===================================================================
RCS file: /cvs/mandoc/cgi.c,v
retrieving revision 1.70
retrieving revision 1.82
diff -u -p -r1.70 -r1.82
--- mandoc/cgi.c	2014/07/13 12:31:23	1.70
+++ mandoc/cgi.c	2014/07/24 20:30:45	1.82
@@ -1,4 +1,4 @@
-/*	$Id: cgi.c,v 1.70 2014/07/13 12:31:23 schwarze Exp $ */
+/*	$Id: cgi.c,v 1.82 2014/07/24 20:30:45 schwarze Exp $ */
 /*
  * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2014 Ingo Schwarze <schwarze@usta.de>
@@ -53,7 +53,6 @@ struct	req {
 };
 
 static	void		 catman(const struct req *, const char *);
-static	int	 	 cmp(const void *, const void *);
 static	void		 format(const struct req *, const char *);
 static	void		 html_print(const char *);
 static	void		 html_printquery(const struct req *);
@@ -64,23 +63,21 @@ static	void		 http_print(const char *);
 static	void 		 http_putchar(char);
 static	void		 http_printquery(const struct req *);
 static	void		 pathgen(struct req *);
+static	void		 pg_error_badrequest(const char *);
+static	void		 pg_error_internal(void);
+static	void		 pg_index(const struct req *);
+static	void		 pg_noresult(const struct req *, const char *);
 static	void		 pg_search(const struct req *);
-static	void		 pg_show(const struct req *, const char *);
+static	void		 pg_searchres(const struct req *,
+				struct manpage *, size_t);
+static	void		 pg_show(struct req *, const char *);
 static	void		 resp_begin_html(int, const char *);
 static	void		 resp_begin_http(int, const char *);
 static	void		 resp_end_html(void);
-static	void		 resp_error_badrequest(const char *);
-static	void		 resp_error_internal(void);
-static	void		 resp_index(const struct req *);
-static	void		 resp_noresult(const struct req *,
-				const char *);
-static	void		 resp_search(const struct req *,
-				struct manpage *, size_t);
 static	void		 resp_searchform(const struct req *);
 static	void		 resp_show(const struct req *, const char *);
 
 static	const char	 *scriptname; /* CGI script name */
-static	const char	 *httphost; /* hostname used in the URIs */
 
 static	const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9};
 static	const char *const sec_numbers[] = {
@@ -251,14 +248,25 @@ http_parse(struct req *req, char *p)
 
 		if (0 == strcmp(key, "query"))
 			req->q.expr = val;
-		else if (0 == strcmp(key, "manpath"))
+		else if (0 == strcmp(key, "manpath")) {
+#ifdef COMPAT_OLDURI
+			if (0 == strncmp(val, "OpenBSD ", 8)) {
+				val[7] = '-';
+				if ('C' == val[8])
+					val[8] = 'c';
+			}
+#endif
 			req->q.manpath = val;
-		else if (0 == strcmp(key, "apropos"))
+		} else if (0 == strcmp(key, "apropos"))
 			req->q.equal = !strcmp(val, "0");
-		else if (0 == strcmp(key, "sec") ||
-			 0 == strcmp(key, "sektion")) {
+		else if (0 == strcmp(key, "sec")) {
 			if (strcmp(val, "0"))
 				req->q.sec = val;
+#ifdef COMPAT_OLDURI
+		} else if (0 == strcmp(key, "sektion")) {
+			if (strcmp(val, "0"))
+				req->q.sec = val;
+#endif
 		} else if (0 == strcmp(key, "arch")) {
 			if (strcmp(val, "default"))
 				req->q.arch = val;
@@ -413,7 +421,11 @@ resp_searchform(const struct req *req)
 
 	/* Write architecture selector. */
 
-	puts("<SELECT NAME=\"arch\">");
+	printf(	"<SELECT NAME=\"arch\">\n"
+		"<OPTION VALUE=\"default\"");
+	if (NULL == req->q.arch)
+		printf(" SELECTED");
+	puts(">All Architectures</OPTION>");
 	for (i = 0; i < arch_MAX; i++) {
 		printf("<OPTION VALUE=\"%s\"", arch_names[i]);
 		if (NULL != req->q.arch &&
@@ -457,8 +469,48 @@ resp_searchform(const struct req *req)
 	puts("<!-- End search form. //-->");
 }
 
+static int
+validate_urifrag(const char *frag)
+{
+
+	while ('\0' != *frag) {
+		if ( ! (isalnum((unsigned char)*frag) ||
+		    '-' == *frag || '.' == *frag ||
+		    '/' == *frag || '_' == *frag))
+			return(0);
+		frag++;
+	}
+	return(1);
+}
+
+static int
+validate_manpath(const struct req *req, const char* manpath)
+{
+	size_t	 i;
+
+	if ( ! strcmp(manpath, "mandoc"))
+		return(1);
+
+	for (i = 0; i < req->psz; i++)
+		if ( ! strcmp(manpath, req->p[i]))
+			return(1);
+
+	return(0);
+}
+
+static int
+validate_filename(const char *file)
+{
+
+	if ('.' == file[0] && '/' == file[1])
+		file += 2;
+
+	return ( ! (strstr(file, "../") || strstr(file, "/..") ||
+	    (strncmp(file, "man", 3) && strncmp(file, "cat", 3))));
+}
+
 static void
-resp_index(const struct req *req)
+pg_index(const struct req *req)
 {
 
 	resp_begin_html(200, NULL);
@@ -475,7 +527,7 @@ resp_index(const struct req *req)
 }
 
 static void
-resp_noresult(const struct req *req, const char *msg)
+pg_noresult(const struct req *req, const char *msg)
 {
 	resp_begin_html(200, NULL);
 	resp_searchform(req);
@@ -486,7 +538,7 @@ resp_noresult(const struct req *req, const char *msg)
 }
 
 static void
-resp_error_badrequest(const char *msg)
+pg_error_badrequest(const char *msg)
 {
 
 	resp_begin_html(400, "Bad Request");
@@ -500,7 +552,7 @@ resp_error_badrequest(const char *msg)
 }
 
 static void
-resp_error_internal(void)
+pg_error_internal(void)
 {
 	resp_begin_html(500, "Internal Server Error");
 	puts("<P>Internal Server Error</P>");
@@ -508,12 +560,23 @@ resp_error_internal(void)
 }
 
 static void
-resp_search(const struct req *req, struct manpage *r, size_t sz)
+pg_searchres(const struct req *req, struct manpage *r, size_t sz)
 {
+	char		*arch, *archend;
 	size_t		 i, iuse, isec;
+	int		 archprio, archpriouse;
 	int		 prio, priouse;
 	char		 sec;
 
+	for (i = 0; i < sz; i++) {
+		if (validate_filename(r[i].file))
+			continue;
+		fprintf(stderr, "invalid filename %s in %s database\n",
+		    r[i].file, req->q.manpath);
+		pg_error_internal();
+		return;
+	}
+
 	if (1 == sz) {
 		/*
 		 * If we have just one result, then jump there now
@@ -521,7 +584,7 @@ resp_search(const struct req *req, struct manpage *r, 
 		 */
 		printf("Status: 303 See Other\r\n");
 		printf("Location: http://%s%s/%s/%s?",
-		    httphost, scriptname, req->q.manpath, r[0].file);
+		    HTTP_HOST, scriptname, req->q.manpath, r[0].file);
 		http_printquery(req);
 		printf("\r\n"
 		     "Content-Type: text/html; charset=utf-8\r\n"
@@ -529,8 +592,6 @@ resp_search(const struct req *req, struct manpage *r, 
 		return;
 	}
 
-	qsort(r, sz, sizeof(struct manpage), cmp);
-
 	resp_begin_html(200, NULL);
 	resp_searchform(req);
 	puts("<DIV CLASS=\"results\">");
@@ -564,12 +625,30 @@ resp_search(const struct req *req, struct manpage *r, 
 		puts("<HR>");
 		iuse = 0;
 		priouse = 10;
+		archpriouse = 3;
 		for (i = 0; i < sz; i++) {
 			isec = strcspn(r[i].file, "123456789");
 			sec = r[i].file[isec];
 			if ('\0' == sec)
 				continue;
 			prio = sec_prios[sec - '1'];
+			if (NULL == req->q.arch) {
+				archprio =
+				    (NULL == (arch = strchr(
+					r[i].file + isec, '/'))) ? 3 :
+				    (NULL == (archend = strchr(
+					arch + 1, '/'))) ? 0 :
+				    strncmp(arch, "amd64/",
+					archend - arch) ? 2 : 1;
+				if (archprio < archpriouse) {
+					archpriouse = archprio;
+					priouse = prio;
+					iuse = i;
+					continue;
+				}
+				if (archprio > archpriouse)
+					continue;
+			}
 			if (prio >= priouse)
 				continue;
 			priouse = prio;
@@ -737,19 +816,20 @@ format(const struct req *req, const char *file)
 	if (rc >= MANDOCLEVEL_FATAL) {
 		fprintf(stderr, "fatal mandoc error: %s/%s\n",
 		    req->q.manpath, file);
-		resp_error_internal();
+		pg_error_internal();
 		return;
 	}
 
-	snprintf(opts, sizeof(opts),
-	    "fragment,man=%s?query=%%N&amp;sec=%%S",
-	    scriptname);
+	snprintf(opts, sizeof(opts), "fragment,man=%s?"
+	    "manpath=%s&amp;query=%%N&amp;sec=%%S&amp;arch=%s",
+	    scriptname, req->q.manpath,
+	    req->q.arch ? req->q.arch : "");
 
 	mparse_result(mp, &mdoc, &man, NULL);
 	if (NULL == man && NULL == mdoc) {
 		fprintf(stderr, "fatal mandoc error: %s/%s\n",
 		    req->q.manpath, file);
-		resp_error_internal();
+		pg_error_internal();
 		mparse_free(mp);
 		return;
 	}
@@ -769,6 +849,9 @@ static void
 resp_show(const struct req *req, const char *file)
 {
 
+	if ('.' == file[0] && '/' == file[1])
+		file += 2;
+
 	if ('c' == *file)
 		catman(req, file);
 	else
@@ -776,17 +859,23 @@ resp_show(const struct req *req, const char *file)
 }
 
 static void
-pg_show(const struct req *req, const char *path)
+pg_show(struct req *req, const char *path)
 {
 	char		*sub;
 
 	if (NULL == path || NULL == (sub = strchr(path, '/'))) {
-		resp_error_badrequest(
+		pg_error_badrequest(
 		    "You did not specify a page to show.");
 		return;
 	} 
 	*sub++ = '\0';
 
+	if ( ! validate_manpath(req, path)) {
+		pg_error_badrequest(
+		    "You specified an invalid manpath.");
+		return;
+	}
+
 	/*
 	 * Begin by chdir()ing into the manpath.
 	 * This way we can pick up the database files, which are
@@ -794,11 +883,21 @@ pg_show(const struct req *req, const char *path)
 	 */
 
 	if (-1 == chdir(path)) {
-		resp_error_badrequest(
-		    "You specified an invalid manpath.");
+		fprintf(stderr, "chdir %s: %s\n",
+		    path, strerror(errno));
+		pg_error_internal();
 		return;
 	}
 
+	if ( ! validate_filename(sub)) {
+		pg_error_badrequest(
+		    "You specified an invalid manual file.");
+		return;
+	}
+
+	if (strcmp(path, "mandoc"))
+		req->q.manpath = path;
+
 	resp_begin_html(200, NULL);
 	resp_searchform(req);
 	resp_show(req, sub);
@@ -823,8 +922,9 @@ pg_search(const struct req *req)
 	 */
 
 	if (-1 == (chdir(req->q.manpath))) {
-		resp_error_badrequest(
-		    "You specified an invalid manpath.");
+		fprintf(stderr, "chdir %s: %s\n",
+		    req->q.manpath, strerror(errno));
+		pg_error_internal();
 		return;
 	}
 
@@ -861,11 +961,11 @@ pg_search(const struct req *req)
 	}
 
 	if (0 == mansearch(&search, &paths, sz, cp, "Nd", &res, &ressz))
-		resp_noresult(req, "You entered an invalid query.");
+		pg_noresult(req, "You entered an invalid query.");
 	else if (0 == ressz)
-		resp_noresult(req, "No results found.");
+		pg_noresult(req, "No results found.");
 	else
-		resp_search(req, res, ressz);
+		pg_searchres(req, res, ressz);
 
 	for (i = 0; i < sz; i++)
 		free(cp[i]);
@@ -895,8 +995,12 @@ main(void)
 	if (NULL == (scriptname = getenv("SCRIPT_NAME")))
 		scriptname = "";
 
-	if (NULL == (httphost = getenv("HTTP_HOST")))
-		httphost = "localhost";
+	if ( ! validate_urifrag(scriptname)) {
+		fprintf(stderr, "unsafe SCRIPT_NAME \"%s\"\n",
+		    scriptname);
+		pg_error_internal();
+		return(EXIT_FAILURE);
+	}
 
 	/*
 	 * First we change directory into the MAN_DIR so that
@@ -907,7 +1011,7 @@ main(void)
 	if (-1 == chdir(MAN_DIR)) {
 		fprintf(stderr, "MAN_DIR: %s: %s\n",
 		    MAN_DIR, strerror(errno));
-		resp_error_internal();
+		pg_error_internal();
 		return(EXIT_FAILURE);
 	} 
 
@@ -919,6 +1023,18 @@ main(void)
 	if (NULL != (querystring = getenv("QUERY_STRING")))
 		http_parse(&req, querystring);
 
+	if ( ! validate_manpath(&req, req.q.manpath)) {
+		pg_error_badrequest(
+		    "You specified an invalid manpath.");
+		return(EXIT_FAILURE);
+	}
+
+	if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) {
+		pg_error_badrequest(
+		    "You specified an invalid architecture.");
+		return(EXIT_FAILURE);
+	}
+
 	/* Dispatch to the three different pages. */
 
 	path = getenv("PATH_INFO");
@@ -932,7 +1048,7 @@ main(void)
 	else if (NULL != req.q.expr)
 		pg_search(&req);
 	else
-		resp_index(&req);
+		pg_index(&req);
 
 	for (i = 0; i < (int)req.psz; i++)
 		free(req.p[i]);
@@ -940,14 +1056,6 @@ main(void)
 	return(EXIT_SUCCESS);
 }
 
-static int
-cmp(const void *p1, const void *p2)
-{
-
-	return(strcasecmp(((const struct manpage *)p1)->names,
-	    ((const struct manpage *)p2)->names));
-}
-
 /*
  * Scan for indexable paths.
  */
@@ -958,14 +1066,37 @@ pathgen(struct req *req)
 	char	*dp;
 	size_t	 dpsz;
 
-	if (NULL == (fp = fopen("manpath.conf", "r")))
-		return;
+	if (NULL == (fp = fopen("manpath.conf", "r"))) {
+		fprintf(stderr, "%s/manpath.conf: %s\n",
+			MAN_DIR, strerror(errno));
+		pg_error_internal();
+		exit(EXIT_FAILURE);
+	}
 
 	while (NULL != (dp = fgetln(fp, &dpsz))) {
 		if ('\n' == dp[dpsz - 1])
 			dpsz--;
 		req->p = mandoc_realloc(req->p,
 		    (req->psz + 1) * sizeof(char *));
-		req->p[req->psz++] = mandoc_strndup(dp, dpsz);
+		dp = mandoc_strndup(dp, dpsz);
+		if ( ! validate_urifrag(dp)) {
+			fprintf(stderr, "%s/manpath.conf contains "
+			    "unsafe path \"%s\"\n", MAN_DIR, dp);
+			pg_error_internal();
+			exit(EXIT_FAILURE);
+		}
+		if (NULL != strchr(dp, '/')) {
+			fprintf(stderr, "%s/manpath.conf contains "
+			    "path with slash \"%s\"\n", MAN_DIR, dp);
+			pg_error_internal();
+			exit(EXIT_FAILURE);
+		}
+		req->p[req->psz++] = dp;
+	}
+
+	if ( req->p == NULL ) {
+		fprintf(stderr, "%s/manpath.conf is empty\n", MAN_DIR);
+		pg_error_internal();
+		exit(EXIT_FAILURE);
 	}
 }