diff options
| author | Max | 2018-02-06 00:35:39 +0100 |
|---|---|---|
| committer | Max | 2018-02-06 00:35:39 +0100 |
| commit | 71950479fbd6088f249e5fda3b180f294d1d745d (patch) | |
| tree | 06f360a7e02b7e0011bda815fa102ec54ae8d0ec /whiskysitenl.go | |
| parent | 13a807854bf4d0258723ec3152b217ed4cf8e051 (diff) | |
| download | alkobote-71950479fbd6088f249e5fda3b180f294d1d745d.tar.gz | |
Moves crawler to designated directory.
Diffstat (limited to 'whiskysitenl.go')
| -rw-r--r-- | whiskysitenl.go | 59 |
1 files changed, 0 insertions, 59 deletions
diff --git a/whiskysitenl.go b/whiskysitenl.go deleted file mode 100644 index c8b35a2..0000000 --- a/whiskysitenl.go +++ /dev/null @@ -1,59 +0,0 @@ -package main - -import ( - "log" - "regexp" - "strings" - - "github.com/gocolly/colly" -) - -func ScrapeWhiskysitenl(shop Shop) []Angebot { - Whiskys := []Angebot{} - - c := colly.NewCollector( - colly.AllowedDomains("whiskysite.nl"), - colly.AllowedDomains("www.whiskysite.nl"), - ) - - c.OnHTML(".product-block", func(e *colly.HTMLElement) { - - W := Angebot{} - - whisky_name := e.ChildAttr("img", "alt") - whisky_url := e.ChildAttr("a", "href") - - W.Name = whisky_name - W.Url = whisky_url - - regular_price := e.ChildText(".price-old") - price_discount_noisy := e.ChildText(".product-block-price") - r, err := regexp.Compile("[0-9]+(,[0-9]{1,2})") - if err != nil { - log.Fatal(err) - } - discounted_price := r.FindString(strings.Trim(strings.TrimPrefix(price_discount_noisy, regular_price), "")) - - W.Original_price, err = sanitize_price(regular_price) - if err != nil { - //log.Println(W.Name, err) - return - } - W.Discounted_price, err = sanitize_price(discounted_price) - if err != nil { - //log.Println(W.Name, err) - return - } - - W.Image_url = e.ChildAttr("img", "src") - - W.Shop = shop.Id - W.Spirit_type = "Whisky" - - Whiskys = append(Whiskys, W) - }) - - c.Visit("https://www.whiskysite.nl/en/specials/?limit=100") - - return Whiskys -} |
