Python parse_html Examples

Example #1

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

    def test_it_parses_element(self):
        assert parse_html('<p></p>') == [OpenTag(P(1, 1), 'p', space=''),
                                         CloseTag(P(1, 4), 'p')]
        assert parse_html('<p>a</p>') == [OpenTag(P(1, 1), 'p', space=''),
                                          Text(P(1, 4), 'a'),
                                          CloseTag(P(1, 5), 'p')]

        assert parse_html('<html:p></html:p>') == [
            OpenTag(P(1, 1), 'html:p', space=''),
            CloseTag(P(1, 9), 'html:p')]

Example #2

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def test_it_parses_implicit_cdata(self):
     p = parse_html('<script>console && console.log("ni!")</script>')
     assert p == [
         OpenTag(P(1, 1),
                 'script',
                 space='',
                 attrs=OrderedDict()),
         Text(P(1, 9), content='console && console.log("ni!")', cdata=True),
         CloseTag(P(1, 38), 'script')]

Example #3

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def test_it_parses_attributes(self):
     assert parse_html('<p data-foo="bar"></p>') == [
         OpenTag(P(1, 1),
                 'p',
                 space=' ',
                 attrs=OrderedDict([('data-foo', Attribute(pos=P(1, 4),
                                                           name='data-foo',
                                                           value='bar',
                                                           value_pos=P(1, 14),
                                                           quote='"',
                                                           space1='',
                                                           space2='',
                                                           space3='',))])),
         CloseTag(P(1, 19), 'p')]

Example #4

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def test_it_preserves_whitespace(self):
     assert parse_html('<a\n\thref =  "foo" ></a>') == [
         OpenTag(P(1, 1),
                 'a',
                 space='\n\t',
                 attrs=OrderedDict(href=Attribute(pos=P(2, 2),
                                                  name='href',
                                                  value='foo',
                                                  value_pos=P(2, 11),
                                                  quote='"',
                                                  space1=' ',
                                                  space2='  ',
                                                  space3=' '))),
         CloseTag(P(2, 17), 'a')
     ]

Example #5

File: test_compilexml.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

def c(s):
    return cx.compile_intermediate(parse_html(s)).children

Example #6

File: template.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def compile_intermediate(self, src):
     parsed = parse_html(src)
     return compilexml.compile_intermediate(parsed)

Example #7

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def test_it_parses_numeric_entity(self):
     assert parse_html(' ') == [Entity(P(1, 1), reference=' ')]

Example #8

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def test_it_parses_entity(self):
     assert parse_html(' ') == [Entity(P(1, 1), reference=' ')]

Example #9

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def test_it_parses_pi(self):
     assert parse_html('<?php ?>') == [PI(P(1, 1), target='php', content=' ')]

Example #10

File: test_parse.py Project: Quantum-Machine-Learning-Initiative/MInimization_GOAT

 def test_it_parses_comment(self):
     assert parse_html('<!-- x -->') == [Comment(P(1, 1), content=' x ')]

Example #11

def tosrc(s):
    return compile_to_source(cx.compile_intermediate(parse_html(s)))