示例#1
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <head wp-name="head">
              <title wp-name="title"/>
            </head>
            </html>
            """

        self.scraper = weakscraper.WeakScraper(template_string)
示例#2
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <body>
              <tag1>a</tag1>
              <tag2>b</tag2>
            </body>
            </html>
            """

        self.scraper = weakscraper.WeakScraper(template_string)
示例#3
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <body>
              <tag1>begining<wp-nugget wp-name="info1"/>middle<wp-nugget wp-name="info2"/>end</tag1>
              <tag2><wp-nugget wp-name="info3"/>end</tag2>
            </body>
            </html>
            """

        self.scraper = weakscraper.WeakScraper(template_string)
示例#4
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <head>
              <title>Title</title>
            </head>
            <body attr1="val1" attr2="val2">
              <div>Hi !</div>
            </body>
            </html>
            """

        self.scraper = weakscraper.WeakScraper(template_string)
示例#5
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <body>
              <tag1 wp-optional>
                some text
              </tag1>
              <tag2 wp-name="tag2" wp-optional/>
            </body>
            </html>
            """

        self.scraper = weakscraper.WeakScraper(template_string)
示例#6
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <body>
              <tag1>a</tag1>
              <wp-ignore wp-until="tag4"/>
              <tag4 attr1="value" wp-ignore-attrs>b</tag4>
              <tag5 wp-ignore-content/>
              <tag6 wp-ignore/>
              <wp-ignore/>
            </body>
            </html>
            """

        self.scraper = weakscraper.WeakScraper(template_string)
示例#7
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <head wp-function="sum">
              <number wp-name="a" wp-function="int"/>
              <number wp-name="b" wp-function="int"/>
            </head>
            </html>
            """

        functions = {
            'int': int,
            'sum': (lambda dic: dic['a'] + dic['b'])
        }

        self.scraper = weakscraper.WeakScraper(template_string, functions)
示例#8
0
    def setUp(self):
        template_string = """
            <!DOCTYPE html>
            <body>
              <div wp-function="sum" wp-list>
                <number wp-name="number" wp-function="int"/>
              </div>
              <number wp-function="str_sum" wp-list/>
            </body>
            </html>
            """

        functions = {
            'int': int,
            'sum': (lambda l: sum([n['sum']['number'] for n in l])),
            'str_sum': (lambda l: sum([int(n['str_sum']) for n in l]))
        }


        self.scraper = weakscraper.WeakScraper(template_string, functions)