UNC digesto via demiurge

gistfile1.py

from demiurge import Item, TextField, AttributeValueField, RelatedItem


class Documento(Item):
    area = TextField(selector="#breadcrumbs-2 a")
    tipo = TextField(selector="#parent-fieldname-kind")
    titulo = TextField(selector="#parent-fieldname-title")
    numero = TextField(selector='#parent-fieldname-number')
    descripcion = TextField(selector="#parent-fieldname-description")
    pdf = AttributeValueField(selector="#archetypes-fieldname-file span a", attr='href')
    emisor = TextField(selector="span#parent-fieldname-kind")
    fecha = TextField(selector="#parent-fieldname-date")
    expedientes = TextField(selector="#archetypes-fieldname-dossierNumber span:gt(0)")

    class Meta:
        selector = '#portal-column-content'


class IndexEntry(Item):
    titulo = TextField(selector='a.url')
    detalle = RelatedItem(Documento, selector='a.url', attr='href')

    class Meta:
        selector = 'span.summary'


class Index(Item):
    documentos = RelatedItem(IndexEntry, selector='div#content-core')

    class Meta:
        base_url = 'http://www.digesto.unc.edu.ar/'