xml: handle unexpected EOF while parsing and fix a bug in name

mustgetc reports unexpected EOF as SyntaxError.  using
mustgetc seems to be a better approach than letting the
caller handle unexpected EOF every time.

name: the second if statement should explicitly return
ok==false.

R=rsc
https://golang.org/cl/174083

diff --git a/src/pkg/xml/xml.go b/src/pkg/xml/xml.go
index 1a8d89ab8983039446333337f7c4f0f0a5467c29..360a39863cae3b7721a4d0731cb959a75d03362d 100644 (file)
--- a/src/pkg/xml/xml.go
+++ b/src/pkg/xml/xml.go
@@ -409,7 +409,7 @@ func (p *Parser) RawToken() (Token, os.Error) {
                return CharData(data), nil;
        }
 
-       if b, ok = p.getc(); !ok {
+       if b, ok = p.mustgetc(); !ok {
                return nil, p.err
        }
        switch b {
@@ -423,7 +423,7 @@ func (p *Parser) RawToken() (Token, os.Error) {
                        return nil, p.err;
                }
                p.space();
-               if b, ok = p.getc(); !ok {
+               if b, ok = p.mustgetc(); !ok {
                        return nil, p.err
                }
                if b != '>' {
@@ -438,17 +438,17 @@ func (p *Parser) RawToken() (Token, os.Error) {
                // the version is 1.0 and the encoding is UTF-8.
                var target string;
                if target, ok = p.name(); !ok {
-                       return nil, p.err
+                       if p.err == nil {
+                               p.err = SyntaxError("expected target name after <?")
+                       }
+                       return nil, p.err;
                }
                p.space();
                p.buf.Reset();
                var b0 byte;
                for {
-                       if b, ok = p.getc(); !ok {
-                               if p.err == os.EOF {
-                                       p.err = SyntaxError("unterminated <? directive")
-                               }
-                               return nil, p.err;
+                       if b, ok = p.mustgetc(); !ok {
+                               return nil, p.err
                        }
                        p.buf.WriteByte(b);
                        if b0 == '?' && b == '>' {
@@ -462,13 +462,13 @@ func (p *Parser) RawToken() (Token, os.Error) {
 
        case '!':
                // <!: Maybe comment, maybe CDATA.
-               if b, ok = p.getc(); !ok {
+               if b, ok = p.mustgetc(); !ok {
                        return nil, p.err
                }
                switch b {
                case '-':       // <!-
                        // Probably <!-- for a comment.
-                       if b, ok = p.getc(); !ok {
+                       if b, ok = p.mustgetc(); !ok {
                                return nil, p.err
                        }
                        if b != '-' {
@@ -479,11 +479,8 @@ func (p *Parser) RawToken() (Token, os.Error) {
                        p.buf.Reset();
                        var b0, b1 byte;
                        for {
-                               if b, ok = p.getc(); !ok {
-                                       if p.err == os.EOF {
-                                               p.err = SyntaxError("unterminated <!-- comment")
-                                       }
-                                       return nil, p.err;
+                               if b, ok = p.mustgetc(); !ok {
+                                       return nil, p.err
                                }
                                p.buf.WriteByte(b);
                                if b0 == '-' && b1 == '-' && b == '>' {
@@ -498,7 +495,7 @@ func (p *Parser) RawToken() (Token, os.Error) {
                case '[':       // <![
                        // Probably <![CDATA[.
                        for i := 0; i < 6; i++ {
-                               if b, ok = p.getc(); !ok {
+                               if b, ok = p.mustgetc(); !ok {
                                        return nil, p.err
                                }
                                if b != "CDATA["[i] {
@@ -519,7 +516,7 @@ func (p *Parser) RawToken() (Token, os.Error) {
                p.buf.Reset();
                p.buf.WriteByte(b);
                for {
-                       if b, ok = p.getc(); !ok {
+                       if b, ok = p.mustgetc(); !ok {
                                return nil, p.err
                        }
                        if b == '>' {
@@ -548,12 +545,12 @@ func (p *Parser) RawToken() (Token, os.Error) {
        attr = make([]Attr, 0, 4);
        for {
                p.space();
-               if b, ok = p.getc(); !ok {
+               if b, ok = p.mustgetc(); !ok {
                        return nil, p.err
                }
                if b == '/' {
                        empty = true;
-                       if b, ok = p.getc(); !ok {
+                       if b, ok = p.mustgetc(); !ok {
                                return nil, p.err
                        }
                        if b != '>' {
@@ -584,7 +581,7 @@ func (p *Parser) RawToken() (Token, os.Error) {
                        return nil, p.err;
                }
                p.space();
-               if b, ok = p.getc(); !ok {
+               if b, ok = p.mustgetc(); !ok {
                        return nil, p.err
                }
                if b != '=' {
@@ -592,7 +589,7 @@ func (p *Parser) RawToken() (Token, os.Error) {
                        return nil, p.err;
                }
                p.space();
-               if b, ok = p.getc(); !ok {
+               if b, ok = p.mustgetc(); !ok {
                        return nil, p.err
                }
                if b != '"' && b != '\'' {
@@ -652,6 +649,19 @@ func (p *Parser) getc() (b byte, ok bool) {
        return b, true;
 }
 
+// Must read a single byte.
+// If there is no byte to read,
+// set p.err to SyntaxError("unexpected EOF")
+// and return ok==false
+func (p *Parser) mustgetc() (b byte, ok bool) {
+       if b, ok = p.getc(); !ok {
+               if p.err == os.EOF {
+                       p.err = SyntaxError("unexpected EOF")
+               }
+       }
+       return;
+}
+
 // Unread a single byte.
 func (p *Parser) ungetc(b byte) {
        if b == '\n' {
@@ -678,7 +688,7 @@ func (p *Parser) text(quote int, cdata bool) []byte {
        p.buf.Reset();
 Input:
        for {
-               b, ok := p.getc();
+               b, ok := p.mustgetc();
                if !ok {
                        return nil
                }
@@ -717,7 +727,10 @@ Input:
                        for i = 0; i < len(p.tmp); i++ {
                                p.tmp[i], p.err = p.r.ReadByte();
                                if p.err != nil {
-                                       return nil
+                                       if p.err == os.EOF {
+                                               p.err = SyntaxError("unexpected EOF")
+                                       }
+                                       return nil;
                                }
                                c := p.tmp[i];
                                if c == ';' {
@@ -819,22 +832,23 @@ func (p *Parser) nsname() (name Name, ok bool) {
 }
 
 // Get name: /first(first|second)*/
-// Do not set p.err if the name is missing: let the caller provide better context.
+// Do not set p.err if the name is missing (unless unexpected EOF is received):
+// let the caller provide better context.
 func (p *Parser) name() (s string, ok bool) {
        var b byte;
-       if b, ok = p.getc(); !ok {
+       if b, ok = p.mustgetc(); !ok {
                return
        }
 
        // As a first approximation, we gather the bytes [A-Za-z_:.-\x80-\xFF]*
        if b < utf8.RuneSelf && !isNameByte(b) {
                p.ungetc(b);
-               return;
+               return "", false;
        }
        p.buf.Reset();
        p.buf.WriteByte(b);
        for {
-               if b, ok = p.getc(); !ok {
+               if b, ok = p.mustgetc(); !ok {
                        return
                }
                if b < utf8.RuneSelf && !isNameByte(b) {

diff --git a/src/pkg/xml/xml_test.go b/src/pkg/xml/xml_test.go
index 837385b6b2eb6f149e8ab9b6293ec7ad0814d955..e689949afe2ee014c1ff95aa9ae03d3d8d66a978 100644 (file)
--- a/src/pkg/xml/xml_test.go
+++ b/src/pkg/xml/xml_test.go
@@ -94,6 +94,57 @@ var cookedTokens = []Token{
        Comment(strings.Bytes(" missing final newline ")),
 }
 
+var xmlInput = []string{
+       // unexpected EOF cases
+       "<",
+       "<t",
+       "<t ",
+       "<t/",
+       "<t/>c",
+       "<!",
+       "<!-",
+       "<!--",
+       "<!--c-",
+       "<!--c--",
+       "<!d",
+       "<t></",
+       "<t></t",
+       "<?",
+       "<?p",
+       "<t a",
+       "<t a=",
+       "<t a='",
+       "<t a=''",
+       "<t/><![",
+       "<t/><![C",
+       "<t/><![CDATA[d",
+       "<t/><![CDATA[d]",
+       "<t/><![CDATA[d]]",
+
+       // other Syntax errors
+       " ",
+       ">",
+       "<>",
+       "<t/a",
+       "<0 />",
+       "<?0 >",
+       //      "<!0 >",        // let the Token() caller handle
+       "</0>",
+       "<t 0=''>",
+       "<t a='&'>",
+       "<t a='<'>",
+       "<t>&nbspc;</t>",
+       "<t a>",
+       "<t a=>",
+       "<t a=v>",
+       //      "<![CDATA[d]]>",        // let the Token() caller handle
+       "cdata",
+       "<t></e>",
+       "<t></>",
+       "<t></t!",
+       "<t>cdata]]></t>",
+}
+
 type stringReader struct {
        s       string;
        off     int;
@@ -149,3 +200,15 @@ func TestToken(t *testing.T) {
                }
        }
 }
+
+func TestSyntax(t *testing.T) {
+       for i := range xmlInput {
+               p := NewParser(StringReader(xmlInput[i]));
+               var err os.Error;
+               for _, err = p.Token(); err == nil; _, err = p.Token() {
+               }
+               if _, ok := err.(SyntaxError); !ok {
+                       t.Fatalf(`xmlInput "%s": expected SyntaxError not received`, xmlInput[i])
+               }
+       }
+}