geomyidae

commit a291bc8c99309fcbfd1a7688988ba60a2e5bf042
parent b12a77acd24fc170b1ad047986ffaf13592fb326
Author: Christoph Lohmann <20h@r-36.net>
Date:   Sat, 22 Jul 2023 15:28:45 +0200

Fix traversal handling, add selector and traversal to CGI environment.

* Add raw selector to CGI scripts.
* Add traversal to CGI scripts.
* Add both to manpages.

Diffstat:
M
geomyidae.8
 | 
9
++++++---
M
handlr.c
 | 
56
++++++++++++++++++++++++++++----------------------------
M
handlr.h
 | 
18
+++++++++++-------
M
ind.c
 | 
21
+++++----------------
M
ind.h
 | 
4
++--
M
main.c
 | 
102
+++++++++++++++++++++++++++++++++++++++----------------------------------------

6 files changed, 102 insertions(+), 108 deletions(-)
diff --git a/geomyidae.8 b/geomyidae.8
@@ -352,9 +352,11 @@ executable.[d]cgi $search $arguments $host $port
 where
 .Bd -literal -offset indent
 search = query string (type 7) or "" (type 0)
-arguments = string after "?" in the path, the remaining path or ""
+arguments = string behind "?" in selector or ""
 host = server's hostname ("localhost" by default)
 port = server's port ("70" by default)
+traversal = remaining path from path traversal
+selector = raw selector
 .Ed
 .Pp
 All terms are tab-separated (per gopher protocol) which can cause some
@@ -377,8 +379,9 @@ GATEWAY_INTERFACE = `CGI/1.1'
 PATH_INFO = script which is executed
 PATH_TRANSLATED = absolute path with script which is executed
 QUERY_STRING = arguments (See above.)
-SELECTOR = arguments (For backwards compatibility.)
-REQUEST = arguments (For backwards compatibility.)
+SELECTOR = raw selector
+REQUEST = raw selector
+TRAVERSAL = traversel (See above.)
 REMOTE_ADDR = IP of the client
 REMOTE_HOST = REMOTE_ADDR
 REQUEST_METHOD = `GET'
diff --git a/handlr.c b/handlr.c
@@ -24,7 +24,8 @@
 
 void
 handledir(int sock, char *path, char *port, char *base, char *args,
-		char *sear, char *ohost, char *chost, char *bhost, int istls)
+		char *sear, char *ohost, char *chost, char *bhost, int istls,
+		char *sel, char *traverse)
 {
 	char *pa, *file, *e, *par;
 	struct dirent **dirent;
@@ -35,12 +36,8 @@ handledir(int sock, char *path, char *port, char *base, char *args,
 	USED(args);
 	USED(sear);
 	USED(bhost);
-
-	printf("handledir:\n");
-	printf("sock = %d; path = %s; port = %s; base = %s; args = %s;\n",
-			sock, path, port, base, args);
-	printf("sear = %s; ohost = %s; chost = %s; bhost = %s; istls = %d;\n",
-			sear, ohost, chost, bhost, istls);
+	USED(sel);
+	USED(traverse);
 
 	pa = xstrdup(path);
 
@@ -70,7 +67,6 @@ handledir(int sock, char *path, char *port, char *base, char *args,
 					pa,
 					pa[strlen(pa)-1] == '/'? "" : "/",
 					dirent[i]->d_name);
-			printf("handledir: smprintf file = %s\n", file);
 			if (stat(file, &st) >= 0 && S_ISDIR(st.st_mode))
 				type = gettype("index.gph");
 			ret = dprintf(sock,
@@ -93,7 +89,8 @@ handledir(int sock, char *path, char *port, char *base, char *args,
 
 void
 handlegph(int sock, char *file, char *port, char *base, char *args,
-		char *sear, char *ohost, char *chost, char *bhost, int istls)
+		char *sear, char *ohost, char *chost, char *bhost, int istls,
+		char *sel, char *traverse)
 {
 	gphindex *act;
 	int i, ret = 0;
@@ -101,12 +98,8 @@ handlegph(int sock, char *file, char *port, char *base, char *args,
 	USED(args);
 	USED(sear);
 	USED(bhost);
-
-	printf("handlegph:\n");
-	printf("sock = %d; file = %s; port = %s; base = %s; args = %s;\n",
-			sock, file, port, base, args);
-	printf("sear = %s; ohost = %s; chost = %s; bhost = %s; istls = %d;\n",
-			sear, ohost, chost, bhost, istls);
+	USED(sel);
+	USED(traverse);
 
 	act = gph_scanfile(file);
 	if (act != NULL) {
@@ -124,7 +117,8 @@ handlegph(int sock, char *file, char *port, char *base, char *args,
 
 void
 handlebin(int sock, char *file, char *port, char *base, char *args,
-		char *sear, char *ohost, char *chost, char *bhost, int istls)
+		char *sear, char *ohost, char *chost, char *bhost, int istls,
+		char *sel, char *traverse)
 {
 	int fd;
 
@@ -134,6 +128,8 @@ handlebin(int sock, char *file, char *port, char *base, char *args,
 	USED(sear);
 	USED(ohost);
 	USED(bhost);
+	USED(sel);
+	USED(traverse);
 
 	fd = open(file, O_RDONLY);
 	if (fd >= 0) {
@@ -145,9 +141,10 @@ handlebin(int sock, char *file, char *port, char *base, char *args,
 
 void
 handlecgi(int sock, char *file, char *port, char *base, char *args,
-		char *sear, char *ohost, char *chost, char *bhost, int istls)
+		char *sear, char *ohost, char *chost, char *bhost, int istls,
+		char *sel, char *traverse)
 {
-	char *script, *path;
+	char *script, *path, *filec;
 
 	USED(base);
 	USED(port);
@@ -157,9 +154,10 @@ handlecgi(int sock, char *file, char *port, char *base, char *args,
 			sock, file, port, base, args);
 	printf("sear = %s; ohost = %s; chost = %s; bhost = %s; istls = %d;\n",
 			sear, ohost, chost, bhost, istls);
+	printf("sel = %s; traverse = %s;\n", sel, traverse);
 
-	path = xstrdup(file);
-	path = dirname(path);
+	filec = xstrdup(file);
+	path = dirname(filec);
 	script = path + strlen(path) + 1;
 	printf("path = %s\n", path);
 	printf("script = %s\n", script);
@@ -180,7 +178,7 @@ handlecgi(int sock, char *file, char *port, char *base, char *args,
 		}
 
 		setcgienviron(script, file, port, base, args, sear, ohost, chost,
-				bhost, istls);
+				bhost, istls, sel, traverse);
 
 		if (execl(file, script, sear, args, ohost, port,
 				(char *)NULL) == -1) {
@@ -192,17 +190,18 @@ handlecgi(int sock, char *file, char *port, char *base, char *args,
 		break;
 	default:
 		wait(NULL);
-		free(path);
+		free(filec);
 		break;
 	}
 }
 
 void
 handledcgi(int sock, char *file, char *port, char *base, char *args,
-		char *sear, char *ohost, char *chost, char *bhost, int istls)
+		char *sear, char *ohost, char *chost, char *bhost, int istls,
+		char *sel, char *traverse)
 {
 	FILE *fp;
-	char *script, *path, *ln = NULL;
+	char *script, *path, *filec, *ln = NULL;
 	size_t linesiz = 0;
 	ssize_t n;
 	int outsocks[2], ret = 0;
@@ -213,12 +212,13 @@ handledcgi(int sock, char *file, char *port, char *base, char *args,
 			sock, file, port, base, args);
 	printf("sear = %s; ohost = %s; chost = %s; bhost = %s; istls = %d;\n",
 			sear, ohost, chost, bhost, istls);
+	printf("sel = %s; traverse = %s;\n", sel, traverse);
 
 	if (socketpair(AF_LOCAL, SOCK_STREAM, 0, outsocks) < 0)
 		return;
 
-	path = xstrdup(file);
-	path = dirname(path);
+	filec = xstrdup(file);
+	path = dirname(filec);
 	script = path + strlen(path) + 1;
 
 	if (sear == NULL)
@@ -238,7 +238,7 @@ handledcgi(int sock, char *file, char *port, char *base, char *args,
 		}
 
 		setcgienviron(script, file, port, base, args, sear, ohost, chost,
-				bhost, istls);
+				bhost, istls, sel, traverse);
 
 		if (execl(file, script, sear, args, ohost, port,
 				(char *)NULL) == -1) {
@@ -277,7 +277,7 @@ handledcgi(int sock, char *file, char *port, char *base, char *args,
 		free(ln);
 		fclose(fp);
 		wait(NULL);
-		free(path);
+		free(filec);
 		break;
 	}
 }
diff --git a/handlr.h b/handlr.h
@@ -23,7 +23,7 @@
  * base .... base path of geomyidae, never ends in '/', so chroot is ''
  *   Sample: /var/gopher
  * args .... Gives all variable input from the selector in some way.
- *   Sample: /with/dirs////?key=value
+ *   Sample: key=value
  * sear .... search part of request
  *   Sample: search what?
  * ohost ... host of geomyidae (See -h in geomyidae(8))
@@ -34,25 +34,29 @@
  *   Sample: 78.46.175.99
  * istls ... set to 1, if TLS was used for thr request
  *   Sample: 1
+ * sel ..... Gives the raw selector after processing.
+ *   Sample: /get/some/script/with/dirs////?key=value
+ * traversal ..... Gives the raw selector after processing.
+ *   Sample: /with/dirs////
  */
 
 void handledir(int sock, char *path, char *port, char *base, char *args,
 			char *sear, char *ohost, char *chost, char *bhost,
-			int istls);
+			int istls, char *sel, char *traverse);
 void handlegph(int sock, char *file, char *port, char *base, char *args,
 			char *sear, char *ohost, char *chost, char *bhost,
-			int istls);
+			int istls, char *sel, char *traverse);
 void handlebin(int sock, char *file, char *port, char *base, char *args,
 			char *sear, char *ohost, char *chost, char *bhost,
-			int istls);
+			int istls, char *sel, char *traverse);
 void handletxt(int sock, char *file, char *port, char *base, char *args,
 			char *sear, char *ohost, char *chost, char *bhost,
-			int istls);
+			int istls, char *sel, char *traverse);
 void handlecgi(int sock, char *file, char *port, char *base, char *args,
 			char *sear, char *ohost, char *chost, char *bhost,
-			int istls);
+			int istls, char *sel, char *traverse);
 void handledcgi(int sock, char *file, char *port, char *base, char *args,
 			char *sear, char *ohost, char *chost, char *bhost,
-			int istls);
+			int istls, char *sel, char *traverse);
 
 #endif
diff --git a/ind.c b/ind.c
@@ -535,7 +535,8 @@ reverselookup(char *host)
 
 void
 setcgienviron(char *file, char *path, char *port, char *base, char *args,
-		char *sear, char *ohost, char *chost, char *bhost, int istls)
+		char *sear, char *ohost, char *chost, char *bhost, int istls,
+		char *sel, char *traverse)
 {
 	/*
 	 * TODO: Clean environment from possible unsafe environment variables.
@@ -547,26 +548,19 @@ setcgienviron(char *file, char *path, char *port, char *base, char *args,
 	setenv("GATEWAY_INTERFACE", "CGI/1.1", 1);
 	/* TODO: Separate, if run like rest.dcgi. */
 	setenv("PATH_INFO", file, 1);
-	printf("PATH_INFO = %s\n", file);
 	setenv("PATH_TRANSLATED", path, 1);
-	printf("PATH_TRANSLATED = %s\n", path);
 
 	setenv("QUERY_STRING", args, 1);
-	printf("QUERY_STRING = %s\n", args);
-	/* legacy compatibility */
-	setenv("SELECTOR", args, 1);
-	printf("SELECTOR = %s\n", args);
-	setenv("REQUEST", args, 1);
-	printf("REQUEST = %s\n", args);
+	setenv("SELECTOR", sel, 1);
+	setenv("REQUEST", sel, 1);
+	setenv("TRAVERSAL", traverse, 1);
 
 	setenv("REMOTE_ADDR", chost, 1);
-	printf("REMOTE_ADDR = %s\n", chost);
 	/*
 	 * Don't do a reverse lookup on every call. Only do when needed, in
 	 * the script. The RFC allows us to set the IP to the value.
 	 */
 	setenv("REMOTE_HOST", chost, 1);
-	printf("REMOTE_HOST = %s\n", chost);
 	/* Please do not implement identd here. */
 	unsetenv("REMOTE_IDENT");
 	unsetenv("REMOTE_USER");
@@ -578,12 +572,9 @@ setcgienviron(char *file, char *path, char *port, char *base, char *args,
 	 */
 	setenv("REQUEST_METHOD", "GET", 1);
 	setenv("SCRIPT_NAME", file, 1);
-	printf("SCRIPT_NAME = %s\n", file);
 	setenv("SERVER_NAME", ohost, 1);
-	printf("SERVER_PORT = %s\n", port);
 	setenv("SERVER_PORT", port, 1);
 	setenv("SERVER_LISTEN_NAME", bhost, 1);
-	printf("SERVER_LISTEN_NAME = %s\n", bhost);
 	if (istls) {
 		setenv("SERVER_PROTOCOL", "gophers/1.0", 1);
 	} else {
@@ -592,10 +583,8 @@ setcgienviron(char *file, char *path, char *port, char *base, char *args,
 	setenv("SERVER_SOFTWARE", "geomyidae", 1);
 
 	setenv("X_GOPHER_SEARCH", sear, 1);
-	printf("X_GOPHER_SEARCH = %s\n", sear);
 	/* legacy compatibility */
 	setenv("SEARCHREQUEST", sear, 1);
-	printf("SEARCHREQUEST = %s\n", sear);
 
 	if (istls) {
 		setenv("GOPHERS", "on", 1);
diff --git a/ind.h b/ind.h
@@ -15,7 +15,7 @@ struct filetype {
         char *end;
         char *type;
         void (* f)(int, char *, char *, char *, char *, char *, char *,
-		char *, char *, int);
+		char *, char *, int, char *, char *);
 };
 
 filetype *gettype(char *filename);
@@ -51,7 +51,7 @@ char *smprintf(char *fmt, ...);
 char *reverselookup(char *host);
 void setcgienviron(char *file, char *path, char *port, char *base,
 		char *args, char *sear, char *ohost, char *chost,
-		char *bhost, int istls);
+		char *bhost, int istls, char *sel, char *traverse);
 char *humansize(off_t n);
 char *humantime(const time_t *clock);
 
diff --git a/main.c b/main.c
@@ -137,8 +137,9 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 	      char *serverp, int nocgi, int istls)
 {
 	struct stat dir;
-	char recvc[1025], recvb[1025], path[1025], args[1025], argsc[1025],
-		*sear, *c, *sep, *recvbp;
+	char recvc[1025], recvb[1025], path[PATH_MAX+1], rpath[PATH_MAX+1], args[1025],
+		argsc[1025], traverse[1025], traversec[1025],
+		*sear, *sep, *recvbp, *c;
 	int len = 0, fd, i, maxrecv, pathfallthrough = 0;
 	filetype *type;
 
@@ -172,6 +173,7 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 	memset(recvc, 0, sizeof(recvc));
 	memset(args, 0, sizeof(args));
 	memset(argsc, 0, sizeof(argsc));
+	memset(traverse, 0, sizeof(argsc));
 
 	maxrecv = sizeof(recvb) - 1;
 	if (rlen > maxrecv || rlen < 0)
@@ -241,18 +243,19 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 	c = strchr(recvb, '?');
 	if (c != NULL) {
 		*c++ = '\0';
-		snprintf(args, sizeof(args), "?%s", c);
+		snprintf(args, sizeof(args), "%s", c);
 	}
 	printf("args = %s\n", args);
 	printf("recvb = %s\n", recvb);
 
 	/* Strip '/' at the end of the request. */
 	for (c = recvb + strlen(recvb) - 1; c >= recvb && c[0] == '/'; c--) {
-		/* Prepend to args. */
-		snprintf(args, sizeof(args), "/%s", args);
+		memmove(traversec, traverse, strlen(traverse));
+		/* Prepend to traverse. */
+		snprintf(traverse, sizeof(traverse), "/%s", traversec);
 		c[0] = '\0';
 	}
-	printf("args = %s\n", args);
+	printf("traverse = %s\n", traverse);
 
 	/* Do not allow requests including "..". */
 	if (strstr(recvb, "..")) {
@@ -284,8 +287,7 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 	 *	$args = $rest_of_path + "?" + $args
 	 */
 	if (stat(path, &dir) == -1) {
-		printf("Not found. Try backtraversal.\n");
-		memmove(argsc, args, strlen(args));
+		memmove(traversec, traverse, strlen(traverse));
 		snprintf(path, sizeof(path), "%s", base);
 		recvbp = recvb;
 
@@ -295,44 +297,42 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 		 * etc.
 		 */
 		while (recvbp != NULL) {
-			/* Traverse multiple / in selector. */
-			for (sep = recvbp; sep != recvbp && sep != recvbp+1;
-				sep = strsep(&recvbp, "/"));
-			printf("traversal directory = %s\n", sep);
+			/* Traverse multiple empty / in selector. */
+			while(recvbp[0] == '/')
+				recvbp++;
+			sep = strchr(recvbp, '/');
+			if (sep != NULL)
+				*sep++ = '\0';
 
-			/* Append found directory to path. */
 			snprintf(path+strlen(path), sizeof(path)-strlen(path),
-				"/%s", sep);
+				"/%s", recvbp);
 			/* path is now always at least '/' */
-			printf("full traversal path = %s\n", path);
-
 			if (stat(path, &dir) == -1) {
-				/*
-				 * Current try was not found. Go back one
-				 * step and finish.
-				 */
-				c = strrchr(path, '/');
-				if (c != NULL) {
-					*c++ = '\0';
-					snprintf(args, sizeof(args),
-						"/%s%s%s%s",
-						c,
-						(recvbp != NULL)? "/" : "",
-						(recvbp != NULL)? recvbp : "",
-						(argsc[0] != '\0')? argsc : ""
-					);
-					printf("args = %s\n", args);
-				}
+				path[strlen(path)-strlen(recvbp)-1] = '\0';
+				snprintf(traverse, sizeof(traverse),
+					"/%s%s%s%s",
+					recvbp,
+					(sep != NULL)? "/" : "",
+					(sep != NULL)? sep : "",
+					(traversec[0] != '\0')? traversec : ""
+				);
 				/* path fallthrough */
 				pathfallthrough = 1;
 				printf("pathfallthrough = 1\n");
 				break;
 			}
+			/* Append found directory to path. */
+			recvbp = sep;
 		}
 	}
 
-	printf("path = %s\n", path);
-	if (stat(path, &dir) != -1) {
+	if (realpath(path, (char *)&rpath) == NULL) {
+		dprintf(sock, notfounderr, recvc);
+		if (loglvl & ERRORS)
+			logentry(clienth, clientp, recvc, "not found");
+	}
+	printf("rpath = %s\n", rpath);
+	if (stat(rpath, &dir) != -1) {
 		/*
 		 * If sticky bit is set, only serve if this is encrypted.
 		 */
@@ -349,9 +349,9 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 			printf("S_ISDIR\n");
 			for (i = 0; i < sizeof(indexf)/sizeof(indexf[0]);
 					i++) {
-				len = strlen(path);
-				if (len + strlen(indexf[i]) + (path[len-1] == '/')? 0 : 1
-						>= sizeof(path)) {
+				len = strlen(rpath);
+				if (len + strlen(indexf[i]) + (rpath[len-1] == '/')? 0 : 1
+						>= sizeof(rpath)) {
 					if (loglvl & ERRORS) {
 						logentry(clienth, clientp,
 							recvc,
@@ -359,21 +359,19 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 					}
 					return;
 				}
-				sprintf(path, "%s%s%s",
-					path,
-					(path[len-1] == '/')? "" : "/",
-					indexf[i]);
-				printf("path index = %s\n", path);
-				fd = open(path, O_RDONLY);
+				if (rpath[len-1] != '/')
+					strcat(rpath, "/");
+				strcat(rpath, indexf[i]);
+				printf("path index = %s\n", rpath);
+				fd = open(rpath, O_RDONLY);
 				if (fd >= 0)
 					break;
 
 				/* Not found. Clear path from indexf. */
-				printf("len = %d\n", len);
-				path[len] = '\0';
+				rpath[len] = '\0';
 			}
 		} else {
-			fd = open(path, O_RDONLY);
+			fd = open(rpath, O_RDONLY);
 			if (fd < 0) {
 				dprintf(sock, notfounderr, recvc);
 				if (loglvl & ERRORS) {
@@ -389,9 +387,9 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 	if (fd >= 0) {
 		close(fd);
 
-		c = strrchr(path, '/');
+		c = strrchr(rpath, '/');
 		if (c == NULL)
-			c = path;
+			c = rpath;
 		type = gettype(c);
 
 		/*
@@ -416,8 +414,8 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 			if (loglvl & FILES)
 				logentry(clienth, clientp, recvc, "serving");
 
-			type->f(sock, path, port, base, args, sear, ohost,
-				clienth, serverh, istls);
+			type->f(sock, rpath, port, base, args, sear, ohost,
+				clienth, serverh, istls, recvc, traverse);
 		}
 	} else {
 		if (pathfallthrough && S_ISDIR(dir.st_mode)) {
@@ -429,8 +427,8 @@ handlerequest(int sock, char *req, int rlen, char *base, char *ohost,
 		}
 
 		if (!pathfallthrough && S_ISDIR(dir.st_mode)) {
-			handledir(sock, path, port, base, args, sear, ohost,
-				clienth, serverh, istls);
+			handledir(sock, rpath, port, base, args, sear, ohost,
+				clienth, serverh, istls, recvc, traverse);
 			if (loglvl & DIRS) {
 				logentry(clienth, clientp, recvc,
 							"dir listing");