Exemple #1
0
 def _download_file(self, f):
     l = util.make_url(self._file_playlist.url, f['file'])
     name = urlparse.urlparse(f['file']).path.split('/')[-1]
     path = os.path.join(self.path, name)
     d = self._download_page(l, path)
     d.addCallback(self._got_file, l, f)
     return d
Exemple #2
0
 def _playlist_updated(self, pl):
     if pl.has_programs():
         # if we got a program playlist, save it and start a program
         self._program_playlist = pl
         (program_url, _) = pl.get_program_playlist(self.program, self.bitrate)
         l = util.make_url(self.url, program_url)
         return self._reload_playlist(M3U8(l))
     elif pl.has_files():
         # we got sequence playlist, start reloading it regularly, and get files
         self._file_playlist = pl
         if not self._files:
             self._files = pl.iter_files()
         if not pl.endlist():
             reactor.callLater(pl.reload_delay(), self._reload_playlist, pl)
         if self._file_playlisted:
             self._file_playlisted.callback(pl)
             self._file_playlisted = None
     else:
         raise
     return pl
Exemple #3
0
    def detail_parse(self, response):
        cate = response.meta['cate']

        img = response.css('div.cont img::attr(src)').extract()
        img.append(
            response.css(
                'div.detailArea div.keyImg img::attr("src")').extract_first())

        price = response.css(
            'meta[property="product:price:amount"]::attr(content)'
        ).extract_first()
        price_unit = response.css(
            'meta[property="product:price:currency"]::attr(content)'
        ).extract_first()

        url = response.url
        product_no = url[42 + len('product_no='):].split('&')[0]
        image_url = response.css(
            'meta[property="og:image"]::attr(content)').extract()[1]
        if image_url not in inserted_image:
            yield {
                'host_url':
                self.domain,
                'tag':
                cate,
                'product_name':
                response.css(
                    'meta[property="og:title"]::attr(content)').extract()[1],
                'image_url':
                make_url(self.domain, image_url),
                'product_price':
                price,
                'currency_unit':
                price_unit,
                'product_url':
                url,
                'product_no':
                product_no,
                'main':
                '1',
                'nation':
                'ko'
            }
            inserted_image.append(image_url)

        for im in img:
            if im not in inserted_image:
                yield {
                    'host_url':
                    self.domain,
                    'tag':
                    cate,
                    'product_name':
                    response.css('meta[property="og:title"]::attr(content)').
                    extract()[1],
                    'image_url':
                    make_url(self.domain, im),
                    'product_price':
                    price,
                    'currency_unit':
                    price_unit,
                    'product_url':
                    url,
                    'product_no':
                    product_no,
                    'main':
                    '0',
                    'nation':
                    'ko'
                }
                inserted_image.append(im)