Python scrape_title示例

示例#1

    def test_keeps_non_tag_text_after_dash(self):
        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Pumpkin Seed Bark &#x2013; Dark Chocolate and Sea Salt</h1>"""
                                  )),
            u'Pumpkin Seed Bark \u2013 Dark Chocolate and Sea Salt')

        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Keto Flatbread Recipe &#x2013; Low Carb, Gluten Free</h1>"""
                                  )), u'Keto Flatbread Recipe')

示例#2

    def test_strips_bracketed_text(self):
        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Low Carb Moscow Mule &#8211; [Keto, Alcohol, Sugar Free]</h1>"""
                                  )), 'Low Carb Moscow Mule')

示例#3

    def test_strips_tags_after_dash(self):
        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Spicy Chicken Sausage &#x2013; Low Carb, Gluten-Free</h1>"""
                                  )), u'Spicy Chicken Sausage')

示例#4

    def test_strips_tags_after_pipe(self):
        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Spanish Cauliflower Rice | Low Carb</h1>""")),
            'Spanish Cauliflower Rice')

示例#5

    def test_strips_tags_after_colon(self):
        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Vodka Mojito: Low Carb and Sugar-Free</h1>""")),
            'Vodka Mojito')
        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Hot Buttered Rum Recipe: Low Carb, Sugar Free</h1>""")
            ), 'Hot Buttered Rum Recipe')
        self.assertEqual(
            ketogasm.scrape_title(
                http.TextResponse(url='',
                                  body="""
<h1 class="entry-title">Gin Fizz Cocktail Recipe &#8211; Low Carb &#038; Sugar Free!</h1>
""")), 'Gin Fizz Cocktail Recipe')