Beispiel #1
0
 def test_divs_class(self):
     file_list = os.listdir('tests/content_class')
     for f in file_list:
         br = Briticle()
         br.open(file_='tests/content_class/' + f)
         self.assertTrue('bbbbb' not in br.text)
         self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
         self.assertTrue(len(br.html) > len(br.text))
Beispiel #2
0
 def test_divs_class(self):
     file_list = os.listdir('tests/content_class')
     for f in file_list:
         br = Briticle()
         br.open(file_='tests/content_class/' + f)
         self.assertTrue('bbbbb' not in br.text)
         self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
         self.assertTrue(len(br.html) > len(br.text))
Beispiel #3
0
    def test_special_sites(self):
        file_list = (
            ("tests/github_project.html", 1273),
            ("tests/bs4_doc.html", 5555),
            ("tests/wikipedia1.html", 29911),
            ("tests/wikipedia2.html", 1865),
        )

        for f, count in file_list:
            br = Briticle()
            br.open(file_=f)
            self.assertOk(len(br.text.replace('\n', '')), count, delta=10)
            self.assertTrue(len(br.html) > len(br.text))
Beispiel #4
0
    def test_special_sites(self):
        file_list = (
            ("tests/github_project.html", 1273),
            ("tests/bs4_doc.html", 5555),
            ("tests/wikipedia1.html", 29911),
            ("tests/wikipedia2.html", 1865),
        )

        for f, count in file_list:
            br = Briticle()
            br.open(file_=f)
            self.assertOk(len(br.text.replace('\n', '')), count, delta=10)
            self.assertTrue(len(br.html) > len(br.text))
Beispiel #5
0
 def test_always_return_div(self):
     br = Briticle()
     br.open(file_="./tests/functions/always_return_div1.html")
     self.assertTrue("<td>" not in br.html)
     br = Briticle()
     br.open(file_="./tests/functions/always_return_div2.html")
     self.assertTrue("<body>" not in br.html)
Beispiel #6
0
 def test_search_divs_with_h1(self):
     br = Briticle()
     br.open(file_="./tests/functions/divs_with_h1_1.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
     br = Briticle()
     br.open(file_="./tests/functions/divs_with_h1_2.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 2800, delta=10)
Beispiel #7
0
 def test_always_return_div(self):
     br = Briticle()
     br.open(file_="./tests/functions/always_return_div1.html")
     self.assertTrue("<td>" not in br.html)
     br = Briticle()
     br.open(file_="./tests/functions/always_return_div2.html")
     self.assertTrue("<body>" not in br.html)
Beispiel #8
0
 def test_search_divs_with_h1(self):
     br = Briticle()
     br.open(file_="./tests/functions/divs_with_h1_1.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
     br = Briticle()
     br.open(file_="./tests/functions/divs_with_h1_2.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 2800, delta=10)
Beispiel #9
0
 def test_search_p_biggest_parent(self):
     br = Briticle()
     br.open(file_="./tests/functions/p_biggest_parent.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
Beispiel #10
0
 def test_remove_font_size(self):
     br = Briticle()
     br.open(file_="./tests/functions/remove_font_size.html")
     self.assertTrue("font size=" not in '{}'.format(br.html))
Beispiel #11
0
 def test_article_html5_tag(self):
     br = Briticle()
     br.open(file_="./tests/functions/article_html5_tag.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
Beispiel #12
0
 def test_remove_meta_info(self):
     br = Briticle()
     br.open(file_="./tests/functions/remove_meta_info.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
Beispiel #13
0
 def test_remove_font_size(self):
     br = Briticle()
     br.open(file_="./tests/functions/remove_font_size.html")
     self.assertTrue("font size=" not in '{}'.format(br.html))
Beispiel #14
0
 def test_title(self):
     br = Briticle()
     br.open(file_="./tests/functions/get_title1.html")
     self.assertEqual(br.title, u'KISS\u539f\u5219')
     br.open(file_="./tests/functions/get_title2.html")
     self.assertEqual(br.title, u'abc')
Beispiel #15
0
 def test_title(self):
     br = Briticle()
     br.open(file_="./tests/functions/get_title1.html")
     self.assertEqual(br.title, u'KISS\u539f\u5219')
     br.open(file_="./tests/functions/get_title2.html")
     self.assertEqual(br.title, u'abc')
Beispiel #16
0
 def test_search_p_biggest_parent(self):
     br = Briticle()
     br.open(file_="./tests/functions/p_biggest_parent.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
Beispiel #17
0
 def test_article_html5_tag(self):
     br = Briticle()
     br.open(file_="./tests/functions/article_html5_tag.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)
Beispiel #18
0
 def test_remove_meta_info(self):
     br = Briticle()
     br.open(file_="./tests/functions/remove_meta_info.html")
     self.assertTrue("bbbb" not in br.text)
     self.assertTrue("bbbb" not in br.html)
     self.assertOk(len(br.text.replace('\n', '')), 1400, delta=10)