async def _setProductThumbUrl(self, soup: BeautifulSoup, product: Product) -> ProductChanged: """ Search for the following JS code and extract URL: <div class="more-views mobilehidden"> <ul> <li> <a href="https:// ... """ isProductChanged = False try: urlThumb = soup.find("div", {"class": "product-img-box"}) \ .find("div", {"class": "more-views mobilehidden"}) \ .find_next("a").get("href") except AttributeError as e: logger.warning("Failed finding product image url. %s. %s", e, product.url) self._failCount += 1 else: logger.debug("Found product image url. %s", product.url) if not product.urlThumb or product.urlThumb != urlThumb: product.urlThumb = urlThumb isProductChanged = True # note that we never switch back to False finally: return isProductChanged
async def _setProductThumbUrl(self, soup: BeautifulSoup, product: Product) -> ProductChanged: isProductChanged = False try: urlThumb = soup.find("div", class_="thumbnail-1") \ .find("div", class_="wrap") \ .img["src"] if not urlThumb: raise AttributeError("No matches in HTML tree.") except AttributeError as e: logger.warning("Failed finding product image url. %s. %s", e, product.url) self._failCount += 1 else: logger.debug("Found product image url. %s", product.url) # Their image URL is relative so we have to concat the shop's base URL with it. absoluteImageURL = urllib.parse.urljoin(self.URL, urlThumb) if not product.urlThumb or product.urlThumb != absoluteImageURL: product.urlThumb = absoluteImageURL isProductChanged = True # note that we never switch back to False finally: return isProductChanged
async def _setProductThumbUrl(self, soup: BeautifulSoup, product: Product) -> ProductChanged: isProductChanged = False try: urlThumb = soup.find("li", class_="thumbnail-1") \ .find("div", class_="wrap") \ .img["src"] # Below code possibly superfluous. The above urlThumb is already valid, # but for mighty reasons linking to it in Discord seems to get blocked by BSTN. There's # a way to shorten the URL, as the last parts of the URL don't seem to be relevant. # Anyway, this won't fix the blocked link in Discord. # Example # https://www.bstn.com/media/140801/w/280/h/280/n/adidas-zx-8000-fv3269-1.jpg # is converted to # https://www.bstn.com/media/140801/w/280/h/280/n/0.jpg # match = re.search(r".*/h/280/n/", urlThumb) # urlThumb = f"{match.group()}0.jpg" if not urlThumb: raise AttributeError("No matches in HTML tree.") except AttributeError as e: logger.warning("Failed finding product image url. %s. %s", e, product.url) self._failCount += 1 else: logger.debug("Found product image url. %s", product.url) if not product.urlThumb or product.urlThumb != urlThumb: product.urlThumb = urlThumb isProductChanged = True # note that we never switch back to False finally: return isProductChanged
async def _setProductThumbUrl(self, soup: BeautifulSoup, product: Product) -> ProductChanged: isProductChanged = False try: urlThumb = soup.find("div", {"class": "b-pdp-product-preview-wrapper"}) \ .find("div", {"class": "b-pdp-carousel-item"}) \ .div["data-default-src"] if not urlThumb: raise AttributeError("No matches in HTML tree.") except AttributeError as e: logger.warning("Failed finding product image url. %s. %s", e, product.url) self._failCount += 1 else: logger.debug("Found product image url. %s", product.url) if not product.urlThumb or product.urlThumb != urlThumb: product.urlThumb = urlThumb isProductChanged = True # note that we never switch back to False finally: return isProductChanged
def create2Shops(self) -> None: """ Caution: Do not change values as they are tested against :return: None """ size01 = Size() size01.uid = str( uuid.UUID(hex="1528dae6-188f-4d7e-8a6c-5af44ce5c222", version=4)) size01.sizeEU = "40 1/3" size01.isInStock = True size01.price = 56.99 size01.url = "http://oneshop.com/bottles/92743867ACTFGJ-UTU/40.1.3.htm" size01.urlAddToCart = "http://oneshop.com/bottles/atc/40.1.3-92743867ACTFGJ-UTU.htm" size02 = Size() size02.uid = str( uuid.UUID(hex="5f561c62-8502-4ec1-8f46-f0adb5e8254c", version=4)) size02.sizeEU = "43" size02.isInStock = False size02.price = 54.99 size02.url = "http://oneshop.com/bottles/92743867ACTFGJ-UTU/43.htm" size02.urlAddToCart = "http://oneshop.com/bottles/atc/43-92743867ACTFGJ-UTU.htm" size03 = Size() size03.uid = str( uuid.UUID(hex="e070b0c9-769d-4c13-a208-f7207f0970db", version=4)) size03.sizeEU = "44.5" size03.isInStock = True size03.price = 189.50 size03.url = "https://megashop.com/shoes/44.5-9a734hd78.htm" size03.urlAddToCart = "https://megashop.com/shoes/atc/44.5-9a734hd78#g89.php" product01 = Product() product01.uid = str( uuid.UUID(hex="2857027b-cf25-4639-965e-0e22f9f4c755", version=4)) product01.url = "http://oneshop.com/bottles/92743867ACTFGJ-UTU" product01.name = "Biggest Corona Bottle ever" product01.setReleaseDate(dtt.datetime(2020, 9, 30, 13, 50, 59), timezone="Europe/Berlin") product01.basePrice = 55.49 product01.lastScanStamp = 1588548868.304869 # setLastScanNow() product01.sizes = [size01, size02] product02 = Product() product02.uid = str( uuid.UUID(hex="9cab557a-419a-4883-8287-f09f7244b225", version=4)) product02.url = "http://oneshop.com/bottles/1362836400447GT-UTU" product02.name = "Neck Bottle" product02.setReleaseDate(dtt.datetime(2023, 1, 30, 15, 40, 35), timezone="Europe/Berlin") product02.basePrice = 3.22 product02.lastScanStamp = 1588548911.230381 # setLastScanNow() product02.sizes = [] product03 = Product() product03.uid = str( uuid.UUID(hex="f0700293-693c-48a6-8f01-014e07151d99", version=4)) product03.url = "https://www.megashop.com/shoes/9a734hd78.html" product03.urlThumb = "https://www.megashop.com/shoes/thumb-9a734hd78.html" product03.name = "Hey Bro Male" product03.setReleaseDate(dtt.datetime(2028, 11, 1, 8, 2, 40), timezone="Europe/Berlin") product03.basePrice = 190 product03.lastScanStamp = 1588548274.102859 # setLastScanNow() product03.sizes = [size03] shop01 = Shop() # ID is usually set by the DBMS, we explicitly set it for equality checks, too: shop01.uid = str( uuid.UUID(hex="73f9cac8-ebdc-4d9b-8163-d04d09f06cd9", version=4)) shop01.name = "Bottle shop" shop01.url = "http://oneshop.com/bottles" shop01.products = [product01, product02] shop02 = Shop() # ID is usually set by the DBMS, we explicitly set it for equality checks, too: shop02.uid = str( uuid.UUID(hex="69ec8e1b-8812-4413-ad72-b74364e2fa7a", version=4)) shop02.name = "Megashop" shop02.url = "https://www.megashop.com/shoes" shop02.products = [product03] self.shops = [shop01, shop02]