def test_selectors(all_options): html = (u'<span><span id="extract-me">text<a>more</a>' '</span>and more text <a> and some more</a> <a></a> </span>') # Selector sel = cleaned_selector(html) assert selector_to_text( sel, **all_options) == 'text more and more text and some more' # SelectorList subsel = sel.xpath('//span[@id="extract-me"]') assert selector_to_text(subsel, **all_options) == 'text more' subsel = sel.xpath('//a') assert selector_to_text(subsel, **all_options) == 'more and some more' subsel = sel.xpath('//a[@id="extract-me"]') assert selector_to_text(subsel, **all_options) == '' subsel = sel.xpath('//foo') assert selector_to_text(subsel, **all_options) == ''
def test_selector(all_options): html = '<div><div id="extract-me">text<div>more</div></div>and more text</div>' sel = cleaned_selector(html) assert selector_to_text(sel, **all_options) == 'text more and more text' subsel = sel.xpath('//div[@id="extract-me"]')[0] assert selector_to_text(subsel, **all_options) == 'text more'