http_parser.rb 0.6.0.beta.1 → 0.6.0.beta.2

data/ext/ruby_http_parser/vendor/http-parser-java/src/impl/http_parser/lolevel/HTTPParser.java

@@ -1,6 +1,5 @@
 package http_parser.lolevel;
 
-import java.io.UnsupportedEncodingException;
 import java.nio.ByteBuffer;
 import http_parser.HTTPException;
 import http_parser.HTTPMethod;
@@ -18,7 +17,7 @@ public class  HTTPParser {
 	HState header_state;
   boolean strict;
 
-	int index; 
+	int index;
 	int flags; // TODO
 
 	int nread;
@@ -37,25 +36,25 @@ public class  HTTPParser {
    * Should be checked when http_parser_execute() returns in addition to
    * error checking.
    */
-  public boolean upgrade; 
+  public boolean upgrade;
 
   /** PUBLIC **/
 	// TODO : this is used in c to maintain application state.
 	// is this even necessary? we have state in java ?
-	// consider 
+	// consider
   // Object data; /* A pointer to get hook to the "connection" or "socket" object */
-			
 
-  /* 
+
+  /*
    * technically we could combine all of these (except for url_mark) into one
    * variable, saving stack space, but it seems more clear to have them
-   * separated. 
+   * separated.
    */
   int header_field_mark = -1;
   int header_value_mark = -1;
   int url_mark = -1;
   int body_mark = -1;
-	
+
   /**
    * Construct a Parser for ParserType.HTTP_BOTH, meaning it
    * determines whether it's parsing a request or a response.
@@ -63,9 +62,9 @@ public class  HTTPParser {
 	public HTTPParser() {
 		this(ParserType.HTTP_BOTH);
 	}
-	
+
   /**
-   * Construct a Parser and initialise it to parse either 
+   * Construct a Parser and initialise it to parse either
    * requests or responses.
    */
 	public HTTPParser(ParserType type) {
@@ -84,7 +83,7 @@ public class  HTTPParser {
 				throw new HTTPException("can't happen, invalid ParserType enum");
 		}
 	}
-	
+
   /*
    * Utility to facilitate System.out.println style debugging (the way god intended)
    */
@@ -267,12 +266,12 @@ public class  HTTPParser {
     int p     = data.position();
     this.p_start = p; // this is used for pretty printing errors.
                       // and returning the amount of processed bytes.
-    
+
 
     // In case the headers don't provide information about the content
     // length, `execute` needs to be called with an empty buffer to
     // indicate that all the data has been send be the client/server,
-    // else there is no way of knowing the message is complete. 
+    // else there is no way of knowing the message is complete.
     int len = (data.limit() - data.position());
     if (0 == len) {
 //      			if (State.body_identity_eof == state) {
@@ -295,7 +294,7 @@ public class  HTTPParser {
       }
     }
 
-		
+
     // in case the _previous_ call to the parser only has data to get to
     // the middle of certain fields, we need to update marks to point at
     // the beginning of the current buffer.
@@ -375,13 +374,16 @@ public class  HTTPParser {
 
           if (H == ch) {
             state = State.res_or_resp_H;
-            settings.call_on_message_begin(this);
           } else {
             type   = ParserType.HTTP_REQUEST;
-            state  = State.start_req;
-            index = 1;
-            reexecute = true;
+            method = start_req_method_assign(ch);
+            if (null == method) {
+              return error(settings, "invalid method", data);
+            }
+            index  = 1;
+            state  = State.req_method;
           }
+          settings.call_on_message_begin(this);
           break;
 
 
@@ -474,7 +476,7 @@ return error(settings, "Not a digit", data);
 return error(settings, "invalid http major version: ", data);
           }
           break;
-          
+
         /* first digit of minor HTTP version */
         case res_first_http_minor:
           if (!isDigit(ch)) {
@@ -535,18 +537,22 @@ return error(settings, "not a valid status code", data);
           if (status_code > 999) {
 return error(settings, "ridiculous status code:", data);
           }
+
+          if (status_code > 99) {
+            settings.call_on_status_complete(this);
+          }
           break;
 
         case res_status:
           /* the human readable status. e.g. "NOT FOUND"
-           * we are not humans so just ignore this 
+           * we are not humans so just ignore this
            * we are not men, we are devo. */
 
            if (CR == ch) {
             state = State.res_line_almost_done;
             break;
            }
-           if (LF == ch) { 
+           if (LF == ch) {
             state = State.header_field_start;
             break;
            }
@@ -581,14 +587,14 @@ return error(settings, "not LF", data);
 
           settings.call_on_message_begin(this);
           break;
-        
+
 
 
         case req_method:
           if (0 == ch) {
             return error(settings, "NULL in method", data);
           }
-          
+
           byte [] arr = method.bytes;
 
           if (SPACE == ch && index == arr.length) {
@@ -637,7 +643,7 @@ return error(settings, "not LF", data);
 
           ++index;
           break;
-      
+
 
 
         /******************* URL *******************/
@@ -687,7 +693,7 @@ return error(settings, "not LF", data);
         case req_fragment_start:
         case req_fragment:
           switch (ch) {
-            case SPACE: 
+            case SPACE:
               settings.call_on_url(this, data, url_mark, p-url_mark);
               settings.call_on_path(this, data, url_mark, p - url_mark);
               url_mark = -1;
@@ -781,7 +787,7 @@ return error(settings, "non digit in http major", data);
 return error(settings, "ridiculous http major", data);
           };
           break;
-        
+
         /* first digit of minor HTTP version */
         case req_first_http_minor:
           if (!isDigit(ch)) {
@@ -811,11 +817,11 @@ return error(settings, "non digit in http minor", data);
           http_minor *= 10;
           http_minor += (int)ch - 0x30;
 
-         
+
           if (http_minor > 999) {
 return error(settings, "ridiculous http minor", data);
           };
-   
+
           break;
 
         /* end of request line */
@@ -860,7 +866,7 @@ return error(settings, "missing LF after request line", data);
           state = State.header_field;
 
           switch (c) {
-            case C: 
+            case C:
               header_state = HState.C;
               break;
 
@@ -888,7 +894,7 @@ return error(settings, "missing LF after request line", data);
         case header_field:
         {
           c = token(ch);
-          if (0 != c) {  
+          if (0 != c) {
             switch (header_state) {
               case general:
                 break;
@@ -997,7 +1003,7 @@ return error(settings, "Unknown Header State", data);
           if (CR == ch) {
             state = State.header_almost_done;
             settings.call_on_header_field(this, data, header_field_mark, p-header_field_mark);
-            
+
             header_field_mark = -1;
             break;
           }
@@ -1005,7 +1011,7 @@ return error(settings, "Unknown Header State", data);
           if (ch == LF) {
             settings.call_on_header_field(this, data, header_field_mark, p-header_field_mark);
             header_field_mark = -1;
-            
+
             state = State.header_field_start;
             break;
           }
@@ -1037,7 +1043,7 @@ return error(settings, "invalid header field", data);
           if (LF == ch) {
             settings.call_on_header_value(this, data, header_value_mark, p-header_value_mark);
             header_value_mark = -1;
-            
+
             state = State.header_field_start;
             break;
           }
@@ -1063,7 +1069,7 @@ return error(settings, "invalid header field", data);
             case content_length:
               if (!isDigit(ch)) {
 return error(settings, "Content-Length not numeric", data);
-              } 
+              }
               content_length = (int)ch - 0x30;
               break;
 
@@ -1122,7 +1128,7 @@ return error(settings, "Shouldn't be here", data);
               }
               if (!isDigit(ch)) {
 return error(settings, "Content-Length not numeric", data);
-              } 
+              }
 
               long t = content_length; 
               t *= 10;
@@ -1218,7 +1224,7 @@ return error(settings, "Content-Length not numeric", data);
           /* Here we call the headers_complete callback. This is somewhat
           * different than other callbacks because if the user returns 1, we
           * will interpret that as saying that this message has no body. This
-          * is needed for the annoying case of receiving a response to a HEAD
+          * is needed for the annoying case of recieving a response to a HEAD
           * request.
           */
 
@@ -1236,27 +1242,27 @@ return error(settings, "Content-Length not numeric", data);
           * parsingHeader) implementation ...
           */
 
-//          /*TODO TODO TODO TODO TODO TODO TODO TODO TODO TODO */
-//          if (null != settings.on_headers_complete) {
-//            settings.call_on_headers_complete(this);
-//            //return;
-//          }
-
-        if (null != settings.on_headers_complete) {
-          switch (settings.on_headers_complete.cb(this)) {
-            case 0:
-              break;
-
-            case 1:
-              flags |= F_SKIPBODY;
-              break;
-
-            default:
-              return error(settings, "HPE_CB_headers_complete", data); /* Error */
-          }
-        }
-        reexecute = true;
-        break;
+          /*TODO TODO TODO TODO TODO TODO TODO TODO TODO TODO */
+          if (null != settings.on_headers_complete) {
+            settings.call_on_headers_complete(this);
+            //return;
+          }
+
+          //        if (null != settings.on_headers_complete) {
+          //          switch (settings.on_headers_complete.cb(parser)) {
+          //            case 0:
+          //              break;
+          //
+          //            case 1:
+          //              flags |= F_SKIPBODY;
+          //              break;
+          //
+          //            default:
+          //              return p - data; /* Error */ // TODO // RuntimeException ?
+          //          }
+          //        }
+          reexecute = true;
+          break;
 
         case headers_done:
           if (strict && (LF != ch)) {
@@ -1316,7 +1322,6 @@ return error(settings, "Content-Length not numeric", data);
 
             if (content_length == 0) {
               state = message_done;
-              p += to_read;
               reexecute = true;
             }
           }
@@ -1344,7 +1349,7 @@ return error(settings, "Content-Length not numeric", data);
         case chunk_size_start:
           if (1 != this.nread) {
 return error(settings, "nread != 1 (chunking)", data);
-          
+
           }
           if (0 == (flags & F_CHUNKED)) {
 return error(settings, "not chunked", data);
@@ -1401,7 +1406,7 @@ return error(settings, "not chunked", data);
             break;
           }
           break;
-          
+
 
 
         case chunk_size_almost_done:
@@ -1469,12 +1474,12 @@ return error(settings, "chunk data terminated incorrectly, expected LF", data);
           state = State.chunk_size_start;
           break;
         /******************* Chunk *******************/
-    
-        
-        
+
+
+
         default:
 return error(settings, "unhandled state", data);
-          
+
       } // switch
     } // while
 
@@ -1483,13 +1488,13 @@ return error(settings, "unhandled state", data);
 
     /* Reaching this point assumes that we only received part of a
      * message, inform the callbacks about the progress made so far*/
-    
+
 	  settings.call_on_header_field(this, data, header_field_mark, p-header_field_mark);
     settings.call_on_header_value(this, data, header_value_mark, p-header_value_mark);
     settings.call_on_url         (this, data, url_mark,          p-url_mark);
     settings.call_on_path        (this, data, url_mark,          p-url_mark);
     
-    return data.position()-this.p_start;	
+    return data.position()-this.p_start;
   } // execute
 
   int error (ParserSettings settings, String mes, ByteBuffer data) {
@@ -1645,33 +1650,19 @@ return error(settings, "unhandled state", data);
   int strtoi(ByteBuffer data, int start_pos) {
     data.position(start_pos);
     byte ch;
-    int start = data.position();
-    int end = data.limit();
+    String str = "";
     while(data.position() < data.limit()) {
       ch = data.get();
       if(Character.isWhitespace((char)ch)){
-        start++;
         continue;
       }
       if(isDigit(ch)){
-        continue;
+        str = str + (char)ch; //TODO replace with something less hacky
       }else{
-        end = data.position() - 1;
         break;
       }
     }
-    byte[] s = new byte[end - start];
-    int cur_pos = data.position();
-    data.position(start);
-    data.get(s);
-    int result = 0;
-    try{
-      result = Integer.parseInt(new String(s, "UTF8"));
-    } catch (UnsupportedEncodingException e) {
-      result = 0;
-    }
-    data.position(cur_pos);
-    return result;
+    return Integer.parseInt(str);
   }
   
   boolean isDigit(byte b) {
@@ -1727,18 +1718,18 @@ return error(settings, "unhandled state", data);
   HTTPMethod start_req_method_assign(byte c){
     switch (c) {
       case C: return HTTPMethod.HTTP_CONNECT;  /* or COPY, CHECKOUT */
-      case D: return HTTPMethod.HTTP_DELETE;  
-      case G: return HTTPMethod.HTTP_GET;     
-      case H: return HTTPMethod.HTTP_HEAD;    
-      case L: return HTTPMethod.HTTP_LOCK;    
+      case D: return HTTPMethod.HTTP_DELETE;
+      case G: return HTTPMethod.HTTP_GET;
+      case H: return HTTPMethod.HTTP_HEAD;
+      case L: return HTTPMethod.HTTP_LOCK;
       case M: return HTTPMethod.HTTP_MKCOL;    /* or MOVE, MKACTIVITY, MERGE, M-SEARCH */
-      case N: return HTTPMethod.HTTP_NOTIFY; 
-      case O: return HTTPMethod.HTTP_OPTIONS; 
+      case N: return HTTPMethod.HTTP_NOTIFY;
+      case O: return HTTPMethod.HTTP_OPTIONS;
       case P: return HTTPMethod.HTTP_POST;     /* or PROPFIND|PROPPATCH|PUT|PATCH|PURGE */
       case R: return HTTPMethod.HTTP_REPORT;
       case S: return HTTPMethod.HTTP_SUBSCRIBE;
-      case T: return HTTPMethod.HTTP_TRACE;   
-      case U: return HTTPMethod.HTTP_UNLOCK; /* or UNSUBSCRIBE */ 
+      case T: return HTTPMethod.HTTP_TRACE;
+      case U: return HTTPMethod.HTTP_UNLOCK; /* or UNSUBSCRIBE */
     }
     return null; // ugh.
   }
@@ -1777,6 +1768,7 @@ return error(settings, "unhandled state", data);
   final int min (int a, long b) {
     return a < b ? a : (int)b;
   }
+
   /* probably not the best place to hide this ... */
 	public boolean HTTP_PARSER_STRICT;
   State new_message() {
@@ -1787,7 +1779,7 @@ return error(settings, "unhandled state", data);
     }
 
   }
-	
+
   State start_state() {
     return type == ParserType.HTTP_REQUEST ? State.start_req : State.start_res;
   }
@@ -1838,28 +1830,28 @@ return error(settings, "unhandled state", data);
       0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,  0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
     };
     static final byte [] CONNECTION = {
-      0x43, 0x4f, 0x4e, 0x4e, 0x45, 0x43, 0x54, 0x49, 0x4f, 0x4e, 
+      0x43, 0x4f, 0x4e, 0x4e, 0x45, 0x43, 0x54, 0x49, 0x4f, 0x4e,
     };
     static final byte [] PROXY_CONNECTION = {
-      0x50, 0x52, 0x4f, 0x58, 0x59, 0x2d, 0x43, 0x4f, 0x4e, 0x4e, 0x45, 0x43, 0x54, 0x49, 0x4f, 0x4e, 
+      0x50, 0x52, 0x4f, 0x58, 0x59, 0x2d, 0x43, 0x4f, 0x4e, 0x4e, 0x45, 0x43, 0x54, 0x49, 0x4f, 0x4e,
     };
     static final byte [] CONTENT_LENGTH = {
-      0x43, 0x4f, 0x4e, 0x54, 0x45, 0x4e, 0x54, 0x2d, 0x4c, 0x45, 0x4e, 0x47, 0x54, 0x48, 
+      0x43, 0x4f, 0x4e, 0x54, 0x45, 0x4e, 0x54, 0x2d, 0x4c, 0x45, 0x4e, 0x47, 0x54, 0x48,
     };
     static final byte [] TRANSFER_ENCODING = {
-      0x54, 0x52, 0x41, 0x4e, 0x53, 0x46, 0x45, 0x52, 0x2d, 0x45, 0x4e, 0x43, 0x4f, 0x44, 0x49, 0x4e, 0x47, 
+      0x54, 0x52, 0x41, 0x4e, 0x53, 0x46, 0x45, 0x52, 0x2d, 0x45, 0x4e, 0x43, 0x4f, 0x44, 0x49, 0x4e, 0x47,
     };
     static final byte [] UPGRADE = {
-      0x55, 0x50, 0x47, 0x52, 0x41, 0x44, 0x45, 
+      0x55, 0x50, 0x47, 0x52, 0x41, 0x44, 0x45,
     };
     static final byte [] CHUNKED = {
-      0x43, 0x48, 0x55, 0x4e, 0x4b, 0x45, 0x44, 
+      0x43, 0x48, 0x55, 0x4e, 0x4b, 0x45, 0x44,
     };
     static final byte [] KEEP_ALIVE = {
-      0x4b, 0x45, 0x45, 0x50, 0x2d, 0x41, 0x4c, 0x49, 0x56, 0x45, 
+      0x4b, 0x45, 0x45, 0x50, 0x2d, 0x41, 0x4c, 0x49, 0x56, 0x45,
     };
     static final byte [] CLOSE = {
-      0x43, 0x4c, 0x4f, 0x53, 0x45, 
+      0x43, 0x4c, 0x4f, 0x53, 0x45,
     };
 
     /* Tokens as defined by rfc 2616. Also lowercases them.
@@ -1973,6 +1965,29 @@ return error(settings, "unhandled state", data);
      true,    true,    true,    true,    true,    true,    true,    true,
 /* 120  x   121  y   122  z   123  {   124  |   125  }   126  ~   127 del */
      true,    true,    true,    true,    true,    true,    true,   false,
+
+/*    hi bit set, not ascii                                                  */
+/*    Remainder of non-ASCII range are accepted as-is to support implicitly UTF-8
+ *    encoded paths. This is out of spec, but clients generate this and most other
+ *    HTTP servers support it. We should, too. */
+
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+     true,    true,    true,    true,    true,    true,    true,    true,
+
     };
 
     public static final byte A = 0x41;
@@ -2019,7 +2034,7 @@ return error(settings, "unhandled state", data);
 
   enum State {
 
-    dead               
+    dead
 
     , start_req_or_res
     , res_or_resp_H