package main import ( "strings" "github.com/gocolly/colly" ) func (app *App) ScrapeWhiskyworld(shop Shop) []Angebot { Shop_urls := []string{"https://www.whiskyworld.de/themen/sonderangebote?ft=%257B%2522produkt_kategorie%2522:%2522Blended%2BMalt%2522%257D", "https://www.whiskyworld.de/themen/sonderangebote?ft=%257B%2522produkt_kategorie%2522:%2522Blended%2BWhiskies%2522%257D", "https://www.whiskyworld.de/themen/sonderangebote?ft=%257B%2522produkt_kategorie%2522:%2522Single%2BMalt%2522%257D", } Whiskys := []Angebot{} c := app.customCollector([]string{"whiskyworld.de", "www.whiskyworld.de"}) c.OnHTML(".product-item", func(e *colly.HTMLElement) { if !stringInSlice(e.Request.URL.String(), Shop_urls) { return } W := Angebot{} W.Shop = shop.Id W.Spirit_type = "Whisky" whisky_name_part1 := e.ChildText(".item-brand") whisky_name_part2 := e.ChildText(".item-description") W.Name = whisky_name_part1 + " " + whisky_name_part2 W.Url = "https://www.whiskyworld.de/" + e.ChildAttr("a", "href") regular_price_noisy := e.ChildText(".offer") regular_price := strings.TrimSuffix(strings.TrimPrefix(regular_price_noisy, "statt "), " €*") var err error W.Original_price, err = convert_price(regular_price) if err != nil { W.error_msg = err.Error() W.error_ctx = regular_price PrintlnOffer(W, "Whiskyworld: Converting original price failed") return } W.Discounted_price, err = convert_price(e.ChildText(".uvp")) if err != nil { W.error_msg = err.Error() W.error_ctx = e.ChildText(".uvp") PrintlnOffer(W, "Whiskyworld: Converting discounted price failed") return } e.ForEach(".product-infobox", func(i int, e *colly.HTMLElement) { text_noisy := e.ChildText(".item-inh") W.Volume, err = extract_volume(text_noisy) if err != nil { W.error_msg = err.Error() W.error_ctx = text_noisy PrintlnOffer(W, "Whiskyworld: Extracting volume failed") return } if W.Volume == 0 { W.error_msg = "Whiskyworld: Volume is zero" W.error_ctx = text_noisy PrintlnOffer(W, "Whiskyworld: Volume is zero") return } abv_noisy := strings.TrimSpace(strings.SplitAfter(text_noisy, "Liter")[1]) abv_noisy = strings.TrimPrefix(abv_noisy, "/") W.Abv, err = extract_abv(abv_noisy) if err != nil { W.error_msg = err.Error() W.error_ctx = abv_noisy PrintlnOffer(W, "Whiskyworld: Extracting abv failed") return } if W.Abv == 0 { W.error_msg = "Whiskyworld: Abv is zero" W.error_ctx = abv_noisy PrintlnOffer(W, "Whiskyworld: Abv is zero") return } }) e.ForEach(".price", func(i int, e *colly.HTMLElement) { base_price_noisy := e.ChildText(".unit") if strings.Contains(base_price_noisy, "Liter") { base_price_noisy = strings.TrimSpace(strings.SplitAfter(base_price_noisy, "Liter")[0]) W.Base_price, err = sanitize_base_price(base_price_noisy) if err != nil { W.error_msg = err.Error() W.error_ctx = base_price_noisy PrintlnOffer(W, "Whiskyworld: Sanitizing base price failed") return } } }) W.Image_url = e.ChildAttr("img", "data-src") if !strings.HasPrefix(W.Image_url, "https:") { W.Image_url = "https:" + W.Image_url } e.Request.Visit(W.Url) W.Website = e.Request.Ctx.Get("website") //Debug(nil, W.Website) Whiskys = append(Whiskys, W) }) c.OnHTML("body", func(e *colly.HTMLElement) { if stringInSlice(e.Request.URL.String(), Shop_urls) { return } e.Request.Ctx.Put("website", string(e.Response.Body)) }) for _, url := range Shop_urls { err := c.Visit(url) if err != nil { Warn(nil, shop.Name+": "+err.Error()) } } return Whiskys }