Bugs in xmerl_sax_parser?

[Tony Garnock-Jones-2]

Hi all,

I've just been playing with xmerl_sax_parser and I think I've found two
bugs:

 1. xmerl_sax_parser:file/2 and :stream/2 behave differently. Create a
    file, "t.xml", containing

<elem attr='123' x:attr='234' xmlns='http://lshift.net/d'
xmlns:x='http://lshift.net/x' />

    and then run the two expressions

    xmerl_sax_parser:file("t.xml", [{event_fun, fun (E,_,_) ->
io:format("~p~n", [E]), ok end}]).
    xmerl_sax_parser:stream("<elem attr='123' x:attr='234'
xmlns='http://lshift.net/d' xmlns:x='http://lshift.net/x' />",
[{event_fun, fun (E,_,_) -> io:format("~p~n", [E]), ok end}]).

    Note that they print the same event sequences, but the first results
    in {ok,ok,<<>>}, while the second results in {fatal_error,...}!

 2. The namespace URI supplied on unprefixed attributes in startElement
    tuples is the same as the URI for the default namespace. According
    to http://www.w3.org/TR/xml-names/#defaulting, "The namespace name
    for an unprefixed attribute name always has no value."

    For example, running one of the expressions above causes the
    following to be printed:

    {startElement,"http://lshift.net/d","elem",
                  {[],"elem"},
                  [{"http://lshift.net/d",[],"attr","123"},
                   {"http://lshift.net/x","x","attr","234"}]}

    While applications can work around this by ignoring the attribute's
    namespace name in cases where the prefix == "", I wonder if it
    wouldn't be better to supply "" for the namespace name for
    unprefixed attributes, like this:

    {startElement,"http://lshift.net/d","elem",
                  {[],"elem"},
                  [{[],[],"attr","123"},
                   {"http://lshift.net/x","x","attr","234"}]}

Regards,
  Tony
--
 [][][] Tony Garnock-Jones     | Mob: [phone]
   [][] LShift Ltd             | Tel: [phone]
 []  [] http://www.lshift.net/ | Email: [hidden email]

________________________________________________________________
erlang-questions mailing list. See http://www.erlang.org/faq.html
erlang-questions (at) erlang.org

[Lars Thorsen]

Hi Tony,
thank you for your information about the errors.

1. xmerl_sax_parse:stream/2 failed with {fatal_error,_, "Continuation
function undefined, and more data needed",_,_} when no continuation
function was defined even though it was a complete document as input.

2. The namespace URI supplied on unprefixed attributes in startElement
tuples is the same as the URI for the default namespace. According to
the standard the namespace for an unprefixed attribute should always has
no value.

I've fixed the errors and a patch can be fetched at
"http://www.erlang.org/download/patches/".

The files are named:
otp_src_R13B02_OTP-8213_OTP-8214.patch
otp_src_R13B02_OTP-8213_OTP-8214.readme

Regards Lars Thorsen
OTP Development

Tony Garnock-Jones wrote:

> Hi all,
>
> I've just been playing with xmerl_sax_parser and I think I've found two
> bugs:
>
>  1. xmerl_sax_parser:file/2 and :stream/2 behave differently. Create a
>     file, "t.xml", containing
>
> <elem attr='123' x:attr='234' xmlns='http://lshift.net/d'
> xmlns:x='http://lshift.net/x' />
>
>     and then run the two expressions
>
>     xmerl_sax_parser:file("t.xml", [{event_fun, fun (E,_,_) ->
> io:format("~p~n", [E]), ok end}]).
>     xmerl_sax_parser:stream("<elem attr='123' x:attr='234'
> xmlns='http://lshift.net/d' xmlns:x='http://lshift.net/x' />",
> [{event_fun, fun (E,_,_) -> io:format("~p~n", [E]), ok end}]).
>
>     Note that they print the same event sequences, but the first results
>     in {ok,ok,<<>>}, while the second results in {fatal_error,...}!
>
>  2. The namespace URI supplied on unprefixed attributes in startElement
>     tuples is the same as the URI for the default namespace. According
>     to http://www.w3.org/TR/xml-names/#defaulting, "The namespace name
>     for an unprefixed attribute name always has no value."
>
>     For example, running one of the expressions above causes the
>     following to be printed:
>
>     {startElement,"http://lshift.net/d","elem",
>                   {[],"elem"},
>                   [{"http://lshift.net/d",[],"attr","123"},
>                    {"http://lshift.net/x","x","attr","234"}]}
>
>     While applications can work around this by ignoring the attribute's
>     namespace name in cases where the prefix == "", I wonder if it
>     wouldn't be better to supply "" for the namespace name for
>     unprefixed attributes, like this:
>
>     {startElement,"http://lshift.net/d","elem",
>                   {[],"elem"},
>                   [{[],[],"attr","123"},
>                    {"http://lshift.net/x","x","attr","234"}]}
>
> Regards,
>   Tony

________________________________________________________________
erlang-questions mailing list. See http://www.erlang.org/faq.html
erlang-questions (at) erlang.org