tarvitz / composite Goto Github PK

Сейчас это воткнуто непосредственно в Document

There's iterate through lxml document that looks like this:

class XMLDocumentBuilder(BaseDocumentBuilder):
    # ...
    @classmethod
    def iterate(cls, source):
        """
        iterate through source document

        :param lxml.etree.Element source: xml document
        :rtype: generator
        :return: tuple[node name, node]
        """
        if isinstance(source, etree._Element):
            for node in source.getchildren():
                yield (node.tag, node)
        else:
            for name, item in source.items():
                yield (name, item)

I suppose builder should not include check the type of object it iterates. Possible it could be resolved by somehow.

There's some magic numbers inside of code should be resolved or stay.

Think about it, resolve if needed.

After making #6 there should be easily add other formats without making dependency hell.

#4

to и некоторые from, а также то что используется внутри будут являться лишними, так как их ответственность выполняет builder классы

В процессе работы с библиотекой выяснилось, что обращение к дискриптору __dict__ инстанса класса Attributes происходит с ошибкой:
>>>offer.attributes.__dict__
*** TypeError: descriptor '__dict__' for 'Attributes' objects doesn't apply to 'Attributes' object
>>> type(offer.attributes)
<class '__main__.Attributes'>

using Field inside Attribute subclass would cause a problem with xml parser, this is not bug just because AttributeField should be use instead. Lock or think something suitable solution for that programming interface lock

from composite import Document
from composite.fields import Field, AttributeField

class UserAttributes(Document):
    class Attribute:
        age = Field('age', int)  #: should be AttributeField

Now lxml is hard requirement, some would need to make xml-> conversions, some won't.

There should be possibility to weaken dependency coupling for lxml.