http-parser-lite 0.4.0 → 0.5.0

data/CHANGELOG

@@ -1,3 +1,7 @@
+== 0.5.0 (2012-07-21)
+
+* merged in pull/118 upstream to support url parsing with username and password.
+
 == 0.4.0 (2012-06-30)
 
 * added missing on_headers_complete callback.

data/README.md

@@ -22,7 +22,7 @@ parser.on_message_complete do
 end
 
 parser.on_headers_complete do
-  puts "value: #{value}"
+  puts "headers complete"
 end
 
 parser.on_url do |url|

data/ext/http-parser/http_parser.c

@@ -254,12 +254,8 @@ 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_host_with_at
   , s_req_path
   , s_req_query_string_start
   , s_req_query_string
@@ -337,6 +333,19 @@ enum header_states
   , h_connection_close
   };
 
+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
+  , 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'
@@ -346,6 +355,12 @@ enum header_states
 #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])
@@ -455,62 +470,28 @@ 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;
+    case s_req_host_with_at:
+      if (ch == '@') {
+        return s_dead;
       }
 
-      break;
-
+    /* FALLTHROUGH */
+    case s_req_host_start:
     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;
+      if (ch == '/') {
+        return s_req_path;
       }
 
-      /* FALLTHROUGH */
-    case s_req_host_v6_start:
-      if (IS_HEX(ch) || ch == ':') {
-        return s_req_host_v6;
+      if (ch == '?') {
+        return s_req_query_string_start;
       }
-      break;
-
-    case s_req_port:
-      switch (ch) {
-        case '/':
-          return s_req_path;
 
-        case '?':
-          return s_req_query_string_start;
+      if (IS_USERINFO_CHAR(ch) || ch == '[' || ch == ']') {
+        return s_req_host;
       }
 
-      /* FALLTHROUGH */
-    case s_req_port_start:
-      if (IS_NUM(ch)) {
-        return s_req_port;
+      if (ch == '@') {
+        return s_req_host_with_at;
       }
 
       break;
@@ -633,12 +614,8 @@ 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_host_with_at:
   case s_req_query_string_start:
   case s_req_query_string:
   case s_req_fragment_start:
@@ -1015,9 +992,6 @@ 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 +1012,7 @@ 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_host_with_at:
       case s_req_path:
       case s_req_query_string_start:
       case s_req_query_string:
@@ -1938,6 +1911,144 @@ 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_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;
+      }
+
+      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;
+}
+
+static int
+http_parse_host(const char * buf, struct http_parser_url *u, int found_at) {
+  enum http_host_state s;
+
+  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;
+
+  s = found_at ? s_http_userinfo_start : s_http_host_start;
+
+  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;
+    }
+
+    switch(new_s) {
+      case s_http_host:
+        if (s != s_http_host) {
+          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 = 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 = p - buf;
+          u->field_data[UF_PORT].len = 0;
+          u->field_set |= (1 << UF_PORT);
+        }
+        u->field_data[UF_PORT].len ++;
+        break;
+
+      case s_http_userinfo:
+        if (s != s_http_userinfo) {
+          u->field_data[UF_USERINFO].off = p - buf ;
+          u->field_data[UF_USERINFO].len = 0;
+          u->field_set |= (1 << UF_USERINFO);
+        }
+        u->field_data[UF_USERINFO].len ++;
+        break;
+
+      default:
+        break;
+    }
+    s = new_s;
+  }
+
+  /* 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;
+  }
+
+  return 0;
+}
+
 int
 http_parser_parse_url(const char *buf, size_t buflen, int is_connect,
                       struct http_parser_url *u)
@@ -1945,6 +2056,7 @@ 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;
@@ -1962,9 +2074,6 @@ 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;
@@ -1973,15 +2082,14 @@ 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:
-      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 +2120,16 @@ 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;
+  /* 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;
+    }
   }
 
-  /* 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)) {

data/ext/http-parser/http_parser.h

@@ -141,7 +141,7 @@ enum flags
 
 
 /* Map for errno-related constants
- * 
+ *
  * The provided argument should be a macro that takes 2 arguments.
  */
 #define HTTP_ERRNO_MAP(XX)                                           \
@@ -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
   };
 
 

data/lib/http-parser-lite.rb

@@ -0,0 +1 @@
+require 'http-parser'

data/test/test_http_parser.rb

@@ -154,4 +154,17 @@ describe 'http-parser' do
     assert !parser.error?
     assert !parser.error
   end
+
+  it 'should parser urls with user:pass' do
+    parser.reset(HTTP::Parser::TYPE_REQUEST)
+
+    url  = 'http://foo:bar@example.org/test.cgi?param1=1'
+    data = []
+    parser.on_url {|url| data << url}
+
+    parser << "GET #{url} HTTP/1.0\r\n\r\n"
+
+    assert !parser.error?
+    assert_equal url, data.first
+  end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: http-parser-lite
 version: !ruby/object:Gem::Version
-  version: 0.4.0
+  version: 0.5.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-06-30 00:00:00.000000000 Z
+date: 2012-07-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -35,13 +35,14 @@ extensions:
 - ext/http-parser/extconf.rb
 extra_rdoc_files: []
 files:
-- ext/http-parser/http_parser.c
 - ext/http-parser/ruby_http_parser.c
+- ext/http-parser/http_parser.c
 - ext/http-parser/http_parser.h
 - ext/http-parser/extconf.rb
 - test/helper.rb
 - test/test_http_parser.rb
 - lib/http-parser.rb
+- lib/http-parser-lite.rb
 - README.md
 - CHANGELOG
 homepage: http://github.com/deepfryed/http-parser-lite
@@ -70,3 +71,4 @@ signing_key:
 specification_version: 3
 summary: Simple wrapper around Joyent http-parser
 test_files: []
+has_rdoc: