class Article(Entity):
title = StringField('./h1/text()', xpath=True)
content = StringField('./p/text()', xpath=True, all=True)
authors = EntityField(Author,
'./div[@class="author"]',
xpath=True,
all=True)
class BookChapter(Entity):
"""An entity with various different field types."""
book = StringField('body>h1::text')
number = IntField('.chnum::text')
price = FloatField('.cost::text', re='\$(.+)')
public = BoolField('p::text', re='Public: (.+)')
updated = DateTimeField('p::text', re='Last updated on (.+).')
next_url = StringField('#next::attr("href")', lower=True)
class SimpleContent(Entity):
"""A simple entity defined using XPath expressions."""
title = StringField('//div/h1/text()', xpath=True)
link_text = StringField('//div/a/text()', xpath=True)
link_url = StringField('//div/a/@href', xpath=True)
urls = StringField('/html/body/div/a/@href',
xpath=True,
all=True,
lower=True)
in_divs = StringField('//div/div', xpath=True, all=True)
process_title = Chain(six.text_type.capitalize, RAdd('!'))
class ArticleC(Entity):
title = StringField('h1::text')
content = StringField('p::text', all=True)
authors = EntityField(AuthorC, 'div.author', all=True)
class AuthorC(Entity):
firstname = StringField('span.firstname::text')
lastname = StringField('span.lastname::text')
class Author(Entity):
firstname = StringField('./span[@class="firstname"]/text()', xpath=True)
lastname = StringField('./span[@class="lastname"]/text()', xpath=True)
class Blog(Entity):
title = StringField('./h1/text()', xpath=True)
content = StringField('./p/text()', xpath=True, all=True)