From 9dfaa67f2b74a11d7cc2a58d28961547b19462fc Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Sat, 7 Jul 2012 22:14:12 +0200
Subject: [PATCH 01/14] Check host on url with hostname:port

---
 test.c | 141 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 141 insertions(+)

diff --git a/test.c b/test.c
index 6d8c004..23bcf5d 100644
--- a/test.c
+++ b/test.c
@@ -50,6 +50,7 @@ struct message {
   char query_string[MAX_ELEMENT_SIZE];
   char body[MAX_ELEMENT_SIZE];
   size_t body_size;
+  const char *host;
   uint16_t port;
   int num_headers;
   enum { NONE=0, FIELD, VALUE } last_header_element;
@@ -649,6 +650,7 @@ const struct message requests[] =
   ,.fragment= ""
   ,.request_path= ""
   ,.request_url= "http://hypnotoad.org:1234?hail=all"
+  ,.host= "hypnotoad.org"
   ,.port= 1234
   ,.num_headers= 0
   ,.headers= { }
@@ -669,6 +671,7 @@ const struct message requests[] =
   ,.fragment= ""
   ,.request_path= ""
   ,.request_url= "http://hypnotoad.org:1234"
+  ,.host= "hypnotoad.org"
   ,.port= 1234
   ,.num_headers= 0
   ,.headers= { }
@@ -1794,6 +1797,10 @@ message_eq (int index, const struct message *expected)
       abort();
     }
 
+    if (expected->host) {
+      MESSAGE_CHECK_URL_EQ(&u, expected, m, host, UF_HOST);
+    }
+
     m->port = (u.field_set & (1 << UF_PORT)) ?
       u.port : 0;
 
@@ -1971,6 +1978,24 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
+, {.name="proxy request with port"
+  ,.url="http://hostname:444/"
+  ,.is_connect=0
+  ,.u=
+    {.field_set=(1 << UF_SCHEMA) | (1 << UF_HOST) | (1 << UF_PORT) | (1 << UF_PATH)
+    ,.port=444
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{  7,  8 } /* UF_HOST */
+      ,{ 16,  3 } /* UF_PORT */
+      ,{ 19,  1 } /* UF_PATH */
+      ,{  0,  0 } /* UF_QUERY */
+      ,{  0,  0 } /* UF_FRAGMENT */
+      }
+    }
+  ,.rv=0
+  }
+
 , {.name="CONNECT request"
   ,.url="hostname:443"
   ,.is_connect=1
@@ -2007,6 +2032,24 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
+, {.name="proxy ipv6 request with port"
+  ,.url="http://[1:2::3:4]:67/"
+  ,.is_connect=0
+  ,.u=
+    {.field_set=(1 << UF_SCHEMA) | (1 << UF_HOST) | (1 << UF_PORT) | (1 << UF_PATH)
+    ,.port=67
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{  8,  8 } /* UF_HOST */
+      ,{ 18,  2 } /* UF_PORT */
+      ,{ 20,  1 } /* UF_PATH */
+      ,{  0,  0 } /* UF_QUERY */
+      ,{  0,  0 } /* UF_FRAGMENT */
+      }
+    }
+  ,.rv=0
+  }
+
 , {.name="CONNECT ipv6 address"
   ,.url="[1:2::3:4]:443"
   ,.is_connect=1
@@ -2043,6 +2086,79 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
+, {.name="space URL encoded"
+  ,.url="/toto.html?toto=a%20b"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_PATH) | (1<<UF_QUERY)
+    ,.port=0
+    ,.field_data=
+      {{  0,  0 } /* UF_SCHEMA */
+      ,{  0,  0 } /* UF_HOST */
+      ,{  0,  0 } /* UF_PORT */
+      ,{  0, 10 } /* UF_PATH */
+      ,{ 11, 10 } /* UF_QUERY */
+      ,{  0,  0 } /* UF_FRAGMENT */
+      }
+    }
+  ,.rv=0
+  }
+
+
+, {.name="URL fragment"
+  ,.url="/toto.html#titi"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_PATH) | (1<<UF_FRAGMENT)
+    ,.port=0
+    ,.field_data=
+      {{  0,  0 } /* UF_SCHEMA */
+      ,{  0,  0 } /* UF_HOST */
+      ,{  0,  0 } /* UF_PORT */
+      ,{  0, 10 } /* UF_PATH */
+      ,{  0,  0 } /* UF_QUERY */
+      ,{ 11,  4 } /* UF_FRAGMENT */
+      }
+    }
+  ,.rv=0
+  }
+
+, {.name="complex URL fragment"
+  ,.url="http://www.webmasterworld.com/r.cgi?f=21&d=8405&url=http://www.example.com/index.html?foo=bar&hello=world#midpage"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH) | (1<<UF_QUERY) | (1<<UF_FRAGMENT)
+    ,.port=0
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{  7, 22 } /* UF_HOST */
+      ,{  0,  0 } /* UF_PORT */
+      ,{ 29,  6 } /* UF_PATH */
+      ,{ 36, 69 } /* UF_QUERY */
+      ,{106,  7 } /* UF_FRAGMENT */
+      }
+    }
+  ,.rv=0
+  }
+
+, {.name="complex URL from node js url parser doc"
+  ,.url="http://host.com:8080/p/a/t/h?query=string#hash"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PORT) | (1<<UF_PATH) | (1<<UF_QUERY) | (1<<UF_FRAGMENT)
+    ,.port=8080
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{  7,  8 } /* UF_HOST */
+      ,{ 16,  4 } /* UF_PORT */
+      ,{ 20,  8 } /* UF_PATH */
+      ,{ 29, 12 } /* UF_QUERY */
+      ,{ 42,  4 } /* UF_FRAGMENT */
+      }
+    }
+  ,.rv=0
+  }
+
 , {.name="proxy empty host"
   ,.url="http://:443/"
   ,.is_connect=0
@@ -2083,11 +2199,36 @@ const struct url_test url_tests[] =
   ,.rv=1 /* s_dead */
   }
 
+, {.name="double : in URL"
+  ,.url="http://hostname::443/"
+  ,.rv=1 /* s_dead */
+  }
+
 , {.name="line feed in URL"
   ,.url="/foo\nbar/"
   ,.rv=1 /* s_dead */
   }
 
+, {.name="line feed in hostname"
+  ,.url="http://host\name/fo"
+  ,.rv=1 /* s_dead */
+  }
+
+, {.name="% in hostname"
+  ,.url="http://host%name/fo"
+  ,.rv=1 /* s_dead */
+  }
+
+, {.name="; in hostname"
+  ,.url="http://host;ame/fo"
+  ,.rv=1 /* s_dead */
+  }
+
+, {.name="= in URL"
+  ,.url="http://host=ame/fo"
+  ,.rv=1 /* s_dead */
+  }
+
 #if HTTP_PARSER_STRICT
 
 , {.name="tab in URL"

From 624a4b44a92f0da3e3ceb3592a191371ded2476f Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Sat, 7 Jul 2012 23:29:43 +0200
Subject: [PATCH 02/14] Add url_parser tool

---
 .gitignore   |  1 +
 Makefile     |  8 +++++++-
 url_parser.c | 47 +++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 55 insertions(+), 1 deletion(-)
 create mode 100644 url_parser.c

diff --git a/.gitignore b/.gitignore
index 8631bd8..3b868c5 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,6 +4,7 @@ tags
 test
 test_g
 test_fast
+url_parser
 *.mk
 *.Makefile
 *.so
diff --git a/Makefile b/Makefile
index 8d90f8d..d9bf839 100644
--- a/Makefile
+++ b/Makefile
@@ -31,6 +31,12 @@ test_fast: http_parser.o test.o http_parser.h
 test.o: test.c http_parser.h Makefile
 	$(CC) $(CPPFLAGS_FAST) $(CFLAGS_FAST) -c test.c -o $@
 
+url_parser: http_parser_g.o url_parser.o
+	$(CC) $(CFLAGS_DEBUG) $(LDFLAGS) http_parser_g.o url_parser.o -o $@
+
+url_parser.o: url_parser.c http_parser.h Makefile
+	$(CC) $(CPPFLAGS_DEBUG) $(CFLAGS_DEBUG) -c url_parser.c -o $@
+
 http_parser.o: http_parser.c http_parser.h Makefile
 	$(CC) $(CPPFLAGS_FAST) $(CFLAGS_FAST) -c http_parser.c
 
@@ -53,6 +59,6 @@ tags: http_parser.c http_parser.h test.c
 	ctags $^
 
 clean:
-	rm -f *.o *.a test test_fast test_g http_parser.tar tags libhttp_parser.so libhttp_parser.o
+	rm -f *.o *.a test test_fast test_g url_parser http_parser.tar tags libhttp_parser.so libhttp_parser.o
 
 .PHONY: clean package test-run test-run-timed test-valgrind
diff --git a/url_parser.c b/url_parser.c
new file mode 100644
index 0000000..c1f3ef0
--- /dev/null
+++ b/url_parser.c
@@ -0,0 +1,47 @@
+#include "http_parser.h"
+#include <stdio.h>
+#include <string.h>
+
+void
+dump_url (const char *url, const struct http_parser_url *u)
+{
+  char part[512];
+  unsigned int i;
+
+  printf("\tfield_set: 0x%x, port: %u\n", u->field_set, u->port);
+  for (i = 0; i < UF_MAX; i++) {
+    if ((u->field_set & (1 << i)) == 0) {
+      printf("\tfield_data[%u]: unset\n", i);
+      continue;
+    }
+
+    memcpy(part, url + u->field_data[i].off, u->field_data[i].len);
+    part[u->field_data[i].len] = '\0';
+
+    printf("\tfield_data[%u]: off: %u len: %u part: \"%s\"\n",
+           i,
+           u->field_data[i].off,
+           u->field_data[i].len,
+           part);
+  }
+}
+
+int main(int argc, char ** argv) {
+  if (argc != 3) {
+    printf("Syntax : %s connect|get url\n", argv[0]);
+    return 1;
+  }
+  struct http_parser_url u;
+  int len = strlen(argv[2]);
+  int connect = strcmp("connect", argv[1]) == 0 ? 1 : 0;
+  printf("Parsing %s, connect %d\n", argv[2], connect);
+
+  int result = http_parser_parse_url(argv[2], len, connect, &u);
+  if (result != 0) {
+    printf("Parse error : %d\n", result);
+    return result;
+  }
+  printf("Parse ok, result : \n");
+  dump_url(argv[2], &u);
+  return 0;
+}
\ No newline at end of file

From ed8475d49f8d60a61f684c802a7c4c0d8d539fca Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Sun, 8 Jul 2012 01:10:43 +0200
Subject: [PATCH 03/14] Refactor host parsing to allow basic auth management

---
 http_parser.c | 224 +++++++++++++++++++++++++++++++-------------------
 1 file changed, 139 insertions(+), 85 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index 0c11eb8..7125665 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -254,12 +254,7 @@ enum state
   , s_req_schema_slash
   , s_req_schema_slash_slash
   , s_req_host_start
-  , s_req_host_v6_start
-  , s_req_host_v6
-  , s_req_host_v6_end
   , s_req_host
-  , s_req_port_start
-  , s_req_port
   , s_req_path
   , s_req_query_string_start
   , s_req_query_string
@@ -337,6 +332,17 @@ enum header_states
   , h_connection_close
   };
 
+enum http_host_state
+  {
+    s_http_host_dead = 1
+  , s_http_host_start
+  , s_http_host_v6_start
+  , s_http_host
+  , s_http_host_v6
+  , s_http_host_v6_end
+  , s_http_host_port_start
+  , s_http_host_port
+};
 
 /* Macros for character classes; depends on strict-mode  */
 #define CR                  '\r'
@@ -456,61 +462,17 @@ parse_url_char(enum state s, const char ch)
       break;
 
     case s_req_host_start:
-      if (ch == '[') {
-        return s_req_host_v6_start;
-      }
-
-      if (IS_HOST_CHAR(ch)) {
-        return s_req_host;
-      }
-
-      break;
-
     case s_req_host:
-      if (IS_HOST_CHAR(ch)) {
-        return s_req_host;
-      }
-
-      /* FALLTHROUGH */
-    case s_req_host_v6_end:
-      switch (ch) {
-        case ':':
-          return s_req_port_start;
-
-        case '/':
-          return s_req_path;
-
-        case '?':
-          return s_req_query_string_start;
-      }
-
-      break;
-
-    case s_req_host_v6:
-      if (ch == ']') {
-        return s_req_host_v6_end;
-      }
-
-      /* FALLTHROUGH */
-    case s_req_host_v6_start:
-      if (IS_HEX(ch) || ch == ':') {
-        return s_req_host_v6;
+      if (ch == '/') {
+        return s_req_path;
       }
-      break;
-
-    case s_req_port:
-      switch (ch) {
-        case '/':
-          return s_req_path;
 
-        case '?':
-          return s_req_query_string_start;
+      if (ch == '?') {
+        return s_req_query_string_start;
       }
 
-      /* FALLTHROUGH */
-    case s_req_port_start:
-      if (IS_NUM(ch)) {
-        return s_req_port;
+      if (IS_ALPHANUM(ch) || ch == '_' || ch == '@' || ch == '[' || ch == ']' || ch == '.' || ch == '-' || ch == '%' || ch == ';' || ch == ':' || ch == '&' || ch == '=' || ch ==  '+' || ch ==  '$' || ch == ',') {
+        return s_req_host;
       }
 
       break;
@@ -633,12 +595,7 @@ size_t http_parser_execute (http_parser *parser,
   case s_req_schema_slash:
   case s_req_schema_slash_slash:
   case s_req_host_start:
-  case s_req_host_v6_start:
-  case s_req_host_v6:
-  case s_req_host_v6_end:
   case s_req_host:
-  case s_req_port_start:
-  case s_req_port:
   case s_req_query_string_start:
   case s_req_query_string:
   case s_req_fragment_start:
@@ -1015,9 +972,6 @@ size_t http_parser_execute (http_parser *parser,
       case s_req_schema_slash:
       case s_req_schema_slash_slash:
       case s_req_host_start:
-      case s_req_host_v6_start:
-      case s_req_host_v6:
-      case s_req_port_start:
       {
         switch (ch) {
           /* No whitespace allowed here */
@@ -1038,8 +992,6 @@ size_t http_parser_execute (http_parser *parser,
       }
 
       case s_req_host:
-      case s_req_host_v6_end:
-      case s_req_port:
       case s_req_path:
       case s_req_query_string_start:
       case s_req_query_string:
@@ -1938,6 +1890,122 @@ http_errno_description(enum http_errno err) {
   return http_strerror_tab[err].description;
 }
 
+static enum http_host_state http_parse_host_char(enum http_host_state s, const char ch) {
+  switch(s) {
+    case s_http_host_start:
+      if (ch == '[') {
+        return s_http_host_v6_start;
+      }
+
+      if (IS_HOST_CHAR(ch)) {
+        return s_http_host;
+      }
+
+      break;
+
+    case s_http_host:
+
+      if (IS_HOST_CHAR(ch)) {
+        return s_http_host;
+      }
+
+    /* FALLTHROUGH */
+    case s_http_host_v6_end:
+      if (ch == ':') {
+        return s_http_host_port_start;
+      }
+
+      break;
+
+    case s_http_host_v6:
+      if (ch == ']') {
+        return s_http_host_v6_end;
+      }
+
+    /* FALLTHROUGH */
+    case s_http_host_v6_start:
+      if (IS_HEX(ch) || ch == ':') {
+        return s_http_host_v6;
+      }
+
+      break;
+
+    case s_http_host_port:
+    case s_http_host_port_start:
+      if (IS_NUM(ch)) {
+        return s_http_host_port;
+      }
+
+      break;
+
+    default:
+      break;
+  }
+  return s_http_host_dead;
+}
+
+int http_parse_host(const char * buf, struct http_parser_url *u) {
+  enum http_host_state s;
+
+  int i;
+  int start = u->field_data[UF_HOST].off;
+  int len = u->field_data[UF_HOST].len;
+
+  u->field_data[UF_HOST].len = 0;
+
+  i = 0;
+  s = s_http_host_start;
+
+  for(; i < len; i ++) {
+    char ch = buf[start + i];
+    enum http_host_state new_s = http_parse_host_char(s, ch);
+    if (new_s == s_http_host_dead) {
+      return 1;
+    }
+
+    switch(new_s) {
+      case s_http_host:
+        if (s != s_http_host) {
+          u->field_data[UF_HOST].off = start + i;
+        }
+        u->field_data[UF_HOST].len ++;
+        break;
+
+      case s_http_host_v6:
+        if (s != s_http_host_v6) {
+          u->field_data[UF_HOST].off = start + i;
+        }
+        u->field_data[UF_HOST].len ++;
+        break;
+
+      case s_http_host_port:
+        if (s != s_http_host_port) {
+          u->field_data[UF_PORT].off = start + i;
+          u->field_data[UF_PORT].len = 0;
+          u->field_set |= (1 << UF_PORT);
+        }
+        u->field_data[UF_PORT].len ++;
+        break;
+
+      default:
+        break;
+    }
+    s = new_s;
+  }
+
+  /* Make sure we don't end somewhere unexpected */
+  switch (s) {
+    case s_http_host_v6_start:
+    case s_http_host_v6:
+    case s_http_host_port_start:
+      return 1;
+    default:
+      break;
+  }
+
+  return 0;
+}
+
 int
 http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
                       struct http_parser_url *u)
@@ -1962,9 +2030,6 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
       case s_req_schema_slash:
       case s_req_schema_slash_slash:
       case s_req_host_start:
-      case s_req_host_v6_start:
-      case s_req_host_v6_end:
-      case s_req_port_start:
       case s_req_query_string_start:
       case s_req_fragment_start:
         continue;
@@ -1974,14 +2039,9 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
         break;
 
       case s_req_host:
-      case s_req_host_v6:
         uf = UF_HOST;
         break;
 
-      case s_req_port:
-        uf = UF_PORT;
-        break;
-
       case s_req_path:
         uf = UF_PATH;
         break;
@@ -2012,21 +2072,15 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
     old_uf = uf;
   }
 
-  /* CONNECT requests can only contain "hostname:port" */
-  if (is_connect && u->field_set != ((1 << UF_HOST)|(1 << UF_PORT))) {
-    return 1;
+  if ((u->field_set & (1 << UF_HOST)) != 0) {
+    if (http_parse_host(buf, u) != 0) {
+      return 1;
+    }
   }
 
-  /* Make sure we don't end somewhere unexpected */
-  switch (s) {
-  case s_req_host_v6_start:
-  case s_req_host_v6:
-  case s_req_host_v6_end:
-  case s_req_host:
-  case s_req_port_start:
+  /* CONNECT requests can only contain "hostname:port" */
+  if (is_connect && u->field_set != ((1 << UF_HOST)|(1 << UF_PORT))) {
     return 1;
-  default:
-    break;
   }
 
   if (u->field_set & (1 << UF_PORT)) {

From 79650962763cd4b849c9c9689953b9d848732f0e Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Sun, 8 Jul 2012 01:44:31 +0200
Subject: [PATCH 04/14] User info implementation

---
 http_parser.c | 54 +++++++++++++++++++++++++++++++++++++++++----------
 http_parser.h |  3 ++-
 test.c        | 13 +++++++++++++
 3 files changed, 59 insertions(+), 11 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index 7125665..5ea352d 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -335,6 +335,8 @@ enum header_states
 enum http_host_state
   {
     s_http_host_dead = 1
+  , s_http_userinfo_start
+  , s_http_userinfo
   , s_http_host_start
   , s_http_host_v6_start
   , s_http_host
@@ -352,6 +354,8 @@ enum http_host_state
 #define IS_NUM(c)           ((c) >= '0' && (c) <= '9')
 #define IS_ALPHANUM(c)      (IS_ALPHA(c) || IS_NUM(c))
 #define IS_HEX(c)           (IS_NUM(c) || (LOWER(c) >= 'a' && LOWER(c) <= 'f'))
+#define IS_MARK(c)          ((c) == '-' || (c) == '_' || (c) == '.' || (c) == '!' || (c) == '~' || (c) == '*' || (c) == '\'' || (c) == '(' || (c) == ')')
+#define IS_USERINFO_CHAR(c) (IS_ALPHANUM(c) || IS_MARK(c) || (c) == '%' || (c) == ';' || (c) == ':' || (c) == '&' || (c) == '=' || (c) == '+' || (c) == '$' || (c) == ',')
 
 #if HTTP_PARSER_STRICT
 #define TOKEN(c)            (tokens[(unsigned char)c])
@@ -408,7 +412,7 @@ int http_message_needs_eof(http_parser *parser);
  * URL and non-URL states by looking for these.
  */
 static enum state
-parse_url_char(enum state s, const char ch)
+parse_url_char(enum state s, const char ch, int * found_at)
 {
   if (ch == ' ' || ch == '\r' || ch == '\n') {
     return s_dead;
@@ -471,7 +475,14 @@ parse_url_char(enum state s, const char ch)
         return s_req_query_string_start;
       }
 
-      if (IS_ALPHANUM(ch) || ch == '_' || ch == '@' || ch == '[' || ch == ']' || ch == '.' || ch == '-' || ch == '%' || ch == ';' || ch == ':' || ch == '&' || ch == '=' || ch ==  '+' || ch ==  '$' || ch == ',') {
+      if (IS_USERINFO_CHAR(ch) || ch == '[' || ch == ']') {
+        return s_req_host;
+      }
+
+      if (ch == '@') {
+        if (found_at) {
+          *found_at = 1;
+        }
         return s_req_host;
       }
 
@@ -959,7 +970,7 @@ size_t http_parser_execute (http_parser *parser,
           parser->state = s_req_host_start;
         }
 
-        parser->state = parse_url_char((enum state)parser->state, ch);
+        parser->state = parse_url_char((enum state)parser->state, ch, 0);
         if (parser->state == s_dead) {
           SET_ERRNO(HPE_INVALID_URL);
           goto error;
@@ -981,7 +992,7 @@ size_t http_parser_execute (http_parser *parser,
             SET_ERRNO(HPE_INVALID_URL);
             goto error;
           default:
-            parser->state = parse_url_char((enum state)parser->state, ch);
+            parser->state = parse_url_char((enum state)parser->state, ch, 0);
             if (parser->state == s_dead) {
               SET_ERRNO(HPE_INVALID_URL);
               goto error;
@@ -1013,7 +1024,7 @@ size_t http_parser_execute (http_parser *parser,
             CALLBACK_DATA(url);
             break;
           default:
-            parser->state = parse_url_char((enum state)parser->state, ch);
+            parser->state = parse_url_char((enum state)parser->state, ch, 0);
             if (parser->state == s_dead) {
               SET_ERRNO(HPE_INVALID_URL);
               goto error;
@@ -1890,8 +1901,20 @@ http_errno_description(enum http_errno err) {
   return http_strerror_tab[err].description;
 }
 
-static enum http_host_state http_parse_host_char(enum http_host_state s, const char ch) {
+static enum http_host_state
+http_parse_host_char(enum http_host_state s, const char ch) {
   switch(s) {
+    case s_http_userinfo:
+    case s_http_userinfo_start:
+      if (ch == '@') {
+        return s_http_host_start;
+      }
+
+      if (IS_USERINFO_CHAR(ch)) {
+        return s_http_userinfo;
+      }
+      break;
+
     case s_http_host_start:
       if (ch == '[') {
         return s_http_host_v6_start;
@@ -1944,7 +1967,7 @@ static enum http_host_state http_parse_host_char(enum http_host_state s, const c
   return s_http_host_dead;
 }
 
-int http_parse_host(const char * buf, struct http_parser_url *u) {
+int http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
   enum http_host_state s;
 
   int i;
@@ -1954,11 +1977,12 @@ int http_parse_host(const char * buf, struct http_parser_url *u) {
   u->field_data[UF_HOST].len = 0;
 
   i = 0;
-  s = s_http_host_start;
+  s = found_at ? s_http_userinfo_start : s_http_host_start;
 
   for(; i < len; i ++) {
     char ch = buf[start + i];
     enum http_host_state new_s = http_parse_host_char(s, ch);
+
     if (new_s == s_http_host_dead) {
       return 1;
     }
@@ -1987,6 +2011,15 @@ int http_parse_host(const char * buf, struct http_parser_url *u) {
         u->field_data[UF_PORT].len ++;
         break;
 
+      case s_http_userinfo:
+        if (s != s_http_userinfo) {
+          u->field_data[UF_USERINFO].off = start + i;
+          u->field_data[UF_USERINFO].len = 0;
+          u->field_set |= (1 << UF_USERINFO);
+        }
+        u->field_data[UF_USERINFO].len ++;
+        break;
+
       default:
         break;
     }
@@ -2013,13 +2046,14 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
   enum state s;
   const char *p;
   enum http_parser_url_fields uf, old_uf;
+  int found_at = 0;
 
   u->port = u->field_set = 0;
   s = is_connect ? s_req_host_start : s_req_spaces_before_url;
   uf = old_uf = UF_MAX;
 
   for (p = buf; p < buf + buflen; p++) {
-    s = parse_url_char(s, *p);
+    s = parse_url_char(s, *p, &found_at);
 
     /* Figure out the next field that we're operating on */
     switch (s) {
@@ -2073,7 +2107,7 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
   }
 
   if ((u->field_set & (1 << UF_HOST)) != 0) {
-    if (http_parse_host(buf, u) != 0) {
+    if (http_parse_host(buf, u, found_at) != 0) {
       return 1;
     }
   }
diff --git a/http_parser.h b/http_parser.h
index 8ed4180..dc15473 100644
--- a/http_parser.h
+++ b/http_parser.h
@@ -256,7 +256,8 @@ enum http_parser_url_fields
   , UF_PATH             = 3
   , UF_QUERY            = 4
   , UF_FRAGMENT         = 5
-  , UF_MAX              = 6
+  , UF_USERINFO         = 6
+  , UF_MAX              = 7
   };
 
 
diff --git a/test.c b/test.c
index 23bcf5d..f7dfcfe 100644
--- a/test.c
+++ b/test.c
@@ -1973,6 +1973,7 @@ const struct url_test url_tests[] =
       ,{ 15,  1 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -1991,6 +1992,7 @@ const struct url_test url_tests[] =
       ,{ 19,  1 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2009,6 +2011,7 @@ const struct url_test url_tests[] =
       ,{  0,  0 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2027,6 +2030,7 @@ const struct url_test url_tests[] =
       ,{ 17,  1 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2045,6 +2049,7 @@ const struct url_test url_tests[] =
       ,{ 20,  1 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2063,6 +2068,7 @@ const struct url_test url_tests[] =
       ,{  0,  0 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2081,6 +2087,7 @@ const struct url_test url_tests[] =
       ,{ 17, 12 } /* UF_PATH */
       ,{ 30,187 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2099,6 +2106,7 @@ const struct url_test url_tests[] =
       ,{  0, 10 } /* UF_PATH */
       ,{ 11, 10 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2118,6 +2126,7 @@ const struct url_test url_tests[] =
       ,{  0, 10 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{ 11,  4 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2136,6 +2145,7 @@ const struct url_test url_tests[] =
       ,{ 29,  6 } /* UF_PATH */
       ,{ 36, 69 } /* UF_QUERY */
       ,{106,  7 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2154,6 +2164,7 @@ const struct url_test url_tests[] =
       ,{ 20,  8 } /* UF_PATH */
       ,{ 29, 12 } /* UF_QUERY */
       ,{ 42,  4 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2254,6 +2265,7 @@ const struct url_test url_tests[] =
       ,{  0,  9 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0
@@ -2270,6 +2282,7 @@ const struct url_test url_tests[] =
       ,{  0,  9 } /* UF_PATH */
       ,{  0,  0 } /* UF_QUERY */
       ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
       }
     }
   ,.rv=0

From fe31fe8bf0ff8028123698d3097347ffdff3c214 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Sun, 8 Jul 2012 02:00:31 +0200
Subject: [PATCH 05/14] Add tests on user info

---
 test.c | 127 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 127 insertions(+)

diff --git a/test.c b/test.c
index f7dfcfe..a9fc05f 100644
--- a/test.c
+++ b/test.c
@@ -51,6 +51,7 @@ struct message {
   char body[MAX_ELEMENT_SIZE];
   size_t body_size;
   const char *host;
+  const char *userinfo;
   uint16_t port;
   int num_headers;
   enum { NONE=0, FIELD, VALUE } last_header_element;
@@ -631,6 +632,7 @@ const struct message requests[] =
   ,.fragment= ""
   ,.request_path= ""
   ,.request_url= "http://hypnotoad.org?hail=all"
+  ,.host= "hypnotoad.org"
   ,.num_headers= 0
   ,.headers= { }
   ,.body= ""
@@ -873,6 +875,28 @@ const struct message requests[] =
   ,.body= ""
   }
 
+#define PROXY_WITH_BASIC_AUTH 33
+, {.name= "host:port and basic_auth"
+  ,.type= HTTP_REQUEST
+  ,.raw= "GET http://a%12:b!&*$@hypnotoad.org:1234/toto HTTP/1.1\r\n"
+         "\r\n"
+  ,.should_keep_alive= TRUE
+  ,.message_complete_on_eof= FALSE
+  ,.http_major= 1
+  ,.http_minor= 1
+  ,.method= HTTP_GET
+  ,.fragment= ""
+  ,.request_path= "/toto"
+  ,.request_url= "http://a%12:b!&*$@hypnotoad.org:1234/toto"
+  ,.host= "hypnotoad.org"
+  ,.userinfo= "a%12:b!&*$"
+  ,.port= 1234
+  ,.num_headers= 0
+  ,.headers= { }
+  ,.body= ""
+  }
+
+
 , {.name= NULL } /* sentinel */
 };
 
@@ -1801,6 +1825,10 @@ message_eq (int index, const struct message *expected)
       MESSAGE_CHECK_URL_EQ(&u, expected, m, host, UF_HOST);
     }
 
+    if (expected->userinfo) {
+      MESSAGE_CHECK_URL_EQ(&u, expected, m, userinfo, UF_USERINFO);
+    }
+
     m->port = (u.field_set & (1 << UF_PORT)) ?
       u.port : 0;
 
@@ -2170,6 +2198,25 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
+, {.name="complex URL with basic auth from node js url parser doc"
+  ,.url="http://a:b@host.com:8080/p/a/t/h?query=string#hash"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PORT) | (1<<UF_PATH) | (1<<UF_QUERY) | (1<<UF_FRAGMENT) | (1<<UF_USERINFO)
+    ,.port=8080
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{ 11,  8 } /* UF_HOST */
+      ,{ 20,  4 } /* UF_PORT */
+      ,{ 24,  8 } /* UF_PATH */
+      ,{ 33, 12 } /* UF_QUERY */
+      ,{ 46,  4 } /* UF_FRAGMENT */
+      ,{  7,  3 } /* UF_USERINFO */
+      }
+    }
+  ,.rv=0
+  }
+
 , {.name="proxy empty host"
   ,.url="http://:443/"
   ,.is_connect=0
@@ -2182,6 +2229,12 @@ const struct url_test url_tests[] =
   ,.rv=1
   }
 
+, {.name="CONNECT with basic auth"
+  ,.url="a:b@hostname:443"
+  ,.is_connect=1
+  ,.rv=1
+  }
+
 , {.name="CONNECT empty host"
   ,.url=":443"
   ,.is_connect=1
@@ -2205,6 +2258,25 @@ const struct url_test url_tests[] =
   ,.rv=1 /* s_dead */
   }
 
+, {.name="basic auth with space url encoded"
+  ,.url="http://a%20:b@host.com/"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH) | (1<<UF_USERINFO)
+    ,.port=0
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{ 14,  8 } /* UF_HOST */
+      ,{  0,  0 } /* UF_PORT */
+      ,{ 22,  1 } /* UF_PATH */
+      ,{  0,  0 } /* UF_QUERY */
+      ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  7,  6 } /* UF_USERINFO */
+      }
+    }
+  ,.rv=0
+  }
+
 , {.name="carriage return in URL"
   ,.url="/foo\rbar/"
   ,.rv=1 /* s_dead */
@@ -2215,11 +2287,47 @@ const struct url_test url_tests[] =
   ,.rv=1 /* s_dead */
   }
 
+, {.name="basic auth with double :"
+  ,.url="http://a::b@host.com/"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH) | (1<<UF_USERINFO)
+    ,.port=0
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{ 12,  8 } /* UF_HOST */
+      ,{  0,  0 } /* UF_PORT */
+      ,{ 20,  1 } /* UF_PATH */
+      ,{  0,  0 } /* UF_QUERY */
+      ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  7,  4 } /* UF_USERINFO */
+      }
+    }
+  ,.rv=0
+  }
+
 , {.name="line feed in URL"
   ,.url="/foo\nbar/"
   ,.rv=1 /* s_dead */
   }
 
+, {.name="empty basic auth"
+  ,.url="http://@hostname/fo"
+  ,.u=
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH)
+    ,.port=0
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{  8,  8 } /* UF_HOST */
+      ,{  0,  0 } /* UF_PORT */
+      ,{ 16,  3 } /* UF_PATH */
+      ,{  0,  0 } /* UF_QUERY */
+      ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  0,  0 } /* UF_USERINFO */
+      }
+    }
+  ,.rv=0
+  }
 , {.name="line feed in hostname"
   ,.url="http://host\name/fo"
   ,.rv=1 /* s_dead */
@@ -2235,6 +2343,25 @@ const struct url_test url_tests[] =
   ,.rv=1 /* s_dead */
   }
 
+, {.name="basic auth with unreservedchars"
+  ,.url="http://a!;-_!=+$@host.com/"
+  ,.is_connect=0
+  ,.u=
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH) | (1<<UF_USERINFO)
+    ,.port=0
+    ,.field_data=
+      {{  0,  4 } /* UF_SCHEMA */
+      ,{ 17,  8 } /* UF_HOST */
+      ,{  0,  0 } /* UF_PORT */
+      ,{ 25,  1 } /* UF_PATH */
+      ,{  0,  0 } /* UF_QUERY */
+      ,{  0,  0 } /* UF_FRAGMENT */
+      ,{  7,  9 } /* UF_USERINFO */
+      }
+    }
+  ,.rv=0
+  }
+
 , {.name="= in URL"
   ,.url="http://host=ame/fo"
   ,.rv=1 /* s_dead */

From f6f761596ed58525ef17db5a54f80247a00518b7 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Sun, 8 Jul 2012 11:58:18 +0200
Subject: [PATCH 06/14] Small refactoring, add edge cases

---
 http_parser.c | 30 +++++++++++++++++-------------
 test.c        | 28 +++++++++++++++++++---------
 2 files changed, 36 insertions(+), 22 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index 5ea352d..99a6b62 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -1967,21 +1967,21 @@ http_parse_host_char(enum http_host_state s, const char ch) {
   return s_http_host_dead;
 }
 
-int http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
+#include <stdio.h>
+
+static int
+http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
   enum http_host_state s;
 
-  int i;
-  int start = u->field_data[UF_HOST].off;
-  int len = u->field_data[UF_HOST].len;
+  const char *p;
+  size_t buflen = u->field_data[UF_HOST].off + u->field_data[UF_HOST].len;
 
   u->field_data[UF_HOST].len = 0;
 
-  i = 0;
   s = found_at ? s_http_userinfo_start : s_http_host_start;
 
-  for(; i < len; i ++) {
-    char ch = buf[start + i];
-    enum http_host_state new_s = http_parse_host_char(s, ch);
+  for (p = buf + u->field_data[UF_HOST].off; p < buf + buflen; p++) {
+    enum http_host_state new_s = http_parse_host_char(s, *p);
 
     if (new_s == s_http_host_dead) {
       return 1;
@@ -1990,21 +1990,21 @@ int http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
     switch(new_s) {
       case s_http_host:
         if (s != s_http_host) {
-          u->field_data[UF_HOST].off = start + i;
+          u->field_data[UF_HOST].off = p - buf;
         }
         u->field_data[UF_HOST].len ++;
         break;
 
       case s_http_host_v6:
         if (s != s_http_host_v6) {
-          u->field_data[UF_HOST].off = start + i;
+          u->field_data[UF_HOST].off = p - buf;
         }
         u->field_data[UF_HOST].len ++;
         break;
 
       case s_http_host_port:
         if (s != s_http_host_port) {
-          u->field_data[UF_PORT].off = start + i;
+          u->field_data[UF_PORT].off = p - buf;
           u->field_data[UF_PORT].len = 0;
           u->field_set |= (1 << UF_PORT);
         }
@@ -2013,7 +2013,7 @@ int http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
 
       case s_http_userinfo:
         if (s != s_http_userinfo) {
-          u->field_data[UF_USERINFO].off = start + i;
+          u->field_data[UF_USERINFO].off = p - buf ;
           u->field_data[UF_USERINFO].len = 0;
           u->field_set |= (1 << UF_USERINFO);
         }
@@ -2028,9 +2028,12 @@ int http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
 
   /* Make sure we don't end somewhere unexpected */
   switch (s) {
+    case s_http_host_start:
     case s_http_host_v6_start:
     case s_http_host_v6:
     case s_http_host_port_start:
+    case s_http_userinfo:
+    case s_http_userinfo_start:
       return 1;
     default:
       break;
@@ -2106,7 +2109,8 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
     old_uf = uf;
   }
 
-  if ((u->field_set & (1 << UF_HOST)) != 0) {
+  // host must be present if there is a schema
+  if ((u->field_set & ((1 << UF_SCHEMA) | (1 << UF_HOST))) != 0) {
     if (http_parse_host(buf, u, found_at) != 0) {
       return 1;
     }
diff --git a/test.c b/test.c
index a9fc05f..f50ece7 100644
--- a/test.c
+++ b/test.c
@@ -2258,7 +2258,7 @@ const struct url_test url_tests[] =
   ,.rv=1 /* s_dead */
   }
 
-, {.name="basic auth with space url encoded"
+, {.name="proxy basic auth with space url encoded"
   ,.url="http://a%20:b@host.com/"
   ,.is_connect=0
   ,.u=
@@ -2282,12 +2282,12 @@ const struct url_test url_tests[] =
   ,.rv=1 /* s_dead */
   }
 
-, {.name="double : in URL"
+, {.name="proxy double : in URL"
   ,.url="http://hostname::443/"
   ,.rv=1 /* s_dead */
   }
 
-, {.name="basic auth with double :"
+, {.name="proxy basic auth with double :"
   ,.url="http://a::b@host.com/"
   ,.is_connect=0
   ,.u=
@@ -2311,7 +2311,7 @@ const struct url_test url_tests[] =
   ,.rv=1 /* s_dead */
   }
 
-, {.name="empty basic auth"
+, {.name="proxy empty basic auth"
   ,.url="http://@hostname/fo"
   ,.u=
     {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH)
@@ -2328,22 +2328,22 @@ const struct url_test url_tests[] =
     }
   ,.rv=0
   }
-, {.name="line feed in hostname"
+, {.name="proxy line feed in hostname"
   ,.url="http://host\name/fo"
   ,.rv=1 /* s_dead */
   }
 
-, {.name="% in hostname"
+, {.name="proxy % in hostname"
   ,.url="http://host%name/fo"
   ,.rv=1 /* s_dead */
   }
 
-, {.name="; in hostname"
+, {.name="proxy ; in hostname"
   ,.url="http://host;ame/fo"
   ,.rv=1 /* s_dead */
   }
 
-, {.name="basic auth with unreservedchars"
+, {.name="proxy basic auth with unreservedchars"
   ,.url="http://a!;-_!=+$@host.com/"
   ,.is_connect=0
   ,.u=
@@ -2362,7 +2362,17 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
-, {.name="= in URL"
+, {.name="proxy only basic auth"
+  ,.url="http://@/fo"
+  ,.rv=1 /* s_dead */
+  }
+
+, {.name="proxy emtpy hostname"
+  ,.url="http:///fo"
+  ,.rv=1 /* s_dead */
+  }
+
+, {.name="proxy = in URL"
   ,.url="http://host=ame/fo"
   ,.rv=1 /* s_dead */
   }

From e42aa93b5264c1f4e1d3201e484db26f7500aeaf Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Sun, 8 Jul 2012 16:06:22 +0200
Subject: [PATCH 07/14] Add tests

---
 test.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/test.c b/test.c
index f50ece7..c85e95c 100644
--- a/test.c
+++ b/test.c
@@ -2045,6 +2045,12 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
+, {.name="CONNECT request but not connect"
+  ,.url="hostname:443"
+  ,.is_connect=0
+  ,.rv=1
+  }
+
 , {.name="proxy ipv6 request"
   ,.url="http://[1:2::3:4]/"
   ,.is_connect=0
@@ -2362,11 +2368,16 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
-, {.name="proxy only basic auth"
+, {.name="proxy only empty basic auth"
   ,.url="http://@/fo"
   ,.rv=1 /* s_dead */
   }
 
+, {.name="proxy only basic auth"
+  ,.url="http://toto@/fo"
+  ,.rv=1 /* s_dead */
+  }
+
 , {.name="proxy emtpy hostname"
   ,.url="http:///fo"
   ,.rv=1 /* s_dead */

From bb29f437411d8aa6403bc4e1d913320b783e6417 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Tue, 17 Jul 2012 08:37:39 +0200
Subject: [PATCH 08/14] Coding style improvment

---
 http_parser.c | 21 +++++++++++----------
 test.c        | 16 +++++++++++-----
 2 files changed, 22 insertions(+), 15 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index 99a6b62..ed5477f 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -354,8 +354,12 @@ enum http_host_state
 #define IS_NUM(c)           ((c) >= '0' && (c) <= '9')
 #define IS_ALPHANUM(c)      (IS_ALPHA(c) || IS_NUM(c))
 #define IS_HEX(c)           (IS_NUM(c) || (LOWER(c) >= 'a' && LOWER(c) <= 'f'))
-#define IS_MARK(c)          ((c) == '-' || (c) == '_' || (c) == '.' || (c) == '!' || (c) == '~' || (c) == '*' || (c) == '\'' || (c) == '(' || (c) == ')')
-#define IS_USERINFO_CHAR(c) (IS_ALPHANUM(c) || IS_MARK(c) || (c) == '%' || (c) == ';' || (c) == ':' || (c) == '&' || (c) == '=' || (c) == '+' || (c) == '$' || (c) == ',')
+#define IS_MARK(c)          ((c) == '-' || (c) == '_' || (c) == '.' || \
+  (c) == '!' || (c) == '~' || (c) == '*' || (c) == '\'' || (c) == '(' || \
+  (c) == ')')
+#define IS_USERINFO_CHAR(c) (IS_ALPHANUM(c) || IS_MARK(c) || (c) == '%' || \
+  (c) == ';' || (c) == ':' || (c) == '&' || (c) == '=' || (c) == '+' || \
+  (c) == '$' || (c) == ',')
 
 #if HTTP_PARSER_STRICT
 #define TOKEN(c)            (tokens[(unsigned char)c])
@@ -412,7 +416,7 @@ int http_message_needs_eof(http_parser *parser);
  * URL and non-URL states by looking for these.
  */
 static enum state
-parse_url_char(enum state s, const char ch, int * found_at)
+parse_url_char(enum state s, const char ch, int *found_at)
 {
   if (ch == ' ' || ch == '\r' || ch == '\n') {
     return s_dead;
@@ -970,7 +974,7 @@ size_t http_parser_execute (http_parser *parser,
           parser->state = s_req_host_start;
         }
 
-        parser->state = parse_url_char((enum state)parser->state, ch, 0);
+        parser->state = parse_url_char((enum state)parser->state, ch, NULL);
         if (parser->state == s_dead) {
           SET_ERRNO(HPE_INVALID_URL);
           goto error;
@@ -992,7 +996,7 @@ size_t http_parser_execute (http_parser *parser,
             SET_ERRNO(HPE_INVALID_URL);
             goto error;
           default:
-            parser->state = parse_url_char((enum state)parser->state, ch, 0);
+            parser->state = parse_url_char((enum state)parser->state, ch, NULL);
             if (parser->state == s_dead) {
               SET_ERRNO(HPE_INVALID_URL);
               goto error;
@@ -1024,7 +1028,7 @@ size_t http_parser_execute (http_parser *parser,
             CALLBACK_DATA(url);
             break;
           default:
-            parser->state = parse_url_char((enum state)parser->state, ch, 0);
+            parser->state = parse_url_char((enum state)parser->state, ch, NULL);
             if (parser->state == s_dead) {
               SET_ERRNO(HPE_INVALID_URL);
               goto error;
@@ -1927,7 +1931,6 @@ http_parse_host_char(enum http_host_state s, const char ch) {
       break;
 
     case s_http_host:
-
       if (IS_HOST_CHAR(ch)) {
         return s_http_host;
       }
@@ -1967,8 +1970,6 @@ http_parse_host_char(enum http_host_state s, const char ch) {
   return s_http_host_dead;
 }
 
-#include <stdio.h>
-
 static int
 http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
   enum http_host_state s;
@@ -2109,7 +2110,7 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
     old_uf = uf;
   }
 
-  // host must be present if there is a schema
+  /* host must be present if there is a schema */
   if ((u->field_set & ((1 << UF_SCHEMA) | (1 << UF_HOST))) != 0) {
     if (http_parse_host(buf, u, found_at) != 0) {
       return 1;
diff --git a/test.c b/test.c
index c85e95c..787beab 100644
--- a/test.c
+++ b/test.c
@@ -2109,7 +2109,9 @@ const struct url_test url_tests[] =
   }
 
 , {.name="extra ? in query string"
-  ,.url="http://a.tbcdn.cn/p/fp/2010c/??fp-header-min.css,fp-base-min.css,fp-channel-min.css,fp-product-min.css,fp-mall-min.css,fp-category-min.css,fp-sub-min.css,fp-gdp4p-min.css,fp-css3-min.css,fp-misc-min.css?t=20101022.css"
+  ,.url="http://a.tbcdn.cn/p/fp/2010c/??fp-header-min.css,fp-base-min.css,"
+  "fp-channel-min.css,fp-product-min.css,fp-mall-min.css,fp-category-min.css,"
+  "fp-sub-min.css,fp-gdp4p-min.css,fp-css3-min.css,fp-misc-min.css?t=20101022.css"
   ,.is_connect=0
   ,.u=
     {.field_set=(1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH) | (1<<UF_QUERY)
@@ -2167,10 +2169,12 @@ const struct url_test url_tests[] =
   }
 
 , {.name="complex URL fragment"
-  ,.url="http://www.webmasterworld.com/r.cgi?f=21&d=8405&url=http://www.example.com/index.html?foo=bar&hello=world#midpage"
+  ,.url="http://www.webmasterworld.com/r.cgi?f=21&d=8405&url="
+    "http://www.example.com/index.html?foo=bar&hello=world#midpage"
   ,.is_connect=0
   ,.u=
-    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH) | (1<<UF_QUERY) | (1<<UF_FRAGMENT)
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PATH) | (1<<UF_QUERY) |\
+      (1<<UF_FRAGMENT)
     ,.port=0
     ,.field_data=
       {{  0,  4 } /* UF_SCHEMA */
@@ -2189,7 +2193,8 @@ const struct url_test url_tests[] =
   ,.url="http://host.com:8080/p/a/t/h?query=string#hash"
   ,.is_connect=0
   ,.u=
-    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PORT) | (1<<UF_PATH) | (1<<UF_QUERY) | (1<<UF_FRAGMENT)
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PORT) | (1<<UF_PATH) |\
+      (1<<UF_QUERY) | (1<<UF_FRAGMENT)
     ,.port=8080
     ,.field_data=
       {{  0,  4 } /* UF_SCHEMA */
@@ -2208,7 +2213,8 @@ const struct url_test url_tests[] =
   ,.url="http://a:b@host.com:8080/p/a/t/h?query=string#hash"
   ,.is_connect=0
   ,.u=
-    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PORT) | (1<<UF_PATH) | (1<<UF_QUERY) | (1<<UF_FRAGMENT) | (1<<UF_USERINFO)
+    {.field_set= (1<<UF_SCHEMA) | (1<<UF_HOST) | (1<<UF_PORT) | (1<<UF_PATH) |\
+      (1<<UF_QUERY) | (1<<UF_FRAGMENT) | (1<<UF_USERINFO)
     ,.port=8080
     ,.field_data=
       {{  0,  4 } /* UF_SCHEMA */

From d2ce562338b999ede72a9cd3476f8b33e1d63c73 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Tue, 17 Jul 2012 08:48:46 +0200
Subject: [PATCH 09/14] Use new state instead of pointer

---
 http_parser.c | 28 +++++++++++++++++++---------
 test.c        |  6 ++++++
 2 files changed, 25 insertions(+), 9 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index ed5477f..8784af6 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -255,6 +255,7 @@ enum state
   , s_req_schema_slash_slash
   , s_req_host_start
   , s_req_host
+  , s_req_host_with_at
   , s_req_path
   , s_req_query_string_start
   , s_req_query_string
@@ -416,7 +417,7 @@ int http_message_needs_eof(http_parser *parser);
  * URL and non-URL states by looking for these.
  */
 static enum state
-parse_url_char(enum state s, const char ch, int *found_at)
+parse_url_char(enum state s, const char ch)
 {
   if (ch == ' ' || ch == '\r' || ch == '\n') {
     return s_dead;
@@ -469,6 +470,12 @@ parse_url_char(enum state s, const char ch, int *found_at)
 
       break;
 
+    case s_req_host_with_at:
+      if (ch == '@') {
+        return s_dead;
+      }
+
+    /* FALLTHROUGH */
     case s_req_host_start:
     case s_req_host:
       if (ch == '/') {
@@ -484,10 +491,7 @@ parse_url_char(enum state s, const char ch, int *found_at)
       }
 
       if (ch == '@') {
-        if (found_at) {
-          *found_at = 1;
-        }
-        return s_req_host;
+        return s_req_host_with_at;
       }
 
       break;
@@ -611,6 +615,7 @@ size_t http_parser_execute (http_parser *parser,
   case s_req_schema_slash_slash:
   case s_req_host_start:
   case s_req_host:
+  case s_req_host_with_at:
   case s_req_query_string_start:
   case s_req_query_string:
   case s_req_fragment_start:
@@ -974,7 +979,7 @@ size_t http_parser_execute (http_parser *parser,
           parser->state = s_req_host_start;
         }
 
-        parser->state = parse_url_char((enum state)parser->state, ch, NULL);
+        parser->state = parse_url_char((enum state)parser->state, ch);
         if (parser->state == s_dead) {
           SET_ERRNO(HPE_INVALID_URL);
           goto error;
@@ -996,7 +1001,7 @@ size_t http_parser_execute (http_parser *parser,
             SET_ERRNO(HPE_INVALID_URL);
             goto error;
           default:
-            parser->state = parse_url_char((enum state)parser->state, ch, NULL);
+            parser->state = parse_url_char((enum state)parser->state, ch);
             if (parser->state == s_dead) {
               SET_ERRNO(HPE_INVALID_URL);
               goto error;
@@ -1007,6 +1012,7 @@ size_t http_parser_execute (http_parser *parser,
       }
 
       case s_req_host:
+      case s_req_host_with_at:
       case s_req_path:
       case s_req_query_string_start:
       case s_req_query_string:
@@ -1028,7 +1034,7 @@ size_t http_parser_execute (http_parser *parser,
             CALLBACK_DATA(url);
             break;
           default:
-            parser->state = parse_url_char((enum state)parser->state, ch, NULL);
+            parser->state = parse_url_char((enum state)parser->state, ch);
             if (parser->state == s_dead) {
               SET_ERRNO(HPE_INVALID_URL);
               goto error;
@@ -2057,7 +2063,7 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
   uf = old_uf = UF_MAX;
 
   for (p = buf; p < buf + buflen; p++) {
-    s = parse_url_char(s, *p, &found_at);
+    s = parse_url_char(s, *p);
 
     /* Figure out the next field that we're operating on */
     switch (s) {
@@ -2076,6 +2082,10 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
         uf = UF_SCHEMA;
         break;
 
+      case s_req_host_with_at:
+        found_at = 1;
+
+      /* FALLTROUGH */
       case s_req_host:
         uf = UF_HOST;
         break;
diff --git a/test.c b/test.c
index 787beab..73b0289 100644
--- a/test.c
+++ b/test.c
@@ -2229,6 +2229,12 @@ const struct url_test url_tests[] =
   ,.rv=0
   }
 
+, {.name="double @"
+  ,.url="http://a:b@@hostname:443/"
+  ,.is_connect=0
+  ,.rv=1
+  }
+
 , {.name="proxy empty host"
   ,.url="http://:443/"
   ,.is_connect=0

From 1d3a0ec75250272979cc7e77ceed721c98307297 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Wed, 25 Jul 2012 00:12:40 +0200
Subject: [PATCH 10/14] Use fixed_length printf instead of copying data and do
 null-termination

---
 test.c       | 9 +++------
 url_parser.c | 9 +++------
 2 files changed, 6 insertions(+), 12 deletions(-)

diff --git a/test.c b/test.c
index 73b0289..1ca8f05 100644
--- a/test.c
+++ b/test.c
@@ -2453,7 +2453,6 @@ const struct url_test url_tests[] =
 void
 dump_url (const char *url, const struct http_parser_url *u)
 {
-  char part[512];
   unsigned int i;
 
   printf("\tfield_set: 0x%x, port: %u\n", u->field_set, u->port);
@@ -2463,14 +2462,12 @@ dump_url (const char *url, const struct http_parser_url *u)
       continue;
     }
 
-    memcpy(part, url + u->field_data[i].off, u->field_data[i].len);
-    part[u->field_data[i].len] = '\0';
-
-    printf("\tfield_data[%u]: off: %u len: %u part: \"%s\"\n",
+    printf("\tfield_data[%u]: off: %u len: %u part: \"%.*s\n",
            i,
            u->field_data[i].off,
            u->field_data[i].len,
-           part);
+           u->field_data[i].len,
+           url + u->field_data[i].off);
   }
 }
 
diff --git a/url_parser.c b/url_parser.c
index c1f3ef0..b1f9c97 100644
--- a/url_parser.c
+++ b/url_parser.c
@@ -5,7 +5,6 @@
 void
 dump_url (const char *url, const struct http_parser_url *u)
 {
-  char part[512];
   unsigned int i;
 
   printf("\tfield_set: 0x%x, port: %u\n", u->field_set, u->port);
@@ -15,14 +14,12 @@ dump_url (const char *url, const struct http_parser_url *u)
       continue;
     }
 
-    memcpy(part, url + u->field_data[i].off, u->field_data[i].len);
-    part[u->field_data[i].len] = '\0';
-
-    printf("\tfield_data[%u]: off: %u len: %u part: \"%s\"\n",
+    printf("\tfield_data[%u]: off: %u len: %u part: \"%.*s\n",
            i,
            u->field_data[i].off,
            u->field_data[i].len,
-           part);
+           u->field_data[i].len,
+           url + u->field_data[i].off);
   }
 }
 

From 7f1b191d6fce976cd44ff4463bca7c28762e4561 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Wed, 25 Jul 2012 00:13:54 +0200
Subject: [PATCH 11/14] Minor speed improvment

---
 http_parser.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index 8784af6..3671df2 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -486,14 +486,14 @@ parse_url_char(enum state s, const char ch)
         return s_req_query_string_start;
       }
 
-      if (IS_USERINFO_CHAR(ch) || ch == '[' || ch == ']') {
-        return s_req_host;
-      }
-
       if (ch == '@') {
         return s_req_host_with_at;
       }
 
+      if (IS_USERINFO_CHAR(ch) || ch == '[' || ch == ']') {
+        return s_req_host;
+      }
+
       break;
 
     case s_req_path:

From 148984cd8ddcf8c4f9c2c1baf22b6df5c7848c03 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Wed, 25 Jul 2012 00:16:53 +0200
Subject: [PATCH 12/14] Rename s_req_host* to be compliant with RFC 2396

---
 http_parser.c | 40 ++++++++++++++++++++--------------------
 1 file changed, 20 insertions(+), 20 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index 3671df2..380026e 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -253,9 +253,9 @@ enum state
   , s_req_schema
   , s_req_schema_slash
   , s_req_schema_slash_slash
-  , s_req_host_start
-  , s_req_host
-  , s_req_host_with_at
+  , s_req_server_start
+  , s_req_server
+  , s_req_server_with_at
   , s_req_path
   , s_req_query_string_start
   , s_req_query_string
@@ -465,19 +465,19 @@ parse_url_char(enum state s, const char ch)
 
     case s_req_schema_slash_slash:
       if (ch == '/') {
-        return s_req_host_start;
+        return s_req_server_start;
       }
 
       break;
 
-    case s_req_host_with_at:
+    case s_req_server_with_at:
       if (ch == '@') {
         return s_dead;
       }
 
     /* FALLTHROUGH */
-    case s_req_host_start:
-    case s_req_host:
+    case s_req_server_start:
+    case s_req_server:
       if (ch == '/') {
         return s_req_path;
       }
@@ -487,11 +487,11 @@ parse_url_char(enum state s, const char ch)
       }
 
       if (ch == '@') {
-        return s_req_host_with_at;
+        return s_req_server_with_at;
       }
 
       if (IS_USERINFO_CHAR(ch) || ch == '[' || ch == ']') {
-        return s_req_host;
+        return s_req_server;
       }
 
       break;
@@ -613,9 +613,9 @@ size_t http_parser_execute (http_parser *parser,
   case s_req_schema:
   case s_req_schema_slash:
   case s_req_schema_slash_slash:
-  case s_req_host_start:
-  case s_req_host:
-  case s_req_host_with_at:
+  case s_req_server_start:
+  case s_req_server:
+  case s_req_server_with_at:
   case s_req_query_string_start:
   case s_req_query_string:
   case s_req_fragment_start:
@@ -976,7 +976,7 @@ size_t http_parser_execute (http_parser *parser,
 
         MARK(url);
         if (parser->method == HTTP_CONNECT) {
-          parser->state = s_req_host_start;
+          parser->state = s_req_server_start;
         }
 
         parser->state = parse_url_char((enum state)parser->state, ch);
@@ -991,7 +991,7 @@ size_t http_parser_execute (http_parser *parser,
       case s_req_schema:
       case s_req_schema_slash:
       case s_req_schema_slash_slash:
-      case s_req_host_start:
+      case s_req_server_start:
       {
         switch (ch) {
           /* No whitespace allowed here */
@@ -1011,8 +1011,8 @@ size_t http_parser_execute (http_parser *parser,
         break;
       }
 
-      case s_req_host:
-      case s_req_host_with_at:
+      case s_req_server:
+      case s_req_server_with_at:
       case s_req_path:
       case s_req_query_string_start:
       case s_req_query_string:
@@ -2059,7 +2059,7 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
   int found_at = 0;
 
   u->port = u->field_set = 0;
-  s = is_connect ? s_req_host_start : s_req_spaces_before_url;
+  s = is_connect ? s_req_server_start : s_req_spaces_before_url;
   uf = old_uf = UF_MAX;
 
   for (p = buf; p < buf + buflen; p++) {
@@ -2073,7 +2073,7 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
       /* Skip delimeters */
       case s_req_schema_slash:
       case s_req_schema_slash_slash:
-      case s_req_host_start:
+      case s_req_server_start:
       case s_req_query_string_start:
       case s_req_fragment_start:
         continue;
@@ -2082,11 +2082,11 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
         uf = UF_SCHEMA;
         break;
 
-      case s_req_host_with_at:
+      case s_req_server_with_at:
         found_at = 1;
 
       /* FALLTROUGH */
-      case s_req_host:
+      case s_req_server:
         uf = UF_HOST;
         break;
 

From 50faa793f47025032021b7c75fd1eb3f7010f873 Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Wed, 25 Jul 2012 00:20:03 +0200
Subject: [PATCH 13/14] Coding style : remove space before ++

---
 http_parser.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/http_parser.c b/http_parser.c
index 380026e..b8c092a 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -1999,14 +1999,14 @@ http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
         if (s != s_http_host) {
           u->field_data[UF_HOST].off = p - buf;
         }
-        u->field_data[UF_HOST].len ++;
+        u->field_data[UF_HOST].len++;
         break;
 
       case s_http_host_v6:
         if (s != s_http_host_v6) {
           u->field_data[UF_HOST].off = p - buf;
         }
-        u->field_data[UF_HOST].len ++;
+        u->field_data[UF_HOST].len++;
         break;
 
       case s_http_host_port:
@@ -2015,7 +2015,7 @@ http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
           u->field_data[UF_PORT].len = 0;
           u->field_set |= (1 << UF_PORT);
         }
-        u->field_data[UF_PORT].len ++;
+        u->field_data[UF_PORT].len++;
         break;
 
       case s_http_userinfo:
@@ -2024,7 +2024,7 @@ http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
           u->field_data[UF_USERINFO].len = 0;
           u->field_set |= (1 << UF_USERINFO);
         }
-        u->field_data[UF_USERINFO].len ++;
+        u->field_data[UF_USERINFO].len++;
         break;
 
       default:

From a828edaf6af085e8fe9ad4a38267a1e75eb3269a Mon Sep 17 00:00:00 2001
From: Bertrand Paquet <bpaquet@octo.com>
Date: Wed, 25 Jul 2012 00:36:04 +0200
Subject: [PATCH 14/14] Add a comment

---
 http_parser.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/http_parser.c b/http_parser.c
index b8c092a..8122376 100644
--- a/http_parser.c
+++ b/http_parser.c
@@ -2121,6 +2121,7 @@ http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
   }
 
   /* host must be present if there is a schema */
+  /* parsing http:///toto will fail */
   if ((u->field_set & ((1 << UF_SCHEMA) | (1 << UF_HOST))) != 0) {
     if (http_parse_host(buf, u, found_at) != 0) {
       return 1;