from wex.extractor import named
from wex.etree import xpath, text
extract = named(name = xpath('//h1') | text,
country = xpath('//dd[@id="country"]') | text,
region = xpath('//dd[@id="region"]') | text)
def test_xpath_re_match():
f = (e.xpath('re:match(//body, "\s+is\s+(some)\s+text", "gi")/text()')
| list)
assert f(create_response(example)) == ['some']
def test_drop_tree():
f = (e.xpath('//*[@id="drop-tree"]') | e.drop_tree(e.css('script'))
| e.xpath('string()'))
assert f(create_response(example)) == ['Drop this please.']
def test_xpath():
f = e.xpath('//h1/text()') | list
assert f(create_response(example)) == ['hi']
def test_xpath_re():
f = e.xpath('//*[re:test(text(), "SOME", "i")]/text()') | list
assert f(create_response(example)) == ['some ']
def region(response):
return text(xpath('//dd[@id="region"]')(response))
def name(response):
return text(xpath('//h1')(response))
def test_drop_tree():
f = (e.xpath('//*[@id="drop-tree"]') |
e.drop_tree(e.css('script')) |
e.xpath('string()'))
assert f(create_response(example)) == ['Drop this please.']
def country(response):
return text(xpath('//dd[@id="country"]')(response))
def test_xpath_re():
f = e.xpath('//*[re:test(text(), "SOME", "i")]/text()') | list
assert f(create_response(example)) == ['some ']
def test_xpath_re_match():
f = (e.xpath('re:match(//body, "\s+is\s+(some)\s+text", "gi")/text()') |
list)
assert f(create_response(example)) == ['some']
def test_xpath():
f = e.xpath('//h1/text()') | list
assert f(create_response(example)) == ['hi']
from wex.extractor import named, labelled
from wex.iterable import one
from wex.etree import xpath, text
cheeses = xpath('//h1[@data-icin]')
icin_attr = xpath('@data-icin') | one
attrs = named(name = text,
country = xpath('following::dd[@id="country"][1]') | text,
region = xpath('following::dd[@id="region"][1]') | text)
extract_cheese = labelled(icin_attr, attrs)
def extract(response):
for cheese in cheeses(response):
for item in extract_cheese(cheese):
yield item
def name(response):
return text(xpath('//h1')(response))
def region(response):
return text(xpath('//dd[@id="region"]')(response))
def country(response):
return text(xpath('//dd[@id="country"]')(response))
