tags_test.py

 # coding: utf-8
import lxml.html
from .abstract import get_strategy

import pytest


class TestTags:

    @pytest.mark.parametrize('start_urls_url, actual_url',
                             [('http://foo.bar/api', 'http://foo.bar/api'),
                              ('http://foo.bar/api', 'http://foo.bar/api/test'),
                              ('http://foo.bar/.*', 'http://foo.bar/api/test')])
    def test_regex_start_urls_params(self, start_urls_url, actual_url):
        # Given
        # Stub ENV variables read by ConfigLoader
        strategy = get_strategy({
            'start_urls': [
                {
                    'url': start_urls_url,
                    'tags': ["test"]
                }
            ],
            'selectors': {
                "lvl0": "h1",
                "lvl1": "h2",
                "lvl2": "h3",
                "content": "p"
            }
        })

        strategy.dom = lxml.html.fromstring("""
        <html><body>
         <h1>Foo</h1>
        </body></html>
        """)

        # When
        actual = strategy.get_records_from_dom(actual_url)

        # Then
        assert actual[0]['tags'] == ["test"]