From 86dd720a666aca5b95c4f913c3f705d6b7a8f17d Mon Sep 17 00:00:00 2001
From: Pavel Plesov <pavel.plesov@gmail.com>
Date: Sun, 8 Aug 2010 16:46:39 +0400
Subject: [PATCH 01/11] Introduce absolute URI parsing helpers.

See evhttp_uri_parse(), evhttp_uri_free() and evhttp_uri_join() for details.
---
 http.c                       | 160 +++++++++++++++++++++++++++++++++++
 include/event2/http.h        |  32 +++++++
 include/event2/http_struct.h |  14 +++
 test/regress_http.c          |  76 +++++++++++++++++
 4 files changed, 282 insertions(+)

diff --git a/http.c b/http.c
index 6aca33c3f7..37936566cf 100644
--- a/http.c
+++ b/http.c
@@ -3334,3 +3334,163 @@ bind_socket(const char *address, ev_uint16_t port, int reuse)
 	return (fd);
 }
 
+struct evhttp_uri *evhttp_uri_parse(const char *source_uri)
+{
+	char *readbuf = 0, *readp = 0, *token = 0, *query = 0, *host = 0, *port = 0;
+
+	struct evhttp_uri *uri = calloc(1, sizeof(*uri));
+	if (uri == NULL) {
+		event_err(1, "%s: calloc", __func__);
+		return NULL;
+	}
+
+	readbuf = strdup(source_uri);
+	if (readbuf == NULL) {
+		event_err(1, "%s: strdup", __func__);
+		free(uri);
+		return NULL;
+	}
+
+	readp = readbuf;
+	token = NULL;
+
+	/* 1. scheme:// */
+	token = strstr(readp, "://");
+	if (!token) {
+		/* unsupported uri */
+		free(readbuf);
+		free(uri);
+		return NULL;
+	}
+
+	*token = '\0';
+	uri->scheme = strdup(readp);
+
+	readp = token;
+	readp += 3; /* eat :// */
+
+	/* 2. query */
+	query = strchr(readp, '/');
+	if (query) {
+		char *fragment = strchr(query, '#');
+		if (fragment) {
+			*fragment++ = '\0'; /* eat '#' */
+			uri->fragment = strdup(fragment);
+		}
+
+		uri->query = strdup(query);
+		*query = '\0'; /* eat '/' */
+	}
+
+	/* 3. user:pass@host:port */
+	host = strchr(readp, '@');
+	if (host) {
+		char *pass = 0;
+		/* got user:pass@host:port */
+		*host++ = '\0'; /* eat @ */;
+		pass = strchr(readp, ':');
+		if (pass) {
+			*pass++ = '\0'; /* eat ':' */
+			uri->pass = strdup(pass);
+		}
+
+		uri->user = strdup(readp);
+		readp = host;
+	}
+
+	/* 4. host:port */
+	port = strchr(readp, ':');
+	if (port) {
+		*port++ = '\0'; /* eat ':' */
+		uri->port = atoi(port);
+	}
+
+	/* 5. host */
+	uri->host = strdup(readp);
+
+	free(readbuf);
+
+	return uri;
+}
+
+void evhttp_uri_free(struct evhttp_uri *uri)
+{
+	if (uri == NULL)
+		return;
+
+#define _URI_FREE_STR(f)		\
+	if (uri->f) {			\
+		free(uri->f);		\
+	}
+
+	_URI_FREE_STR(scheme);
+	_URI_FREE_STR(user);
+	_URI_FREE_STR(pass);
+	_URI_FREE_STR(host);
+	_URI_FREE_STR(query);
+	_URI_FREE_STR(fragment);
+
+	free(uri);
+
+#undef _URI_FREE_STR
+}
+
+char *evhttp_uri_join(struct evhttp_uri *uri, void *buf, size_t limit)
+{
+	struct evbuffer *tmp = 0;
+	unsigned char *joined = 0;
+	size_t joined_size = 0;
+
+#define _URI_ADD(f)	evbuffer_add(tmp, uri->f, strlen(uri->f))
+	if (!uri || !uri->scheme || !buf || !limit)
+		return NULL;
+
+	tmp = evbuffer_new();
+	if (!tmp)
+		return NULL;
+
+	_URI_ADD(scheme);
+	evbuffer_add(tmp, "://", 3);
+	if (uri->host && *uri->host) {
+		if (uri->user && *uri->user) {
+			_URI_ADD(user);
+			if (uri->pass && *uri->pass) {
+				evbuffer_add(tmp, ":", 1);
+				_URI_ADD(pass);
+			}
+			evbuffer_add(tmp, "@", 1);
+		}
+
+		_URI_ADD(host);
+
+		if (uri->port > 0)
+			evbuffer_add_printf(tmp,":%u", uri->port);
+	}
+
+	if (uri->query && *uri->query)
+		_URI_ADD(query);
+
+	if (uri->fragment && *uri->fragment) {
+		if (!uri->query || !*uri->query)
+			evbuffer_add(tmp, "/", 1);
+
+		evbuffer_add(tmp, "#", 1);
+		_URI_ADD(fragment);
+	}
+
+	evbuffer_add(tmp, "\0", 1); /* NUL */
+
+	joined = evbuffer_pullup(tmp, -1);
+	joined_size = evbuffer_get_length(tmp);
+
+	if (joined_size < limit)
+		memcpy(buf, joined, joined_size);
+	else {
+		memcpy(buf, joined, limit-1);
+		*((char *)buf+ limit - 1) = '\0';
+	}
+	evbuffer_free(tmp);
+
+	return (char *)buf;
+#undef _URI_ADD
+}
diff --git a/include/event2/http.h b/include/event2/http.h
index d872a18977..9293270d17 100644
--- a/include/event2/http.h
+++ b/include/event2/http.h
@@ -607,6 +607,38 @@ int evhttp_parse_query__checked_20(const char *uri, struct evkeyvalq *headers);
  */
 char *evhttp_htmlescape(const char *html);
 
+struct evhttp_uri;
+
+/**
+   Helper function to parse out uri.
+
+   Parsing a uri like
+
+      scheme://[[user[:pass]@]foo.com[:port]]/[path][?q=test&s=some+thing][#fragment]
+
+   @param source_uri the request URI
+   @return uri container to hold parsed data, or NULL if there is error
+   @see evhttp_uri_free()
+ */
+struct evhttp_uri *evhttp_uri_parse(const char *source_uri);
+
+/**
+ * Free the memory allocated for the uri and parsed data
+ * @param uri container with parsed data
+   @see evhttp_uri_parse()
+ */
+void evhttp_uri_free(struct evhttp_uri *uri);
+
+/**
+ * Join together the uri parts from parsed data
+ * @param uri container with parsed data
+ * @param buf destination buffer
+ * @param limit destination buffer size
+ * @return an joined uri as string or NULL on error
+   @see evhttp_uri_parse()
+ */
+char *evhttp_uri_join(struct evhttp_uri *uri, void *buf, size_t limit);
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/include/event2/http_struct.h b/include/event2/http_struct.h
index a3664e0448..168f5aca60 100644
--- a/include/event2/http_struct.h
+++ b/include/event2/http_struct.h
@@ -118,6 +118,20 @@ struct {
 	void (*chunk_cb)(struct evhttp_request *, void *);
 };
 
+/**
+ * structure to hold parsed uri
+ */
+struct evhttp_uri {
+	char *scheme; /* scheme; e.g http, ftp etc */
+	char *host; /* hostname, or NULL */
+	char *user; /* usename, or NULL */
+	char *pass; /* password, or NULL */
+	int port; /* port, or zero */
+	char *query; /* path + query: e.g. /path/to?param=foo, or NULL */
+	char *fragment; /* fragment or NULL */
+};
+
+
 #ifdef __cplusplus
 }
 #endif
diff --git a/test/regress_http.c b/test/regress_http.c
index e2fd987b91..170432195e 100644
--- a/test/regress_http.c
+++ b/test/regress_http.c
@@ -1707,6 +1707,81 @@ http_parse_query_test(void *ptr)
 	evhttp_clear_headers(&headers);
 }
 
+static void
+http_parse_uri_test(void *ptr)
+{
+	struct evhttp_uri *uri = NULL;
+	char url_tmp[4096];
+
+#define TT_URI(want) do { 						\
+	char *ret = evhttp_uri_join(uri, url_tmp, sizeof(url_tmp));	\
+	tt_want(ret != NULL);						\
+	tt_want(ret == url_tmp);					\
+	tt_want(strcmp(ret, want) == 0);				\
+	} while(0)
+
+	tt_want(evhttp_uri_join(0, 0, 0) == NULL);
+	tt_want(evhttp_uri_join(0, url_tmp, 0) == NULL);
+	tt_want(evhttp_uri_join(0, url_tmp, sizeof(url_tmp)) == NULL);
+	tt_want(evhttp_uri_join(uri, url_tmp, sizeof(url_tmp)) == NULL);
+
+	tt_want(evhttp_uri_parse("mailto:foo@bar") == NULL);
+
+	uri = evhttp_uri_parse("http://www.test.com/?q=test");
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "www.test.com") == 0);
+	tt_want(strcmp(uri->query, "/?q=test") == 0);
+	tt_want(uri->user == NULL);
+	tt_want(uri->pass == NULL);
+	tt_want(uri->port == 0);
+	tt_want(uri->fragment == NULL);
+	TT_URI("http://www.test.com/?q=test");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("ftp://www.test.com/?q=test");
+	tt_want(strcmp(uri->scheme, "ftp") == 0);
+	tt_want(strcmp(uri->host, "www.test.com") == 0);
+	tt_want(strcmp(uri->query, "/?q=test") == 0);
+	tt_want(uri->user == NULL);
+	tt_want(uri->pass == NULL);
+	tt_want(uri->port == 0);
+	tt_want(uri->fragment == NULL);
+	TT_URI("ftp://www.test.com/?q=test");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("scheme://user:pass@foo.com:42/?q=test&s=some+thing#fragment");
+	tt_want(strcmp(uri->scheme, "scheme") == 0);
+	tt_want(strcmp(uri->user, "user") == 0);
+	tt_want(strcmp(uri->pass, "pass") == 0);
+	tt_want(strcmp(uri->host, "foo.com") == 0);
+	tt_want(uri->port == 42);
+	tt_want(strcmp(uri->query, "/?q=test&s=some+thing") == 0);
+	tt_want(strcmp(uri->fragment, "fragment") == 0);
+	TT_URI("scheme://user:pass@foo.com:42/?q=test&s=some+thing#fragment");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("scheme://user@foo.com/#fragment");
+	tt_want(strcmp(uri->scheme, "scheme") == 0);
+	tt_want(strcmp(uri->user, "user") == 0);
+	tt_want(uri->pass == NULL);
+	tt_want(strcmp(uri->host, "foo.com") == 0);
+	tt_want(uri->port == 0);
+	tt_want(strcmp(uri->query, "/") == 0);
+	tt_want(strcmp(uri->fragment, "fragment") == 0);
+	TT_URI("scheme://user@foo.com/#fragment");
+	evhttp_uri_free(uri);
+	uri = evhttp_uri_parse("file:///some/path/to/the/file");
+	tt_want(strcmp(uri->scheme, "file") == 0);
+	tt_want(uri->user == NULL);
+	tt_want(uri->pass == NULL);
+	tt_want(strcmp(uri->host, "") == 0);
+	tt_want(uri->port == 0);
+	tt_want(strcmp(uri->query, "/some/path/to/the/file") == 0);
+	tt_want(uri->fragment == NULL);
+	TT_URI("file:///some/path/to/the/file");
+	evhttp_uri_free(uri);
+}
+
 static void
 http_uriencode_test(void *ptr)
 {
@@ -2801,6 +2876,7 @@ struct testcase_t http_testcases[] = {
 	{ "base", http_base_test, TT_FORK|TT_NEED_BASE, NULL, NULL },
 	{ "bad_headers", http_bad_header_test, 0, NULL, NULL },
 	{ "parse_query", http_parse_query_test, 0, NULL, NULL },
+	{ "parse_uri", http_parse_uri_test, 0, NULL, NULL },
 	{ "uriencode", http_uriencode_test, 0, NULL, NULL },
 	HTTP_LEGACY(basic),
 	HTTP_LEGACY(cancel),

From 86212341c5f38a973e3190a2c5c36ea947ef7253 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Mon, 18 Oct 2010 14:34:20 -0400
Subject: [PATCH 02/11] Make evhttp_uri_parse and friends conform to memory
 management standards

---
 http.c | 34 ++++++++++++++++++----------------
 1 file changed, 18 insertions(+), 16 deletions(-)

diff --git a/http.c b/http.c
index 37936566cf..7677e493e1 100644
--- a/http.c
+++ b/http.c
@@ -3334,20 +3334,21 @@ bind_socket(const char *address, ev_uint16_t port, int reuse)
 	return (fd);
 }
 
-struct evhttp_uri *evhttp_uri_parse(const char *source_uri)
+struct evhttp_uri *
+evhttp_uri_parse(const char *source_uri)
 {
 	char *readbuf = 0, *readp = 0, *token = 0, *query = 0, *host = 0, *port = 0;
 
-	struct evhttp_uri *uri = calloc(1, sizeof(*uri));
+	struct evhttp_uri *uri = mm_calloc(1, sizeof(struct evhttp_uri));
 	if (uri == NULL) {
 		event_err(1, "%s: calloc", __func__);
 		return NULL;
 	}
 
-	readbuf = strdup(source_uri);
+	readbuf = mm_strdup(source_uri);
 	if (readbuf == NULL) {
 		event_err(1, "%s: strdup", __func__);
-		free(uri);
+		mm_free(uri);
 		return NULL;
 	}
 
@@ -3358,13 +3359,13 @@ struct evhttp_uri *evhttp_uri_parse(const char *source_uri)
 	token = strstr(readp, "://");
 	if (!token) {
 		/* unsupported uri */
-		free(readbuf);
-		free(uri);
+		mm_free(readbuf);
+		mm_free(uri);
 		return NULL;
 	}
 
 	*token = '\0';
-	uri->scheme = strdup(readp);
+	uri->scheme = mm_strdup(readp);
 
 	readp = token;
 	readp += 3; /* eat :// */
@@ -3375,10 +3376,10 @@ struct evhttp_uri *evhttp_uri_parse(const char *source_uri)
 		char *fragment = strchr(query, '#');
 		if (fragment) {
 			*fragment++ = '\0'; /* eat '#' */
-			uri->fragment = strdup(fragment);
+			uri->fragment = mm_strdup(fragment);
 		}
 
-		uri->query = strdup(query);
+		uri->query = mm_strdup(query);
 		*query = '\0'; /* eat '/' */
 	}
 
@@ -3391,10 +3392,10 @@ struct evhttp_uri *evhttp_uri_parse(const char *source_uri)
 		pass = strchr(readp, ':');
 		if (pass) {
 			*pass++ = '\0'; /* eat ':' */
-			uri->pass = strdup(pass);
+			uri->pass = mm_strdup(pass);
 		}
 
-		uri->user = strdup(readp);
+		uri->user = mm_strdup(readp);
 		readp = host;
 	}
 
@@ -3406,9 +3407,9 @@ struct evhttp_uri *evhttp_uri_parse(const char *source_uri)
 	}
 
 	/* 5. host */
-	uri->host = strdup(readp);
+	uri->host = mm_strdup(readp);
 
-	free(readbuf);
+	mm_free(readbuf);
 
 	return uri;
 }
@@ -3420,7 +3421,7 @@ void evhttp_uri_free(struct evhttp_uri *uri)
 
 #define _URI_FREE_STR(f)		\
 	if (uri->f) {			\
-		free(uri->f);		\
+		mm_free(uri->f);		\
 	}
 
 	_URI_FREE_STR(scheme);
@@ -3430,12 +3431,13 @@ void evhttp_uri_free(struct evhttp_uri *uri)
 	_URI_FREE_STR(query);
 	_URI_FREE_STR(fragment);
 
-	free(uri);
+	mm_free(uri);
 
 #undef _URI_FREE_STR
 }
 
-char *evhttp_uri_join(struct evhttp_uri *uri, void *buf, size_t limit)
+char *
+evhttp_uri_join(struct evhttp_uri *uri, void *buf, size_t limit)
 {
 	struct evbuffer *tmp = 0;
 	unsigned char *joined = 0;

From 7d45431e15dc0d8aa88fddc4fc406e19fe54d004 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Mon, 18 Oct 2010 14:38:48 -0400
Subject: [PATCH 03/11] Do not silently truncate URIs in evhttp_uri_join. Also
 avoid evbuffer_pullup.

---
 http.c                | 15 +++++++--------
 include/event2/http.h |  2 +-
 2 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/http.c b/http.c
index 7677e493e1..ba2d528b9b 100644
--- a/http.c
+++ b/http.c
@@ -3437,10 +3437,9 @@ void evhttp_uri_free(struct evhttp_uri *uri)
 }
 
 char *
-evhttp_uri_join(struct evhttp_uri *uri, void *buf, size_t limit)
+evhttp_uri_join(struct evhttp_uri *uri, char *buf, size_t limit)
 {
 	struct evbuffer *tmp = 0;
-	unsigned char *joined = 0;
 	size_t joined_size = 0;
 
 #define _URI_ADD(f)	evbuffer_add(tmp, uri->f, strlen(uri->f))
@@ -3482,15 +3481,15 @@ evhttp_uri_join(struct evhttp_uri *uri, void *buf, size_t limit)
 
 	evbuffer_add(tmp, "\0", 1); /* NUL */
 
-	joined = evbuffer_pullup(tmp, -1);
 	joined_size = evbuffer_get_length(tmp);
 
-	if (joined_size < limit)
-		memcpy(buf, joined, joined_size);
-	else {
-		memcpy(buf, joined, limit-1);
-		*((char *)buf+ limit - 1) = '\0';
+	if (joined_size > limit) {
+		/* It doesn't fit. */
+		evbuffer_free(tmp);
+		return NULL;
 	}
+       	evbuffer_remove(tmp, buf, joined_size);
+
 	evbuffer_free(tmp);
 
 	return (char *)buf;
diff --git a/include/event2/http.h b/include/event2/http.h
index 9293270d17..655ff25494 100644
--- a/include/event2/http.h
+++ b/include/event2/http.h
@@ -637,7 +637,7 @@ void evhttp_uri_free(struct evhttp_uri *uri);
  * @return an joined uri as string or NULL on error
    @see evhttp_uri_parse()
  */
-char *evhttp_uri_join(struct evhttp_uri *uri, void *buf, size_t limit);
+char *evhttp_uri_join(struct evhttp_uri *uri, char *buf, size_t limit);
 
 #ifdef __cplusplus
 }

From fadbfd4e6e2d9badd4b61d5f847c589ed84dd2c8 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Mon, 18 Oct 2010 14:43:54 -0400
Subject: [PATCH 04/11] Clean up error handling in uri_parse a little

---
 http.c | 23 ++++++++++++-----------
 1 file changed, 12 insertions(+), 11 deletions(-)

diff --git a/http.c b/http.c
index ba2d528b9b..87dc1f51e3 100644
--- a/http.c
+++ b/http.c
@@ -3337,19 +3337,18 @@ bind_socket(const char *address, ev_uint16_t port, int reuse)
 struct evhttp_uri *
 evhttp_uri_parse(const char *source_uri)
 {
-	char *readbuf = 0, *readp = 0, *token = 0, *query = 0, *host = 0, *port = 0;
+	char *readbuf = NULL, *readp = NULL, *token = NULL, *query = NULL, *host = NULL, *port = NULL;
 
 	struct evhttp_uri *uri = mm_calloc(1, sizeof(struct evhttp_uri));
 	if (uri == NULL) {
 		event_err(1, "%s: calloc", __func__);
-		return NULL;
+		goto err;
 	}
 
 	readbuf = mm_strdup(source_uri);
 	if (readbuf == NULL) {
 		event_err(1, "%s: strdup", __func__);
-		mm_free(uri);
-		return NULL;
+		goto err;
 	}
 
 	readp = readbuf;
@@ -3359,9 +3358,7 @@ evhttp_uri_parse(const char *source_uri)
 	token = strstr(readp, "://");
 	if (!token) {
 		/* unsupported uri */
-		mm_free(readbuf);
-		mm_free(uri);
-		return NULL;
+		goto err;
 	}
 
 	*token = '\0';
@@ -3412,13 +3409,17 @@ evhttp_uri_parse(const char *source_uri)
 	mm_free(readbuf);
 
 	return uri;
+err:
+	if (uri)
+		evhttp_uri_free(uri);
+	if (readbuf)
+		mm_free(readbuf);
+	return NULL;
 }
 
-void evhttp_uri_free(struct evhttp_uri *uri)
+void
+evhttp_uri_free(struct evhttp_uri *uri)
 {
-	if (uri == NULL)
-		return;
-
 #define _URI_FREE_STR(f)		\
 	if (uri->f) {			\
 		mm_free(uri->f);		\

From eaa5f1d9edd8ae2a5f637a3a07928dfee113accc Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Tue, 19 Oct 2010 11:26:59 -0400
Subject: [PATCH 05/11] Revise evhttp_uri_parse implementation to handle more
 of RFC3986

---
 http.c                       | 354 ++++++++++++++++++++++++++++-------
 include/event2/http_struct.h |   8 +-
 test/regress_http.c          | 143 ++++++++++++--
 3 files changed, 414 insertions(+), 91 deletions(-)

diff --git a/http.c b/http.c
index 87dc1f51e3..883c675f5c 100644
--- a/http.c
+++ b/http.c
@@ -2317,6 +2317,9 @@ static const char uri_chars[256] = {
 	0, 0, 0, 0, 0, 0, 0, 0,   0, 0, 0, 0, 0, 0, 0, 0,
 };
 
+#define CHAR_IS_UNRESERVED(c)			\
+	(uri_chars[(unsigned char)(c)])
+
 /*
  * Helper functions to encode/decode a string for inclusion in a URI.
  * The returned string must be freed by the caller.
@@ -2337,7 +2340,7 @@ evhttp_uriencode(const char *uri, ev_ssize_t len, int space_as_plus)
 		end = uri+strlen(uri);
 
 	for (p = uri; p < end; p++) {
-		if (uri_chars[(unsigned char)(*p)]) {
+		if (CHAR_IS_UNRESERVED(*p)) {
 			evbuffer_add(buf, p, 1);
 		} else if (*p == ' ' && space_as_plus) {
 			evbuffer_add(buf, "+", 1);
@@ -3334,16 +3337,208 @@ bind_socket(const char *address, ev_uint16_t port, int reuse)
 	return (fd);
 }
 
+/* Return true of the string starting at s and ending immediately before eos
+ * is a valid URI scheme according to RFC3986
+ */
+static int
+scheme_ok(const char *s, const char *eos)
+{
+	/* scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) */
+	EVUTIL_ASSERT(eos >= s);
+	if (s == eos)
+		return 0;
+	if (!EVUTIL_ISALPHA(*s))
+		return 0;
+	while (++s < eos) {
+		if (! EVUTIL_ISALNUM(*s) &&
+		    *s != '+' && *s != '-' && *s != '.')
+			return 0;
+	}
+	return 1;
+}
+
+#define SUBDELIMS "!$&'()*+,;="
+
+/* Return true iff [s..eos) is a valid userinfo */
+static int
+userinfo_ok(const char *s, const char *eos)
+{
+	while (s < eos) {
+		if (CHAR_IS_UNRESERVED(*s) ||
+		    strchr(SUBDELIMS, *s) ||
+		    *s == ':')
+			++s;
+		else if (*s == '%' && s+2 < eos &&
+		    EVUTIL_ISXDIGIT(s[1]) &&
+		    EVUTIL_ISXDIGIT(s[2]))
+			s += 3;
+		else
+			return 0;
+	}
+	return 1;
+}
+
+static int
+regname_ok(const char *s, const char *eos)
+{
+	while (s && s<eos) {
+		if (CHAR_IS_UNRESERVED(*s) ||
+		    strchr(SUBDELIMS, *s))
+			++s;
+		else if (*s == '%' &&
+		    EVUTIL_ISXDIGIT(s[1]) &&
+		    EVUTIL_ISXDIGIT(s[2]))
+			s += 3;
+		else
+			return 0;
+	}
+	return 1;
+}
+
+static int
+parse_port(const char *s, const char *eos)
+{
+	int portnum = 0;
+	if (s == eos)
+		return 0; /* The RFC allows an empty port. */
+	while (s < eos) {
+		if (! EVUTIL_ISDIGIT(*s))
+			return -1;
+		portnum = (portnum * 10) + (*s - '0');
+		++s;
+	}
+	return portnum;
+}
+
+/* returns 0 for bad, 1 for ipv6, 2 for IPvFuture */
+static int
+bracket_addr_ok(const char *s, const char *eos)
+{
+	if (s + 3 > eos || *s != '[' || *(eos-1) != ']')
+		return 0;
+	if (s[1] == 'v') {
+		/* IPvFuture, or junk.
+		   "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" )
+		 */
+		s += 2; /* skip [v */
+		--eos;
+		if (!EVUTIL_ISXDIGIT(*s)) /*require at least one*/
+			return 0;
+		while (s < eos && *s != '.') {
+			if (EVUTIL_ISXDIGIT(*s))
+				++s;
+			else
+				return 0;
+		}
+		if (*s != '.')
+			return 0;
+		++s;
+		while (s < eos) {
+			if (CHAR_IS_UNRESERVED(*s) ||
+			    strchr(SUBDELIMS, *s) ||
+			    *s == ':')
+				++s;
+			else
+				return 0;
+		}
+		return 2;
+	} else {
+		/* IPv6, or junk */
+		char buf[64];
+		int n_chars = eos-s-2;
+		struct in6_addr in6;
+		if (n_chars >= 64) /* way too long */
+			return 0;
+		memcpy(buf, s+1, n_chars);
+		buf[n_chars]='\0';
+		return (evutil_inet_pton(AF_INET6,buf,&in6)==1) ? 1 : 0;
+	}
+}
+
+static int
+parse_authority(struct evhttp_uri *uri, char *s, char *eos)
+{
+	char *cp, *port;
+	EVUTIL_ASSERT(eos);
+	if (eos == s) {
+		uri->host = mm_strdup("");
+		return 0;
+	}
+
+	/* Optionally, we start with "userinfo@" */
+
+	cp = strchr(s, '@');
+	if (cp && cp < eos) {
+		if (! userinfo_ok(s,cp))
+			return -1;
+		*cp++ = '\0';
+		uri->userinfo = mm_strdup(s);
+	} else {
+		cp = s;
+	}
+	/* Optionally, we end with ":port" */
+	for (port=eos-1; port >= cp && EVUTIL_ISDIGIT(*port); --port)
+		;
+	if (port >= cp && *port == ':') {
+		if ((uri->port = parse_port(port+1, eos))<0)
+			return -1;
+		eos = port;
+	}
+	/* Now, cp..eos holds the "host" port, which can be an IPv4Address,
+	 * an IP-Literal, or a reg-name */
+	EVUTIL_ASSERT(eos >= cp);
+	if (*cp == '[' && eos >= cp+2 && *(eos-1) == ']') {
+		/* IPv6address, IP-Literal, or junk. */
+		if (! bracket_addr_ok(cp, eos))
+			return -1;
+	} else {
+		/* Make sure the host part is ok. */
+		if (! regname_ok(cp,eos)) /* Match IPv4Address or reg-name */
+			return -1;
+	}
+	uri->host = mm_malloc(eos-cp+1);
+	memcpy(uri->host, cp, eos-cp);
+	uri->host[eos-cp] = '\0';
+	return 0;
+
+}
+
+/* Return the character after the longest prefix of 'cp' that matches...
+ *   *pchar / "/" if allow_qchars is false, or
+ *   *(pchar / "/" / "?") if allow_chars is true.
+ */
+static char *
+end_of_path(char *cp, int allow_qchars)
+{
+	while (*cp) {
+		if (CHAR_IS_UNRESERVED(*cp) ||
+		    strchr(SUBDELIMS, *cp) ||
+		    *cp == ':' || *cp == '@' || *cp == '/')
+			++cp;
+		else if (*cp == '%' && EVUTIL_ISXDIGIT(cp[1]) &&
+		    EVUTIL_ISXDIGIT(cp[2]))
+			cp += 3;
+		else if (*cp == '?' && allow_qchars)
+			++cp;
+		else
+			return cp;
+	}
+	return cp;
+}
+
 struct evhttp_uri *
 evhttp_uri_parse(const char *source_uri)
 {
-	char *readbuf = NULL, *readp = NULL, *token = NULL, *query = NULL, *host = NULL, *port = NULL;
+	char *readbuf = NULL, *readp = NULL, *token = NULL, *query = NULL;
+	char *path = NULL, *fragment = NULL;
+	int got_authority = 0;
 
 	struct evhttp_uri *uri = mm_calloc(1, sizeof(struct evhttp_uri));
 	if (uri == NULL) {
 		event_err(1, "%s: calloc", __func__);
 		goto err;
 	}
+	uri->port = -1;
 
 	readbuf = mm_strdup(source_uri);
 	if (readbuf == NULL) {
@@ -3354,57 +3549,79 @@ evhttp_uri_parse(const char *source_uri)
 	readp = readbuf;
 	token = NULL;
 
-	/* 1. scheme:// */
-	token = strstr(readp, "://");
-	if (!token) {
-		/* unsupported uri */
-		goto err;
-	}
+	/* We try to follow RFC3986 here as much as we can, and match
+	   the productions
 
-	*token = '\0';
-	uri->scheme = mm_strdup(readp);
+	      URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
 
-	readp = token;
-	readp += 3; /* eat :// */
+              relative-ref  = relative-part [ "?" query ] [ "#" fragment ]
 
-	/* 2. query */
-	query = strchr(readp, '/');
-	if (query) {
-		char *fragment = strchr(query, '#');
-		if (fragment) {
-			*fragment++ = '\0'; /* eat '#' */
-			uri->fragment = mm_strdup(fragment);
-		}
+	 */
 
-		uri->query = mm_strdup(query);
-		*query = '\0'; /* eat '/' */
-	}
-
-	/* 3. user:pass@host:port */
-	host = strchr(readp, '@');
-	if (host) {
-		char *pass = 0;
-		/* got user:pass@host:port */
-		*host++ = '\0'; /* eat @ */;
-		pass = strchr(readp, ':');
-		if (pass) {
-			*pass++ = '\0'; /* eat ':' */
-			uri->pass = mm_strdup(pass);
-		}
+	/* 1. scheme: */
+	token = strchr(readp, ':');
+	if (token && scheme_ok(readp,token)) {
+		*token = '\0';
+		uri->scheme = mm_strdup(readp);
+
+		readp = token+1; /* eat : */
+	}
 
-		uri->user = mm_strdup(readp);
-		readp = host;
+	/* 2. Optionally, "//" then an 'authority' part. */
+	if (readp[0]=='/' && readp[1] == '/') {
+		char *authority;
+		readp += 2;
+		authority = readp;
+		path = strchr(readp, '/'); /*XXXX path can be empty; we can
+					    * have a query though */
+		if (!path)
+			path = strchr(readp, '\0');
+		if (parse_authority(uri, authority, path) < 0)
+			goto err;
+		readp = path;
+		got_authority = 1;
 	}
 
-	/* 4. host:port */
-	port = strchr(readp, ':');
-	if (port) {
-		*port++ = '\0'; /* eat ':' */
-		uri->port = atoi(port);
+	/* 3. Query: path-abempty, path-absolute, path-rootless, or path-empty
+	 */
+	path = readp;
+	readp = end_of_path(path, 0);
+
+	/* Query */
+	if (*readp == '?') {
+		*readp = '\0';
+		++readp;
+		query = readp;
+		readp = end_of_path(readp, 1);
+	}
+	/* fragment */
+	if (*readp == '#') {
+		*readp = '\0';
+		++readp;
+		fragment = readp;
+		readp = end_of_path(readp, 1);
+	}
+	if (*readp != '\0') {
+		goto err;
 	}
 
-	/* 5. host */
-	uri->host = mm_strdup(readp);
+	/* If you didn't get an authority, the path can't begin with "//" */
+	if (!got_authority && path[0]=='/' && path[1]=='/')
+		goto err;
+	/* If you did get an authority, the path must begin with "/" or be
+	 * empty. */
+	if (got_authority && path[0] != '/' && path[0] != '\0')
+		goto err;
+
+	if (path)
+		uri->path = mm_strdup(path);
+	else
+		uri->path = mm_strdup("");
+
+	if (query)
+		uri->query = mm_strdup(query);
+	if (fragment)
+		uri->fragment = mm_strdup(fragment);
 
 	mm_free(readbuf);
 
@@ -3426,8 +3643,7 @@ evhttp_uri_free(struct evhttp_uri *uri)
 	}
 
 	_URI_FREE_STR(scheme);
-	_URI_FREE_STR(user);
-	_URI_FREE_STR(pass);
+	_URI_FREE_STR(userinfo);
 	_URI_FREE_STR(host);
 	_URI_FREE_STR(query);
 	_URI_FREE_STR(fragment);
@@ -3442,40 +3658,42 @@ evhttp_uri_join(struct evhttp_uri *uri, char *buf, size_t limit)
 {
 	struct evbuffer *tmp = 0;
 	size_t joined_size = 0;
+	char *output = NULL;
 
 #define _URI_ADD(f)	evbuffer_add(tmp, uri->f, strlen(uri->f))
-	if (!uri || !uri->scheme || !buf || !limit)
+
+	if (!uri || !buf || !limit)
 		return NULL;
 
 	tmp = evbuffer_new();
 	if (!tmp)
 		return NULL;
 
-	_URI_ADD(scheme);
-	evbuffer_add(tmp, "://", 3);
-	if (uri->host && *uri->host) {
-		if (uri->user && *uri->user) {
-			_URI_ADD(user);
-			if (uri->pass && *uri->pass) {
-				evbuffer_add(tmp, ":", 1);
-				_URI_ADD(pass);
-			}
-			evbuffer_add(tmp, "@", 1);
-		}
-
+	if (uri->scheme) {
+		_URI_ADD(scheme);
+		evbuffer_add(tmp, ":", 1);
+	}
+	if (uri->host) {
+		evbuffer_add(tmp, "//", 2);
+		if (uri->userinfo)
+			evbuffer_add_printf(tmp,"%s@", uri->userinfo);
 		_URI_ADD(host);
+		if (uri->port >= 0)
+			evbuffer_add_printf(tmp,":%d", uri->port);
 
-		if (uri->port > 0)
-			evbuffer_add_printf(tmp,":%u", uri->port);
+		if (uri->path && uri->path[0] != '/' && uri->path[0] != '\0')
+			goto err;
 	}
 
-	if (uri->query && *uri->query)
-		_URI_ADD(query);
+	if (uri->path)
+		_URI_ADD(path);
 
-	if (uri->fragment && *uri->fragment) {
-		if (!uri->query || !*uri->query)
-			evbuffer_add(tmp, "/", 1);
+	if (uri->query) {
+		evbuffer_add(tmp, "?", 1);
+		_URI_ADD(query);
+	}
 
+	if (uri->fragment) {
 		evbuffer_add(tmp, "#", 1);
 		_URI_ADD(fragment);
 	}
@@ -3491,8 +3709,10 @@ evhttp_uri_join(struct evhttp_uri *uri, char *buf, size_t limit)
 	}
        	evbuffer_remove(tmp, buf, joined_size);
 
+	output = buf;
+err:
 	evbuffer_free(tmp);
 
-	return (char *)buf;
+	return output;
 #undef _URI_ADD
 }
diff --git a/include/event2/http_struct.h b/include/event2/http_struct.h
index 168f5aca60..264877990d 100644
--- a/include/event2/http_struct.h
+++ b/include/event2/http_struct.h
@@ -123,11 +123,11 @@ struct {
  */
 struct evhttp_uri {
 	char *scheme; /* scheme; e.g http, ftp etc */
-	char *host; /* hostname, or NULL */
-	char *user; /* usename, or NULL */
-	char *pass; /* password, or NULL */
+	char *host; /* hostname, IP address, or NULL */
+	char *userinfo; /* userinfo (typically username:pass), or NULL */
 	int port; /* port, or zero */
-	char *query; /* path + query: e.g. /path/to?param=foo, or NULL */
+	char *path; /* path, or NULL */
+	char *query; /* query, or NULL */
 	char *fragment; /* fragment or NULL */
 };
 
diff --git a/test/regress_http.c b/test/regress_http.c
index 170432195e..fee9a263a3 100644
--- a/test/regress_http.c
+++ b/test/regress_http.c
@@ -1725,15 +1725,23 @@ http_parse_uri_test(void *ptr)
 	tt_want(evhttp_uri_join(0, url_tmp, sizeof(url_tmp)) == NULL);
 	tt_want(evhttp_uri_join(uri, url_tmp, sizeof(url_tmp)) == NULL);
 
-	tt_want(evhttp_uri_parse("mailto:foo@bar") == NULL);
+	uri = evhttp_uri_parse("mailto:foo@bar");
+	tt_want(uri != NULL);
+	tt_want(uri->host == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(!strcmp(uri->scheme, "mailto"));
+	tt_want(!strcmp(uri->path, "foo@bar"));
+	tt_want(uri->query == NULL);
+	tt_want(uri->fragment == NULL);
 
 	uri = evhttp_uri_parse("http://www.test.com/?q=test");
 	tt_want(strcmp(uri->scheme, "http") == 0);
 	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want(strcmp(uri->query, "/?q=test") == 0);
-	tt_want(uri->user == NULL);
-	tt_want(uri->pass == NULL);
-	tt_want(uri->port == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(strcmp(uri->query, "q=test") == 0);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
 	tt_want(uri->fragment == NULL);
 	TT_URI("http://www.test.com/?q=test");
 	evhttp_uri_free(uri);
@@ -1741,45 +1749,140 @@ http_parse_uri_test(void *ptr)
 	uri = evhttp_uri_parse("ftp://www.test.com/?q=test");
 	tt_want(strcmp(uri->scheme, "ftp") == 0);
 	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want(strcmp(uri->query, "/?q=test") == 0);
-	tt_want(uri->user == NULL);
-	tt_want(uri->pass == NULL);
-	tt_want(uri->port == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(strcmp(uri->query, "q=test") == 0);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
 	tt_want(uri->fragment == NULL);
 	TT_URI("ftp://www.test.com/?q=test");
 	evhttp_uri_free(uri);
 
+	uri = evhttp_uri_parse("ftp://[::1]:999/?q=test");
+	tt_want(strcmp(uri->scheme, "ftp") == 0);
+	tt_want(strcmp(uri->host, "[::1]") == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(strcmp(uri->query, "q=test") == 0);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == 999);
+	tt_want(uri->fragment == NULL);
+	TT_URI("ftp://[::1]:999/?q=test");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("ftp://[ff00::127.0.0.1]/?q=test");
+	tt_want(strcmp(uri->scheme, "ftp") == 0);
+	tt_want(strcmp(uri->host, "[ff00::127.0.0.1]") == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(strcmp(uri->query, "q=test") == 0);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(uri->fragment == NULL);
+	TT_URI("ftp://[ff00::127.0.0.1]/?q=test");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("ftp://[v99.not_anytime_soon]/?q=test");
+	tt_want(strcmp(uri->scheme, "ftp") == 0);
+	tt_want(strcmp(uri->host, "[v99.not_anytime_soon]") == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(strcmp(uri->query, "q=test") == 0);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(uri->fragment == NULL);
+	TT_URI("ftp://[v99.not_anytime_soon]/?q=test");
+	evhttp_uri_free(uri);
+
 	uri = evhttp_uri_parse("scheme://user:pass@foo.com:42/?q=test&s=some+thing#fragment");
 	tt_want(strcmp(uri->scheme, "scheme") == 0);
-	tt_want(strcmp(uri->user, "user") == 0);
-	tt_want(strcmp(uri->pass, "pass") == 0);
+	tt_want(strcmp(uri->userinfo, "user:pass") == 0);
 	tt_want(strcmp(uri->host, "foo.com") == 0);
 	tt_want(uri->port == 42);
-	tt_want(strcmp(uri->query, "/?q=test&s=some+thing") == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(strcmp(uri->query, "q=test&s=some+thing") == 0);
 	tt_want(strcmp(uri->fragment, "fragment") == 0);
 	TT_URI("scheme://user:pass@foo.com:42/?q=test&s=some+thing#fragment");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("scheme://user@foo.com/#fragment");
 	tt_want(strcmp(uri->scheme, "scheme") == 0);
-	tt_want(strcmp(uri->user, "user") == 0);
-	tt_want(uri->pass == NULL);
+	tt_want(strcmp(uri->userinfo, "user") == 0);
 	tt_want(strcmp(uri->host, "foo.com") == 0);
-	tt_want(uri->port == 0);
-	tt_want(strcmp(uri->query, "/") == 0);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(uri->query == NULL);
 	tt_want(strcmp(uri->fragment, "fragment") == 0);
 	TT_URI("scheme://user@foo.com/#fragment");
 	evhttp_uri_free(uri);
+
 	uri = evhttp_uri_parse("file:///some/path/to/the/file");
 	tt_want(strcmp(uri->scheme, "file") == 0);
-	tt_want(uri->user == NULL);
-	tt_want(uri->pass == NULL);
+	tt_want(uri->userinfo == NULL);
 	tt_want(strcmp(uri->host, "") == 0);
-	tt_want(uri->port == 0);
-	tt_want(strcmp(uri->query, "/some/path/to/the/file") == 0);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "/some/path/to/the/file") == 0);
+	tt_want(uri->query == NULL);
 	tt_want(uri->fragment == NULL);
 	TT_URI("file:///some/path/to/the/file");
 	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("///some/path/to/the-file");
+	tt_want(uri != NULL);
+	tt_want(uri->scheme == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(strcmp(uri->host, "") == 0);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "/some/path/to/the-file") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(uri->fragment == NULL);
+	TT_URI("///some/path/to/the-file");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("/s:ome/path/to/the-file?q=99#fred");
+	tt_want(uri != NULL);
+	tt_want(uri->scheme == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->host == NULL);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "/s:ome/path/to/the-file") == 0);
+	tt_want(strcmp(uri->query, "q=99") == 0);
+	tt_want(strcmp(uri->fragment, "fred") == 0);
+	TT_URI("/s:ome/path/to/the-file?q=99#fred");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("relative/path/with/co:lon");
+	tt_want(uri != NULL);
+	tt_want(uri->scheme == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->host == NULL);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "relative/path/with/co:lon") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(uri->fragment == NULL);
+	TT_URI("relative/path/with/co:lon");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("bob?q=99&q2=q?33#fr?ed");
+	tt_want(uri != NULL);
+	tt_want(uri->scheme == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->host == NULL);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "bob") == 0);
+	tt_want(strcmp(uri->query, "q=99&q2=q?33") == 0);
+	tt_want(strcmp(uri->fragment, "fr?ed") == 0);
+	TT_URI("bob?q=99&q2=q?33#fr?ed");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("#fr?ed");
+	tt_want(uri != NULL);
+	tt_want(uri->scheme == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->host == NULL);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(strcmp(uri->fragment, "fr?ed") == 0);
+	TT_URI("#fr?ed");
+	evhttp_uri_free(uri);
+
 }
 
 static void

From ad923a11f1e110b3bf03b7179cdcd15ccfcfedd6 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Tue, 19 Oct 2010 12:33:50 -0400
Subject: [PATCH 06/11] Improvements to tinytest_macros.h

First, handle cases where we have %s in a tt_want or tt_assert.

Second, add tt_want_*_op that do a tt_*_op test, but do not exit the
test on failure.

We should push these upstream to tinytest some time.
---
 test/tinytest_macros.h | 39 ++++++++++++++++++++++++++++-----------
 1 file changed, 28 insertions(+), 11 deletions(-)

diff --git a/test/tinytest_macros.h b/test/tinytest_macros.h
index a7fa64a824..032393ccf7 100644
--- a/test/tinytest_macros.h
+++ b/test/tinytest_macros.h
@@ -90,10 +90,10 @@
 	TT_STMT_BEGIN					\
 	if (!(b)) {					\
 		_tinytest_set_test_failed();		\
-		TT_GRIPE((msg));			\
+		TT_GRIPE(("%s",msg));			\
 		fail;					\
 	} else {					\
-		TT_BLATHER((msg));			\
+		TT_BLATHER(("%s",msg));			\
 	}						\
 	TT_STMT_END
 
@@ -111,7 +111,7 @@
 #define tt_assert(b) tt_assert_msg((b), "assert("#b")")
 
 #define tt_assert_test_fmt_type(a,b,str_test,type,test,printf_type,printf_fmt, \
-				setup_block,cleanup_block)		\
+    setup_block,cleanup_block,die_on_fail)				\
 	TT_STMT_BEGIN							\
 	type _val1 = (type)(a);						\
 	type _val2 = (type)(b);						\
@@ -135,33 +135,50 @@
 		cleanup_block;						\
 		if (!_tt_status) {					\
 			_tinytest_set_test_failed();			\
-			TT_EXIT_TEST_FUNCTION;				\
+			die_on_fail ;					\
 		}							\
 	}								\
 	TT_STMT_END
 
-#define tt_assert_test_type(a,b,str_test,type,test,fmt)			\
+#define tt_assert_test_type(a,b,str_test,type,test,fmt,die_on_fail)	\
 	tt_assert_test_fmt_type(a,b,str_test,type,test,type,fmt,	\
-				{_print=_value;},{})
+	    {_print=_value;},{},die_on_fail)
 
 /* Helper: assert that a op b, when cast to type.  Format the values with
  * printf format fmt on failure. */
 #define tt_assert_op_type(a,op,b,type,fmt)				\
-	tt_assert_test_type(a,b,#a" "#op" "#b,type,(_val1 op _val2),fmt)
+	tt_assert_test_type(a,b,#a" "#op" "#b,type,(_val1 op _val2),fmt, \
+	    TT_EXIT_TEST_FUNCTION)
 
 #define tt_int_op(a,op,b)			\
-	tt_assert_test_type(a,b,#a" "#op" "#b,long,(_val1 op _val2),"%ld")
+	tt_assert_test_type(a,b,#a" "#op" "#b,long,(_val1 op _val2), \
+	    "%ld",TT_EXIT_TEST_FUNCTION)
 
 #define tt_uint_op(a,op,b)						\
 	tt_assert_test_type(a,b,#a" "#op" "#b,unsigned long,		\
-			    (_val1 op _val2),"%lu")
+	    (_val1 op _val2),"%lu",TT_EXIT_TEST_FUNCTION)
 
 #define tt_ptr_op(a,op,b)						\
 	tt_assert_test_type(a,b,#a" "#op" "#b,void*,			\
-			    (_val1 op _val2),"%p")
+	    (_val1 op _val2),"%p",TT_EXIT_TEST_FUNCTION)
 
 #define tt_str_op(a,op,b)						\
 	tt_assert_test_type(a,b,#a" "#op" "#b,const char *,		\
-			    (strcmp(_val1,_val2) op 0),"<%s>")
+	    (strcmp(_val1,_val2) op 0),"<%s>",TT_EXIT_TEST_FUNCTION)
+
+#define tt_want_int_op(a,op,b)						\
+	tt_assert_test_type(a,b,#a" "#op" "#b,long,(_val1 op _val2),"%ld",(void)0)
+
+#define tt_want_uint_op(a,op,b)						\
+	tt_assert_test_type(a,b,#a" "#op" "#b,unsigned long,		\
+	    (_val1 op _val2),"%lu",(void)0)
+
+#define tt_want_ptr_op(a,op,b)						\
+	tt_assert_test_type(a,b,#a" "#op" "#b,void*,			\
+	    (_val1 op _val2),"%p",(void)0)
+
+#define tt_want_str_op(a,op,b)						\
+	tt_assert_test_type(a,b,#a" "#op" "#b,const char *,		\
+	    (strcmp(_val1,_val2) op 0),"<%s>",(void)0)
 
 #endif

From a5a76e689c185fb9f6f430fd8e989645b627bdb3 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Tue, 19 Oct 2010 12:35:50 -0400
Subject: [PATCH 07/11] Add a huge pile of tests for the new URI functions, and
 make them pass.

---
 http.c              |  52 +++++++++++---
 test/regress_http.c | 162 +++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 193 insertions(+), 21 deletions(-)

diff --git a/http.c b/http.c
index 883c675f5c..2bf48498a8 100644
--- a/http.c
+++ b/http.c
@@ -3399,12 +3399,12 @@ static int
 parse_port(const char *s, const char *eos)
 {
 	int portnum = 0;
-	if (s == eos)
-		return 0; /* The RFC allows an empty port. */
 	while (s < eos) {
 		if (! EVUTIL_ISDIGIT(*s))
 			return -1;
 		portnum = (portnum * 10) + (*s - '0');
+		if (portnum < 0)
+			return -1;
 		++s;
 	}
 	return portnum;
@@ -3480,7 +3480,10 @@ parse_authority(struct evhttp_uri *uri, char *s, char *eos)
 	for (port=eos-1; port >= cp && EVUTIL_ISDIGIT(*port); --port)
 		;
 	if (port >= cp && *port == ':') {
-		if ((uri->port = parse_port(port+1, eos))<0)
+		if (port+1 == eos) /* Leave port unspecified; the RFC allows a
+				    * nil port */
+			uri->port = -1;
+		else if ((uri->port = parse_port(port+1, eos))<0)
 			return -1;
 		eos = port;
 	}
@@ -3503,6 +3506,17 @@ parse_authority(struct evhttp_uri *uri, char *s, char *eos)
 
 }
 
+static char *
+end_of_authority(char *cp)
+{
+	while (*cp) {
+		if (*cp == '?' || *cp == '#' || *cp == '/')
+			return cp;
+		++cp;
+	}
+	return cp;
+}
+
 /* Return the character after the longest prefix of 'cp' that matches...
  *   *pchar / "/" if allow_qchars is false, or
  *   *(pchar / "/" / "?") if allow_chars is true.
@@ -3526,6 +3540,19 @@ end_of_path(char *cp, int allow_qchars)
 	return cp;
 }
 
+static int
+path_matches_noscheme(const char *cp)
+{
+	while (*cp) {
+		if (*cp == ':')
+			return 0;
+		else if (*cp == '/')
+			return 1;
+		++cp;
+	}
+	return 1;
+}
+
 struct evhttp_uri *
 evhttp_uri_parse(const char *source_uri)
 {
@@ -3572,10 +3599,7 @@ evhttp_uri_parse(const char *source_uri)
 		char *authority;
 		readp += 2;
 		authority = readp;
-		path = strchr(readp, '/'); /*XXXX path can be empty; we can
-					    * have a query though */
-		if (!path)
-			path = strchr(readp, '\0');
+		path = end_of_authority(readp);
 		if (parse_authority(uri, authority, path) < 0)
 			goto err;
 		readp = path;
@@ -3605,6 +3629,8 @@ evhttp_uri_parse(const char *source_uri)
 		goto err;
 	}
 
+	/* These next two cases may be unreachable; I'm leaving them
+	 * in to be defensive. */
 	/* If you didn't get an authority, the path can't begin with "//" */
 	if (!got_authority && path[0]=='/' && path[1]=='/')
 		goto err;
@@ -3612,11 +3638,15 @@ evhttp_uri_parse(const char *source_uri)
 	 * empty. */
 	if (got_authority && path[0] != '/' && path[0] != '\0')
 		goto err;
+	/* (End of maybe-unreachable cases) */
 
-	if (path)
-		uri->path = mm_strdup(path);
-	else
-		uri->path = mm_strdup("");
+	/* If there was no scheme, the first part of the path (if any) must
+	 * have no colon in it. */
+	if (! uri->scheme && !path_matches_noscheme(path))
+		goto err;
+
+	EVUTIL_ASSERT(path);
+	uri->path = mm_strdup(path);
 
 	if (query)
 		uri->query = mm_strdup(query);
diff --git a/test/regress_http.c b/test/regress_http.c
index fee9a263a3..34d2781e1e 100644
--- a/test/regress_http.c
+++ b/test/regress_http.c
@@ -1717,13 +1717,66 @@ http_parse_uri_test(void *ptr)
 	char *ret = evhttp_uri_join(uri, url_tmp, sizeof(url_tmp));	\
 	tt_want(ret != NULL);						\
 	tt_want(ret == url_tmp);					\
-	tt_want(strcmp(ret, want) == 0);				\
+	if (strcmp(ret,want) != 0)					\
+		TT_FAIL(("\"%s\" != \"%s\"",ret,want));			\
 	} while(0)
 
-	tt_want(evhttp_uri_join(0, 0, 0) == NULL);
-	tt_want(evhttp_uri_join(0, url_tmp, 0) == NULL);
-	tt_want(evhttp_uri_join(0, url_tmp, sizeof(url_tmp)) == NULL);
+	tt_want(evhttp_uri_join(NULL, 0, 0) == NULL);
+	tt_want(evhttp_uri_join(NULL, url_tmp, 0) == NULL);
+	tt_want(evhttp_uri_join(NULL, url_tmp, sizeof(url_tmp)) == NULL);
+
+	/* bad URIs: parsing */
+#define BAD(s) do {							\
+		if (evhttp_uri_parse(s) != NULL)			\
+			TT_FAIL(("Expected error parsing \"%s\"",s));	\
+	} while(0)
+	BAD("http://www.test.com/ why hello");
+	BAD("http://www.test.com/why-hello\x01");
+	BAD("http://www.test.com/why-hello?\x01");
+	BAD("http://www.test.com/why-hello#\x01");
+	BAD("http://www.\x01.test.com/why-hello");
+	BAD("http://www.%7test.com/why-hello");
+	BAD("http://www.test.com/why-hell%7o");
+	BAD("h%3ttp://www.test.com/why-hello");
+	BAD("http://www.test.com/why-hello%7");
+	BAD("http://www.test.com/why-hell%7o");
+	BAD("http://www.test.com/foo?ba%r");
+	BAD("http://www.test.com/foo#ba%r");
+	BAD("99:99/foo");
+	BAD("http://www.test.com:999x/");
+	BAD("http://www.test.com:x/");
+	BAD("http://[hello-there]/");
+	BAD("http://[::1]]/");
+	BAD("http://[::1/");
+	BAD("http://[foob/");
+	BAD("http://[/");
+	BAD("http://[ffff:ffff:ffff:ffff:Ffff:ffff:ffff:"
+	            "ffff:ffff:ffff:ffff:ffff:ffff:ffff]/");
+	BAD("http://[vX.foo]/");
+	BAD("http://[vX.foo]/");
+	BAD("http://[v.foo]/");
+	BAD("http://[v5.fo%o]/");
+	BAD("http://[v5X]/");
+	BAD("http://[v5]/");
+	BAD("http://[]/");
+	BAD("http://f\x01red@www.example.com/");
+	BAD("http://f%0red@www.example.com/");
+	BAD("http://www.example.com:9999999999999999999999999999999999999/");
+	BAD("http://www.example.com:hihi/");
+	BAD("://www.example.com/");
+
+	/* bad URIs: joining */
+	uri = calloc(sizeof(struct evhttp_uri),1);
+	uri->host = (char*)"www.example.com";
+	tt_want(evhttp_uri_join(uri, url_tmp, sizeof(url_tmp)) != NULL);
+	/* not enough space: */
+	tt_want(evhttp_uri_join(uri, url_tmp, 3) == NULL);
+	/* host is set, but path doesn't start with "/": */
+	uri->path = (char*)"hi_mom";
 	tt_want(evhttp_uri_join(uri, url_tmp, sizeof(url_tmp)) == NULL);
+	tt_want(evhttp_uri_join(uri, NULL, sizeof(url_tmp))==NULL);
+	tt_want(evhttp_uri_join(uri, url_tmp, 0)==NULL);
+	free(uri);
 
 	uri = evhttp_uri_parse("mailto:foo@bar");
 	tt_want(uri != NULL);
@@ -1734,16 +1787,95 @@ http_parse_uri_test(void *ptr)
 	tt_want(!strcmp(uri->path, "foo@bar"));
 	tt_want(uri->query == NULL);
 	tt_want(uri->fragment == NULL);
+	TT_URI("mailto:foo@bar");
+	evhttp_uri_free(uri);
 
-	uri = evhttp_uri_parse("http://www.test.com/?q=test");
+	uri = evhttp_uri_parse("http://www.test.com/?q=t%33est");
 	tt_want(strcmp(uri->scheme, "http") == 0);
 	tt_want(strcmp(uri->host, "www.test.com") == 0);
 	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(strcmp(uri->query, "q=t%33est") == 0);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(uri->fragment == NULL);
+	TT_URI("http://www.test.com/?q=t%33est");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("http://%77ww.test.com");
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "%77ww.test.com") == 0);
+	tt_want(strcmp(uri->path, "") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(uri->fragment == NULL);
+	TT_URI("http://%77ww.test.com");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("http://www.test.com?q=test");
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "www.test.com") == 0);
+	tt_want(strcmp(uri->path, "") == 0);
 	tt_want(strcmp(uri->query, "q=test") == 0);
 	tt_want(uri->userinfo == NULL);
 	tt_want(uri->port == -1);
 	tt_want(uri->fragment == NULL);
-	TT_URI("http://www.test.com/?q=test");
+	TT_URI("http://www.test.com?q=test");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("http://www.test.com#fragment");
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "www.test.com") == 0);
+	tt_want(strcmp(uri->path, "") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want_str_op(uri->fragment, ==, "fragment");
+	TT_URI("http://www.test.com#fragment");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("http://8000/");
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "8000") == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(uri->fragment == NULL);
+	TT_URI("http://8000/");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("http://:8000/");
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "") == 0);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == 8000);
+	tt_want(uri->fragment == NULL);
+	TT_URI("http://:8000/");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("http://www.test.com:/"); /* empty port */
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "www.test.com") == 0);
+	tt_want_str_op(uri->path, ==, "/");
+	tt_want(uri->query == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(uri->fragment == NULL);
+	TT_URI("http://www.test.com/");
+	evhttp_uri_free(uri);
+
+	uri = evhttp_uri_parse("http://www.test.com:"); /* empty port 2 */
+	tt_want(strcmp(uri->scheme, "http") == 0);
+	tt_want(strcmp(uri->host, "www.test.com") == 0);
+	tt_want(strcmp(uri->path, "") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(uri->userinfo == NULL);
+	tt_want(uri->port == -1);
+	tt_want(uri->fragment == NULL);
+	TT_URI("http://www.test.com");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("ftp://www.test.com/?q=test");
@@ -1779,15 +1911,15 @@ http_parse_uri_test(void *ptr)
 	TT_URI("ftp://[ff00::127.0.0.1]/?q=test");
 	evhttp_uri_free(uri);
 
-	uri = evhttp_uri_parse("ftp://[v99.not_anytime_soon]/?q=test");
+	uri = evhttp_uri_parse("ftp://[v99.not_(any:time)_soon]/?q=test");
 	tt_want(strcmp(uri->scheme, "ftp") == 0);
-	tt_want(strcmp(uri->host, "[v99.not_anytime_soon]") == 0);
+	tt_want(strcmp(uri->host, "[v99.not_(any:time)_soon]") == 0);
 	tt_want(strcmp(uri->path, "/") == 0);
 	tt_want(strcmp(uri->query, "q=test") == 0);
 	tt_want(uri->userinfo == NULL);
 	tt_want(uri->port == -1);
 	tt_want(uri->fragment == NULL);
-	TT_URI("ftp://[v99.not_anytime_soon]/?q=test");
+	TT_URI("ftp://[v99.not_(any:time)_soon]/?q=test");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("scheme://user:pass@foo.com:42/?q=test&s=some+thing#fragment");
@@ -1812,6 +1944,17 @@ http_parse_uri_test(void *ptr)
 	TT_URI("scheme://user@foo.com/#fragment");
 	evhttp_uri_free(uri);
 
+	uri = evhttp_uri_parse("scheme://%75ser@foo.com/#frag@ment");
+	tt_want(strcmp(uri->scheme, "scheme") == 0);
+	tt_want(strcmp(uri->userinfo, "%75ser") == 0);
+	tt_want(strcmp(uri->host, "foo.com") == 0);
+	tt_want(uri->port == -1);
+	tt_want(strcmp(uri->path, "/") == 0);
+	tt_want(uri->query == NULL);
+	tt_want(strcmp(uri->fragment, "frag@ment") == 0);
+	TT_URI("scheme://%75ser@foo.com/#frag@ment");
+	evhttp_uri_free(uri);
+
 	uri = evhttp_uri_parse("file:///some/path/to/the/file");
 	tt_want(strcmp(uri->scheme, "file") == 0);
 	tt_want(uri->userinfo == NULL);
@@ -1882,7 +2025,6 @@ http_parse_uri_test(void *ptr)
 	tt_want(strcmp(uri->fragment, "fr?ed") == 0);
 	TT_URI("#fr?ed");
 	evhttp_uri_free(uri);
-
 }
 
 static void

From 3a33462827ec1c31083c297a16aa0287c92a8bfe Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Tue, 19 Oct 2010 13:02:18 -0400
Subject: [PATCH 08/11] Document behavior of URI parsing more thoroughly.

Also, move evhttp_uri struct into http.h, since it is part of the API.
---
 include/event2/http.h        | 65 ++++++++++++++++++++++++++++--------
 include/event2/http_struct.h | 14 --------
 2 files changed, 52 insertions(+), 27 deletions(-)

diff --git a/include/event2/http.h b/include/event2/http.h
index 655ff25494..e2c6942e3c 100644
--- a/include/event2/http.h
+++ b/include/event2/http.h
@@ -607,30 +607,69 @@ int evhttp_parse_query__checked_20(const char *uri, struct evkeyvalq *headers);
  */
 char *evhttp_htmlescape(const char *html);
 
-struct evhttp_uri;
-
 /**
-   Helper function to parse out uri.
-
-   Parsing a uri like
-
-      scheme://[[user[:pass]@]foo.com[:port]]/[path][?q=test&s=some+thing][#fragment]
+ * A structure to hold a parsed URI.
+ */
+struct evhttp_uri {
+	char *scheme; /* scheme; e.g http, ftp etc */
+	char *host; /* hostname, IP address, or NULL */
+	char *userinfo; /* userinfo (typically username:pass), or NULL */
+	int port; /* port, or zero */
+	char *path; /* path, or "". */
+	char *query; /* query, or NULL */
+	char *fragment; /* fragment or NULL */
+};
 
-   @param source_uri the request URI
-   @return uri container to hold parsed data, or NULL if there is error
-   @see evhttp_uri_free()
+/**
+ * Helper function to parse a URI-Reference as specified by RFC3986.
+ *
+ * This function matches the URI-Reference production from RFC3986,
+ * which includes both URIs like
+ *
+ *    scheme://[[userinfo]@]foo.com[:port]]/[path][?query][#fragment]
+ *
+ *  and relative-refs like
+ *
+ *    [path][?query][#fragment]
+ *
+ * Any optional elements portions not present in the original URI are
+ * left set to NULL in the resulting evhttp_uri.  If no port is
+ * specified, the port is set to -1.
+ *
+ * Note that no decoding is performed on percent-escaped characters in
+ * the string; if you want to parse them, use evhttp_uridecode as
+ * appropriate.
+ *
+ * Note also that most URI schemes will have additional constraints that
+ * this function does not know about, and cannot check.  For example,
+ * mailto://www.example.com/cgi-bin/fortune.pl is not a reasonable
+ * mailto url, http://www.example.com:99999/ is not a reasonable HTTP
+ * URL, and ftp:username@example.com is not a reasonable FTP URL.
+ * Nevertheless, all of these URLs conform to RFC3986, and this function
+ * accepts all of them as valid.
+ *
+ * @param source_uri the request URI
+ * @return uri container to hold parsed data, or NULL if there is error
+ * @see evhttp_uri_free()
  */
 struct evhttp_uri *evhttp_uri_parse(const char *source_uri);
 
 /**
- * Free the memory allocated for the uri and parsed data
+ * Free all memory allocated for a parsed uri.  Only use this for URIs
+ * generated by evhttp_uri_parse.
+ *
  * @param uri container with parsed data
-   @see evhttp_uri_parse()
+ * @see evhttp_uri_parse()
  */
 void evhttp_uri_free(struct evhttp_uri *uri);
 
 /**
- * Join together the uri parts from parsed data
+ * Join together the uri parts from parsed data to form a URI-Reference.
+ *
+ * Note that no escaping of reserved characters is done on the members
+ * of the evhttp_uri, so the generated string might not be a valid URI
+ * unless the members of evhttp_uri are themselves valid.
+ *
  * @param uri container with parsed data
  * @param buf destination buffer
  * @param limit destination buffer size
diff --git a/include/event2/http_struct.h b/include/event2/http_struct.h
index 264877990d..a3664e0448 100644
--- a/include/event2/http_struct.h
+++ b/include/event2/http_struct.h
@@ -118,20 +118,6 @@ struct {
 	void (*chunk_cb)(struct evhttp_request *, void *);
 };
 
-/**
- * structure to hold parsed uri
- */
-struct evhttp_uri {
-	char *scheme; /* scheme; e.g http, ftp etc */
-	char *host; /* hostname, IP address, or NULL */
-	char *userinfo; /* userinfo (typically username:pass), or NULL */
-	int port; /* port, or zero */
-	char *path; /* path, or NULL */
-	char *query; /* query, or NULL */
-	char *fragment; /* fragment or NULL */
-};
-
-
 #ifdef __cplusplus
 }
 #endif

From 2075fbcff0d8a9e06e2c3287b1cf4d400394c285 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Tue, 19 Oct 2010 13:15:48 -0400
Subject: [PATCH 09/11] Add evhttp_parse_query_str to be used with
 evhttp_uri_parse.

The old evhttp_parse_query() doesn't work well with struct
evhttp_uri.query, since it expects to get whole URIs, rather than
just the query portion.
---
 http.c                | 42 ++++++++++++++++++++++++++----------------
 include/event2/http.h | 34 +++++++++++++++++++++++++++++-----
 2 files changed, 55 insertions(+), 21 deletions(-)

diff --git a/http.c b/http.c
index 2bf48498a8..e4b8e31a3a 100644
--- a/http.c
+++ b/http.c
@@ -2449,30 +2449,39 @@ evhttp_uridecode(const char *uri, int decode_plus, size_t *size_out)
  */
 
 int
-evhttp_parse_query__checked_20(const char *uri, struct evkeyvalq *headers)
+evhttp_parse_query__checked_20(const char *str, struct evkeyvalq *headers,
+    int is_whole_uri)
 {
-	char *line;
+	char *line=NULL;
 	char *argument;
 	char *p;
+	const char *query_part;
 	int result = -1;
+	struct evhttp_uri *uri=NULL;
 
 	TAILQ_INIT(headers);
 
+	if (is_whole_uri) {
+		uri = evhttp_uri_parse(str);
+		if (!uri)
+			goto error;
+		query_part = uri->query;
+	} else {
+		query_part = str;
+	}
+
 	/* No arguments - we are done */
-	if (strchr(uri, '?') == NULL)
-		return 0;
+	if (!query_part || !strlen(query_part)) {
+		result = 0;
+		goto done;
+	}
 
-	if ((line = mm_strdup(uri)) == NULL) {
+	if ((line = mm_strdup(query_part)) == NULL) {
 		event_warn("%s: strdup", __func__);
-		return -1;
+		goto error;
 	}
 
-	argument = line;
-
-	/* We already know that there has to be a ? */
-	strsep(&argument, "?");
-
-	p = argument;
+	p = argument = line;
 	while (p != NULL && *p != '\0') {
 		char *key, *value, *decoded_value;
 		argument = strsep(&p, "&");
@@ -2499,7 +2508,10 @@ evhttp_parse_query__checked_20(const char *uri, struct evkeyvalq *headers)
 error:
 	evhttp_clear_headers(headers);
 done:
-	mm_free(line);
+	if (line)
+		mm_free(line);
+	if (uri)
+		evhttp_uri_free(uri);
 	return result;
 }
 
@@ -2512,11 +2524,9 @@ void evhttp_parse_query(const char *uri, struct evkeyvalq *headers);
 void
 evhttp_parse_query(const char *uri, struct evkeyvalq *headers)
 {
-	evhttp_parse_query__checked_20(uri, headers);
+	evhttp_parse_query__checked_20(uri, headers, 1);
 }
 
-
-
 static struct evhttp_cb *
 evhttp_dispatch_callback(struct httpcbq *callbacks, struct evhttp_request *req)
 {
diff --git a/include/event2/http.h b/include/event2/http.h
index e2c6942e3c..d882061c73 100644
--- a/include/event2/http.h
+++ b/include/event2/http.h
@@ -573,7 +573,7 @@ char *evhttp_uridecode(const char *uri, int decode_plus,
 /**
    Helper function to parse out arguments in a query.
 
-   Parsing a uri like
+   Parsing a URI like
 
       http://foo.com/?q=test&s=some+thing
 
@@ -582,17 +582,41 @@ char *evhttp_uridecode(const char *uri, int decode_plus,
    The first entry is: key="q", value="test"
    The second entry is: key="s", value="some thing"
 
+   @deprecated This function is deprecated as of Libevent 2.0.9.  Use
+     evhttp_uri_parse and evhttp_parse_query_str instead.
+
    @param uri the request URI
    @param headers the head of the evkeyval queue
    @return 0 on success, -1 on failure
  */
 #define evhttp_parse_query(uri, headers) \
-	evhttp_parse_query__checked_20((uri), (headers))
+	evhttp_parse_query__checked_20((uri), (headers), 1)
+
+/**
+   Helper function to parse out arguments from the query portion of an
+   HTTP URI.
+
+   Parsing a query string like
+
+     q=test&s=some+thing
+
+   will result in two entries in the key value queue.
+
+   The first entry is: key="q", value="test"
+   The second entry is: key="s", value="some thing"
+
+   @param query_parse the query portion of the URI
+   @param headers the head of the evkeyval queue
+   @return 0 on success, -1 on failure
+ */
+#define evhttp_parse_query_str(query, headers)			\
+	evhttp_parse_query__checked_20((uri), (headers), 0)
 
 /* Do not call this function directly; it is a temporary alias introduced
  * to avoid changing the old signature for evhttp_parse_query
  */
-int evhttp_parse_query__checked_20(const char *uri, struct evkeyvalq *headers);
+int evhttp_parse_query__checked_20(const char *uri, struct evkeyvalq *headers,
+    int is_whole_url);
 
 /**
  * Escape HTML character entities in a string.
@@ -637,8 +661,8 @@ struct evhttp_uri {
  * specified, the port is set to -1.
  *
  * Note that no decoding is performed on percent-escaped characters in
- * the string; if you want to parse them, use evhttp_uridecode as
- * appropriate.
+ * the string; if you want to parse them, use evhttp_uridecode or
+ * evhttp_parse_query_str as appropriate.
  *
  * Note also that most URI schemes will have additional constraints that
  * this function does not know about, and cannot check.  For example,

From 45f6869c753218a72dd44660a880237aa8daa386 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Thu, 21 Oct 2010 14:41:12 -0400
Subject: [PATCH 10/11] Make evhttp_uri non-public, and give it accessor
 functions.

---
 http.c                | 140 ++++++++++++++++++-
 include/event2/http.h |  75 ++++++++--
 test/regress_http.c   | 317 +++++++++++++++++++++---------------------
 3 files changed, 360 insertions(+), 172 deletions(-)

diff --git a/http.c b/http.c
index e4b8e31a3a..321920072b 100644
--- a/http.c
+++ b/http.c
@@ -2465,7 +2465,7 @@ evhttp_parse_query__checked_20(const char *str, struct evkeyvalq *headers,
 		uri = evhttp_uri_parse(str);
 		if (!uri)
 			goto error;
-		query_part = uri->query;
+		query_part = evhttp_uri_get_query(uri);
 	} else {
 		query_part = str;
 	}
@@ -3347,6 +3347,25 @@ bind_socket(const char *address, ev_uint16_t port, int reuse)
 	return (fd);
 }
 
+struct evhttp_uri {
+	char *scheme; /* scheme; e.g http, ftp etc */
+	char *userinfo; /* userinfo (typically username:pass), or NULL */
+	char *host; /* hostname, IP address, or NULL */
+	int port; /* port, or zero */
+	char *path; /* path, or "". */
+	char *query; /* query, or NULL */
+	char *fragment; /* fragment or NULL */
+};
+
+struct evhttp_uri *
+evhttp_uri_new(void)
+{
+	struct evhttp_uri *uri = mm_calloc(sizeof(struct evhttp_uri), 1);
+	if (uri)
+		uri->port = -1;
+	return uri;
+}
+
 /* Return true of the string starting at s and ending immediately before eos
  * is a valid URI scheme according to RFC3986
  */
@@ -3689,7 +3708,6 @@ evhttp_uri_free(struct evhttp_uri *uri)
 	_URI_FREE_STR(fragment);
 
 	mm_free(uri);
-
 #undef _URI_FREE_STR
 }
 
@@ -3756,3 +3774,121 @@ evhttp_uri_join(struct evhttp_uri *uri, char *buf, size_t limit)
 	return output;
 #undef _URI_ADD
 }
+
+const char *
+evhttp_uri_get_scheme(const struct evhttp_uri *uri)
+{
+	return uri->scheme;
+}
+const char *
+evhttp_uri_get_userinfo(const struct evhttp_uri *uri)
+{
+	return uri->userinfo;
+}
+const char *
+evhttp_uri_get_host(const struct evhttp_uri *uri)
+{
+	return uri->host;
+}
+int
+evhttp_uri_get_port(const struct evhttp_uri *uri)
+{
+	return uri->port;
+}
+const char *
+evhttp_uri_get_path(const struct evhttp_uri *uri)
+{
+	return uri->path;
+}
+const char *
+evhttp_uri_get_query(const struct evhttp_uri *uri)
+{
+	return uri->query;
+}
+const char *
+evhttp_uri_get_fragment(const struct evhttp_uri *uri)
+{
+	return uri->fragment;
+}
+
+#define _URI_SET_STR(f) do {					\
+	if (uri->f)						\
+		mm_free(uri->f);				\
+	if (f) {						\
+		if ((uri->f = mm_strdup(f)) == NULL) {		\
+			event_warn("%s: strdup()", __func__);	\
+			return -1;				\
+		}						\
+	} else {						\
+		uri->f = NULL;					\
+	}							\
+	} while(0)
+
+int
+evhttp_uri_set_scheme(struct evhttp_uri *uri, const char *scheme)
+{
+	if (scheme && !scheme_ok(scheme, scheme+strlen(scheme)))
+		return -1;
+
+	_URI_SET_STR(scheme);
+	return 0;
+}
+int
+evhttp_uri_set_userinfo(struct evhttp_uri *uri, const char *userinfo)
+{
+	if (userinfo && !userinfo_ok(userinfo, userinfo+strlen(userinfo)))
+		return -1;
+	_URI_SET_STR(userinfo);
+	return 0;
+}
+int
+evhttp_uri_set_host(struct evhttp_uri *uri, const char *host)
+{
+	if (host) {
+		if (host[0] == '[') {
+			if (! bracket_addr_ok(host, host+strlen(host)))
+				return -1;
+		} else {
+			if (! regname_ok(host, host+strlen(host)))
+				return -1;
+		}
+	}
+
+	_URI_SET_STR(host);
+	return 0;
+}
+int
+evhttp_uri_set_port(struct evhttp_uri *uri, int port)
+{
+	if (port < -1)
+		return -1;
+	uri->port = port;
+	return 0;
+}
+#define end_of_cpath(cp,aq) ((const char*)(end_of_path(((char*)(cp)), (aq))))
+
+int
+evhttp_uri_set_path(struct evhttp_uri *uri, const char *path)
+{
+	if (path && end_of_cpath(path, 0) != path+strlen(path))
+		return -1;
+
+	_URI_SET_STR(path);
+	return 0;
+}
+int
+evhttp_uri_set_query(struct evhttp_uri *uri, const char *query)
+{
+	if (query && end_of_cpath(query, 1) != query+strlen(query))
+		return -1;
+	_URI_SET_STR(query);
+	return 0;
+}
+int
+evhttp_uri_set_fragment(struct evhttp_uri *uri, const char *fragment)
+{
+	if (fragment && end_of_cpath(fragment, 1) != fragment+strlen(fragment))
+		return -1;
+	_URI_SET_STR(fragment);
+	return 0;
+}
diff --git a/include/event2/http.h b/include/event2/http.h
index d882061c73..3fdd1f9883 100644
--- a/include/event2/http.h
+++ b/include/event2/http.h
@@ -632,17 +632,70 @@ int evhttp_parse_query__checked_20(const char *uri, struct evkeyvalq *headers,
 char *evhttp_htmlescape(const char *html);
 
 /**
- * A structure to hold a parsed URI.
- */
-struct evhttp_uri {
-	char *scheme; /* scheme; e.g http, ftp etc */
-	char *host; /* hostname, IP address, or NULL */
-	char *userinfo; /* userinfo (typically username:pass), or NULL */
-	int port; /* port, or zero */
-	char *path; /* path, or "". */
-	char *query; /* query, or NULL */
-	char *fragment; /* fragment or NULL */
-};
+ * A structure to hold a parsed URI or Relative-Ref conforming to RFC3986.
+ */
+struct evhttp_uri;
+
+/**
+ * Return a new empty evhttp_uri with no fields set.
+ */
+struct evhttp_uri *evhttp_uri_new(void);
+
+/** Return the scheme of an evhttp_uri, or NULL if there is no scheme has
+ * been set and the evhttp_uri contains a Relative-Ref. */
+const char *evhttp_uri_get_scheme(const struct evhttp_uri *uri);
+/**
+ * Return the userinfo part of an evhttp_uri, or NULL if it has no userinfo
+ * set.
+ */
+const char *evhttp_uri_get_userinfo(const struct evhttp_uri *uri);
+/**
+ * Return the host part of an evhttp_uri, or NULL if it has no host set.
+ * The host may either be a regular hostname (conforming to the RFC 3986
+ * "regname" production), or an IPv4 address, or the empty string, or a
+ * bracketed IPv6 address, or a bracketed 'IP-Future' address.
+ *
+ * Note that having a NULL host means that the URI has no authority
+ * section, but having an empty-string host means that the URI has an
+ * authority section with no host part.  For example,
+ * "mailto:user@example.com" has a host of NULL, but "file:///etc/motd"
+ * has a host of "".
+ */
+const char *evhttp_uri_get_host(const struct evhttp_uri *uri);
+/** Return the port part of an evhttp_uri, or -1 if there is no port set. */
+int evhttp_uri_get_port(const struct evhttp_uri *uri);
+/** Return the path part of an evhttp_uri, or NULL if it has no path set */
+const char *evhttp_uri_get_path(const struct evhttp_uri *uri);
+/** Return the query part of an evhttp_uri (excluding the leading "?"), or
+ * NULL if it has no query set */
+const char *evhttp_uri_get_query(const struct evhttp_uri *uri);
+/** Return the fragment part of an evhttp_uri (excluding the leading "#"),
+ * or NULL if it has no fragment set */
+const char *evhttp_uri_get_fragment(const struct evhttp_uri *uri);
+
+/** Set the scheme of an evhttp_uri, or clear the scheme if scheme==NULL.
+ * Returns 0 on success, -1 if scheme is not well-formed. */
+int evhttp_uri_set_scheme(struct evhttp_uri *uri, const char *scheme);
+/** Set the userinfo of an evhttp_uri, or clear the userinfo if userinfo==NULL.
+ * Returns 0 on success, -1 if userinfo is not well-formed. */
+int evhttp_uri_set_userinfo(struct evhttp_uri *uri, const char *userinfo);
+/** Set the host of an evhttp_uri, or clear the host if host==NULL.
+ * Returns 0 on success, -1 if host is not well-formed. */
+int evhttp_uri_set_host(struct evhttp_uri *uri, const char *host);
+/** Set the port of an evhttp_uri, or clear the port if port==-1.
+ * Returns 0 on success, -1 if port is not well-formed. */
+int evhttp_uri_set_port(struct evhttp_uri *uri, int port);
+/** Set the path of an evhttp_uri, or clear the path if path==NULL.
+ * Returns 0 on success, -1 if path is not well-formed. */
+int evhttp_uri_set_path(struct evhttp_uri *uri, const char *path);
+/** Set the query of an evhttp_uri, or clear the query if query==NULL.
+ * The query should not include a leading "?".
+ * Returns 0 on success, -1 if query is not well-formed. */
+int evhttp_uri_set_query(struct evhttp_uri *uri, const char *query);
+/** Set the fragment of an evhttp_uri, or clear the fragment if fragment==NULL.
+ * The fragment should not include a leading "#".
+ * Returns 0 on success, -1 if fragment is not well-formed. */
+int evhttp_uri_set_fragment(struct evhttp_uri *uri, const char *fragment);
 
 /**
  * Helper function to parse a URI-Reference as specified by RFC3986.
diff --git a/test/regress_http.c b/test/regress_http.c
index 34d2781e1e..f90e98c38e 100644
--- a/test/regress_http.c
+++ b/test/regress_http.c
@@ -1766,263 +1766,262 @@ http_parse_uri_test(void *ptr)
 	BAD("://www.example.com/");
 
 	/* bad URIs: joining */
-	uri = calloc(sizeof(struct evhttp_uri),1);
-	uri->host = (char*)"www.example.com";
+	uri = evhttp_uri_new();
+	tt_want(0==evhttp_uri_set_host(uri, "www.example.com"));
 	tt_want(evhttp_uri_join(uri, url_tmp, sizeof(url_tmp)) != NULL);
 	/* not enough space: */
 	tt_want(evhttp_uri_join(uri, url_tmp, 3) == NULL);
 	/* host is set, but path doesn't start with "/": */
-	uri->path = (char*)"hi_mom";
+	tt_want(0==evhttp_uri_set_path(uri, "hi_mom"));
 	tt_want(evhttp_uri_join(uri, url_tmp, sizeof(url_tmp)) == NULL);
 	tt_want(evhttp_uri_join(uri, NULL, sizeof(url_tmp))==NULL);
 	tt_want(evhttp_uri_join(uri, url_tmp, 0)==NULL);
-	free(uri);
-
+	evhttp_uri_free(uri);
 	uri = evhttp_uri_parse("mailto:foo@bar");
 	tt_want(uri != NULL);
-	tt_want(uri->host == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(!strcmp(uri->scheme, "mailto"));
-	tt_want(!strcmp(uri->path, "foo@bar"));
-	tt_want(uri->query == NULL);
-	tt_want(uri->fragment == NULL);
+	tt_want(evhttp_uri_get_host(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(!strcmp(evhttp_uri_get_scheme(uri), "mailto"));
+	tt_want(!strcmp(evhttp_uri_get_path(uri), "foo@bar"));
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("mailto:foo@bar");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://www.test.com/?q=t%33est");
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(strcmp(uri->query, "q=t%33est") == 0);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "www.test.com") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=t%33est") == 0);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("http://www.test.com/?q=t%33est");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://%77ww.test.com");
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "%77ww.test.com") == 0);
-	tt_want(strcmp(uri->path, "") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "%77ww.test.com") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("http://%77ww.test.com");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://www.test.com?q=test");
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want(strcmp(uri->path, "") == 0);
-	tt_want(strcmp(uri->query, "q=test") == 0);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "www.test.com") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=test") == 0);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("http://www.test.com?q=test");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://www.test.com#fragment");
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want(strcmp(uri->path, "") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want_str_op(uri->fragment, ==, "fragment");
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "www.test.com") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want_str_op(evhttp_uri_get_fragment(uri), ==, "fragment");
 	TT_URI("http://www.test.com#fragment");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://8000/");
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "8000") == 0);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "8000") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("http://8000/");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://:8000/");
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "") == 0);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == 8000);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == 8000);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("http://:8000/");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://www.test.com:/"); /* empty port */
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want_str_op(uri->path, ==, "/");
-	tt_want(uri->query == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "www.test.com") == 0);
+	tt_want_str_op(evhttp_uri_get_path(uri), ==, "/");
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("http://www.test.com/");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("http://www.test.com:"); /* empty port 2 */
-	tt_want(strcmp(uri->scheme, "http") == 0);
-	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want(strcmp(uri->path, "") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "www.test.com") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("http://www.test.com");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("ftp://www.test.com/?q=test");
-	tt_want(strcmp(uri->scheme, "ftp") == 0);
-	tt_want(strcmp(uri->host, "www.test.com") == 0);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(strcmp(uri->query, "q=test") == 0);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "ftp") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "www.test.com") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=test") == 0);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("ftp://www.test.com/?q=test");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("ftp://[::1]:999/?q=test");
-	tt_want(strcmp(uri->scheme, "ftp") == 0);
-	tt_want(strcmp(uri->host, "[::1]") == 0);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(strcmp(uri->query, "q=test") == 0);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == 999);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "ftp") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "[::1]") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=test") == 0);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == 999);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("ftp://[::1]:999/?q=test");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("ftp://[ff00::127.0.0.1]/?q=test");
-	tt_want(strcmp(uri->scheme, "ftp") == 0);
-	tt_want(strcmp(uri->host, "[ff00::127.0.0.1]") == 0);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(strcmp(uri->query, "q=test") == 0);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "ftp") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "[ff00::127.0.0.1]") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=test") == 0);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("ftp://[ff00::127.0.0.1]/?q=test");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("ftp://[v99.not_(any:time)_soon]/?q=test");
-	tt_want(strcmp(uri->scheme, "ftp") == 0);
-	tt_want(strcmp(uri->host, "[v99.not_(any:time)_soon]") == 0);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(strcmp(uri->query, "q=test") == 0);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->port == -1);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "ftp") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "[v99.not_(any:time)_soon]") == 0);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=test") == 0);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("ftp://[v99.not_(any:time)_soon]/?q=test");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("scheme://user:pass@foo.com:42/?q=test&s=some+thing#fragment");
-	tt_want(strcmp(uri->scheme, "scheme") == 0);
-	tt_want(strcmp(uri->userinfo, "user:pass") == 0);
-	tt_want(strcmp(uri->host, "foo.com") == 0);
-	tt_want(uri->port == 42);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(strcmp(uri->query, "q=test&s=some+thing") == 0);
-	tt_want(strcmp(uri->fragment, "fragment") == 0);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "scheme") == 0);
+	tt_want(strcmp(evhttp_uri_get_userinfo(uri), "user:pass") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "foo.com") == 0);
+	tt_want(evhttp_uri_get_port(uri) == 42);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=test&s=some+thing") == 0);
+	tt_want(strcmp(evhttp_uri_get_fragment(uri), "fragment") == 0);
 	TT_URI("scheme://user:pass@foo.com:42/?q=test&s=some+thing#fragment");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("scheme://user@foo.com/#fragment");
-	tt_want(strcmp(uri->scheme, "scheme") == 0);
-	tt_want(strcmp(uri->userinfo, "user") == 0);
-	tt_want(strcmp(uri->host, "foo.com") == 0);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(strcmp(uri->fragment, "fragment") == 0);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "scheme") == 0);
+	tt_want(strcmp(evhttp_uri_get_userinfo(uri), "user") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "foo.com") == 0);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(strcmp(evhttp_uri_get_fragment(uri), "fragment") == 0);
 	TT_URI("scheme://user@foo.com/#fragment");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("scheme://%75ser@foo.com/#frag@ment");
-	tt_want(strcmp(uri->scheme, "scheme") == 0);
-	tt_want(strcmp(uri->userinfo, "%75ser") == 0);
-	tt_want(strcmp(uri->host, "foo.com") == 0);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "/") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(strcmp(uri->fragment, "frag@ment") == 0);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "scheme") == 0);
+	tt_want(strcmp(evhttp_uri_get_userinfo(uri), "%75ser") == 0);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "foo.com") == 0);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(strcmp(evhttp_uri_get_fragment(uri), "frag@ment") == 0);
 	TT_URI("scheme://%75ser@foo.com/#frag@ment");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("file:///some/path/to/the/file");
-	tt_want(strcmp(uri->scheme, "file") == 0);
-	tt_want(uri->userinfo == NULL);
-	tt_want(strcmp(uri->host, "") == 0);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "/some/path/to/the/file") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->fragment == NULL);
+	tt_want(strcmp(evhttp_uri_get_scheme(uri), "file") == 0);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "") == 0);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/some/path/to/the/file") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("file:///some/path/to/the/file");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("///some/path/to/the-file");
 	tt_want(uri != NULL);
-	tt_want(uri->scheme == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(strcmp(uri->host, "") == 0);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "/some/path/to/the-file") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->fragment == NULL);
+	tt_want(evhttp_uri_get_scheme(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(strcmp(evhttp_uri_get_host(uri), "") == 0);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/some/path/to/the-file") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("///some/path/to/the-file");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("/s:ome/path/to/the-file?q=99#fred");
 	tt_want(uri != NULL);
-	tt_want(uri->scheme == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->host == NULL);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "/s:ome/path/to/the-file") == 0);
-	tt_want(strcmp(uri->query, "q=99") == 0);
-	tt_want(strcmp(uri->fragment, "fred") == 0);
+	tt_want(evhttp_uri_get_scheme(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_host(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "/s:ome/path/to/the-file") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=99") == 0);
+	tt_want(strcmp(evhttp_uri_get_fragment(uri), "fred") == 0);
 	TT_URI("/s:ome/path/to/the-file?q=99#fred");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("relative/path/with/co:lon");
 	tt_want(uri != NULL);
-	tt_want(uri->scheme == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->host == NULL);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "relative/path/with/co:lon") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(uri->fragment == NULL);
+	tt_want(evhttp_uri_get_scheme(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_host(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "relative/path/with/co:lon") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(evhttp_uri_get_fragment(uri) == NULL);
 	TT_URI("relative/path/with/co:lon");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("bob?q=99&q2=q?33#fr?ed");
 	tt_want(uri != NULL);
-	tt_want(uri->scheme == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->host == NULL);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "bob") == 0);
-	tt_want(strcmp(uri->query, "q=99&q2=q?33") == 0);
-	tt_want(strcmp(uri->fragment, "fr?ed") == 0);
+	tt_want(evhttp_uri_get_scheme(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_host(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "bob") == 0);
+	tt_want(strcmp(evhttp_uri_get_query(uri), "q=99&q2=q?33") == 0);
+	tt_want(strcmp(evhttp_uri_get_fragment(uri), "fr?ed") == 0);
 	TT_URI("bob?q=99&q2=q?33#fr?ed");
 	evhttp_uri_free(uri);
 
 	uri = evhttp_uri_parse("#fr?ed");
 	tt_want(uri != NULL);
-	tt_want(uri->scheme == NULL);
-	tt_want(uri->userinfo == NULL);
-	tt_want(uri->host == NULL);
-	tt_want(uri->port == -1);
-	tt_want(strcmp(uri->path, "") == 0);
-	tt_want(uri->query == NULL);
-	tt_want(strcmp(uri->fragment, "fr?ed") == 0);
+	tt_want(evhttp_uri_get_scheme(uri) == NULL);
+	tt_want(evhttp_uri_get_userinfo(uri) == NULL);
+	tt_want(evhttp_uri_get_host(uri) == NULL);
+	tt_want(evhttp_uri_get_port(uri) == -1);
+	tt_want(strcmp(evhttp_uri_get_path(uri), "") == 0);
+	tt_want(evhttp_uri_get_query(uri) == NULL);
+	tt_want(strcmp(evhttp_uri_get_fragment(uri), "fr?ed") == 0);
 	TT_URI("#fr?ed");
 	evhttp_uri_free(uri);
 }

From bc98f5e6bad6a8502895d992cc3413b118969e01 Mon Sep 17 00:00:00 2001
From: Nick Mathewson <nickm@torproject.org>
Date: Thu, 21 Oct 2010 14:53:21 -0400
Subject: [PATCH 11/11] Unit tests for evhttp_uri_set*

---
 test/regress_http.c | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

diff --git a/test/regress_http.c b/test/regress_http.c
index f90e98c38e..95b2763c20 100644
--- a/test/regress_http.c
+++ b/test/regress_http.c
@@ -1789,6 +1789,46 @@ http_parse_uri_test(void *ptr)
 	TT_URI("mailto:foo@bar");
 	evhttp_uri_free(uri);
 
+	uri = evhttp_uri_new();
+	/* Bad URI usage: setting invalid values */
+	tt_want(-1 == evhttp_uri_set_scheme(uri,""));
+	tt_want(-1 == evhttp_uri_set_scheme(uri,"33"));
+	tt_want(-1 == evhttp_uri_set_scheme(uri,"hi!"));
+	tt_want(-1 == evhttp_uri_set_userinfo(uri,"hello@"));
+	tt_want(-1 == evhttp_uri_set_host(uri,"[1.2.3.4]"));
+	tt_want(-1 == evhttp_uri_set_host(uri,"["));
+	tt_want(-1 == evhttp_uri_set_host(uri,"www.[foo].com"));
+	tt_want(-1 == evhttp_uri_set_port(uri,-3));
+	tt_want(-1 == evhttp_uri_set_path(uri,"hello?world"));
+	tt_want(-1 == evhttp_uri_set_query(uri,"hello#world"));
+	tt_want(-1 == evhttp_uri_set_fragment(uri,"hello#world"));
+	/* Valid URI usage: setting valid values */
+	tt_want(0 == evhttp_uri_set_scheme(uri,"http"));
+	tt_want(0 == evhttp_uri_set_scheme(uri,NULL));
+	tt_want(0 == evhttp_uri_set_userinfo(uri,"username:pass"));
+	tt_want(0 == evhttp_uri_set_userinfo(uri,NULL));
+	tt_want(0 == evhttp_uri_set_host(uri,"www.example.com"));
+	tt_want(0 == evhttp_uri_set_host(uri,"1.2.3.4"));
+	tt_want(0 == evhttp_uri_set_host(uri,"[1:2:3:4::]"));
+	tt_want(0 == evhttp_uri_set_host(uri,"[v7.wobblewobble]"));
+	tt_want(0 == evhttp_uri_set_host(uri,NULL));
+	tt_want(0 == evhttp_uri_set_host(uri,""));
+	tt_want(0 == evhttp_uri_set_port(uri, -1));
+	tt_want(0 == evhttp_uri_set_port(uri, 80));
+	tt_want(0 == evhttp_uri_set_port(uri, 65535));
+	tt_want(0 == evhttp_uri_set_path(uri, ""));
+	tt_want(0 == evhttp_uri_set_path(uri, "/documents/public/index.html"));
+	tt_want(0 == evhttp_uri_set_path(uri, NULL));
+	tt_want(0 == evhttp_uri_set_query(uri, "key=val&key2=val2"));
+	tt_want(0 == evhttp_uri_set_query(uri, "keyvalblarg"));
+	tt_want(0 == evhttp_uri_set_query(uri, ""));
+	tt_want(0 == evhttp_uri_set_query(uri, NULL));
+	tt_want(0 == evhttp_uri_set_fragment(uri, ""));
+	tt_want(0 == evhttp_uri_set_fragment(uri, "here?i?am"));
+	tt_want(0 == evhttp_uri_set_fragment(uri, NULL));
+	evhttp_uri_free(uri);
+
+	/* Valid parsing */
 	uri = evhttp_uri_parse("http://www.test.com/?q=t%33est");
 	tt_want(strcmp(evhttp_uri_get_scheme(uri), "http") == 0);
 	tt_want(strcmp(evhttp_uri_get_host(uri), "www.test.com") == 0);