Make URI parser more like the RFC.

sip/uri_parse.rl

@@ -4,7 +4,6 @@ import (
   "bytes"
   "errors"
   "fmt"
-  "strconv"
 )
 
 %% machine uri;
@@ -34,25 +33,62 @@ func ParseURIBytes(data []byte) (uri *URI, err error) {
   var hex byte
 
   %%{
-    action strStart   { amt = 0                          }
-    action strChar    { buf[amt] = fc; amt++             }
-    action strLower   { buf[amt] = fc + 0x20; amt++      }
-    action hexHi      { hex = unhex(fc) * 16             }
-    action hexLo      { hex += unhex(fc)
-                        buf[amt] = hex; amt++            }
-    action user       { uri.User = string(buf[0:amt])    }
-    action pass       { uri.Pass = string(buf[0:amt])    }
-    action host       { uri.Host = string(buf[0:amt])    }
-    action scheme     { uri.Scheme = string(buf[0:amt])  }
-    action b1         { b1 = string(buf[0:amt]); amt = 0 }
-    action b2         { b2 = string(buf[0:amt]); amt = 0 }
+    action start {
+      amt = 0
+    }
+
+    action append {
+      buf[amt] = fc
+      amt++
+    }
+
+    action hexHi {
+      hex = unhex(fc) * 16
+    }
+
+    action hexLo {
+      hex += unhex(fc)
+      buf[amt] = hex
+      amt++
+    }
+
+    action scheme {
+      uri.Scheme = string(buf[0:amt])
+    }
+
+    action user {
+      uri.User = string(buf[0:amt])
+    }
+
+    action pass {
+      uri.Pass = string(buf[0:amt])
+    }
+
+    action host {
+      uri.Host = string(buf[0:amt])
+    }
 
     action port {
-      if amt > 0 {
-        port, err := strconv.ParseUint(string(buf[0:amt]), 10, 16)
-        if err != nil { goto fail }
-        uri.Port = uint16(port)
+      uri.Port = uri.Port * 10 + uint16(fc - 0x30)
+    }
+
+    action b1 {
+      b1 = string(buf[0:amt])
+      amt = 0
+    }
+
+    action b2 {
+      b2 = string(buf[0:amt])
+      amt = 0
+    }
+
+    action lower {
+      if 'A' <= fc && fc <= 'Z' {
+        buf[amt] = fc + 0x20
+      } else {
+        buf[amt] = fc
       }
+      amt++
     }
 
     action param {
@@ -69,53 +105,45 @@ func ParseURIBytes(data []byte) (uri *URI, err error) {
       uri.Headers[b1] = b2
     }
 
-    # TODO(jart): Use BNF from SIP RFC: https://tools.ietf.org/html/rfc3261#section-25.1
-
-    # Define what a single character is allowed to be.
-    toxic         = ( cntrl | 127 ) ;
-    scary         = ( toxic | space | "\"" | "#" | "%" | "<" | ">" | "=" ) ;
-    schmchars     = ( lower | digit | "+" | "-" | "." ) ;
-    authdelims    = ( "/" | "?" | "#" | ":" | "@" | ";" | "[" | "]" | "&" ) ;
-    userchars     = any -- ( authdelims | scary ) ;
-    passchars     = userchars ;
-    hostchars     = passchars -- upper;
-    hostcharsEsc  = ( hostchars | ":" ) -- upper;
-    portchars     = digit ;
-    paramchars    = userchars -- upper ;
-    headerchars   = userchars ;
-
-    # Define how characters trigger actions.
-    escape        = "%" xdigit xdigit ;
-    unescape      = "%" ( xdigit @hexHi ) ( xdigit @hexLo ) ;
-    schmfirst     = ( upper @strLower ) | ( lower @strChar ) ;
-    schmchar      = ( upper @strLower ) | ( schmchars @strChar ) ;
-    userchar      = unescape | ( userchars @strChar ) ;
-    passchar      = unescape | ( passchars @strChar ) ;
-    hostchar      = unescape | ( upper @strLower ) | ( hostchars @strChar ) ;
-    hostcharEsc   = unescape | ( upper @strLower ) | ( hostcharsEsc @strChar ) ;
-    portchar      = unescape | ( portchars @strChar ) ;
-    paramchar     = unescape | ( upper @strLower ) | ( paramchars @strChar ) ;
-    headerchar    = unescape | ( headerchars @strChar ) ;
-
-    # Define multi-character patterns.
-    scheme        = ( schmfirst schmchar* ) >strStart %scheme ;
-    user          = userchar+ >strStart %user ;
-    pass          = passchar+ >strStart %pass ;
-    hostPlain     = hostchar+ >strStart %host ;
-    hostQuoted    = "[" ( hostcharEsc+ >strStart %host ) "]" ;
-    host          = hostQuoted | hostPlain ;
-    port          = portchar* >strStart %port ;
-    paramkey      = paramchar+ >strStart >b2 %b1 ;
-    paramval      = paramchar+ >strStart %b2 ;
-    param         = space* ";" paramkey ( "=" paramval )? %param ;
-    headerkey     = headerchar+ >strStart >b2 %b1 ;
-    headerval     = headerchar+ >strStart %b2 ;
-    header        = headerkey ( "=" headerval )? %header ;
-    headers       = "?" header ( "&" header )* ;
-    userpass      = user ( ":" pass )? ;
-    hostport      = host ( ":" port )? ;
-    uriSansUser  := space* scheme ":" hostport param* space* headers? space* ;
-    uriWithUser  := space* scheme ":" userpass "@" hostport param* space* headers? space* ;
+    # Byte character definitions.
+    mark            = "-" | "_" | "." | "!" | "~" | "*" | "'" | "(" | ")" ;
+    reserved        = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" | "$" | "," ;
+    unreserved      = alpha | digit | mark ;
+    ipv4            = digit | "." ;
+    ipv6            = xdigit | "." | ":" ;
+    hostname        = alpha | digit | "-" | "." ;
+    tel             = digit | "+" | "-" ;
+    schmchars       = alpha | digit | "+" | "-" | "." ;
+    userchars       = unreserved | "&" | "=" | "+" | "$" | "," | ";" | "?" | "/" ;
+    passchars       = unreserved | "&" | "=" | "+" | "$" | "," ;
+    paramchars      = unreserved | "[" | "]" | "/" | ":" | "&" | "+" | "$" ;
+    headerchars     = unreserved | "[" | "]" | "/" | "?" | ":" | "+" | "$" ;
+
+    # Multibyte character definitions.
+    escaped         = "%" ( xdigit @hexHi ) ( xdigit @hexLo ) ;
+    userchar        = escaped | ( userchars @append ) ;
+    passchar        = escaped | ( passchars @append ) ;
+    paramchar       = escaped | ( paramchars @lower ) ;
+    headerchar      = escaped | ( headerchars @append ) ;
+
+    # URI component definitions.
+    scheme          = ( alpha schmchars* ) >start @lower %scheme ;
+    user            = userchar+ >start %user ;
+    pass            = passchar+ >start %pass ;
+    host6           = "[" ( ipv6+ >start @lower %host ) "]" ;
+    host            = host6 | ( ( ipv4 | hostname | tel )+ >start @lower %host ) ;
+    port            = digit+ @port ;
+    paramkey        = paramchar+ >start >b2 %b1 ;
+    paramval        = paramchar+ >start %b2 ;
+    param           = space* ";" paramkey ( "=" paramval )? %param ;
+    headerkey       = headerchar+ >start >b2 %b1 ;
+    headerval       = headerchar+ >start %b2 ;
+    header          = headerkey ( "=" headerval )? %header ;
+    headers         = "?" header ( "&" header )* ;
+    userpass        = user ( ":" pass )? ;
+    hostport        = host ( ":" port )? ;
+    uriSansUser    := space* scheme ":" hostport param* space* headers? space* ;
+    uriWithUser    := space* scheme ":" userpass "@" hostport param* space* headers? space* ;
   }%%
 
   %% write init;
@@ -128,13 +156,10 @@ func ParseURIBytes(data []byte) (uri *URI, err error) {
 
   if cs < uri_first_final {
     if p == pe {
-      return nil, errors.New(fmt.Sprintf("Unexpected EOF: %s", data))
+      return nil, errors.New(fmt.Sprintf("Incomplete URI: %s", data))
     } else {
       return nil, errors.New(fmt.Sprintf("Error in URI at pos %d: %s", p, data))
     }
   }
   return uri, nil
-
-fail:
-  return nil, errors.New(fmt.Sprintf("Bad URI: %s", data))
 }

sip/uri_test.go

@@ -23,7 +23,7 @@ var uriTests = []uriTest{
 
 	uriTest{
 		s: "sip:",
-		e: errors.New("Unexpected EOF: sip:"),
+		e: errors.New("Incomplete URI: sip:"),
 	},
 
 	uriTest{
@@ -39,15 +39,6 @@ var uriTests = []uriTest{
 		},
 	},
 
-	uriTest{
-		s: "sip:example.com:",
-		uri: &sip.URI{
-			Scheme: "sip",
-			Host:   "example.com",
-		},
-		skipFormat: true,
-	},
-
 	uriTest{
 		s: "sip:example.com:5060",
 		uri: &sip.URI{
@@ -204,7 +195,15 @@ var uriTests = []uriTest{
 		},
 	},
 
-	// TODO(jart): sip:alice;day=tuesday@atlanta.com
+	uriTest{
+		s: "sip:alice;day=tuesday@atlanta.com",
+		uri: &sip.URI{
+			Scheme: "sip",
+			User:   "alice;day=tuesday",
+			Host:   "atlanta.com",
+		},
+		skipFormat: true, // TODO(jart): Fix this.
+	},
 }
 
 func TestParseURI(t *testing.T) {