multiple root nodes

[lugensa]

Expat stops extracting after reading the first root element.

    def test_multiple_root_nodes(self): 
        snippet = """\                                                                                                                                                                                              
                <metal:foo>
                  <dummy i18n:translate="">Foo</dummy>
                  <dummy i18n:translate="">Foo</dummy>
                </metal:foo>
                <metal:bla>
                  <dummy i18n:translate="">Bla</dummy>
                  <dummy i18n:translate="">Bla</dummy>
                </metal:bla>
                """
        self.assertEqual(self.extract(snippet),
                [
                    (2, None, u"Foo", []),          
                    (3, None, u"Foo", []),          
                    (6, None, u"Bla", []),          
                    (7, None, u"Bla", []),          
                 ])

[wichert]

lingua expects an XML document as input. What you are providing is an XML fragment, not a document, and as a result it fails to parse. Is there a special reason you can not restructure your input to be a valid XML document?

[lugensa]

Hmm it's just the way of how to define multiple macros in one file. I've added now a node around and it works but this feels still like a hack.

I think this should be at least be documented, so that there is a difference between Chameleon <= 1.99, Zope i18n extraction and lingua.

    def test_multiple_root_nodes(self): 
        snippet = """\
                <tal:def define="foo nothing">                                                                                                                                                                                              
                <metal:foo>
                  <dummy i18n:translate="">Foo</dummy>
                  <dummy i18n:translate="">Foo</dummy>
                </metal:foo>
                <metal:bla>
                  <dummy i18n:translate="">Bla</dummy>
                  <dummy i18n:translate="">Bla</dummy>
                </metal:bla>
                </tal:def>
                """
        self.assertEqual(self.extract(snippet),
                [
                    (3, None, u"Foo", []),          
                    (4, None, u"Foo", []),          
                    (7, None, u"Bla", []),          
                    (8, None, u"Bla", []),          
                 ])

[wichert]

If I remember correctly Chameleon 1.x would not be able to parse your file either since it also used an XML parser. Note that you can simplify your input a bit further: you do need to define any attributes on your root element since elements in the tal or metal namespaces never result in output. My macro templates usually look like this:

<tal:macros xmlns:tal="http://namespaces.zope.org/tal"
            xmlns:metal="http://namespaces.zope.org/metal">
  <metal:navtree define-macro="navtree">
    ....
  </metal:navtree>
</tal:macros>

[lugensa]

It worked with Chameleon 1.x. We had some old repoze.bfg apps witch are fully translated. I've tried it with

<tal:macros xmlns:tal="http://namespaces.zope.org/tal"
            xmlns:metal="http://namespaces.zope.org/metal">
  <metal:navtree define-macro="navtree">
    ....
  </metal:navtree>
</tal:macros>

but than i get an KeyError: "Macro does not exist: 'navbar'." And names is empty

>>> foo = pyramid.renderers.get_renderer('foo:templates/foo.pt').implementation()
>>> foo.macros.names
[]

[wichert]

My mistake - I mistyped the namespaces from memory. They should be http://xml.zope.org/namespaces/tal and http://xml.zope.org/namespaces/metal

[lugensa]

Ahh ok didn't knew that Chameleon is also checking namspaces :)

[wichert]

Chameleon has built-in defaults in case you do not declare namespaces yourself. However if you do declare them properly you must be careful to do it right :). Thecurrent version of lingua shares similar logic for the i18n namespace, but unlike Chameleon 2 it uses a full XML parser, which is why it has stricter requirements on input.

[wichert]

This works correctly in lingua 2. Note that your exact input will still not work since you did not set an i18n:domain, which is required for translation to work.