def test_re_field_get_nothing_with_no_default():
field = RegexField(re_select='nothing to match.')
try:
field.extract(html=HTML)
raise AssertionError
except NothingMatchedError:
pass
def test_re_field_in_dict_format_with_many():
field = RegexField(re_select='<a class="test_link" href="(?P<href>.*?)">(?P<text>.*?)</a>', many=True)
matches = field.extract(html=HTML)
assert len(matches) == 5
assert matches[0]['href'] == 'https://github.com/howie6879/'
assert matches[0]['text'] == 'hello1 github.'
assert matches[4]['href'] == 'https://github.com/howie6879/'
assert matches[4]['text'] == 'hello5 github.'
def test_re_field_with_many():
field = RegexField(re_select='<a class="test_link" href="(.*?)">(.*?)</a>', many=True)
matches = field.extract(html=HTML)
assert len(matches) == 5
href0, text0 = matches[0]
href4, text4 = matches[4]
assert href0 == 'https://github.com/howie6879/'
assert text0 == 'hello1 github.'
assert href4 == 'https://github.com/howie6879/'
assert text4 == 'hello5 github.'
def test_re_field_with_html_element():
field = RegexField(
re_select='<h1><a href="(?P<href>.*?)">(?P<text>.*?)</a></h1>')
result = field.extract(html=html_etree)
assert result["href"] == "https://github.com"
assert result["text"] == "Github"
def test_re_field_get_nothing_with_no_default():
field = RegexField(re_select="nothing to match.")
try:
field.extract(html=HTML)
except Exception as e:
assert isinstance(e, NothingMatchedError)
def test_re_field_with_default():
field = RegexField(re_select="nothing to match.", default="default value")
result = field.extract(html=HTML)
assert result == "default value"
def test_re_field_with_many_groups():
field = RegexField(re_select='<h1><a href="(.*?)">(.*?)</a></h1>')
href, text = field.extract(html=HTML)
assert href == "https://github.com"
assert text == "Github"
def test_re_field_with_no_group():
field = RegexField(re_select="<title>.*?</title>")
href = field.extract(html=HTML)
assert href == "<title>ruia</title>"
def test_re_field_with_one_group():
field = RegexField(re_select="<title>(.*?)</title>")
href = field.extract(html=HTML)
assert href == "ruia"
def test_re_field_with_html_element():
field = RegexField(
re_select='<h1><a href="(?P<href>.*?)">(?P<text>.*?)</a></h1>')
result = field.extract(html=html_etree)
assert result['href'] == 'https://github.com'
assert result['text'] == 'Github'