def test_html_field():
field_en = HtmlField(css_select="div.brand a")
field_zh = HtmlField(css_select="div.brand p")
assert (
field_en.extract(html_etree=html_etree)
== '<a href="https://github.com">Github</a>'
)
assert field_zh.extract(html_etree=html_etree) == "<p>你好</p>\n"
def test_html_field_with_many():
field = HtmlField(css_select="a.test_link", many=True)
values = field.extract(html_etree=html_etree)
assert len(values) == 5
assert values[0] == '<a class="test_link" href="https://github.com/howie6879/">hello1 github.</a>\n'
assert values[4] == '<a class="test_link" href="https://github.com/howie6879/">hello5 github.</a>\n' \
' Some text outside.\n'
def test_html_field():
field = HtmlField(css_select="div.brand a")
assert (field.extract(
html_etree=html_etree) == '<a href="https://github.com">Github</a>')