http_parser: cherry-pick 3258e4a and b89f944 from upstream

Fixes #1569, parse error on multi-line HTTP headers.

diff --git a/deps/http_parser/http_parser.c b/deps/http_parser/http_parser.c
index 1453d41..e8f5364 100644 (file)
--- a/deps/http_parser/http_parser.c
+++ b/deps/http_parser/http_parser.c
@@ -241,6 +241,7 @@ enum state
   , s_header_field
   , s_header_value_start
   , s_header_value
+  , s_header_value_lws
 
   , s_header_almost_done
 
@@ -332,6 +333,7 @@ size_t http_parser_execute (http_parser *parser,
                             size_t len)
 {
   char c, ch;
+  int8_t unhex_val;
   const char *p = data, *pe;
   int64_t to_read;
 
@@ -1039,6 +1041,7 @@ size_t http_parser_execute (http_parser *parser,
       }
 
       case s_header_field_start:
+      header_field_start:
       {
         if (ch == CR) {
           state = s_headers_almost_done;
@@ -1216,7 +1219,7 @@ size_t http_parser_execute (http_parser *parser,
 
       case s_header_value_start:
       {
-        if (ch == ' ') break;
+        if (ch == ' ' || ch == '\t') break;
 
         MARK(header_value);
 
@@ -1359,7 +1362,7 @@ size_t http_parser_execute (http_parser *parser,
       {
         STRICT_CHECK(ch != LF);
 
-        state = s_header_field_start;
+        state = s_header_value_lws;
 
         switch (header_state) {
           case h_connection_keep_alive:
@@ -1377,6 +1380,18 @@ size_t http_parser_execute (http_parser *parser,
         break;
       }
 
+      case s_header_value_lws:
+      {
+        if (ch == ' ' || ch == '\t')
+          state = s_header_value_start;
+        else
+        {
+          state = s_header_field_start;
+          goto header_field_start;
+        }
+        break;
+      }
+
       case s_headers_almost_done:
       headers_almost_done:
       {
@@ -1478,9 +1493,9 @@ size_t http_parser_execute (http_parser *parser,
         assert(nread == 1);
         assert(parser->flags & F_CHUNKED);
 
-        c = unhex[(unsigned char)ch];
-        if (c == -1) goto error;
-        parser->content_length = c;
+        unhex_val = unhex[(unsigned char)ch];
+        if (unhex_val == -1) goto error;
+        parser->content_length = unhex_val;
         state = s_chunk_size;
         break;
       }
@@ -1494,9 +1509,9 @@ size_t http_parser_execute (http_parser *parser,
           break;
         }
 
-        c = unhex[(unsigned char)ch];
+        unhex_val = unhex[(unsigned char)ch];
 
-        if (c == -1) {
+        if (unhex_val == -1) {
           if (ch == ';' || ch == ' ') {
             state = s_chunk_parameters;
             break;
@@ -1505,7 +1520,7 @@ size_t http_parser_execute (http_parser *parser,
         }
 
         parser->content_length *= 16;
-        parser->content_length += c;
+        parser->content_length += unhex_val;
         break;
       }
 

diff --git a/deps/http_parser/test.c b/deps/http_parser/test.c
index a4b80a2..876cacf 100644 (file)
--- a/deps/http_parser/test.c
+++ b/deps/http_parser/test.c
@@ -582,6 +582,35 @@ const struct message requests[] =
   ,.body= ""
   }
 
+#define LINE_FOLDING_IN_HEADER 20
+, {.name= "line folding in header value"
+  ,.type= HTTP_REQUEST
+  ,.raw= "GET / HTTP/1.1\r\n"
+         "Line1:   abc\r\n"
+         "\tdef\r\n"
+         " ghi\r\n"
+         "\t\tjkl\r\n"
+         "  mno \r\n"
+         "\t \tqrs\r\n"
+         "Line2: \t line2\t\r\n"
+         "\r\n"
+  ,.should_keep_alive= TRUE
+  ,.message_complete_on_eof= FALSE
+  ,.http_major= 1
+  ,.http_minor= 1
+  ,.method= HTTP_GET
+  ,.query_string= ""
+  ,.fragment= ""
+  ,.request_path= "/"
+  ,.request_url= "/"
+  ,.num_headers= 2
+  ,.headers= { { "Line1", "abcdefghijklmno qrs" }
+             , { "Line2", "line2\t" }
+             }
+  ,.body= ""
+  }
+
+
 #define QUERY_TERMINATED_HOST 21
 , {.name= "host terminated by a query string"
   ,.type= HTTP_REQUEST
@@ -1943,7 +1972,7 @@ main (void)
     "\tRA==\r\n"
     "\t-----END CERTIFICATE-----\r\n"
     "\r\n";
-  test_simple(dumbfuck2, 0);
+  test_simple(dumbfuck2, 1);
 
 #if 0
   // NOTE(Wed Nov 18 11:57:27 CET 2009) this seems okay. we just read body