X-Git-Url: http://git.tuebingen.mpg.de/?p=paraslash.git;a=blobdiff_plain;f=net.c;h=88b3e3cc8aee0b2fadfd7783c763b062b8ac0d24;hp=60f8e9a1d07cceafa4dcc51bb64ec5ae6b50b82e;hb=d3efa4f48d28ccc4c8f0238178eeb0d2641d6dfb;hpb=ec22c98399a8317cb8310a343ac0030279b58571

diff --git a/net.c b/net.c
index 60f8e9a1..88b3e3cc 100644
--- a/net.c
+++ b/net.c
@@ -1,11 +1,17 @@
 /*
- * Copyright (C) 2005-2008 Andre Noll <maan@systemlinux.org>
+ * Copyright (C) 2005-2009 Andre Noll <maan@systemlinux.org>
  *
  * Licensed under the GPL v2. For licencing details see COPYING.
  */
 
 /** \file net.c Networking-related helper functions. */
 
+/*
+ * Since glibc 2.8, the _GNU_SOURCE feature test macro must be defined in order
+ * to obtain the definition of the ucred structure.
+ */
+#define _GNU_SOURCE
+
 #include <netdb.h>
 
 /* At least NetBSD needs these. */
@@ -20,6 +26,7 @@
 #endif
 
 #include <dirent.h>
+#include <regex.h>
 
 #include "para.h"
 #include "error.h"
@@ -85,6 +92,114 @@ void disable_crypt(int fd)
 	crypt_data_array[fd].private_data = NULL;
 }
 
+/**
+ * Match string as a candidate IPv4 address.
+ *
+ * \param address The string to match.
+ * \return True if \a address has "dot-quad" format.
+ */
+static bool is_v4_dot_quad(const char *address)
+{
+	bool result;
+	regex_t r;
+
+	assert(!regcomp(&r, "^([0-9]+\\.){3}[0-9]+$", REG_EXTENDED|REG_NOSUB));
+	result = regexec(&r, address, 0, NULL, 0) == 0;
+	regfree(&r);
+	return result;
+}
+
+/**
+ * Perform basic syntax checking on the host-part of an URL:
+ *
+ * - Since ':' is invalid in IPv4 addresses and DNS names, the
+ *   presence of ':' causes interpretation as IPv6 address;
+ * - next the first-match-wins algorithm from RFC 3986 is applied;
+ * - else the string is considered as DNS name, to be resolved later.
+ *
+ * \param host The host string to check.
+ * \return True if \a host passes the syntax checks.
+ *
+ * \sa RFC 3986, 3.2.2; RFC 1123, 2.1; RFC 1034, 3.5
+ */
+static bool host_string_ok(const char *host)
+{
+	if (host == NULL || *host == '\0')
+		return false;
+	if (strchr(host, ':') != NULL)
+		return is_valid_ipv6_address(host);
+	if (is_v4_dot_quad(host))
+		return is_valid_ipv4_address(host);
+	return true;
+}
+
+/**
+ * Parse and validate URL string.
+ *
+ * The URL syntax is loosely based on RFC 3986, supporting one of
+ * - "["host"]"[:port] for native IPv6 addresses and
+ * - host[:port] for IPv4 hostnames and DNS names.
+ *
+ * Native IPv6 addresses must be enclosed in square brackets, since
+ * otherwise there is an ambiguity with the port separator `:'.
+ * The 'port' part is always considered to be a number; if absent,
+ * it is set to -1, to indicate that a default port is to be used.
+ *
+ * The following are valid examples:
+ * - 10.10.1.1
+ * - 10.10.1.2:8000
+ * - localhost
+ * - localhost:8001
+ * - [::1]:8000
+ * - [badc0de::1]
+ *
+ * \param url	  The URL string to take apart.
+ * \param host	  To return the copied host part of \a url.
+ * \param hostlen The maximum length of \a host.
+ * \param port	  To return the port number (if any) of \a url.
+ *
+ * \return Pointer to \a host, or NULL if failed.
+ * If NULL is returned, \a host and \a portnum are undefined. If no
+ * port number was present in \a url, \a portnum is set to -1.
+ *
+ * \sa RFC 3986, 3.2.2/3.2.3
+ */
+char *parse_url(const char *url,
+		char	*host, ssize_t hostlen,
+		int32_t	*port)
+{
+	const char *o = url;
+	char *c = host, *end = c + (hostlen - 1);
+
+	*port = -1;
+
+	if (o == NULL || hostlen < 1)
+		goto failed;
+
+	if (*o == '[') {
+		for (++o; (*c = *o == ']' ? '\0' : *o); c++, o++)
+			if (c == end)
+				goto failed;
+
+		if (*o++ != ']' || (*o != '\0' && *o != ':'))
+			goto failed;
+	} else {
+		for (; (*c = *o == ':'? '\0' : *o); c++, o++)
+			if (c == end)
+				goto failed;
+	}
+
+	if (*o == ':')
+		if (para_atoi32(++o, port) < 0 ||
+		    *port < 0 || *port > 0xffff)
+			goto failed;
+
+	if (host_string_ok(host))
+		return host;
+failed:
+	*host = '\0';
+	return NULL;
+}
 
 /**
  * Determine the socket type for a given layer-4 protocol.
@@ -728,10 +843,9 @@ int recv_cred_buffer(int fd, char *buf, size_t size)
  *
  * \return Positive if \a pattern was received, negative otherwise.
  *
- * This function creates a buffer of size \a bufsize and tries
- * to receive at most \a bufsize bytes from file descriptor \a fd.
- * If at least \p strlen(\a pattern) bytes were received, the beginning of
- * the received buffer is compared with \a pattern, ignoring case.
+ * This function tries to receive at most \a bufsize bytes from file descriptor
+ * \a fd. If at least \p strlen(\a pattern) bytes were received, the beginning
+ * of the received buffer is compared with \a pattern, ignoring case.
  *
  * \sa recv_buffer(), \sa strncasecmp(3).
  */
@@ -739,7 +853,7 @@ int recv_pattern(int fd, const char *pattern, size_t bufsize)
 {
 	size_t len = strlen(pattern);
 	char *buf = para_malloc(bufsize + 1);
-	int ret = -E_RECV_PATTERN, n = recv_buffer(fd, buf, bufsize);
+	int ret = -E_RECV_PATTERN, n = recv_buffer(fd, buf, bufsize + 1);
 
 	if (n < len)
 		goto out;