changed copies may not be distributed without
+ * reference to the author.
+ * There isn't any guarantee of usefulness or efficacy for
+ * any particular purpose. Take it as it is.
+ */
+#include <stdio.h>
+#include <string.h>
+#include "syslog.h"
+#include "conf.h"
+
+#ifdef TEST
+#define SITEEXCLUDING
+#endif /*TEST*/
+
+#ifdef SITEEXCLUDING
+typedef struct slist
+{
+	struct slist *next;
+	char *str;
+} SLIST, *SLISTPTR;
+
+#define HASHSIZE 256
+#define HASHFN(x) (((x[1] & 15) << 4)|(x[2] & 15)) /* x is (char *), skip '@' */
+
+#ifdef TEST
+#define SITEFILE "./sites"
+#define DROPFILE "./dropped"
+#define log_msg printf
+#define log_sys printf
+#define debug_flag 1
+#else /*!TEST*/
+extern int debug_flag;
+#ifndef SITEFILE
+#define SITEFILE "/var/lib/news/excludesites"
+#endif /*SITEFILE*/
+#ifndef DROPFILE
+#define DROPFILE "/var/lib/news/droppedarticles"
+#endif /*DROPFILE*/
+#endif /*TEST*/
+
+SLIST hash[HASHSIZE];
+
+static FILE *dropfp = (FILE*)0;
+
+site_init()
+{
+	int i;
+	FILE *fp;
+	char linbuf[BUFSIZ];
+	SLISTPTR ptr;
+	int entries, collisions;
+	entries = 0;
+	collisions = 0;
+	for(i = 0; i < HASHSIZE; i++)
+	{
+		hash[i].next = (SLISTPTR)0;
+		hash[i].str = (char *)0;
+	}
+	if(!(fp = fopen(SITEFILE, "r")))
+	{
+		log_msg("site_init: warning - no site exclude file %s", SITEFILE);
+		return;
+	}
+	if(!(dropfp = fopen(DROPFILE, "a")))
+	{
+		log_msg("site_init: warning - cannot open file %s for append, dropped articles logged instead", DROPFILE);
+	}
+	else
+		fprintf(dropfp, "-----\n");
+	/* File contains lines with complete site names to exclude */
+	/* We add the leading @ and trailing > */
+	linbuf[0] = '@';
+	while(1)
+	{
+        /* Read in a line */
+        (void) fgets(linbuf+1, sizeof (linbuf)-2, fp);
+        if(feof(fp))
+			break;
+        /* If a read error then report it and abort */
+        if (ferror(fp))
+		{
+            log_sys("site_init: error reading %s", SITEFILE);
+            break;
+		}
+		if(linbuf[1])
+		{
+			/* Kill the \n */
+			linbuf[strlen(linbuf)-1] = '>';
+			entries++;
+			ptr = &hash[HASHFN(linbuf)];
+			if(debug_flag)
+				fprintf(stderr, "'%s: %08lx'\n", linbuf, ptr);
+			if(!ptr->str)
+				ptr->str = strdup(linbuf);
+			else
+			{
+				/* Run along to the last in line */
+				while(ptr->next)
+					ptr = ptr->next;
+				ptr->next = (SLISTPTR)malloc(sizeof(SLIST));
+				ptr = ptr->next;
+				ptr->next = (SLISTPTR)0;
+				ptr->str = strdup(linbuf);
+				collisions++;
+			}
+		}
+	}
+	fclose(fp);
+	if(debug_flag)
+		log_msg("site_init: site exclude file %s read: %d entries, %d collisions",
+			SITEFILE, entries, collisions);
+}
+
+site_check(char *site)
+{
+	SLISTPTR ptr;
+	ptr = &hash[HASHFN(site)];
+	if(debug_flag)
+		fprintf(stderr, " '%s: %08lx'", site, ptr);
+	if(!ptr->str)
+		return(0);
+	while(ptr)
+	{
+		if(strcmp(site, ptr->str) == 0)
+			return(1);
+		ptr = ptr->next;
+	}
+	return(0);
+}
+
+site_droparticle(char *id)
+{
+	if(dropfp)
+		fprintf(dropfp, "%s\n", id);
+	else
+		log_msg("Excluded article: %s", id);
+}
+
+site_close()
+{
+	if(dropfp)
+		fclose(dropfp);
+}
+
+#ifdef TEST
+main(int argc, char **argv)
+{
+	site_init();
+	argv++;
+	argc--;
+	while(argc--)
+	{
+		if(site_check(*argv))
+			printf("Found %s\n", *argv);
+		argv++;
+	}
+	return(0);
+}
+#endif /*TEST*/
+#endif /* SITEEXCLUDING */
diff --git a/news/slurp/pkg-descr b/news/slurp/pkg-descr
index 0be45cf..99cea36 100644
--- a/news/slurp/pkg-descr
+++ b/news/slurp/pkg-descr
@@ -6,3 +6,10 @@ forwarding on via UUCP to another news system. It replaces nntpxfer
 from the NNTP 1.5.12 reference implementation and nntpget from the INN
 distribution.
 
+Patches from Peter Fox (fox@roestock.demon.co.uk) have been added to
+avoid downloading articles from certain sites.  To make use of this
+facility, create a file ${PREFIX}/news/lib/excludesites and list the
+sites in it, eg
+
+fritter.spam.com
+lusers.org
diff --git a/news/slurp/scripts/configure b/news/slurp/scripts/configure
new file mode 100644
index 0000000..894bc90
--- /dev/null
+++ b/news/slurp/scripts/configure
@@ -0,0 +1,20 @@
+#!/bin/sh
+#
+# $Id$
+#
+
+echo "s!/usr/lib/news!${PREFIX}/news/lib!" > ${TMPDIR}/tmp.slurp.sed.$$
+
+if [ -d /var/news ]
+then
+	echo 's!/var/spool!/var!' >> ${TMPDIR}/tmp.slurp.sed.$$	
+fi
+
+sed -f ${TMPDIR}/tmp.slurp.sed.$$ < ${WRKSRC}/conf.h > ${WRKSRC}/conf.h.tmp
+mv ${WRKSRC}/conf.h.tmp ${WRKSRC}/conf.h
+rm -f ${TMPDIR}/tmp.slurp.sed.$$
+
+rm -f ${WRKDIR}/syslog.h
+
+cp ${FILESDIR}/dbz.* ${WRKSRC}
+cp ${FILESDIR}/siteexclude.c ${WRKSRC}